Korea Invest Insights

Korea Foreign-Investor Playbook: KOSPI 168, KOSDAQ 355 and the MSCI DM Optionality

Sun, 31 May 2026 23:45:00 +0900

This is a follow-up to Korea Foreign Investor Flows, KOSPI Foreign Ownership vs Samsung Electronics and SK hynix, and Korea ADR 67: Narrow Leadership Across KOSPI and KOSDAQ. Those posts asked how much foreigners sold and why the average Korean stock felt weak. This one asks a more basic portfolio question: which Korean stocks are actually in the foreign-investor playbook?

TL;DR

KOSPI foreign activity is extremely compressed. Based on Research OS local DB, from Jan. 2 to May 29, 2026, 42 A Core KOSPI names captured 73.9% of the foreign-activity proxy. A+B names, only 168 stocks, captured 94.4%.
KOSDAQ is wider but lower-quality. 88 A Core KOSDAQ names captured 55.6% of the proxy, while A+B names, 355 stocks, captured 84.5%. The mix is more theme-rotational: biotech, semiconductor equipment, robotics, batteries, and optical communications.
Korea is still an MSCI Emerging Market as of May 31, 2026. MSCI did not upgrade Korea in the 2025 Market Classification Review and said it would continue monitoring whether Korea’s access reforms replicate the outcomes of fully operational Developed Market access. (MSCI)
Even if Korea’s MSCI Developed Market upgrade eventually happens, the benefit is likely to be selective. It should accrue first to stocks already understandable to DM long-only funds: AI memory, shareholder-return financials, shipbuilding, defense, electrical equipment, and global industrial/NAV names.
The practical rule: before asking whether foreigners are buying or selling a stock, first ask whether the stock is in the foreign-investor playbook at all.

Data basis: Research OS local DB, period 2026-01-02 to 2026-05-29, using investor_flow_raw_daily, prices_daily, and kr_fundamentals_daily. Important caveat: the local DB is based on foreign net buying, not gross foreign buy plus gross foreign sell. Therefore this note uses sum(abs(foreign net buying)) as a conservative proxy for foreign trading interest.

1. MSCI DM Upgrade Is an Option, Not a Done Deal

Korea’s MSCI Developed Market upgrade has been debated for years. But as of May 31, 2026, Korea remains classified as Emerging Market. In the 2025 review, MSCI highlighted ongoing monitoring of Korea’s FX-market reforms, omnibus-account and LEI processes, OTC transaction reporting, investment-instrument access, short-selling rules, stock lending, and settlement stability. (MSCI Market Classification)

That means “MSCI DM upgrade = immediate broad Korea inflow” is too simple. The better framing is:

MSCI DM inclusion is a long-duration discount-rate option for Korea. But the first beneficiaries are likely to be the Korean stocks global long-only managers can already understand, trade, and explain.

The reason is that Korea is already deeply embedded in global AI portfolios through memory. As of April 30, 2026, Samsung Electronics was 6.03% and SK hynix 4.05% of the MSCI EM Index. (MSCI EM Index) In the MSCI Korea Index, Samsung Electronics was 32.24% and SK hynix 21.68%. (MSCI Korea Index)

In other words, the AI-memory rally already behaved like a narrow preview of the DM-upgrade effect. The problem is that the effect arrived first, and intensely, in a very small number of names.

2. Market-Level Picture

Market	Names	Foreign abs net-buy proxy	Foreign net buying	Total trading value	Foreign activity / turnover
KOSPI	835	KRW 38.2T	KRW -9.95T	KRW 319.1T	12.0%
KOSDAQ	1,776	KRW 10.9T	KRW +0.44T	KRW 137.9T	7.9%

KOSPI saw heavy foreign activity but net selling. That selling was concentrated in large winners such as Samsung Electronics, SK hynix, and Hyundai Motor. KOSDAQ saw a smaller foreign-activity base, but positive net buying. So the better interpretation is not “foreigners sold Korea.” It is: foreigners reduced KOSPI mega-cap winners while buying selected KOSDAQ growth and theme names.

3. Foreign Target Classification

Bucket	Definition
A Core	Top 5% by market-level foreign abs net-buy proxy and foreign activity on at least 80% of trading days
B Active	Top 20% and activity on at least 70% of trading days
C Peripheral	Mid-tier, event-driven activity possible
D Non-target	Not a meaningful foreign trading target

Market	Bucket	Names	Foreign activity	Share
KOSPI	A Core	42	KRW 28.2T	73.9%
KOSPI	B Active	126	KRW 7.8T	20.5%
KOSPI	C Peripheral	250	KRW 1.8T	4.7%
KOSPI	D Non-target	417	KRW 0.3T	0.8%
KOSDAQ	A Core	88	KRW 6.1T	55.6%
KOSDAQ	B Active	267	KRW 3.1T	28.9%
KOSDAQ	C Peripheral	530	KRW 1.4T	12.4%
KOSDAQ	D Non-target	891	KRW 0.3T	3.0%

The conclusion is blunt. On KOSPI, only 168 names captured 94.4% of the foreign-activity proxy. On KOSDAQ, 355 names captured 84.5%. Korea’s 2026 market is not a market where foreign money disappeared. It is a market where the gap widened between stocks foreigners can trade and stocks they effectively ignore.

4. KOSPI: The Core Foreign Trading Names

Rank	Name	Foreign activity	Foreign net buying	YTD	Interpretation
1	Samsung Electronics	KRW 9.26T	KRW -5.18T	+146.7%	Largest foreign target, heavy distribution
2	SK hynix	KRW 6.42T	KRW -3.74T	+244.6%	Core AI-memory name, profit-taking dominant
3	Hyundai Motor	KRW 1.52T	KRW -0.96T	+142.2%	Global auto long-only name, but foreign selling
4	Doosan Enerbility	KRW 0.84T	KRW +0.21T	+40.4%	Nuclear, power, turbine theme
5	Samsung SDI	KRW 0.47T	KRW +0.14T	+162.1%	Battery rebound beta
6	Hyundai Mobis	KRW 0.45T	KRW -0.33T	+108.1%	Liquid, but under foreign selling
7	SK Square	KRW 0.44T	KRW -0.24T	+214.5%	SK hynix proxy distribution
8	Hanwha Aerospace	KRW 0.44T	KRW +0.05T	+24.0%	Defense core
9	Samsung Electro-Mechanics	KRW 0.43T	KRW -0.01T	+687.8%	AI components, crowded
10	Hanmi Semiconductor	KRW 0.39T	KRW -0.01T	+95.2%	HBM equipment core

The important point is that foreign net selling is not automatically bearish. Samsung Electronics, SK hynix, and Hyundai Motor were heavily sold, but they are still core foreign book names. A stock that foreigners do not trade at all is in a different category.

KOSPI foreign playbook sectors are:

Theme	Examples
AI, semiconductors, substrates	Samsung Electronics, SK hynix, Samsung Electro-Mechanics, Hanmi Semiconductor, ISU Petasys
Autos	Hyundai Motor, Kia, Hyundai Mobis
Financials, holdings, shareholder returns	KB Financial, Shinhan Financial, Samsung Fire, Samsung C&T, HD Hyundai
Electrical equipment and cables	HD Hyundai Electric, LS ELECTRIC, Hyosung Heavy, Taihan Cable, Sanil Electric
Defense and shipbuilding	Hanwha Aerospace, Hanwha Ocean, HD Hyundai Heavy Industries, Hyundai Rotem, LIG Nex1
Batteries	Samsung SDI, LG Energy Solution, POSCO Holdings, POSCO Future M

5. KOSDAQ: Broader, More Thematic

Rank	Name	Foreign activity	Foreign net buying	YTD	Interpretation
1	EcoPro	KRW 272.5B	KRW +28.4B	+57.4%	Battery beta
2	Woori Technology	KRW 262.0B	KRW -14.9B	+307.2%	Nuclear/power theme, high activity but net selling
3	ABL Bio	KRW 216.3B	KRW -1.4B	-42.8%	Biotech core trading name
4	Alteogen	KRW 204.6B	KRW +14.9B	-19.3%	Biotech leader, consistently traded
5	Rainbow Robotics	KRW 191.0B	KRW +8.8B	+42.2%	Robotics leader
6	Samchundang Pharm	KRW 169.1B	KRW +46.2B	+38.7%	Strong biotech accumulation
7	EcoPro BM	KRW 163.7B	KRW +50.3B	+53.1%	Battery buying
8	Leeno Industrial	KRW 158.0B	KRW -80.2B	+50.1%	Quality semiconductor name under selling
9	Taihan Fiberoptics	KRW 155.4B	KRW -19.1B	+746.1%	Optical-communications theme
10	HPSP	KRW 147.1B	KRW +40.1B	+41.2%	Semiconductor equipment buying

KOSDAQ foreign activity centers on biotech leaders, semiconductor equipment and substrates, robotics, batteries, and optical communications. However, KOSDAQ A/B status is not the same as DM long-only quality status. It often includes event-driven, liquidity-driven, and theme-rotation flows.

6. Why D Group Matters

Market	D names	Share of foreign activity	Median YTD	Median turnover	Median foreign activity
KOSPI	417	0.8%	-6.2%	KRW 56.1B	KRW 6.5B
KOSDAQ	891	3.0%	-8.0%	KRW 35.5B	KRW 3.3B

D group stocks can rise, sometimes sharply. But their rallies are not evidence of foreign long-only rerating. They are more likely to be driven by domestic retail liquidity, local themes, and short-term trading. They should be separated from the MSCI DM-upgrade thesis.

7. The DM Long-Only Basket

If Korea’s access improves further, the first foreign long-only candidates are not obscure theme stocks. They are names a global PM can explain in one sentence.

Category	Core candidates	One-line thesis
Shareholder-return financials	Samsung Fire, KB Financial	Value-Up, dividends, buybacks, ROE visibility
Shipbuilding	HD Hyundai Heavy, Hanwha Ocean, HD Korea Shipbuilding	Global order book, naval shipbuilding, LNG, US shipbuilding cooperation
Defense	LIG Nex1	Missiles, radar, export defense quality
Global industrial / NAV	Samsung C&T, HD Hyundai	Holding-company NAV discount, shareholder returns, shipbuilding and power-equipment exposure

The crowded pullback basket is different.

Category	Pullback candidates	View
Electrical equipment	HD Hyundai Electric, Hyosung Heavy, LS ELECTRIC	Clear AI power-grid thesis, but valuation and profit-taking risk are high
AI components	Samsung Electro-Mechanics, LG Innotek	Strong stories, but after extreme moves the entry efficiency is weaker
Autos	Hyundai Motor, Hyundai Mobis	Global quality names, but foreign distribution remains an overhang

8. Practical Checklist

Step	Question	Portfolio meaning
1	Is it an A/B foreign target?	If not, do not lead with a foreign-rerating thesis.
2	Are foreigners buying or selling?	Buying suggests accumulation; selling may be profit-taking or rebalance.
3	Is foreign activity moving with price?	If yes, it may be leadership. If not, watch hand-off risk.
4	Does the stock hold despite foreign selling?	Domestic institutional or retail absorption may be strong.
5	Is it a D group rally?	Treat it as domestic liquidity/theme trading, not long-only rerating.

This also reframes NAVER. NAVER is in KOSPI A Core, but its YTD return was -5.3% as of May 29. That makes NAVER’s rerating case more interesting: it is not an unknown stock. It is a neglected large platform already inside the foreign playbook.

Final View

MSCI DM inclusion could lower Korea’s market discount over time. But 2026 already shows how selective the effect is likely to be. AI memory ran first, foreigners distributed into strength, and capital rotated into shipbuilding, defense, electrical equipment, shareholder-return financials, and selected KOSDAQ growth themes.

The next stage of Korea rerating is therefore not: “Will foreigners buy Korea?” The better question is:

Which Korean stocks are foreign investors structurally able and willing to buy?

The practical answer is KOSPI 168 and KOSDAQ 355. Stocks outside that universe are not bad by definition. But they should not be framed as foreign long-only rerating candidates until they enter the playbook.

Evidence Ledger

Item	Source	What was verified
Foreign A/B classification	Research OS local DB, `investor_flow_raw_daily`, `prices_daily`, `kr_fundamentals_daily`, accessed 2026-05-31	KOSPI 168 and KOSDAQ 355 A/B foreign-target lists for 2026-01-02 to 2026-05-29
Local output	`/Users/youngseongshin/agents/Stock_Research/outputs/kr_foreign_trading_targets_ytd_20260529.md`	Full markdown output plus KOSPI 168 and KOSDAQ 355 CSV files confirmed
Korea market accessibility	MSCI 2025 Market Classification Review	Korea was not upgraded in 2025; MSCI continues to monitor accessibility reforms
MSCI EM composition	MSCI EM Index	Samsung Electronics 6.03%, SK hynix 4.05% as of April 30, 2026
MSCI Korea composition	MSCI Korea Index	Samsung Electronics 32.24%, SK hynix 21.68% as of April 30, 2026

Fact / Inference / Blocked

[Fact]

42 KOSPI A Core names captured 73.9% of the foreign-activity proxy; KOSPI A+B 168 captured 94.4%.
88 KOSDAQ A Core names captured 55.6%; KOSDAQ A+B 355 captured 84.5%.
MSCI did not reclassify Korea to Developed Market in the 2025 Market Classification Review.

[Inference]

MSCI DM-upgrade upside is likely to concentrate in Korean names already inside the foreign-investor playbook.
KOSPI foreign net selling is better understood as AI-memory mega-cap distribution plus rotation, not a broad Korea exit.
KOSDAQ A/B status confirms foreign attention, but not necessarily long-only quality.

[Blocked]

The local DB does not provide gross foreign buy and sell; this note uses sum(abs(net flow)) as a proxy.
The timing, probability, and exact passive-flow impact of any MSCI DM upgrade remain uncertain.
Actual mandate-level foreign long-only positioning, stock-lending data, short-sale balances, and program-trading details are outside this note.

Disclaimer: This article is for research and information only. It is not investment advice.

Alibaba AI Pay's 300M Transactions: Korea's Real Risk Is Seller OS and Payment Permission, Not a 272T Won E-Commerce Collapse

Sun, 31 May 2026 22:45:00 +0900

Context This is a follow-up to NAVER at KRW 36.7T: Are KRW 8T Dunamu-Combined Cash/BTC and a KRW 3T Mirae Stake Priced In?. That note framed NAVER through SOTP, AI cloud, Npay, Dunamu, and won-stablecoin optionality. This note asks the next question: if commerce becomes agentic, who owns the seller operating system and the payment-permission layer?

TL;DR

The Alibaba/Ant story is not that Chinese platforms will immediately eat Korea’s entire consumer e-commerce market. The more important question is where Korean sellers’ global-export workflows and AI-agent payment permissions will accumulate.

The direct consumer-commerce takeover thesis is overstated. Korea’s 2025 online shopping market was KRW 272.0 trillion, while direct purchases from China were KRW 5.57 trillion, or about 2.05% of the total. But Alipay says AI Pay has processed 300 million transactions, while Alibaba.com has launched Accio Work in Korea for SME seller automation. That combination targets seller OS, cross-border workflow, and payment trust layers before it targets Korea’s whole shopping basket. (BusinessWire, The Pickool)

The investment takeaway:

NAVER must turn SmartStore, Shopping AI, Npay, search, and ads into a Korean seller-agent OS.
NHN KCP, KakaoPay, and Naver Pay should be evaluated less as payment buttons and more as agent authorization, spending-limit, fraud, refund, dispute, and routing infrastructure.
Cafe24, K-beauty brands, and ODMs face both threat and opportunity. Accio Work validates the demand for seller automation, but domestic platforms must capture that workflow before Alibaba does.

My view: this is not a standalone “buy BABA” call. In Korea, the better watchlist is NAVER, NHN KCP, Cafe24, KakaoPay, and the K-beauty export stack. But this is not yet a buy-now setup. We need evidence on paid adoption, GMV uplift, actual order conversion, and AP2/x402-like payment protocol commercialization.

1. What Changed

Two things changed.

First, Alipay announced AI Wallet, Token Pay, and next-generation AI payment infrastructure in May 2026. The company said AI Pay had processed 300 million transactions. AI Wallet is positioned as a consumer control layer for agent-driven payments before, during, and after checkout. Token Pay targets AI model subscriptions, token top-ups, and B2B AI billing. (BusinessWire)

Second, Alibaba.com launched Accio Work in Korea. Accio Work is an agentic AI business team for SMEs and solo entrepreneurs, designed to automate market research, product planning, sourcing, price negotiation, product registration, global marketing, buyer inquiries, and store operations. (The Pickool)

Together, the structure is:

Korean seller / K-brand / SME
→ Accio Work automates research, listing, inquiry, negotiation
→ Alibaba.com connects global buyer demand
→ AI Wallet / AI Pay / Token Pay add payment, authorization, settlement, trust
→ Seller workflow data and cross-border operating data accumulate

This is not merely about consumers buying cheap goods on AliExpress. The real question is which operating desk Korean sellers open every morning when they start selling overseas.

2. Fact Check

Claim	Verdict	Comment
Alipay AI Pay reached 300M transactions	[Fact, company statement]	Confirmed in the BusinessWire release. It is not an independently audited figure. (BusinessWire)
AI Pay exceeded 120M transactions in one week in February 2026	[Fact, company statement]	But the February weekly figure and May cumulative figure should not be annualized mechanically. Sustained run-rate is [Blocked]. (BusinessWire)
Accio Work launched in Korea	[Fact]	Alibaba.com launched it as an agentic AI platform for Korean SMEs. (The Pickool)
The Shinsegae-Alibaba JV can connect Gmarket sellers to Alibaba’s ecosystem	[Fact/Inferred]	The JV was conditionally approved, but Korean customer-data sharing is restricted. (Reuters)
Korea’s whole e-commerce market will be quickly eaten by Alibaba	[Overstated]	Direct purchases from China were only about 2.05% of 2025 Korean online shopping. (Yonhap)
Seller OS and agentic payment layers are real battlegrounds	[Inference]	This follows from the feature set of Accio Work and AI Wallet.

[Blocked] Alipay AI Pay’s actual TPV, average ticket size, take-rate, fraud loss, and dispute cost are not public. Accio Work’s Korean paid conversion, retention, ARPU, and seller-level GMV uplift are also not yet disclosed.

3. Korea: The Problem Is Workflow, Not Just GMV

Korea’s 2025 online shopping transaction value was KRW 272.0398 trillion. Direct purchases from China were KRW 5.5742 trillion. (Yonhap)

China direct purchase / Korea online shopping
= KRW 5.5742T / KRW 272.0398T
= 2.05%

So yes, AliExpress, Temu, and Alibaba-linked platforms are growing. But they are not yet large enough to structurally dominate the whole Korean consumer market. Coupang’s Rocket Delivery, NAVER’s search-review-ad-Npay loop, local easy-payment habits, returns, and customer service remain strong local barriers.

For sellers, the problem is different. Overseas selling requires translation, listings, local pricing, MOQ, buyer inquiries, negotiation emails, compliance, logistics, settlement, and refunds. If AI agents automate those tasks, platform power moves from traffic to workflow.

Old commerce power:
traffic → search/recommendation → payment → logistics

Agentic commerce power:
structured product data → seller workflow automation → AI exposure optimization
→ delegated payment and limits → refund and dispute liability

If Korean platforms miss this, Korean sellers’ overseas operating data will accumulate inside Alibaba.com’s seller OS, not inside Korean platforms.

4. NAVER: The Next Rerating Question Is Seller AI OS

NAVER has both the strongest defense and a meaningful risk.

The defense is its combination of search, Naver Shopping, SmartStore, reviews, ads, Npay, and membership inside one account system. NAVER launched a beta conversational Shopping AI Agent in Naver Plus Store in February 2026, expanding product summaries, comparison, review analysis, and guided discovery. (NAVER)

The risk is overseas seller workflow. A SmartStore merchant may still sell domestically through NAVER, but if cross-border listing, inquiries, negotiation, and buyer conversion happen through Accio Work, NAVER becomes the domestic traffic layer while Alibaba captures global product and buyer-intent data.

NAVER needs more than consumer-facing shopping AI. It needs:

SmartStore Center
+ Shopping AI Agent
+ Npay / stablecoin payment option
+ ad automation
+ global listings, translation, settlement, logistics
+ buyer inquiry and customer service automation
= Korean seller-agent OS

In the previous NAVER rerating note, the focus was Dunamu, Mirae Asset, AI cloud, and stablecoins. Here the question is whether NAVER can own seller workflow as well.

5. NHN KCP and KakaoPay: The Layer Is Permission, Not a Button

Traditional payment asks: which payment method did the user choose?

Agentic payment asks:

How much authority does an AI agent have,
under what conditions,
and with which merchants?

If a user asks an agent to reorder detergent and bottled water under KRW 50,000 with delivery by tomorrow, the core infrastructure is not just card authorization. It includes agent identity, structured purchase conditions, spending limits, merchant trust, user confirmation, audit trail, refunds, disputes, and fraud detection.

Google’s Agent Payments Protocol(AP2) is an attempt to standardize mandates, authorization, and payment interoperability for agent-to-merchant transactions. Google’s partner list includes NHN KCP among global payment players. (Google Cloud)

Mastercard and Visa are moving in the same direction. Mastercard’s Agent Pay emphasizes tokenization, agent registration, user control, and fraud/dispute support. (Mastercard)

The biggest risk for Korean payment companies is not immediate Alipay displacement. It is becoming a simple funding source instead of an agent wallet / payment trust layer.

The opportunity is also clear. PG companies can move above payment authorization into:

Agent identity verification
Mandate management
Spending limit API
Merchant verification
Risk scoring
Refund/dispute API
Cross-border settlement
Stablecoin / card / bank transfer orchestration
Transaction audit trail

For NHN KCP, KakaoPay, Naver Pay, and Toss Payments, TPV growth is no longer enough. The key metrics are agent authorization APIs and dispute/fraud economics.

6. Cafe24 and K-Beauty: Threat and Opportunity

Cafe24 has two-sided exposure.

The threat is that Accio Work becomes the default global operating tool for sellers. Basic store building, templates, and listing support can be commoditized.

The opportunity is that Cafe24 is closer to D2C infrastructure than to a pure marketplace. If Alibaba, Coupang, and NAVER are marketplace-centric, Cafe24 can become an independent brand’s multi-channel global seller OS.

The required functions are:

automated product data generation
multilingual product pages
multi-marketplace listing
TikTok Shop / Amazon / Shopee / Alibaba / Naver / Coupang integration
cross-border payment and settlement
customs, certification, return automation
AI customer service and review analytics

K-beauty has more upside. Korea’s Q1 2026 online direct overseas sales were KRW 1.0599 trillion, up 24.4% year over year. Cosmetics were KRW 633.6 billion, or 59.8% of the total. (Statistics Korea)

Cosmetics share of overseas direct online sales
= KRW 633.6B / KRW 1.0599T
= 59.8%

If Korean sellers’ overseas operating costs fall, K-beauty brands and ODMs benefit. ODMs such as Kolmar Korea and Cosmax can be second-order beneficiaries if more indie brands and SKUs emerge.

But this is not universally bullish. Agentic commerce lowers export barriers, but it also strengthens price comparison and automated negotiation. Undifferentiated sellers may face margin pressure.

7. Threat vs Opportunity Matrix

Segment	Threat	Opportunity	View
Overall Korean consumer e-commerce	Low-medium	Medium	Immediate full-market disruption is overstated.
Low-price cross-border long-tail SKUs	High	Low	Chinese platform advantage is already strong.
Korean marketplace sellers	High	High	They may move to Alibaba, or they may expand globally.
K-beauty, fashion, consumer brands	Medium	High	Global buyer access is the opportunity.
NAVER	Medium	High	Seller AI OS can defend and extend the platform.
Coupang	Low-medium	Medium	Logistics moat is defensive, long-tail SKUs are exposed.
Cafe24	Medium	High	Rerating requires global multi-channel AI OS execution.
Easy-payment wallets	Medium	Medium	Agent wallet status matters more than payment buttons.
PG companies	Low	High	Agentic payment middleware is the upside.

8. Investment Checklist

Checkpoint	Why It Matters
Accio Work paid conversion in Korea	Separates free AI tool from seller OS
Korean seller SKU registration on Alibaba.com	Shows where seller operating data is moving
Buyer inquiry growth and order conversion	Inquiries are vanity; conversion matters
Gmarket seller global integration	Tests whether the JV becomes an export gateway
NAVER SmartStore seller AI functions	Core defense for NAVER
Npay TPV and merchant growth	Base layer for payment optionality
NHN KCP AP2 commercialization	Shows whether protocol exposure becomes revenue
KakaoPay / Naver Pay agent wallet functions	Next layer beyond easy payment
Cafe24 global multi-channel automation	Determines whether Cafe24 can rerate
K-beauty overseas direct sales and ODM customer count	Measures second-order export stack impact

9. Practical View

Buy now: none. The issue is structurally important, but paid conversion, TPV, take-rate, and GMV uplift remain unproven.

Watchlist 1: NAVER NAVER has SmartStore, Shopping AI, Npay, search, ads, and cloud. As argued in the prior NAVER note, it is now a conditional Buy candidate. For this specific theme, the key added metric is seller AI OS execution.

Watchlist 2: NHN KCP AP2 exposure, overseas PG growth, and stablecoin PoC intersect. The question is whether it can become agentic payment middleware rather than a low-margin PG.

Watchlist 3: Cafe24 High-risk, high-beta. If Cafe24 uses the seller OS demand validated by Accio Work to build a domestic D2C and creator-commerce automation layer, rerating is possible.

Watchlist: KakaoPay The AI-wallet narrative is relevant, but valuation and monetization need proof.

K-beauty export stack Brands, ODMs, and cross-border fulfillment are second-order beneficiaries. But margin pressure from automated price comparison is the key risk.

Final View

Alibaba’s threat is not mainly that Korean consumers will buy more on AliExpress. The more important question is whether Korean sellers, when they start selling overseas, open Accio Work and Alipay/Ant infrastructure before they open NAVER, Cafe24, or a domestic PG stack.

Immediate full-market Korean e-commerce disruption is overstated. But if seller operating data and payment-permission layers accumulate overseas, domestic platforms’ ad, commission, and workflow economics can be pressured.

The investment conclusion is simple:

Rather than buying BABA on this headline alone, Korea investors should track how quickly NAVER, NHN KCP, Cafe24, KakaoPay, and the K-beauty export stack internalize the agentic commerce stack.

Evidence Ledger

[Fact]

Alipay announced AI Wallet, Token Pay, and AI payment infrastructure, and cited 300 million cumulative AI Pay transactions. (BusinessWire)
Alibaba.com launched Accio Work in Korea. (The Pickool)
The Shinsegae-Alibaba JV was conditionally approved with customer-data sharing restrictions. (Reuters)
Korea’s 2025 online shopping market was KRW 272.0398 trillion, and direct purchases from China were KRW 5.5742 trillion. (Yonhap)
Google AP2’s partner ecosystem mentions NHN KCP. (Google Cloud)
Korea’s Q1 2026 online direct overseas sales were KRW 1.0599 trillion, with cosmetics at KRW 633.6 billion. (Statistics Korea)

[Inference]

Alibaba’s more important Korea angle is seller OS and global buyer workflow, not immediate full consumer-commerce displacement.
Korean payment companies risk becoming commodity rails if they miss the agent payment protocol layer.
NAVER is the strongest domestic platform candidate to defend the seller OS layer.

[Speculation]

Accio Work could materially raise Korean sellers’ GMV and export conversion.
NHN KCP could monetize AP2-style agentic payment middleware.
NAVER could turn SmartStore Center into a Korean seller-agent OS.

[Blocked]

Alipay AI Pay TPV, average ticket size, take-rate, and fraud loss.
Accio Work Korea paid conversion, retention, ARPU, and seller-level GMV uplift.
Gmarket seller migration rate into Alibaba.com’s buyer network.
Actual revenue contribution by agentic payment products for Korean PG companies.

NAVER at KRW 36.7T: Are KRW 8T Dunamu-Combined Cash/BTC and a KRW 3T Mirae Stake Priced In?

Sun, 31 May 2026 17:40:00 +0900

Context This is a follow-up to Can China and Hong Kong’s AI overheat spill into Korea? and US non-semiconductor rerating and its Korea translation. Those pieces treated NAVER as a sovereign AI and AI software/data-operations candidate. This note focuses on NAVER itself: Dunamu, its Mirae Asset Securities stake, NAVER Cloud, won stablecoin optionality, and global C2C turnaround. The next follow-up is Alibaba AI Pay and Korea seller OS, which extends the NAVER thesis into agentic commerce and seller workflow. Related hubs: Korean AI Companies Hub, Korean Financials Hub, and Korean Stocks for Foreign Investors Hub.

TL;DR

The NAVER rerating question is not simply whether the portal business is cheap. It is whether NAVER can be reclassified as a platform + financial assets + digital assets + integrated AI cloud operator.

The rerating gap is real. In the Research OS local database, NAVER ranked 18th out of 20 major Korean sector leaders by YTD performance as of May 29, 2026. NAVER closed at KRW 234,000, was down 5.3% YTD, down 8.1% over three months, but up 15.0% over the most recent two weeks. The comparison basket median was +32.5%, and the median excluding NAVER and Kakao was +39.2%.

Two corrections matter.

First, “Dunamu has KRW 8 trillion of cash-like assets” is imprecise. The KRW 8.1 trillion figure in NAVER’s Dunamu consolidation material is not Dunamu standalone cash. It is the simple-sum post-transaction consolidated net cash/BTC figure: NAVER net cash KRW 4.2 trillion, Dunamu net cash KRW 1.2 trillion, and BTC KRW 2.7 trillion. (NAVER Dunamu presentation)

Second, the May 29 flow was strong, but not yet conclusive. On that day, institutions bought KRW 205.2 billion net while individuals sold KRW 241.0 billion net. That is a high-quality handoff. However, foreign investors were still net sellers by KRW 124.8 billion over the 10 trading days from May 15 to May 29. This is a rerating start signal, not a completed rerating.

My view: NAVER moves from Watchlist to conditional Buy candidate. Do not chase one large candle. Confirmation requires NAVER holding above KRW 234,000, at least three days of foreign net buying, continued pension/investment-trust buying, Q2 margin recovery, progress on the Dunamu transaction, and clearer Korean stablecoin regulation.

1. Data Basis and Fact Check

Market data, flow, and consensus come from the Research OS local database:

/Users/youngseongshin/agents/Stock_Research/agents/KR_Crawler/data/screener_kr.db

The price and flow date is May 29, 2026. The two-week window covers 10 trading days from May 15 to May 29. I used prices_daily, kr_market_surface_daily, investor_flow_raw_daily, and consensus_daily.

[Blocked] The 20-stock sector-leader basket is an internal Research OS comparison universe. This post uses NAVER’s rank and basket medians, but does not publish the full reproducible member list. Short-sale, program-trading, and stock-loan coverage was incomplete for this exact two-week window, so I do not use it in the conclusion.

NAVER’s Current Position

Item	NAVER
Close	KRW 234,000
YTD return	-5.3%
3-month return	-8.1%
1-month return	+6.4%
2-week return	+15.0%
Below 52-week high	-20.7%
Above 52-week low	+26.1%
FY2026E EPS	KRW 12,628
FY2026E P/E at current price	about 18.5x
Consensus target price	KRW 300,450
Implied upside	+28.4%

NAVER’s official Q1 2026 results were revenue of KRW 3.2411 trillion and operating profit of KRW 541.8 billion. Revenue grew 16.3% year over year and operating profit grew 7.2%. NAVER said AI helped advertising growth, while C2C and Global Initiatives drove segment expansion. (NAVER IR)

These are not weak numbers. The problem is relative opportunity cost. Korean investors have aggressively rerated AI infrastructure, shipbuilding, defense, autos, and financial leaders. NAVER maintained growth, but stayed outside the main rerating wave because of AI infrastructure cost concerns and the old platform-growth-stock discount.

The May 29 Handoff

May 29, 2026	Net Buy
Individuals	-KRW 241.0bn
Foreigners	+KRW 27.8bn
Institutions	+KRW 205.2bn
Financial investment	+KRW 73.7bn
Investment trusts	+KRW 44.0bn
Private funds	+KRW 18.7bn
Pension funds etc.	+KRW 59.1bn
Insurance	+KRW 9.5bn

The quality of this flow is good. It was not just short-term financial-investment buying. Pension funds, investment trusts, private funds, and insurers all bought. Individuals sold heavily, and institutions absorbed the supply.

The 10-day picture is less complete:

Period	Individuals	Foreigners	Institutions	Financial Inv.	Insurance	Inv. Trusts	Private Funds	Pension etc.
May 15-21	+KRW 93.8bn	-KRW 111.5bn	+KRW 19.8bn	+KRW 3.0bn	+KRW 0.9bn	-KRW 1.3bn	+KRW 2.5bn	+KRW 14.5bn
May 22-29	-KRW 194.9bn	-KRW 13.3bn	+KRW 198.5bn	+KRW 57.9bn	+KRW 8.8bn	+KRW 36.9bn	+KRW 21.3bn	+KRW 74.0bn
10-day total	-KRW 101.1bn	-KRW 124.8bn	+KRW 218.3bn	+KRW 60.8bn	+KRW 9.7bn	+KRW 35.6bn	+KRW 23.8bn	+KRW 88.5bn

So far, this is an institution-led handoff. Foreign buying must continue for the rerating to become global capital’s rerating.

2. Thesis Check

Claim	Verdict	Investment Meaning
Dunamu has KRW 8T cash-like assets	Needs correction	KRW 8.1T is combined consolidated net cash/BTC, not Dunamu standalone cash.
Mirae Asset Securities stake is worth around KRW 3T	Mostly right	Conservative value: 47.74M shares x KRW 61,400 = KRW 2.93T. Local May 29 KRW 79,000 price implies KRW 3.77T.
NAVER is becoming an integrated AI cloud operator	True, margin still needs proof	NAVER Cloud has GPUaaS, sovereign AI, physical AI, public/financial/security cloud exposure.
Stablecoin is a future business	Optionality	Npay + Dunamu/Upbit + NAVER Cloud could become a won stablecoin payment infrastructure candidate. Regulation is the gating item.
Global business is turning around	Partly true	C2C is strong, Contents is still weak. The right phrase is “early C2C-led global turnaround.”

3. Dunamu: The Financial-Platform Reclassification Option

The Dunamu transaction does not inject cash into NAVER Corp. directly. It is a share-swap structure in which NAVER Financial secures 100% of Dunamu, while NAVER seeks to retain consolidation through voting-right delegation. NAVER’s material presents Dunamu equity value at KRW 15.1 trillion, NAVER Financial at KRW 4.9 trillion, and NAVER’s post-transaction stake in NAVER Financial at 17.0%, with delegated voting rights supporting consolidation. (NAVER Dunamu presentation)

The key figures:

Item	NAVER Material	Interpretation
Post-deal consolidated net cash/BTC	KRW 8.1T	NAVER KRW 4.2T + Dunamu net cash KRW 1.2T + BTC KRW 2.7T
NAVER standalone FCF	KRW 1.3T	NAVER standalone cash generation
Post-deal consolidated FCF	KRW 2.3T	Simple-sum basis
Post-deal EBIT	KRW 3.5T	Simple-sum basis
Post-deal EBIT margin	25.8%	Reflects Dunamu’s high-margin profile

The investment point is the potential reclassification from advertising/commerce platform to digital-asset financial infrastructure platform. If Npay, Upbit liquidity, and NAVER Cloud security/data infrastructure can be connected, the optionality moves into won stablecoins, tokenization, digital-asset payments, and financial product distribution.

But accounting consolidation is not the same as value attributable to NAVER common shareholders. NAVER’s economic ownership in NAVER Financial falls to 17% after the transaction. The consolidated numbers may improve, but they do not accrue 100% to NAVER common shareholders. That distinction is essential.

4. Mirae Asset Securities Stake: SOTP Downside Support

NAVER owns roughly 47.74 million common shares of Mirae Asset Securities. Public ownership data providers show small differences, generally around 47,740,283 to 47,740,312 shares. (Wisereport)

Conservative value:

Mirae Asset Securities stake
= 47,740,283 shares x KRW 61,400
= about KRW 2.93T

However, the Research OS local Kiwoom surface showed Mirae Asset Securities at KRW 79,000 on May 29, 2026. On that price:

47,740,283 shares x KRW 79,000
= about KRW 3.77T

I treat it as a KRW 3 trillion-class SOTP asset. This does not mean NAVER holds KRW 3 trillion of instantly usable cash. It is a listed strategic financial asset whose value should be discounted for market risk, tax, liquidity, and strategic relationship constraints. Still, it makes pure P/E analysis incomplete.

5. NAVER Cloud: Not Just Search, but Sovereign AI Infrastructure

NAVER Cloud is more than a generic cloud business. NAVER is moving toward a full-stack AI infrastructure role: data centers, cloud, GPUaaS, HyperCLOVA X, public/financial/security cloud, digital twins, and sovereign AI.

In Q1 2026, NAVER’s Enterprise revenue grew 18.8% year over year. NAVER attributed growth to AI and digital twin initiatives and LINE WORKS expansion. (NAVER IR)

NVIDIA’s Korea AI infrastructure release strengthens this reading. NVIDIA said NAVER Cloud is expanding NVIDIA AI infrastructure with over 60,000 GPUs for enterprise and physical AI workloads, and preparing sovereign and physical AI infrastructure including Blackwell-class GPUs. (NVIDIA)

The right peer frame is not OpenAI or Anthropic. NAVER’s more realistic opportunity is sovereign AI operator for Korea and selected overseas markets.

The risk is margin. GPUaaS and AI cloud require heavy upfront infrastructure spending. NAVER’s Q1 operating margin of 16.7% already reflected AI infrastructure investment. NAVER Cloud only becomes a rerating factor if Enterprise/GPUaaS revenue can outrun depreciation, power, and networking costs.

6. Stablecoin: Npay + Upbit as Payment Infrastructure Optionality

NAVER’s stablecoin thesis is not simply that NAVER becomes the sole issuer. The larger question is whether NAVER can become the infrastructure layer that combines payment distribution and digital-asset liquidity.

Asset	Stablecoin Relevance
Npay	Payment volume and real merchant acceptance
Smart Store / commerce	Seller and consumer endpoints
NAVER login / membership	Identity and account layer
Dunamu / Upbit	Digital-asset liquidity and users
NAVER Cloud	Security, data, and compliance infrastructure
Mirae Asset stake	Securities, tokenization, investment-product optionality

NAVER’s Dunamu material explicitly presents payment ecosystem expansion, tokenization, and global expansion as part of the strategic rationale. (NAVER Dunamu presentation)

This remains a regulatory option. If Korea’s stablecoin framework is delayed, or if issuance economics are kept inside bank-led structures, NAVER’s role could be limited to distribution, technology, and payments. Exchange ownership regulation is another core risk.

The conservative sentence is:

NAVER can become a key distribution infrastructure candidate for won stablecoins, but issuance economics and regulatory structure are not yet confirmed.

7. Global Turnaround: C2C Is Strong, Contents Is Still Weak

NAVER’s Global Initiatives revenue was KRW 941.6 billion in Q1 2026, up 18.4% year over year and 5.1% quarter over quarter.

Segment	Q1 2026 Revenue	YoY
C2C	KRW 351.1bn	+57.7%
Contents	KRW 440.1bn	-1.4%
Enterprise	KRW 150.5bn	+18.8%

C2C improved through Wallapop consolidation, Poshmark restructuring, and KREAM/SODA growth. But Contents was still down year over year. So the right framing is early C2C-led global turnaround, not full global turnaround.

This matters because NAVER’s global discount partly came from Poshmark skepticism. If C2C can deliver growth and profit improvement, Global Initiatives can move from value-destructive narrative to optionality.

8. Valuation Bridge: What Would Justify KRW 300,000?

At KRW 234,000 per share and a KRW 36.7 trillion market cap, a KRW 300,000 target implies about KRW 47.1 trillion of market cap.

Target market cap
= KRW 36.7T x 300,000 / 234,000
= about KRW 47.1T

That is roughly KRW 10.4 trillion of required upside. The bridge must come from more than narrative.

Rerating Bridge	Needed Confirmation
Core profit growth	NAVER Platform growth and OPM recovery
Dunamu consolidation	Approval, consolidation, non-controlling interest impact
Mirae Asset stake	SOTP recognition and smaller holding-company discount
Stablecoin option	Legal framework and platform/exchange participation
AI cloud growth	Enterprise/GPUaaS revenue and margin
Global discount reduction	C2C ex-Wallapop growth and Contents recovery

Consensus has not yet followed the narrative. In Research OS local DB, NAVER’s consensus target price fell from KRW 312,450 on May 11 to KRW 300,450 on May 29. FY2026E EPS fell from KRW 12,879 to KRW 12,628. So the May 29 move was not yet analyst estimate revision. It was narrative-driven flow rotation.

9. Trading Strategy

NAVER moves to conditional Buy candidate.

The May 29 move was triggered by expectations around NVIDIA CEO Jensen Huang’s Korea visit and a possible meeting with NAVER founder Lee Hae-jin, tied to physical AI and AI cloud cooperation. (Korea JoongAng Daily)

Entry

Approach	Condition
Aggressive	Small position after KRW 225,000-234,000 support holds
Conservative	Break above KRW 240,000 with simultaneous foreign and institutional buying
Pullback	Below KRW 220,000 if thesis is intact
Earnings-confirmed	Q2 margin recovery and Financial Platform TPV growth above 20%

Add Conditions

Add Trigger	Meaning
Dunamu share-swap approval	Financial-platform reclassification becomes more concrete
Consolidation maintained	Consolidated FCF/EBIT bridge remains intact
Stablecoin law allows platform/exchange participation	Npay+Upbit option expands
Global Opportunities profit improves	Global discount narrows
Enterprise/GPUaaS revenue reaccelerates QoQ	AI cloud thesis becomes measurable

Invalidation

Reduce or discard the rerating thesis if two or more of the following occur:

NAVER Platform growth drops below 10%.
Operating margin stays at or below 16%.
Financial Platform TPV growth slows below 15%.
Dunamu approval is delayed or consolidation is impaired by ownership regulation.
Stablecoin regulation limits platform/exchange participation.
Dunamu profit deteriorates further as trading volume weakens.
Enterprise/GPUaaS costs grow without revenue acceleration.
C2C growth fades after the Wallapop consolidation effect.
Google or Coupang structurally erodes search, shopping, or payment intent.

10. Final View

NAVER has finally entered the rerating candidate zone.

The reasons are straightforward. It was one of the most neglected large Korean platform leaders YTD. May 29 showed a high-quality institutional handoff. Dunamu creates digital-asset financial-platform optionality. The Mirae Asset Securities stake is a KRW 3 trillion-class SOTP asset. NAVER Cloud has over 60,000 GPU infrastructure optionality in sovereign and physical AI. And C2C is showing early global turnaround signs.

But it is not confirmed yet. Foreign investors were still net sellers over the 10-day window, and consensus target price and EPS have not turned upward.

NAVER is now a conditional Buy candidate, not a one-day chase. The confirmation set is foreign flow, price holding above KRW 234,000, Q2 margin recovery, Dunamu approval, stablecoin regulation, and C2C profit improvement.

Evidence Ledger

Item	Source	Evidence
NAVER Q1 2026 results	NAVER official IR	Revenue KRW 3.2411T, OP KRW 541.8bn, Financial Platform, Global Initiatives, C2C/Enterprise growth
Dunamu consolidation	NAVER shareholder presentation, Nov. 2025	Transaction structure, Dunamu KRW 15.1T, NAVER Financial KRW 4.9T, consolidated net cash/BTC KRW 8.1T
NAVER Cloud GPUs	NVIDIA Newsroom PDF	NAVER Cloud over 60,000 GPUs for sovereign/physical AI workloads
Jensen Huang catalyst	Korea JoongAng Daily / other press	Expected Korea visit and AI partnership discussion including NAVER
Price/flow/consensus	Research OS local DB	May 29 close KRW 234,000, institution +KRW 205.2bn, FY2026E EPS KRW 12,628, target KRW 300,450
Mirae Asset stake	Wisereport / ownership datasets / Research OS local DB	NAVER owns roughly 47.74M shares; conservative value KRW 2.93T, local price value KRW 3.77T

US Non-Semiconductor Re-Rating Check: AI Power, Software and Stablecoin Rails Through a Korean Equity Lens

Sun, 31 May 2026 16:20:00 +0900

Connected context This post follows AI token futures and cost per token, whether China and Hong Kong AI overheating can spill into Korea, and HD Hyundai Heavy Industries’ SMR option. The question here is narrower: after the US non-semiconductor re-rating, which Korean second-line stocks can translate the same AI-infrastructure logic into actual price and flow?

TL;DR

Over the last two months, the strongest US re-rating themes outside semiconductors were quantum computing, onsite power and fuel cells, AI software and data apps, crypto and stablecoin rails, and second-line space-defense-drone names.

The most investable baskets are not necessarily the highest-return baskets. My preference is AI power/grid plus AI software/data operations. The first reflects the shift of AI-factory bottlenecks from GPUs to electricity. The second reflects the reversal of the “AI kills SaaS seats” fear into stronger Snowflake and Datadog-style demand for data, observability, security and AI operations.

The Korean translation is different.

US theme	Korean translation	Price reflection
AI power/grid	Electrical equipment, cables, busduct, HVDC, second-line grid names	Leaders already priced, second-lines only
AI software/data apps	ERP, AX, cloud, enterprise data, AI operations platforms	Still underpriced
Stablecoin/payment rails	PGs, banks, wallets, KRW stablecoin payments	Underpriced but policy-driven
Space/defense/drones	Defense electronics, satellites, drones, space systems	Some second-lines underpriced
Quantum/security	Mostly quantum security and PQC, not true quantum compute	Already overheated
Fuel cell/on-site power	Data-center onsite power, SOFC, PAFC	High-beta, stock selection required

The practical Korean watchlist is short.

Rank	Stock	View	Why
1	NHN KCP	Conditional Buy	Stablecoin payment PoC, 1Q earnings growth, low-teens PER, recent foreign and institutional buying
2	Douzone Bizon	Watchlist	Korea’s closest AI software/data operations translation. Price has barely moved while earnings growth is strong
3	Hanwha Systems	Watchlist	Space and defense-electronics option is not fully priced, but valuation and Philly Shipyard risks remain
4	Semyung Electric	High-volatility watchlist	Second-line grid candidate after leader correction. Lower valuation burden but liquidity risk

There is no broad “buy now” call. This is a second-line confirmation market, not a late chase market. The practical trigger is a 20-day moving-average recovery, box breakout, and renewed foreign plus quality-institutional flow.

1. Data Basis

US price data is based on yfinance adjusted close prices, from the March 31, 2026 close to the May 29, 2026 close. Direct semiconductors, semiconductor equipment and memory names are excluded. SOXX and SMH are used only as benchmarks.

Korean price and flow data comes from the local Research OS database:

/Users/youngseongshin/agents/Stock_Research/agents/KR_Crawler/data/screener_kr.db

Korean prices use the same March 31 to May 29 window. Flow data uses the most recent 20 and 5 trading days from investor_flow_raw_daily, with amount_unit='million_krw'.

[Blocked] I did not fully recompute every stock’s latest forward EV/Sales, forward PER, EPS revision history or full DART line items. “Re-rating” here means price, basket performance, flow and verified public catalysts.

2. US Non-Semiconductor Re-Rating Ranking

Rank	Sub-theme	Basket size	2M median	1M median	Median drawdown from high	Leaders
1	Quantum computing	5	+81.9%	+45.9%	-2.8%	IONQ +150.0%, QBTS +108.9%, RGTI +81.9%
2	Clean hydrogen / fuel cell	7	+74.8%	+12.5%	-4.6%	FCEL +231.7%, BLDP +159.9%, BE +110.3%
3	AI software / data apps	10	+34.0%	+30.3%	0.0%	DDOG +109.5%, CRWD +87.2%, SNOW +69.4%
4	Crypto / stablecoin / brokerage rails	9	+27.5%	-1.2%	-10.4%	RIOT +119.3%, CLSK +114.9%, MARA +76.2%
5	Space / defense / drones	10	+22.6%	+14.2%	-1.9%	LUNR +136.2%, RKLB +123.4%, PL +83.0%
6	Software broad	8	+21.3%	+14.8%	0.0%	SKYY +32.2%, IGV +27.0%, CLOU +24.5%
7	Copper / electrification metals	7	+15.4%	+12.5%	-2.1%	TECK +27.8%, BHP +22.2%, XME +15.9%
8	Grid / electrical equipment / EPC	10	+15.0%	-4.9%	-9.9%	GNRC +42.3%, FIX +32.6%, PWR +29.7%
9	AI power / data-center electricity	11	+12.3%	-4.6%	-7.4%	BE +110.3%, OKLO +34.9%, PWR +29.7%

Benchmarks:

Index / ETF	2M	1M	Drawdown
SPY	+16.3%	+5.0%	0.0%
QQQ	+27.9%	+9.5%	0.0%
IWM	+17.1%	+4.0%	-0.5%
SOXX	+73.2%	+22.2%	-0.2%
SMH	+56.2%	+17.5%	-0.5%

Semiconductors remain the strongest broad complex. But outside semis, quantum and fuel cells caught up with the SOXX-style move. That does not automatically make them the best buys. The cleaner re-rating is in AI software and AI power.

3. US Theme Read-Through

Quantum Computing

Quantum was the strongest two-month basket. IonQ’s Q1 2026 filing points to revenue growth, a large cash and investment balance, and FY2026 revenue guidance. (SEC IonQ EX-99.1)

[Inference] The market moved from “this is still a lab theme” to “commercial revenue is arriving faster than expected.” But after 80% to 150% two-month moves in IONQ, QBTS and RGTI, this looks more like late-stage extension than a fresh O’Neil-style first breakout.

View: Watchlist. Do not chase before a new base forms.

Clean Hydrogen and Onsite Fuel Cell Power

Bloom Energy reported Q1 2026 revenue of $751.1 million, up 130.4% year over year, product revenue up 208.4%, and operating income of $72.2 million. It also raised the midpoint of full-year 2026 revenue growth guidance from about 60% to about 80%. (Bloom Energy)

[Inference] This is not a generic hydrogen-economy trade. It is an AI data-center power timing trade. Grids take time. AI campuses need electricity now. Bloom is being reclassified as onsite power for AI infrastructure.

View: Bloom deserves separate treatment. Avoid chasing lower-quality FCEL, BLDP and PLUG after sharp moves.

AI Software and Data Apps

Snowflake reported FY2027 Q1 product revenue of $1.334 billion, up 34%, and RPO of $9.21 billion, up 38%. (Snowflake SEC filing)

Datadog reported Q1 2026 revenue of $1.006 billion, up 32%, about 4,550 customers with $100k+ ARR, and GA releases for GPU Monitoring, Bits AI Security Agent and MCP Server. (Datadog IR)

[Inference] The AI-software re-rating is the reversal of a fear: AI does not simply destroy software seats. It also creates more demand for data control planes, observability, security and AI operations.

View: Highest-quality non-semi re-rating. Wait for post-earnings digestion rather than chase.

Crypto, Stablecoin and Brokerage Rails

Coinbase reported Q1 2026 crypto trading volume market share of 8.6%, average USDC held in Coinbase products of about $19 billion, and Base processing 62% of global onchain stablecoin transaction volume. (Coinbase IR)

AP reported that the GENIUS Act was signed into law on July 18, 2025, creating a US regulatory framework for stablecoins. (AP)

[Inference] The re-rating is broadening beyond Bitcoin beta into stablecoins, prediction markets and agentic commerce rails.

View: COIN and HOOD are higher-quality platforms. Miners remain high-beta trading instruments.

Space, Defense and Drones

Rocket Lab reported Q1 2026 revenue of $200.3 million, up 63.5%, backlog of $2.2 billion, and new Electron/HASTE and Neutron launch contracts. (Rocket Lab IR PDF)

[Inference] The theme is no longer just small launch vehicles. It is space systems, defense payloads and vertical integration.

View: Watch RKLB, AVAV, KTOS and ASTS selectively. Contract confirmation and backlog quality matter more than the theme label.

4. Korea Translation: What Is Priced And What Is Not?

Korean sub-theme	Names	2M median	1M median	Median drawdown	20D foreign	20D institution	Read
Copper/electrification metals	9	+70.2%	-19.0%	-26.0%	-KRW 98.0bn	+KRW 19.4bn	Already re-rated, now correcting
Quantum/security/network	8	+67.4%	-4.3%	-26.5%	+KRW 49.8bn	-KRW 26.6bn	Overheated
AI power/grid/electrical	12	+42.8%	-24.3%	-26.0%	-KRW 1.44tn	-KRW 165.8bn	Leaders priced, second-lines only
On-site power/fuel cell/SMR	11	+26.8%	-19.3%	-23.0%	-KRW 435.6bn	-KRW 404.6bn	Some re-rating, weak flow
Space/defense/drones	12	-0.5%	-19.5%	-24.7%	+KRW 157.9bn	+KRW 10.2bn	Underpriced in second-lines
Crypto/stablecoin/payment rails	12	-3.6%	-17.3%	-25.7%	-KRW 32.6bn	-KRW 1.2bn	Underpriced, stock selection needed
AI software/data apps	12	-3.6%	-14.8%	-21.0%	-KRW 306.4bn	-KRW 25.4bn	Underpriced, quality filter needed

The key translation:

Korean power, copper and quantum-security baskets already front-ran much of the US theme. Korean AI software and stablecoin-payment rails have not.

5. Korean Candidate 1: NHN KCP

View: Conditional Buy Theme: Stablecoin payments, PG, PAYCO, payment infrastructure

NHN KCP is the cleanest Korean stablecoin-payment read-through. Seoul Shinmun reported that the company is running an online and offline stablecoin payment PoC using an Avalanche-based payment-specific mainnet, PAYCO integration, merchant-management tools, and QR approval times of about two seconds. (Seoul Shinmun)

It also signed a digital payment partnership with NH Nonghyup Bank. (Financial News) Yonhap reported Q1 2026 operating profit of KRW 13.8 billion, up 26.4% year over year and 4.4% above consensus. (Yonhap)

Metric	Value
Close	KRW 17,930
2M return	+14.3%
1M return	-10.6%
Drawdown from 2M high	-15.8%
20D foreign flow	+KRW 3.81bn
20D quality-institution flow	+KRW 2.89bn
5D foreign flow	+KRW 1.74bn
5D institutional flow	+KRW 3.66bn
Forward PER	13.0x
TTM PER	14.0x
OP YoY	+24.9%

Why it is not fully priced: The stock is only up 14.3% over two months and down 10.6% over one month. The market still prices it mostly as a low-margin PG stock, not as a possible KRW stablecoin payment-infrastructure option.

Entry: KRW 17,000 to 18,000 support with continued foreign and quality-institution buying, or a KRW 18,500 breakout with turnover expansion.

Invalidation: PoC fails to convert into commercialization discussion, PG operating-profit growth slows below 10%, or KRW 16,500 breaks while foreign flow disappears.

6. Korean Candidate 2: Douzone Bizon

View: Watchlist / breakout candidate Theme: AI ERP, enterprise data, AI software operations

Douzone Bizon is the most natural Korean translation of the Snowflake and Datadog logic. Enterprise AI needs access to ERP, workflow, approval, documents, email, groupware and permissioned internal data. Douzone already sits on those surfaces.

BusinessPost reported that Douzone’s Q1 2026 results were expected to improve on AI and cloud-solution revenue growth plus profitability improvement. (BusinessPost)

Metric	Value
Close	KRW 120,000
2M return	+0.8%
1M return	0.0%
Drawdown from 2M high	-0.2%
20D foreign flow	+KRW 24.37bn
5D foreign flow	+KRW 6.11bn
TTM PER	36.8x
OP YoY	+44.9%

Entry: Hold KRW 118,000 to 120,000 with volume contraction, or break KRW 123,000 to 125,000 with continued foreign buying.

Invalidation: AI/cloud revenue growth slows, operating-profit growth falls below 20%, or the stock breaks KRW 115,000 with foreign buying disappearing.

7. Korean Candidate 3: Hanwha Systems

View: Watchlist / conditional inclusion Theme: Space, defense electronics, satellites, radar, ICT

Chosun Biz reported Hanwha Systems Q1 2026 revenue of KRW 807.1 billion, operating profit of KRW 34.3 billion, and order backlog of KRW 12.1963 trillion. Defense exports and domestic mass-production projects supported growth. (Chosun Biz)

Metric	Value
Close	KRW 105,000
2M return	-8.0%
1M return	-12.3%
Drawdown from 2M high	-23.4%
20D foreign flow	+KRW 62.82bn
5D foreign flow	+KRW 32.94bn
5D quality-institution flow	+KRW 9.81bn
Forward PER	82.0x

Why it is not fully priced: The stock is down over two months, but foreign buyers have returned. The market is still focused on valuation and connected-loss risks rather than defense electronics, satellites, SAR and AI imagery options.

Entry: KRW 102,000 to 105,000 support with continued foreign buying, or KRW 112,000 breakout with institutional selling easing.

Invalidation: Philly Shipyard-related losses worsen in Q2, defense margin slows, or KRW 100,000 breaks with foreign buying stopping.

8. Korean Candidate 4: Semyung Electric

View: High-volatility watchlist Theme: Second-line grid, HVDC, power equipment

Metric	Value
Close	KRW 9,890
2M return	+14.3%
1M return	-37.3%
Drawdown from 2M high	-40.3%
20D foreign flow	+KRW 2.04bn
20D institutional flow	+KRW 0.12bn
TTM PER	11.4x
OP YoY	+393.3%

[Inference] Korean grid leaders have already re-rated. Semyung is a smaller, higher-volatility second-line name with a deeper one-month correction and lower valuation burden. The price setup is interesting, but liquidity risk is real.

Entry: KRW 9,500 to 10,000 support with turnover contraction, or KRW 11,000 recovery with foreign and institutional buying.

Invalidation: HVDC/grid earnings repeatability weakens, or KRW 9,000 breaks with turnover drying up.

9. What Not To Chase

Group	Why
Xgate, Iwin Plus, Dream Security	Up 100% to 278% over two months. This is theme beta, not the same quality as US quantum commercialization
LS Electric, Hyosung Heavy, Sanil Electric	Good businesses, but already re-rated and now testing flow deterioration
Doosan Fuel Cell	Strong fuel-cell move, but valuation and loss risks remain high
Taihan Cable, KBI Metal, Gaon Cable	Copper/cable beta surged and then corrected deeply. Wait for a new base

10. Execution Framework

Buy now: None.

Waitlist conditions

Bucket	Names	Trigger
US AI software	SNOW, DDOG, CRWD, APP, MDB, NET	5 to 10 trading days of post-earnings digestion, 20-day support, volume contraction then re-acceleration
US AI power/grid	GEV, PWR, ETN, BE, CEG, VST	8% to 15% pullback from highs, lower volume, next order or guidance confirmation
Korean stablecoin rails	NHN KCP	KRW 18,500 breakout or KRW 17,000 to 18,000 support plus flow
Korean AI software	Douzone Bizon	KRW 123,000 to 125,000 breakout plus foreign buying
Korean space/defense second-line	Hanwha Systems	KRW 112,000 breakout plus easing institutional selling
Korean power second-line	Semyung Electric	KRW 11,000 recovery plus foreign and institutional buying

11. Final View

The US non-semiconductor re-rating proves that AI does not end with semiconductors. But not every follow-on theme has the same quality.

The important shift is that AI-factory bottlenecks are spreading from GPUs into power, operations software, data-center infrastructure, payment rails and space-defense networks.

In Korea, power, copper and quantum-security leaders already moved first. What remains less priced is stablecoin payments, AI software/data operations, and selected second-line space-defense names.

This is not a market for chasing yesterday’s leaders. It is a market for confirming the first real flows into second-line candidates. My active Korean watchlist is NHN KCP, Douzone Bizon, Hanwha Systems and Semyung Electric.

Evidence Ledger

Item	Source	What it confirms
Price screens	yfinance and Research OS local DB	US and Korean theme-basket returns through May 29, 2026
IonQ	SEC EX-99.1	Q1 2026 revenue, cash and guidance context
Bloom Energy	Company IR	Q1 2026 revenue, growth and raised guidance
Snowflake	SEC EX-99.1	FY2027 Q1 product revenue and RPO
Datadog	Company IR	Q1 revenue, $100k+ ARR customers, GPU/MCP/AI product launches
Coinbase	Company IR	USDC, Base, stablecoin and agentic-commerce metrics
GENIUS Act	AP	US stablecoin regulatory framework
NHN KCP PoC	Seoul Shinmun	Online/offline stablecoin payment PoC, PAYCO integration, QR approval time
NHN KCP earnings	Yonhap	Q1 2026 operating profit growth
Hanwha Systems	Chosun Biz	Q1 2026 revenue, operating profit and backlog

Follow-Up Queue

Compare forward EV/Sales, NTM revenue growth and FCF margins across the AI-software basket.
Compare backlog growth and 2026/2027 EPS revisions across AI power/grid names.
Re-run Korean 5-day and 20-day volume plus investor-flow screens on the four candidates.
Apply late-stage filters to quantum and fuel-cell surge names.
Track NHN KCP’s commercial stablecoin timetable and whether economics accrue to banks, PGs, wallets or platforms.

For research and information purposes only. Not investment advice.

Can China's and Hong Kong's AI Overheat Spill Into Korea?

Sun, 31 May 2026 15:30:00 +0900

Context This note follows our work on Upstage and Korean sovereign AI, SK Telecom as an AI infrastructure operator, and AI token futures and cost per token. The question here is narrower: if Hong Kong and China are willing to price scarce LLM listings aggressively, where can that narrative land in Korea’s listed market? Related hubs: Korean AI Companies Hub and Korea Stocks for Foreign Investors Hub.

TL;DR

Yes, the Hong Kong / China AI heat can spill into Korea. But it is unlikely to arrive as a clean Zhipu- or MiniMax-style listed LLM pure-play trade. Korea almost does not have such a stock.

The spillover is more likely to happen through four proxies.

Spillover path	Korea proxy	View
Sovereign LLM scarcity	NAVER, private Upstage	Possible, but NAVER is a platform with an LLM option, not an LLM pure-play
AI cloud / enterprise AI	LG CNS, Samsung SDS, SK AX	The most realistic route, because revenue and customers are visible
AI data center / GPUaaS	SK Telecom, parts of NAVER Cloud	Direct, but utilization and capex payback must be proven
LLM venture proxy	SK Networks, parts of SK Inc.	Event-driven exposure to Upstage valuation / IPO optionality

The key conclusion:

If China’s and Hong Kong’s AI overheat transmits to Korea, it will probably not be a “buy the Korean MiniMax” trade. It will be a reclassification trade in companies that can turn AI usage into recurring revenue, margin, and capex payback.

The cleanest listed candidates are LG CNS and SK Telecom. LG CNS has the clearest Cloud&AI revenue mix. SK Telecom is beginning to show real AIDC / GPUaaS numbers. Samsung SDS has the strongest narrative, but the numbers need to catch up. NAVER owns the largest sovereign AI option, but it is not a pure-play. SK Networks is an Upstage event proxy.

1. Why Hong Kong and China matter

The core signal from Hong Kong and China is simple.

The market is willing to price scarce foundation-model listings as sovereign AI call options, not as normal near-term earnings stories.

Zhipu / Knowledge Atlas is one of the clearest examples. It owns the GLM model family and monetizes through API and private deployment. Its 2025 revenue was roughly RMB724 million, up 131.9% year on year, but net loss was about RMB4.72 billion. The stock is therefore not about current earnings; it is about scarcity, domestic model capability, and the “China OpenAI” option.

MiniMax follows a similar logic. Its 2025 revenue was reported at about US$79 million, up 158.9% year on year, while net loss remained large. Investors are paying for a global AI app, multimodal model optionality, and LLM scarcity rather than present profitability.

Enterprise AI and AI cloud names are valued differently. Fourth Paradigm, SenseTime, Baidu, Alibaba, and Tencent all sit in different baskets. The market asks whether AI revenue is recurring, whether API usage is scaling, whether gross margin after inference cost is attractive, and whether capex will dilute the economics.

Basket	Multiple logic	Korea translation
LLM / foundation model pure-play	Extreme P/S can appear	Almost absent in listed Korea
AI app / consumer AI	High growth premium possible	Private Upstage, some NAVER service optionality
Enterprise AI platform	Lower P/S or EV/EBIT	LG CNS, Samsung SDS, SK AX
AI cloud / big tech	Normal PER plus capex discount	NAVER, Samsung SDS, parts of SKT
AI infra / GPUaaS	EV/EBITDA and capex payback	SK Telecom, parts of NAVER Cloud

The practical question is therefore not “Does Korea have a listed Zhipu?” It does not. The better question is: which Korean listed companies can borrow part of the China/Hong Kong AI valuation language?

2. Korea’s missing slot: no listed LLM pure-play

Korea has almost no direct Zhipu / MiniMax equivalent in the public market.

NAVER is the only large listed Korean platform with a meaningful proprietary LLM option: HyperCLOVA X, CLOVA Studio, NAVER Cloud, AI search, and AI-driven ad / commerce optimization. But NAVER’s revenue mix is still search, ads, commerce, fintech, content, and C2C. NAVER reported Q1 2026 revenue of KRW3.2411 trillion and operating profit of KRW541.8 billion, and it reclassified revenue into NAVER Platform, Financial Platform, and Global Initiatives. (NAVER)

That makes NAVER a platform company with an LLM option, not an LLM company.

The more realistic public-market route is AI cloud integration. LG CNS and Samsung SDS do not sell foundation models as their core product. They help enterprises deploy AI through cloud migration, MSP, GPUaaS, agent workflows, public-sector AX, and managed platforms.

Category	Listed in Korea?	Representative names	View
LLM pure-play	Almost no	Private Upstage is closest	No direct listed comp
Large listed proprietary LLM owner	Yes	NAVER	Sovereign LLM optionality
AI cloud integrator	Yes	LG CNS, Samsung SDS, SK AX	The most realistic spillover path
AI data center / GPUaaS	Yes	SK Telecom, parts of NAVER Cloud	Asset-heavy AI infra proxy
AI venture / LLM proxy	Yes	SK Networks	Indirect Upstage exposure

3. Company-by-company translation

3.1 NAVER: listed sovereign LLM owner, not a pure-play

NAVER has Korea’s most visible listed LLM option. HyperCLOVA X, CLOVA Studio, NAVER Cloud, AI Search, advertising, and commerce all connect to the thesis. NAVER said AI accounted for more than 50% of advertising revenue growth in Q1 2026, and it also referenced sovereign AI opportunities. (NAVER)

But the right equation is not “LLM revenue x LLM P/S”.

NAVER AI value
= better defense and monetization in search / ads / commerce
+ Enterprise AI / NAVER Cloud revenue
+ sovereign AI optionality
- AI capex and compute cost burden

NAVER can receive a scarcity premium if Korea investors start looking for listed sovereign AI exposure. But the pace of re-rating depends on AI revenue disclosure and margin defense.

3.2 Samsung SDS: KKR-backed full-stack AI integrator, but numbers need to catch up

Samsung SDS is not a model company. It is an AI cloud integrator with a strong narrative: FabriX, Brity Copilot, Brity Automation, Samsung Cloud Platform, GPUaaS, public-sector AX, and KKR-backed capital.

In Q1 2026, Samsung SDS cloud revenue was KRW690.9 billion, up 5.8% year on year. Against KRW1.6105 trillion of IT services revenue, cloud accounted for 42.9%.

Cloud revenue share = KRW690.9bn / KRW1.6105tn = 42.9%

Samsung SDS reported Q1 revenue of KRW3.3529 trillion and operating profit of KRW78.3 billion. Operating profit fell sharply due to a one-time KRW112 billion retirement-benefit provision. The company also laid out a KRW10 trillion investment plan by 2031, including KRW5 trillion for AI infrastructure. (Samsung SDS)

Samsung SDS can attract the strongest AI full-stack narrative in Korea. The constraint is proof: cloud growth, AI SaaS revenue, GPUaaS margin, recurring revenue, and public AX conversion must show up in the numbers.

3.3 LG CNS: the cleanest listed Korean AI cloud integrator

LG CNS is not an LLM company either. But it may be the cleanest listed AI cloud integration proxy.

The Elec reported that LG CNS posted Q1 2026 revenue of KRW1.315 trillion and operating profit of KRW94.2 billion. Its core AI and cloud segment revenue reached KRW765.4 billion, or about 58% of total sales. (The Elec)

Cloud&AI revenue share = KRW765.4bn / KRW1.315tn = 58.2%

Its portfolio includes public-sector, defense, finance, manufacturing, pharma, biotech, and shipbuilding AX projects; global cloud partnerships; OpenAI and Palantir relationships; agentic AI workflows; and data-center DBO.

The upside is not a Zhipu-style LLM multiple. The upside is that LG CNS may deserve a higher multiple than a traditional SI if Cloud&AI revenue is recurring, high-margin, and locked into customer workflows.

3.4 SK Telecom: the most direct AI infrastructure proxy

SK Telecom can be reclassified from telco to AI data center / GPUaaS / sovereign AI infrastructure operator.

In Q1 2026, SKT reported consolidated revenue of KRW4.3923 trillion, operating income of KRW537.6 billion, and net income of KRW316.4 billion. Its AIDC business recorded KRW131.4 billion of revenue, up 89.3% year on year, helped by higher data-center utilization and GPUaaS revenue. (SK Telecom)

Annualized AIDC revenue = KRW131.4bn x 4 = KRW525.6bn
Annualized group revenue = KRW4.3923tn x 4 = KRW17.5692tn
AIDC share = KRW525.6bn / KRW17.5692tn = 3.0%

The share is still small. But it is real and growing. SKT combines AIDC, GPUaaS, the A.X sovereign model, the A. agent, B2B AI, and group synergies with SK hynix, SK ecoplant, and SK Innovation / E&S.

SKT is not CoreWeave. But in the Korean listed market, it is one of the most direct AI infrastructure operators.

3.5 SK Inc.: AI full-stack holding company, not AI software multiple

SK Inc. is not an LLM stock or cloud integrator. It is an AI full-stack holding company.

SK Square and SK hynix provide AI memory exposure. SK Telecom provides AIDC / GPUaaS. SK AX provides enterprise AX. SK ecoplant and materials subsidiaries provide infrastructure and materials. Energy and cooling sit in the wider group. SK Inc.’s Q1 2026 briefing also shows SK hynix HBM / eSSD strength and SK Telecom AIDC / GPUaaS contribution. (SK Inc.)

The right framework is NAV discount, not LLM P/S.

SK Inc. appeal
= AI-related affiliate NAV growth x holding discount compression
- non-AI asset, debt, and overlap risk

3.6 SK Networks: listed Upstage call option

SK Networks is not a fundamental AI operating company. It is a public-market proxy for Upstage optionality.

Maeil Business reported that SK Networks, an existing Upstage investor, is known to have invested an additional KRW50 billion in Upstage’s Series C round. Upstage raised KRW180 billion in the first close and was valued above KRW1 trillion. (MK) VentureSquare separately reported the same KRW180 billion first close and unicorn valuation. (VentureSquare)

What remains unclear is important: exact SK Networks ownership, post-money dilution, fair-value accounting, and IPO timing.

SK Networks is therefore an event proxy for Upstage valuation / IPO optionality, not a core AI operating-company re-rating.

3.7 Upstage: private benchmark for Korean LLM valuation

Upstage is the closest Korean private LLM pure-play. Solar, Document AI, private deployment, and sovereign AI exposure make it the key benchmark.

The public market will care about the following metrics if Upstage moves toward an IPO.

KPI	Why it matters
Model / API revenue	Core LLM pure-play metric
Document AI ARR	Real enterprise monetization
On-prem / private deployment revenue	Regulated sovereign AI demand
Gross margin after inference cost	True LLM economics
Inference cost per 1M tokens	Model efficiency
Customer ARR / NRR / GRR	Proof that it is recurring software, not SI
Overseas revenue share	Global AI app / service optionality

The better framing is not “Korea’s OpenAI”. It is an LLM-native workflow platform for regulated enterprise document and knowledge work.

4. Ranking the Korea proxies

AI cloud / enterprise integration

Rank	Company	Why
1	LG CNS	Cloud&AI is already about 58% of revenue; cleanest listed proxy
2	Samsung SDS	Strong KKR / FabriX / Brity / GPUaaS narrative; needs proof
3	SK AX / SK Inc.	Relevant, but small within SK Inc. NAV

AI infrastructure / GPU cloud

Rank	Company	Why
1	SK Telecom	AIDC revenue +89.3% YoY; GPUaaS demand is visible
2	NAVER	NAVER Cloud, GPUaaS, sovereign AI; smaller within group revenue
3	Samsung SDS	Some SCP / GPUaaS exposure; more integrator than infra pure-play

LLM / sovereign AI

Rank	Company	Why
1	Upstage	Private, but closest Korean LLM pure-play
2	NAVER	Only large listed proprietary LLM platform
3	SK Telecom	A.X exists, but LLM revenue disclosure is limited

Event / proxy

Rank	Company	Why
1	SK Networks	Upstage IPO / valuation proxy
2	SK Inc.	SK Group full-stack AI NAV proxy
3	Samsung SDS	KKR / AI infrastructure investment events

5. Failure conditions

The AI spillover can happen, but it can also fail quickly if the market sees only labels without numbers.

Theme	Spillover probability	Failure condition
LLM pure-play P/S	Low	No listed Korean pure-play
Sovereign LLM scarcity	Medium	Weak NAVER AI revenue disclosure; delayed Upstage IPO
AI cloud integrator	High	AI revenue proves to be low-margin project SI
AI data center / GPUaaS	High	Utilization, margin, and capex payback stay undisclosed
Upstage proxy	Event-driven	Ownership / fair value remains opaque; IPO delayed

The question is not “Does the company do AI?”

The better question is: Can the company convert AI usage into revenue, recurrence, margin, and capex payback?

Conclusion

The Hong Kong / China AI heat can spill into Korea, but the form will differ. Korea lacks a listed Zhipu or MiniMax. The local trade is therefore a reclassification trade in existing listed companies.

LG CNS and SK Telecom are the cleanest structural candidates. Samsung SDS has the strongest narrative but needs numbers. NAVER has the largest listed sovereign AI option but is not a pure-play. SK Networks is an Upstage call option.

One-line version:

In Korean AI stocks, the key is not “who is the LLM?” It is who can translate rising AI usage into recurring revenue, margin, and capex recovery?

Sources

In an Earnings Bubble, Estimates Get Cut Last: AI Infrastructure Crowding and the Value of a Deep-Dive

Sun, 31 May 2026 14:00:00 +0900

This is a methodology note. For the underlying pieces that fed the analysis, it helps to read them alongside: the AI substrate/PCB thesis (the system BOM’s common bottleneck), Goldman’s token demand vs. J.P. Morgan’s memory ASP peak-out, and the Thesis OS public note that explains the structure running all of this work.

TL;DR

In a recent report, BCA Research argues that the AI bubble is not a valuation bubble but an earnings bubble. It is not the P/E that swells but earnings themselves. And like every bubble it deflates eventually, though BCA adds that its own AI demand gauges show no imminent signal yet.
The defining feature of an earnings bubble is the lead-lag. In BCA’s words, in almost every case “stocks began to fall well before profit estimates were cut.” Consensus estimates are a lagging signal.
So precisely in a phase where money is crowding into AI infrastructure like this, what matters more is a deep-dive that reads system structure and leading demand indicators directly, instead of waiting on consensus EPS. Once estimates are cut, it is already too late.
This note sets out, without exaggeration, what such a deep-dive actually looks at, and how we have run it as a structure called Thesis OS.

1. What it means to call the AI bubble an “earnings bubble”

When people say “bubble,” they usually picture P/E ratios shooting up — a valuation bubble, where price rises far faster than earnings. BCA Research sees AI as a somewhat different kind. It is an earnings bubble, where earnings themselves swell rather than price.

This is not a new pattern. Homebuilders and banks did exactly this just before the financial crisis. Their P/E ratios looked low, but only because unsustainable earnings inflated the denominator (E) and made the multiple look cheap. Industries that swing hard with boom and bust — natural resources, airlines, shipping, and today’s semiconductors — are vulnerable to this kind of earnings bubble.

Right now the semiconductor revenue curve resembles that picture.

Source: an approximate reconstruction based on public WSTS annual aggregates, with 2025-2026 as estimates. Illustrative, not investment advice. The shape of the underlying data is in the same vein as the “parabolic semiconductor sales” chart presented in the BCA Research report (2026-05-28).

A revenue curve going parabolic is both good news and a warning. When earnings rise fast, the P/E looks low. But if those earnings are a product of the cycle, the very fact that the multiple looks cheap can become a trap. The old warning of cyclical industries applies here: “the most dangerous moment is when earnings are at their peak."

Make no mistake. Neither BCA nor we are saying “it deflates now.” BCA judges that its AI demand gauges — adoption rates, token spending, AI coding-tool downloads, GPU and memory prices — are mostly still at reassuring levels. The point is not timing but how this bubble behaves.

2. The real trap of an earnings bubble is the “lead-lag”

What makes an earnings bubble dangerous is not that it bursts, but the order in which it bursts.

The core point BCA makes is this: Wall Street analysts are poor at calling when an earnings bubble will deflate. And in almost every case, “stocks began to fall well before profit estimates were cut” (BCA Research, 2026-05-28).

Here is what that sentence means in practice, drawn as a picture.

This is a conceptual diagram, not actual data. It simplifies the lead-lag structure BCA described, in which the price leads and the estimates lag.

The red line (the share price) turns down first. The blue dotted line (consensus earnings estimates) is cut only much later. The gray band in between is the lag. If you hold a rule that says “I’ll sell when analysts cut their target price or estimates,” you will always move late by exactly that lag.

The conclusion follows. Consensus estimates are a lagging signal. The stock does not look most expensive when earnings peak, and by the time estimates are cut the price has already fallen. So if you watch only the estimates, you miss both the peak of the bubble and its inflection point.

3. That is why a deep-dive is needed — what does it look at

If estimates lag, what should you watch to get ahead? A deep-dive looks not at headline EPS but at the structure and leading indicators that produce that EPS. Concretely, four things.

① It reads the system structure. The linear story — “after GPU comes memory, then substrates” — is easy to trade but only half right. Real AI infrastructure is a rack-level system in which GPU, CPU, DPU, NIC, switch ASIC, memory modules, and power boards all scale together. As we laid out in the AI substrate/PCB thesis, substrates and PCBs are not the final stop of a rotation but the common denominator of the entire system bill of materials (BOM). Knowing the structure reveals where the true bottleneck is.

② It separates the variables. Looking at the same AI demand, Goldman tracks token usage (Q) and cost per token (C), while J.P. Morgan tracks the rate of rise in memory prices (P). Decompose the two forecasts into P, Q, and C and it becomes clear that the two views, which looked like a clash, are actually talking about different variables and can hold at the same time. Lumping it all into a single headline number is what hides this.

③ It tracks leading indicators directly. Instead of waiting for consensus EPS to be cut, it watches what moves earlier — HBM long-term contract prices and volumes, server DRAM contract prices, token spending, GPU and memory spot prices, adoption rates. These change direction ahead of the estimates.

④ It separates fact, inference, and speculation. “Officially confirmed fact,” “reasonable inference,” and “mere speculation” do not go in the same box. Things that are unverified — customer names, whether a part was adopted, contract terms — are clearly flagged as inference or speculation. Without that separation, you get swept up in an attractive story and buy speculation as if it were fact.

These four do not wait for estimates to be cut. That is why they suffer less from the lag.

4. Thesis OS — the structure that runs this deep-dive as a system

Doing the four things above well once or twice is not hard. The hard part is doing them every time, with the same discipline. So we entrust this work to a structure rather than to a person’s mood on a given day. That structure is Thesis OS.

Thesis OS is divided into three roles.

Role	What it does
Alpha (알파)	Evidence gathering — market data, screeners, crawlers, fact-checking pipelines
Lattice (격자)	Judgment — weaving evidence into a thesis, building forecasts, stress-testing with the counter-case
Arki (아키)	Governance — keeping the whole consistent through schemas, workflows, and health checks

The point is not flashy automation but the repeatability of discipline. Separating evidence (Alpha) from judgment (Lattice) reduces the chance that a good story runs ahead of the evidence. With governance (Arki) in place, you split fact, inference, and speculation by the same standard each time and keep tracking the leading indicators. Thesis OS is open source, so interested readers can inspect the structure itself directly.

5. Our blog’s work — plainly

We write this as a record, not a boast. The most honest evidence is what pieces this methodology actually produced.

Mapping the system structure: the AI substrate/PCB thesis — seeing AI as a rack-level system and redefining substrates as the common bottleneck.
Separating the variables: Goldman vs. J.P. Morgan — decomposing two seemingly opposed forecasts into P, Q, and C.
Earnings read-through: Marvell Q1 FY2027, Dell Q1 FY2027 — translating U.S. earnings into Korean component and materials bottlenecks.
Tracking the cost structure: AI token futures and cost per token — the axis shifting from a performance race to a cost race.

What these pieces share is that they do not rush to a “buy/sell” conclusion. Instead they map the structure, separate the variables, present the leading indicators, and treat names not as recommendations but as observation points. The aim is to give readers material to judge for themselves. We do not claim to be able to call the market top or the moment the bubble bursts. As BCA concludes, even analysts are poor at that. What we are trying to do is more modest: to understand the structure before estimates are cut, and to make ourselves watch leading signals rather than lagging ones.

6. Closing

In a phase where this much capital has crowded into AI infrastructure, the most dangerous posture is to wait for the consensus to turn for you. In an earnings bubble that signal always arrives late. The share price moves before estimates are cut.

So a deep-dive is not a flashy forecast but preparation to be less late. Understanding the system, separating the variables, watching the leading indicators directly, and distinguishing fact from speculation. To repeat that work not once but every time with the same discipline, we use a structure called Thesis OS. If you are interested, we encourage you to look not only at the conclusions but at the structure and process behind them.

This piece briefly cites, with attribution, the published core argument of BCA Research’s “Earnings Bubbles Are Still Bubbles” (Global Investment Strategy, 2026-05-28), and the charts are self-produced based on public data and concepts. It is not a recommendation to buy or sell any particular security; investment decisions and their consequences rest with the investor.

Is FADU Korea's Sandisk Beta? Foreign Flows and the AI Storage Bottleneck

Sun, 31 May 2026 13:20:00 +0900

Context This note follows SK Hynix vs Micron and the AI Infrastructure Multiple Map. Those notes focused on HBM and mega-cap memory. This one moves down the stack into AI storage, NAND and eSSD controllers. Related hubs: AI HBM Hub and Korea Semiconductor Value Chain Hub.

TL;DR

FADU is starting to behave like a Korean high-beta proxy for Sandisk, but the relationship is still early. The strongest evidence is not daily price correlation; it is foreign flow. Since May 2026, foreigners net bought about KRW 445.1bn of FADU while domestic institutions net sold about KRW 215.2bn. Over the same period, FADU rose +38.1% and Sandisk rose +35.0%.

The price beta is weaker than the SK Hynix-Micron pair. Sandisk’s prior-U.S.-session return versus FADU’s next Korea-session return correlation is +0.16 since May and +0.42 over the latest 10-14 trading days. Micron-to-Hynix is +0.57 and +0.55. Translation: foreign flow says yes, domestic institutions say no, and the broader market pair is still forming.

The conclusion: FADU is a candidate Korean beta for AI storage/NAND, but valuation is already demanding. Structurally, FADU is cleaner. Numerically, Jeju Semiconductor looks cheaper. Globally, Sandisk screens as the cleaner conditional buy candidate for FY2027, while Western Digital already prices in much of the HDD bottleneck.

Data Basis and Coverage

The analysis covers 2026-01-01 to 2026-05-29. FADU and SK Hynix prices and investor flows use Research OS local DB tables prices_daily and investor_flow_daily. Sandisk and Micron prices use Yahoo Finance/yfinance.

Coverage limits matter. Some May FADU individual flow and detailed institution-subtype fields are blank in the local DB, so this note judges FADU mainly through foreign and institution totals. U.S. names do not have Korean-style investor-flow data.

1. Is FADU Following Sandisk?

On price level, yes. FADU rose from KRW 21,250 to KRW 112,100 year-to-date, up about +427.5%. Sandisk also rallied sharply. Both are being repriced around AI storage and NAND scarcity.

But daily return linkage is still modest.

Item	Sandisk → FADU	Micron → SK Hynix	Interpretation
YTD price-level correlation	+0.91	high	Same broad uptrend
Prior U.S. day → next Korea day return	+0.12	+0.46	FADU pair still weak
Since May	+0.16	+0.57	Hynix-Micron is much more structural
Latest 10-14 sessions	+0.42	+0.55	FADU beta has recently strengthened

The key distinction is that “both went up” is not the same as “FADU mechanically tracks Sandisk.” FADU is not yet a mature pair trade. But the recent correlation jump suggests investors are beginning to map the relationship.

2. The Real Signal Is Foreign Flow

Since May, FADU’s flow profile has been unusually clear.

Period	FADU Return	Sandisk Return	FADU Foreign Flow	FADU Institution Flow	Foreign Positive-Day Ratio
2026 YTD	+427.5%	strong rally	+KRW 528.2bn	-KRW 173.2bn	46.4%
Since May	+38.1%	+35.0%	+KRW 445.1bn	-KRW 215.2bn	83.3%
Latest 10-14 sessions	+11.7%	+9.5%	+KRW 346.4bn	-KRW 194.1bn	85.7%

This is not a domestic-institution-led move. It is closer to a foreign-discovery trade in a Korean high-beta proxy for AI storage/NAND.

3. Why It Is Weaker Than SK Hynix-Micron

SK Hynix and Micron are both large-cap global memory stocks tied to the same HBM/DRAM cycle. Their earnings line items, investor base and product overlap are more directly comparable.

FADU and Sandisk are different. Sandisk is a global NAND and data-center SSD company. FADU is closer to an eSSD controller, firmware and enterprise SSD solution fabless company. Both benefit from the AI storage cycle, but the economics and customer-validation risks are not identical.

The clean framing is:

Sandisk = global AI storage/NAND thermometer FADU = foreign-discovered Korean high-beta proxy Key validation = continued foreign buying + breakout above KRW 118,500 + less institution selling

4. FADU and Jeju Through the “Next SEMCO” Lens

The Samsung Electro-Mechanics re-rating showed how a seemingly small component can be reclassified as an AI infrastructure bottleneck. Applying the same frame to memory/storage gives five tests: bottleneck, P up, Q up, customer validation and multiple reclassification.

Company	Bottleneck	Strength	Weakness	View
FADU	AI eSSD controller / firmware / SSD solution	Cleanest structural AI-storage bottleneck in Korea	Valuation already rich	Watch / Wait
Jeju Semiconductor	LPDDR4X, MCP, legacy low-power memory	Looks much cheaper on current numbers	Needs proof that 1Q26 repeats	Tactical Watchlist
Sandisk	NAND / data-center SSD	Global AI storage/NAND thermometer, cheaper FY2027 setup	NAND peak-out risk	Conditional Buy candidate
Western Digital	HDD capacity / nearline storage	Strong HDD cash-flow bottleneck	Much of the premium reflected	Low appeal for fresh entry

5. FADU: Strong Structure, Difficult Price

FADU’s structural thesis is strong. AI data centers are not just a GPU/HBM story. KV-cache, checkpoints, training/inference pipelines, enterprise SSDs, controllers and firmware become parallel bottlenecks. FADU sits near the eSSD controller and firmware qualification layer.

Public reporting says FADU posted Q1 2026 revenue of KRW 59.5bn and operating profit of KRW 7.7bn, returning to profitability, with disclosed new orders reaching KRW 166.3bn by April. (StorageNewsletter)

The issue is price. The input research cites MarketScreener estimates of KRW 323.3bn 2026E revenue and KRW 584.8bn 2027E revenue, with 2027E P/E around 38x. That is not cheap for a company that still needs to convert orders into revenue and margin.

Upgrade Conditions

Watch Item	Buy Upgrade Trigger
Price	Break and hold above KRW 118,500 with turnover
Flow	Continued foreign buying and less institution selling
Earnings	Q2 revenue KRW 70-80bn, OPM above 15%, ideally 20%
Orders	Cumulative new orders above KRW 400bn
Products	Gen6 / CXL / low-latency SSD design wins

Invalidation: foreign buying stops despite Sandisk strength, FADU breaks below KRW 105,900 with foreign selling, orders fail to convert, OPM falls below 10%, or controller mix declines.

6. Jeju: Cheaper Numbers, Higher Repeatability Risk

As discussed in the Jeju Semiconductor 1Q26 note, Jeju is not an HBM company. It benefits from shortages in LPDDR4X, MCP and low-power memory as Samsung, SK Hynix and Micron allocate more capacity toward HBM and server DRAM.

Jeju’s Q1 2026 operating margin was 37.2% on KRW 180.4bn revenue and KRW 67.1bn operating profit. TrendForce expects 2Q26 LPDDR4X ASPs to rise +70-75% QoQ and LPDDR5X +78-83%. (TrendForce)

On simple annualized Q1 numbers, Jeju looks cheaper than FADU. But the question is repeatability. If Q1 was peak margin, it is a trap. If Q2 revenue stays above KRW 180bn and OPM remains above 30%, it becomes a tactical undervaluation candidate.

7. Global Read-Through: Sandisk Screens Best, WDC Less Clean

Sandisk is the global AI storage/NAND thermometer. Fiscal Q2 2026 revenue was $3.025bn, GAAP net income was $803m and non-GAAP EPS was $6.20. The company guided fiscal Q3 revenue to $4.4-4.8bn and non-GAAP EPS to $12-14. (Sandisk)

Western Digital has a strong HDD bottleneck and cash-flow story. Fiscal Q3 2026 revenue was $3.337bn, GAAP gross margin was 50.2% and operating cash flow was $1.12bn. But fresh-entry appeal is lower because much of the HDD scarcity premium is already recognized. (Western Digital)

Company	Core Exposure	View
Sandisk	NAND, data-center SSD	Conditional Buy candidate
Jeju Semiconductor	LPDDR/MCP legacy memory squeeze	Tactical Watchlist
FADU	AI eSSD controller / firmware	Watch / Wait
Western Digital	HDD capacity / cash flow	Less attractive for fresh entry

Final View

FADU is starting to be interpreted as Korea’s Sandisk beta. It is not yet a mature global pair like SK Hynix-Micron. But KRW 445.1bn of foreign net buying since May is too large to dismiss.

Setup	Action
Sandisk strong + FADU foreign buying continues + KRW 118,500 breakout	Recognize FADU as Korean AI-storage beta; conditional Buy upgrade
Sandisk strong but FADU foreign flow slows and institutions keep selling	Beta failure; watch for theme peak
Sandisk corrects but FADU foreign buying persists	Upgrade toward FADU-specific alpha
Break below KRW 105,900 + foreign net selling	Risk management

One sentence:

FADU is being discovered as Korea’s high-beta proxy for AI storage/NAND, but the right stance is Watch / conditional Buy, not blind chase.

Evidence Classes

[Fact]

FADU Q1 2026 revenue was KRW 59.5bn, operating profit was KRW 7.7bn, and disclosed new orders reached KRW 166.3bn by April. (StorageNewsletter)
TrendForce expects LPDDR4X ASP +70-75% QoQ and LPDDR5X +78-83% QoQ in 2Q26. (TrendForce)
Sandisk fiscal Q2 2026 revenue was $3.025bn, GAAP net income was $803m and non-GAAP EPS was $6.20. (Sandisk)
Western Digital fiscal Q3 2026 revenue was $3.337bn, GAAP gross margin was 50.2% and operating cash flow was $1.12bn. (Western Digital)
FADU-Sandisk flow/correlation work is based on Research OS local DB and yfinance data through 2026-05-29.

[Inference]

FADU is not a direct Sandisk tracker yet; it is a foreign-discovered Korean high-beta proxy for AI storage/NAND.
Jeju is a tactical low-power memory squeeze candidate, not the cleanest structural AI-storage bottleneck.

[Blocked]

FADU customer-level orders, controller ASP, gross margin and Gen6/CXL design wins.
U.S.-name investor-flow data comparable to Korean foreign/institution flow.
Jeju 2Q26 repeatability and working-capital quality.

SK Hynix vs Micron: HBM Technology Premium or U.S. Listing Premium?

Sun, 31 May 2026 12:15:00 +0900

Context This is a follow-up to Samsung vs SK Hynix Forward P/E Inversion. That note asked whether SK Hynix can deserve a higher P/E than Samsung. This one asks whether Micron can deserve a higher multiple than SK Hynix. Related hub: AI HBM Hub.

TL;DR

Micron’s premium is not just a technology premium. It is a mix of U.S.-listed AI-memory scarcity premium, a re-rating from long-term supply contracts reducing memory cyclicality, and an AI memory/storage platform story built around HBM4, SOCAMM2 and PCIe Gen6 SSDs.

On technology, references and current HBM dominance, SK Hynix still leads. Micron, however, is the easiest U.S.-listed large-cap AI memory stock to buy, and it can package NVIDIA Vera Rubin HBM4, SOCAMM2 and Gen6 SSD into one clean narrative.

Conclusion: a modest Micron premium can persist, but a large premium is harder to justify. Relative value favors SK Hynix over Micron, while the broader three-stock setup still leaves the most asymmetric catch-up case at Samsung Electronics.

1. Key Data: Same $1 Trillion Zone, Different Premiums

The input snapshot uses 2026-05-29 closes and USD/KRW 1,507.42.

Item	Samsung Electronics	SK Hynix	Micron
Ticker	005930.KS	000660.KS	MU
2026-05-29 close	KRW 317,000	KRW 2,333,000	$971
Market cap	KRW 2,015.39T	KRW 1,662.73T	$1,108.9B
USD market cap	about $1.337T	about $1.103T	about $1.109T
TTM P/E	25.4x	22.1x	45.8x
Forward P/E	7.83x	8.24x	9.87x
Market story	HBM laggard catch-up	HBM bottleneck leader	U.S. AI-memory scarcity

Forward P/E spread:

SK Hynix vs Samsung = 8.24 / 7.83 - 1 = +5.2%
Micron vs SK Hynix = 9.87 / 8.24 - 1 = +19.8%
Micron vs Samsung = 9.87 / 7.83 - 1 = +26.1%

Market multiple ranking is Micron > SK Hynix > Samsung. That is not the same as the technology ranking. The technology and customer-reference ranking is closer to SK Hynix leading, Samsung catching up, Micron pursuing.

2. SK Hynix: The Cleanest HBM Bottleneck Leader

SK Hynix’s strength is straightforward. In 1Q26 it reported KRW 52.6T of revenue, KRW 37.6T of operating profit and a 72% operating margin. Management cited HBM, high-capacity server DRAM and eSSD as drivers. (SK hynix Newsroom)

The HBM share backdrop also favors Hynix. Reuters, citing Counterpoint Research, reported 4Q25 HBM share at SK Hynix 57%, Samsung 22% and Micron 21%. (Reuters)

Variable	Interpretation
P	Premium mix from HBM, DDR5, server DRAM and eSSD
Q	Qualified supplier status at NVIDIA and AI accelerator customers
C	Yield, quality and production stability raise switching cost

It is not irrational that SK Hynix has started to earn a higher forward P/E than Samsung. HBM has become a qualified bottleneck part that can gate GPU shipments. But the market has already begun to price this. Hynix is a strong company, yet fresh buying is cleaner after pullbacks or further EPS revisions.

3. Micron: U.S. AI-Memory Proxy, Not Technology No. 1

Micron’s premium has three pillars:

U.S.-listed large-cap memory scarcity premium.
HBM4 reference for NVIDIA Vera Rubin.
AI memory/storage platform narrative: HBM + SOCAMM2 + PCIe Gen6 SSD.

Micron reported FY2Q26 revenue of $23.86B, GAAP gross margin of 74.4% and diluted EPS of $12.07. FY3Q26 guidance is revenue of $33.5B ± $0.75B, gross margin around 81% and non-GAAP EPS of $19.15 ± $0.40. (Micron Technology)

Micron also announced that HBM4 36GB 12H for NVIDIA Vera Rubin, PCIe Gen6 SSDs and SOCAMM2 were in high-volume production. (Micron Technology)

Run-rate sanity check:

Micron FY3Q26 EPS guide annualized = $19.15 × 4 = $76.60
Micron run-rate P/E = $971 / $76.60 = 12.7x

SK Hynix 1Q26 net income annualized = KRW 40.3459T × 4 = KRW 161.3836T
SK Hynix run-rate P/E = KRW 1,662.73T / KRW 161.3836T = 10.3x

Micron run-rate premium = 12.7 / 10.3 - 1 = 23.0%

That premium is explainable. U.S. investors can buy Micron directly through stocks, options, ETFs and quant factors. It also carries the strategic-industry premium of being the only large U.S. memory manufacturer. But this should not be confused with Micron being ahead of SK Hynix in HBM technology.

4. Why the Micron Premium Exists

4.1 U.S. listing scarcity

There are very few clean U.S.-listed large-cap AI memory proxies. Samsung and SK Hynix are Korean listings with access, FX and market-structure discounts. Micron is U.S.-listed, optionable, ETF-owned and politically easy to frame as a strategic AI infrastructure supplier.

Reuters’ report that SK Hynix is exploring a U.S. listing is itself evidence that the discount exists. Better U.S. access could narrow the Hynix-Micron spread. (Reuters)

4.2 LTAs reduce the memory-cycle discount

The logic behind UBS’s sharp Micron target raise was not merely technology. It was the argument that long-term supply agreements reduce historical earnings volatility in memory. Reuters reported that hyperscalers are choosing supply security over price flexibility, with some volume and pricing commitments improving visibility. (Reuters)

This also applies to SK Hynix. In its 1Q26 results, SK Hynix said HBM, high-capacity server DRAM and eSSD sales drove the profit expansion. (SK hynix Newsroom) The difference is that U.S. markets can price this narrative faster through Micron.

4.3 HBM plus memory/storage stack

Micron is not selling only HBM4. It is selling a Vera Rubin memory stack: HBM4, SOCAMM2 and PCIe Gen6 SSD. As inference grows, KV cache, checkpointing and data movement make memory and storage layers more valuable.

This links to the NVIDIA Vera Rubin inference-stack note. AI inference does not stop at HBM. It extends into SRAM decode tiers, KV-cache storage, high-speed SSDs, networking and orchestration. Micron tells that story well to U.S. investors.

5. Can the Premium Persist?

My base case: a 10-25% Micron forward P/E premium versus SK Hynix can persist. U.S. listing, policy premium, HBM4 earnings-revision beta and the SOCAMM2/eSSD narrative all support it.

But a 30-40%+ premium becomes harder to defend. SK Hynix still has the stronger HBM reference set and current profitability. If Hynix improves U.S. investor access, the capital-market discount can narrow.

Micron Premium vs SK Hynix	Reading	Strategy
0-10%	Low versus U.S. listing premium	Both acceptable
10-25%	Explainable normal range	Hynix relatively more attractive
25-40%	Micron overheating risk rises	Consider Hynix long / Micron neutral
40%+	Too far from technology ranking	Pair-trade candidate

At about 19.8%, the current premium sits near the upper half of the explainable range.

6. Why Samsung Still Matters

This is a Hynix-versus-Micron note, but Samsung is the valuation anchor.

The best HBM company today is SK Hynix. The best U.S. AI-memory proxy is Micron. But the stock with the largest remaining discount is Samsung Electronics.

Samsung reported 1Q26 consolidated revenue of KRW 133.9T and operating profit of KRW 57.2T. DS revenue was KRW 81.7T and DS operating profit was KRW 53.7T. (Samsung Global Newsroom)

Reuters also reported that Samsung shipped 12-high HBM4E samples to global customers. That is not yet qualification or production economics, but it is a trigger for HBM laggard discount compression. (Reuters)

Stock	View	One-line Thesis
Samsung Electronics	Conditional Buy	HBM4E qualification can compress the laggard discount and drive EPS revisions
SK Hynix	Hold-Wait	Best HBM franchise, but already more priced
Micron	Wait	U.S. premium is explainable, but fresh chase is less attractive

Final View

Micron’s premium is a capital-market and narrative premium more than a pure technology premium. U.S. listing scarcity, HBM4 Vera Rubin, SOCAMM2 and Gen6 SSDs explain why investors pay more.

But relative value favors SK Hynix over Micron. In the same $1T market-cap zone, Hynix has stronger HBM references and current profitability, yet a lower forward P/E.

Across all three, the most asymmetric risk/reward still sits at Samsung Electronics. Hynix is already the winner. Micron already has the U.S. premium. Samsung still has HBM4E execution risk, but if that risk clears, the valuation gap can close fastest.

Best HBM company: SK Hynix
Best U.S. AI memory proxy: Micron
Best risk/reward: Samsung Electronics

Sources

Disclaimer: For research and information purposes only. Not investment advice.

Samsung vs SK Hynix Forward P/E Inversion: Is the HBM Premium a New Normal?

Sun, 31 May 2026 11:30:00 +0900

Context This is a bridge post linking the Samsung stock-bonus / memory-supercycle note, the AI infrastructure multiple map, and the NVIDIA Vera Rubin LPX / CMX inference-stack note. The related hub is the AI HBM Hub. The next linked note is SK Hynix vs Micron: HBM Technology Premium or U.S. Listing Premium?.

TL;DR

SK Hynix trading at a higher forward P/E than Samsung Electronics is unusual by historical standards. Memory pure-plays usually carried a discount to Samsung because of cyclicality, customer concentration and weaker business diversification.

But the current move is not simply a valuation error. It reflects a collision between SK Hynix’s HBM bottleneck premium and Samsung’s HBM laggard discount. SK Hynix is being reclassified as an AI-memory bottleneck supplier, while Samsung’s HBM4E catch-up and broad DRAM/NAND price leverage may still be underpriced.

The cleaner alpha is not chasing SK Hynix. It is the Samsung Electronics catch-up trade, conditional on HBM4E customer qualification, 2027 EPS revisions and DS margin durability.

1. Data Frame: Annual P/E Inverted, 12MF P/E Near Parity

A precise monthly three-year 12-month-forward P/E history requires Bloomberg, FactSet, LSEG or FnGuide terminal data. Public sources are enough to frame the current signal, but not to rebuild the full time series.

Basis	Current Reading
FnGuide 2026E P/E	Samsung 6.77x, SK Hynix 6.79x, a slight Hynix inversion
2026-05-29 close and 2026E EPS proxy	Samsung around 7.55x, SK Hynix around 8.02x, a roughly 6.2% Hynix premium
Mirae Asset 2026-05-22 12MF P/E	Samsung 6.2x, SK Hynix 5.9x, meaning 12MF parity is approaching but not cleanly inverted

The conservative sentence is:

Annual 2026E P/E has inverted; 12-month-forward P/E is approaching parity.

That is different from saying Hynix will permanently deserve a higher multiple than Samsung. It says the market is beginning to treat SK Hynix as a scarce AI HBM bottleneck supplier.

2. Why It Is Unusual

Historically, Samsung deserved the higher multiple for four reasons.

It had memory plus mobile, displays, consumer electronics, foundry and system LSI.
Its balance sheet and cash-flow stability were stronger.
SK Hynix was closer to a pure memory-cycle stock.
Memory stocks were usually discounted for oversupply and ASP downside risk.

So “SK Hynix P/E > Samsung P/E” looks strange in the old framework. HBM changes the framework.

HBM is not commodity DRAM. It is a customer-qualified, package-integrated memory product that can gate GPU and AI ASIC shipments. Once qualified, it is hard for a customer to switch suppliers quickly. That gives SK Hynix a credible argument for a higher-quality earnings stream than a traditional DRAM cycle.

3. Why the SK Hynix Premium Makes Sense

SK Hynix’s 1Q26 numbers look nothing like a normal commodity memory company. It reported KRW 52.6T of revenue, KRW 37.6T of operating profit and a 72% operating margin. (SK hynix Newsroom)

That margin reflects more than DRAM pricing. It reflects a mix shift toward HBM and high-value server memory, plus the fact that qualified HBM supply can gate AI accelerator output.

The Hynix premium is justified if these conditions hold:

Condition	Meaning
HBM share stays high	AI accelerator memory bottleneck remains intact
HBM4 / HBM4E leadership continues	Next-generation customer lock-in holds
LTAs expand	Peak-earnings doubt fades
High ROE persists	P/E and P/B premiums are easier to defend
Samsung catch-up is delayed	Relative premium remains intact

The market is effectively asking whether HBM has moved from commodity memory into a contracted high-value memory franchise.

4. Why the Alpha Is Still Samsung Catch-Up

The problem is price. SK Hynix has already started to receive the HBM-leader premium. Samsung still carries an HBM laggard discount.

Samsung’s catch-up trade has three pillars.

4.1 HBM4E 12-high samples

Samsung reportedly shipped HBM4E 12-high samples to major global customers in May 2026. This is not yet revenue certainty. Customer qualification and production yield still matter. But the valuation path is clear. (Reuters)

HBM4E sample
→ customer qualification
→ mass production
→ EPS revision
→ HBM laggard discount compression

The market may not be fully pricing that path yet.

4.2 DRAM and NAND pricing leverage

This memory cycle is broader than HBM. TrendForce reported that Samsung’s memory ASP was up 146% versus its 2025 average in 1Q26, while SK Hynix saw DRAM ASP up in the mid-60% QoQ range and NAND ASP up in the mid-70% QoQ range. (TrendForce)

This matters for Samsung because its memory portfolio is broader: DRAM, NAND, HBM, eSSD and SOCAMM. A broad memory squeeze can drive wider EPS revisions at Samsung than a pure HBM thesis alone suggests.

4.3 DS profit scale and foundry optionality

Samsung’s DS division generated KRW 81.7T of revenue and KRW 53.7T of operating profit in 1Q26. The absolute profit pool is already enormous. (Samsung Global Newsroom)

If HBM4E uses logic base dies and advanced packaging, Samsung has more than a memory-stack story. It has a memory plus base-die plus packaging optionality that is not as cleanly priced as SK Hynix’s pure HBM leadership.

5. Relative-Value Strategy: Long Samsung / Neutral Hynix

This is not an anti-Hynix call. It is a statement that Samsung catch-up now offers the better relative-value setup.

SK Hynix P/E Premium vs Samsung	Interpretation	Strategy
0-5%	Acceptable for HBM leadership	Own both
5-10%	Samsung catch-up becomes more attractive	Add Samsung weight
10-15%	Hynix premium may be stretched	Samsung long / Hynix neutral
15%+ with slowing EPS revisions	Mean reversion risk rises	Prefer Samsung

Using the 2026-05-29 close and the 2026E EPS proxy from the user-provided data, the SK Hynix premium is around 6.2%. That already sits in the Samsung catch-up zone.

6. Samsung Upside Bridge

Using a 2026E Samsung EPS proxy of KRW 41,972 and a 2026-05-29 close of KRW 317,000, the simple P/E is about 7.55x.

EPS Revision	Revised EPS	P/E at KRW 317,000
0%	KRW 41,972	7.55x
+10%	KRW 46,169	6.87x
+15%	KRW 48,268	6.57x
+20%	KRW 50,366	6.29x

If Samsung EPS rises 15% and the market assigns an 8.0x P/E closer to SK Hynix:

KRW 48,268 × 8.0 = KRW 386,144

That is about 21.8% above KRW 317,000. This is not a formal target price. It is a sensitivity showing how powerful EPS revision plus P/E parity can be.

7. What Breaks the Thesis

Risk	Meaning
Samsung HBM4E qualification delay	HBM laggard discount stays
Weak HBM4E yield	EPS revisions do not arrive
DRAM/NAND pricing cools	Samsung portfolio-breadth thesis weakens
Foundry / LSI losses widen	Non-memory drag offsets memory upside
SK Hynix 2027E EPS keeps rising	HBM leader premium remains justified
AI capex slows	Peak-earnings doubt returns

Samsung HBM4E sample shipment is a fact. Customer qualification and mass-production economics remain blocked items.

Final View

SK Hynix trading above Samsung on forward P/E is unusual, but not irrational. HBM has moved closer to an AI infrastructure bottleneck than a commodity DRAM cycle, and SK Hynix deserves credit for leadership.

But the better alpha now is Samsung catch-up:

SK Hynix is already receiving more HBM-leader premium.
Samsung’s HBM4E, DRAM/NAND ASP leverage and DS profit scale may still be under-reflected.
As the P/E spread widens above 5-10%, Samsung becomes increasingly attractive on relative value.

The resulting stance is Samsung Electronics: conditional Buy / SK Hynix: Hold-Wait. The next evidence points are HBM4E qualification, 2027 EPS revisions, DS margins and SK Hynix HBM4 / HBM4E share defense.

Evidence Map

[Fact]

FnGuide 2026E P/E was reported at Samsung 6.77x and SK Hynix 6.79x. (Seoul Economic Daily)
Samsung reportedly shipped 12-high HBM4E samples to major global customers. (Reuters)
1Q26 memory pricing surged across Samsung memory and SK Hynix DRAM/NAND. (TrendForce)
Samsung DS reported KRW 81.7T revenue and KRW 53.7T operating profit in 1Q26. (Samsung Global Newsroom)
SK Hynix reported KRW 52.6T revenue, KRW 37.6T operating profit and 72% operating margin in 1Q26. (SK hynix Newsroom)

[Inference]

SK Hynix is being reclassified from commodity memory pure-play to AI HBM bottleneck supplier.
Samsung can recover P/E parity if its HBM laggard discount compresses.
The better relative-value alpha is Samsung catch-up rather than Hynix chase.

[Blocked]

Exact current monthly 12MF P/E time series.
Samsung HBM4E customer qualification, yield, volume and ASP.
2027 HBM LTA downside-price protection.

Sources

Disclaimer: For research and information purposes only. Not investment advice.

Goldman's Token Demand vs. J.P. Morgan's Memory ASP Peak-Out: Do the Two Calls Really Clash?

Sun, 31 May 2026 11:00:00 +0900

This piece follows on from AI Token Futures and Cost per Token, Samsung Electronics 2026 Deep Dive, SK Hynix: The HBM Heavyweight, and Korea AI Infrastructure After Nvidia Q1 FY27. Where those earlier posts looked separately at cost per token, individual companies, and the spread of AI infrastructure, this one merges two different calls — Goldman on demand and J.P. Morgan on memory pricing — into a single framework and lays out the investment timeline from 2026 to 2030.

TL;DR

Goldman Sachs sees an explosion in token usage (24x by 2030) plus a steep drop in cost per token (60-70% per year). J.P. Morgan sees the year-over-year (YoY) growth rate of DRAM and NAND ASP slowing from 2027. Because they track different variables, the two calls do not clash head-on.
Combine them and a single path appears: 2026 is memory ASP beta, 2027 is a peak-out in price momentum, and 2028-2030 is a rotation of alpha into the components and systems that lower cost per token.
So the investment conclusion is not “buy memory without limit.” In 2026, memory beta such as HBM, server DRAM and eSSD is favorable, but after that you have to be selective about where the bottlenecks are — ASIC, AI networking, optical, HBM leaders, eSSD, advanced packaging, and MLCC/FC-BGA.
The two most common misreadings are: (1) seeing J.P. Morgan’s peak-out and concluding “the AI infrastructure cycle is over,” and (2) seeing Goldman’s demand and concluding “memory prices keep surging through 2030.” Both go too far.

1. Two Giants, Two Seemingly Opposite Calls

Looking at the same AI era, two global investment banks have painted pictures that seem to point in opposite directions.

Goldman Sachs (official article, May 5, 2026) looks at the demand side. The core is two points.

Token usage grows 24x by 2030. The call is that monthly usage reaches 120 quadrillion tokens in 2030; working backward, that puts the 2026 base at roughly 5 quadrillion. That is about 121% average annual growth over four years.
At the same time, inference cost per token falls 60-70% per year. The drivers are improving chip efficiency and better data-center architecture.

Source: a reconstructed chart that simply indexes the figures in Goldman Sachs’s official article (2026-05-05). It is not the original chart but the original numbers — “24x tokens by 2030, cost per token down 60-70% a year” — plotted on a log scale.

In the chart above, the blue line (usage) climbs steeply, and the orange and green lines (cost per token) fall even more steeply. At a 60% annual decline, cost four years out is about 2.6% of the 2026 level (roughly a 39x improvement); at 70%, about 0.8% (roughly a 123x improvement).

By contrast, the J.P. Morgan material looks at price. The picture is that DRAM and NAND ASP rise strongly in 2026 but that, from late 2026 into early 2027, the year-over-year growth rate decelerates quickly.

Source: a session-uploaded image (DRAM from WSTS / J.P. Morgan estimates, NAND from Gartner / J.P. Morgan estimates). Detailed figures such as FY26 DRAM +53% and NAND +30%, FY27 DRAM +1% and NAND -6% are based on a secondary summary; the original table is unverified.

On the surface, “demand explodes” (Goldman) and “memory price gains roll over” (J.P. Morgan) seem to clash. But look closely and the two calls are talking about different things.

2. Why It’s Not a Clash — Separating P, Q and C

A memory company’s profit breaks down into a simple multiplication.

Revenue = shipments (Q) × average selling price (P), and the token economy = usage (Q) × cost per token (C).

Through this lens, it becomes clear that the two calls are watching different variables.

Dimension	Goldman Sachs	J.P. Morgan
Variable watched	Token usage (Q), cost per token (C)	YoY change in DRAM/NAND ASP (P momentum)
Core message	AI usage explodes long term, cost per token plunges	Memory price growth rate slows from 2027
Time horizon	2026-2030	Late 2026 to early 2027

There is a trap that must be flagged here. The vertical axis of J.P. Morgan’s chart is not the absolute level of price but the “year-over-year growth rate (YoY %)." The two are entirely different.

Take an example.

Point in time	ASP index	YoY
4Q25	100	–
4Q26	300	+200%
4Q27	315	+5%

From 4Q26 to 4Q27 the price index still rises, from 300 to 315. Yet the year-over-year growth rate plunges from +200% to +5%. In other words, J.P. Morgan’s 2027 “peak-out” may not mean prices collapse but that the pace of increase slows. The stock market usually prices in this direction of the growth rate before it prices in “record-high profit” itself. That is why share prices can cool first, right when profit is at its peak.

To sum up, Goldman watches Q (usage) and C (cost), while J.P. Morgan watches the pace of P (price). Because these are different variables, both can be right at the same time.

3. Total Inference Cost May Actually Fall

Here a counterintuitive conclusion emerges. Even if usage grows 24x, if cost per token falls 60-70% per year, then the total inference-cost burden could actually be lower than in 2026.

It is simple arithmetic. Assuming the same token mix and the same model complexity, total cost in 2030 is the usage index × the cost-per-token index.

At a 60% annual decline: 24 × 2.6% ≈ 61% of 2026
At a 70% annual decline: 24 × 0.8% ≈ 19% of 2026

This is a heavily simplified sensitivity calculation. It does not reflect token mix, model size, context length, the rise in reasoning tokens, multimodality, redundant processing, latency constraints, or memory-bandwidth bottlenecks. In practice these factors could push cost back up.

This chart is the heart of Goldman’s logic. Whoever lowers cost per token ends up making the money. More than the rise in usage itself, it is the cost reduction that makes that usage affordable which keeps the industry’s cash flow alive. That said, as the caveat notes, if context length or reasoning tokens rise quickly, the real cost curve may fall less than this chart suggests.

4. So the Timeline Shifts

Combine the two calls and the following path is logically consistent.

Period	What happens	Where it’s favorable
2026	Agent usage surges → HBM, server DRAM, eSSD and NAND allocation tightens → DRAM and NAND ASP spike	Memory ASP beta works strongly
2027	A higher base plus some supply additions and long-term agreements (LTAs) slow the YoY ASP growth rate. B2B AI memory stays firm; B2C consumer memory hits price resistance	Share prices price in “slowing growth rate” before “peak profit.” Segment divergence begins
2028-2030	Usage keeps rising, but the drop in cost per token absorbs it	Alpha rotates from broad memory beta to the token-cost-reduction stack

The core message is one. From 2027, “how much further DRAM and NAND prices rise” matters less than “which components and systems lower cost per token."

5. Example Investment Ideas (Observation Points, Not Recommendations)

Below are not stock recommendations but examples of “where to look first” along the timeline above. AI and memory stocks have already re-rated quickly, so this is not a call to buy right now but a map for when the conditions are met.

Example 1 — 2026 Memory ASP Beta

The phase where HBM, server DRAM and eSSD allocation tightens and DRAM and NAND prices spike. Memory large-caps such as SK Hynix, Samsung Electronics and Micron benefit directly. The counterargument is that once the 2027 slowdown in the growth rate begins, the share price can take multiple compression first, even if profit is high.

Example 2 — Token-Cost-Reduction Stack

The real heart of Goldman’s logic is not the rise in usage but the fall in cost per token. Customers will pay more for chips and systems that lower their cost per token. Custom ASICs, AI networking and optical, and HBM leaders fit this best. (The interconnect, substrate and power bottlenecks seen in the Marvell read-through are this same thread.)

Example 3 — eSSD / Enterprise NAND

Agent inference demands far more retrieval (RAG), logs, context, KV cache and checkpoint storage than plain training. If it is true that AI servers use roughly 3x the SSD capacity of ordinary servers, then NAND can be reclassified not as a plain commodity-cycle asset but as an AI storage bottleneck. The counterargument is the possibility that price resistance in consumer SSDs dilutes enterprise strength.

Example 4 — Advanced Packaging / MLCC / FC-BGA

If Goldman’s 2030 token call is right, the complexity of server and rack architecture keeps rising. It is not only GPUs, ASICs and HBM that grow; demand for substrate area, power stabilization, decoupling capacitors and high-speed signal quality grows alongside. High-spec MLCC and FC-BGA suppliers such as Samsung Electro-Mechanics fall here.

Different Peak-Outs by Segment

What matters is not “memory as a whole” but the differences by segment. J.P. Morgan’s peak-out logic does not apply equally to all memory.

Segment	Likelihood peak-out applies	Tension with long-term demand
HBM	Low-Medium	High (demand keeps reinforcing the bandwidth wall)
Server DRAM	Medium	Medium
eSSD / Enterprise NAND	Medium	High (structural demand possible)
Mobile DRAM	High	Low (fast consumer price resistance)
Commodity DRAM/NAND	High	Low (peak-out logic applies best)

Common condition: the examples above need more than “oversold” or “because it’s AI.” A candidate must be somewhere where actual orders, contract prices and earnings estimates are newly rising, and where bottlenecks and entry barriers are confirmed.

6. The Two Most Common Misreadings

In this phase the market is prone to two mistakes pointing in opposite directions.

Misreading 1 — “J.P. Morgan says peak-out, so the AI infrastructure cycle is over." No. A peak-out is a slowdown in the growth rate of price, not a collapse in the level of price. And it applies best to commodity memory and less to bottlenecks such as HBM, eSSD, ASIC and packaging.

Misreading 2 — “Goldman says 24x tokens, so DRAM and NAND prices keep surging through 2030 too." This too goes too far. Goldman says explosive usage and a steep drop in cost per token at the same time. If cost falls quickly, the YoY growth rate of memory prices can slow even as usage rises.

The right reading is in between. A memory super-cycle in 2026, a peak-out in price momentum in 2027, and a long-term expansion of the token-cost-reduction stack in 2028-2030.

7. Fund Manager’s Comment

The two calls are not enemies but two axes of one picture. Goldman speaks to “how much, and how cheaply, we come to use AI”; J.P. Morgan speaks to “how long memory prices keep rising fast in that process.” The investor’s job is not to pick one of the two but to not miss the inflection points on the timeline.

The two most dangerous choices are these.

Throwing out all of AI infrastructure on seeing the single word peak-out — the mistake of giving away even the winners with bottlenecks at a bargain price.
Chasing commodity memory to the bitter end on token demand alone — the mistake of taking multiple compression head-on in the phase where the price growth rate rolls over.

The signals to check now, put simply, are these.

What you watch	Goldman (demand) strengthens	J.P. Morgan (peak-out) strengthens
Token usage	High growth sustained	Growth rate slows
Cost per token	Continues to fall 60%+ a year	Decline rate slows, latency cost rises
HBM long-term agreements	Prices held or raised	Renegotiation, volume deferral
Server DRAM contract price	Further increase	Slowing growth rate, spot discounts
Enterprise SSD	CSP long-term agreements expand	Consumer SSD price resistance spreads in

In short, recognizing memory beta in 2026 while preparing, from 2027, to shift weight toward the bottlenecks that lower cost per token is the most reasonable stance available now. It is neither chasing memory to the end nor throwing it all out in fear of the peak-out.

This piece is an analysis reconstructed from Goldman Sachs’s official article (2026-05-05), J.P. Morgan-related charts and secondary summaries, and TrendForce’s official outlook. The full text of J.P. Morgan’s original report, its quarterly ASP tables, and its segment-level detail could not be verified, and company names are examples illustrating the analytical flow, not investment recommendations. Actual investment decisions and their responsibility rest with the investor.

PharmaResearch Re-Rating Check: Can Rejuran Move From Medical Device Monopoly to U.S. Derma-Cosmetics Platform?

Sun, 31 May 2026 10:00:00 +0900

Context This is a follow-up to Who Owns Rejuran? PharmaResearch and Korea’s Skin-Booster Platform and PharmaResearch: The PDRN Pioneer Behind Rejuran. Those notes explained the PN/PDRN platform and Rejuran brand. This note asks why the market cut the multiple after 1Q26, and whether Rejuran Cosmetics’ U.S. sell-through can reopen the re-rating path. The broader map sits in the Olive Young, PharmaResearch and K-Beauty Hub.

Key Visuals

TL;DR

PharmaResearch’s multiple compression is not just a flow correction. It reflects four issues at once: the 4Q25 earnings miss, a pause in medical-device export growth, tougher domestic skin-booster competition and skepticism about the durability of cosmetics revenue.

But cosmetics is no longer a side business. In 1Q26, PharmaResearch reported cosmetics revenue of KRW 42.2B, roughly 29% of consolidated revenue, with cosmetics exports of KRW 26.9B. Rejuran Cosmetics is already the company’s second growth engine. (PharmaResearch 1Q26 release)

The newest signal is Olive Young U.S. As checked on May 31, 2026, Olive Young U.S. says its best-seller list is updated daily based on sales data. REJURAN Dual Effect Ampoule ranked #2, while REJURAN Turnover Ampoule ranked #7. This is not yet proof of large revenue. It is an early signal that U.S. consumers are buying the Rejuran skincare SKUs. (Olive Young US Best Sellers)

My conclusion is Buy now, but only a first tranche. At KRW 295,000, the stock trades around 14.9x 2026F EPS using Kiwoom’s KRW 19,760 estimate. That already prices in much of the medical-device slowdown concern. A larger position should wait for 2Q26 cosmetics exports above KRW 30B, U.S. channel re-orders and consolidated OPM above 38%.

1. Why the Multiple Fell

The key is reclassification.

PharmaResearch used to be priced as a Rejuran medical-device monopoly growth stock. It created the Korean skin-booster category, had strong PN/PDRN brand recognition and a clinic network. That kind of company can receive 25x or 30x multiples when growth visibility stays clean.

Since late 2025, the market has seen a more complicated story.

Compression Driver	Why It Matters
4Q25 earnings miss	When a high-multiple growth stock needs complicated explanations, the discount rate rises.
Medical-device export pause	1Q26 medical-device exports were KRW 21.1B, flat YoY and down QoQ. The core segment paused.
Domestic skin-booster competition	Rejuran still leads, but the category-monopoly narrative weakened.
Cosmetics quality discount	Cosmetics revenue carries channel fees, promotion, inventory and hit-SKU durability risk.

Kiwoom described 1Q26 as a quarter where cosmetics more than offset the medical-device gap, and its report showed 2026F EPS of KRW 19,760 and 2027F EPS of KRW 24,722. (Kiwoom report)

The multiple fall is therefore not random. It is the market moving PharmaResearch from “medical-device monopoly” to “hybrid beauty-healthcare platform where cosmetics must prove durability.”

2. Cosmetics Is No Longer a Side Business

1Q26 consolidated revenue was KRW 146.1B, operating profit was KRW 57.3B, and OPM was 39.2%. The company called it a record quarterly result. (PharmaResearch 1Q26 release)

Item	1Q26 Revenue	Mix
Consolidated revenue	KRW 146.1B	100.0%
Medical devices	KRW 79.5B	54.4%
Cosmetics	KRW 42.2B	28.9%
Cosmetics exports	KRW 26.9B	18.4% of total, 63.7% of cosmetics
Domestic cosmetics	KRW 15.3B	10.5% of total, 36.3% of cosmetics

The math is simple.

Cosmetics mix = 42.2 / 146.1 = 28.9%
Cosmetics export mix = 26.9 / 42.2 = 63.7%
Cosmetics exports / total revenue = 26.9 / 146.1 = 18.4%

That is large enough to change the investment frame. PharmaResearch is not just a medical-device company selling some cosmetics. It is trying to convert Rejuran’s procedure brand equity into derma-cosmetic SKUs.

If that works, the multiple logic changes. Medical-device revenue is tied to clinic channels, procedure demand and regulatory expansion. Cosmetics revenue can scale through Olive Young, Sephora, Amazon and TikTok Shop. The margin and repeat-purchase quality still need proof.

3. Why Olive Young U.S. Matters

Olive Young U.S. says its best-seller list is “updated daily based on sales data.” As checked on May 31, 2026, medicube Age-R Booster Pro ranked #1, REJURAN Dual Effect Ampoule ranked #2 and REJURAN Turnover Ampoule ranked #7. (Olive Young US Best Sellers)

Rank	Brand / Product	Price
#1	medicube Age-R Booster Pro	$198.90
#2	REJURAN Dual Effect Ampoule 30ml	$44.00
#7	REJURAN Turnover Ampoule 30ml	$41.60

The important part is not only the #2 rank. It is that two REJURAN SKUs were simultaneously in the Top 10. That is stronger than a single promotion-driven spike.

But the interpretation must stay conservative. Olive Young U.S. will not immediately change consolidated earnings by itself. The key question is whether it validates SKU velocity in a U.S. consumer channel.

Olive Young U.S. Top 10
→ Sephora U.S. online + about 380 stores
→ repeat purchases / reviews / re-orders
→ 2Q26 cosmetics exports above KRW 30B
→ estimate revisions
→ multiple re-rating

PharmaResearch announced in March 2026 that Rejuran Cosmetics had entered Sephora U.S. online and about 380 physical stores. Its April English release said the LA pop-up drew more than 5,000 visitors and that Turnover Ampoule and Dual Effect Ampoule saw consistent sell-through during the event. (Sephora launch, LA pop-up)

4. Small Direct EPS Impact, Larger Multiple Impact

The direct earnings impact of an Olive Young U.S. ranking is still small. The re-rating impact can be larger if U.S. cosmetics sales accumulate into estimate revisions.

Assumptions:

EPS uplift = incremental revenue × operating contribution × after-tax conversion / shares

Operating contribution = 35%
After-tax conversion = 80%
Share count = 10.39M

Incremental U.S. cosmetics revenue	EPS uplift	% of 2026F EPS
KRW 10B	+KRW 269	+1.4%
KRW 30B	+KRW 808	+4.1%
KRW 50B	+KRW 1,347	+6.8%
KRW 100B	+KRW 2,695	+13.6%

KRW 10B alone does not move the stock. KRW 30B-50B begins to matter. At that point the story is no longer “cosmetics filled the medical-device gap.” It becomes “Rejuran brand equity can travel into consumer skincare.”

Margins are the test. If consolidated OPM stays near 38-40% while cosmetics grows, the market will stop treating cosmetics as lower-quality revenue. If growth is mostly promotion and channel spend, the re-rating stops.

5. Re-Rating Scenarios

The reference price in the attached report is KRW 295,000 as of May 29, 2026. Using Kiwoom’s 2026F EPS of KRW 19,760, 2026F PER is roughly 14.9x.

2026F PER = 295,000 / 19,760 = 14.93x
2027F PER = 295,000 / 24,722 = 11.93x

Using 2026F EPS of KRW 19,760:

Target PER	Implied Price	Upside vs KRW 295,000
15.0x	KRW 296,000	+0.5%
17.5x	KRW 346,000	+17.2%
20.0x	KRW 395,000	+33.9%
22.5x	KRW 445,000	+50.8%
25.0x	KRW 494,000	+67.5%

The 20x scenario is KRW 19,760 × 20 = KRW 395,200, rounded to KRW 395,000.

Zone	Required Evidence	Interpretation
15x	Current market level	Medical-device slowdown and cosmetics discount reflected
17.5x	Export recovery plus continued cosmetics growth	Partial re-rating
20x	U.S. SKU velocity converts into 2Q26 numbers	Realistic bull case
22.5x	Repeat sales across Sephora / Olive Young / Amazon / TikTok Shop	Global derma-cosmetics platform case
25x	APR-style K-beauty premium partially applied	Still early

6. Strategy

View

Buy now, but only a first tranche.

The current price already reflects a lot of disappointment. Around 15x 2026F EPS looks low relative to PharmaResearch’s profitability, net-cash profile and cosmetics growth. But U.S. sell-through is still an early signal, so a full position is premature.

Add Conditions

Consider adding if at least two of the following happen:

Two REJURAN SKUs stay in Olive Young U.S. Top 20 for more than two weeks.
2Q26 cosmetics exports exceed KRW 30B.
Sephora U.S. reviews, sell-outs and re-stocking show repeat demand.
2Q26 medical-device exports recover QoQ from KRW 21.1B in 1Q26.
Consolidated OPM stays above 38%.

Invalidation

Cut the re-rating thesis if at least two of the following happen:

Core REJURAN SKUs fall out of Olive Young U.S. Top 20 and Sephora traction slows.
2Q26 cosmetics exports fail to exceed 1Q26’s KRW 26.9B.
Medical-device exports fail to recover again in 2Q26.
Cosmetics growth pushes OPM below 35%.
2026F EPS is revised below KRW 17,000.

7. Conclusion

PharmaResearch’s multiple fell for a reason. The market is no longer willing to pay 30x purely for the Rejuran medical-device monopoly story. The 4Q25 miss and 1Q26 medical-device export pause were real.

But the price now reflects much of that disappointment. Cosmetics is already 29% of revenue. If Rejuran Cosmetics proves repeat sell-through at Sephora and Olive Young U.S., PharmaResearch can be reclassified from “slowing medical-device name” into a K-beauty healthcare platform that extends Rejuran across procedures and consumer skincare.

This is not a full-position moment based on story alone. It is a first-tranche moment after disappointment has been priced and the first U.S. sell-through signal has appeared. The next proof points are 2Q26 cosmetics exports, Sephora / Olive Young re-orders and OPM durability.

Evidence Map

[Fact]

PharmaResearch reported 1Q26 revenue of KRW 146.1B, operating profit of KRW 57.3B and OPM of 39.2%. (PharmaResearch)
1Q26 cosmetics revenue was KRW 42.2B, about 29% of revenue; cosmetics exports were KRW 26.9B. (PharmaResearch)
Rejuran Cosmetics entered Sephora U.S. online and about 380 physical stores. (PharmaResearch)
As checked on May 31, 2026, Olive Young U.S. ranked REJURAN Dual Effect Ampoule #2 and REJURAN Turnover Ampoule #7. (Olive Young US)

[Inference]

Cosmetics is becoming a core growth axis rather than a side business.
Multiple compression reflects damaged visibility and a weaker medical-device monopoly narrative.
Olive Young U.S. rankings matter more as a SKU-velocity signal than as direct revenue.
Incremental U.S. cosmetics revenue of KRW 30B-50B begins to support estimate upgrades and multiple expansion.

[Speculation]

Rejuran Cosmetics may earn part of the global K-beauty premium currently attached to APR-like names.
Consumer skincare awareness may feed back into demand for Rejuran procedures.
A 20-22.5x PER re-rating is possible if 2Q26 proof points arrive.

[Blocked]

Olive Young U.S. unit sales, revenue and inventory data.
Store-level Sephora sell-through and re-order data.
Pure U.S. contribution inside Rejuran cosmetics exports.
Confirmed 2Q26 cosmetics exports and consolidated OPM.

Disclaimer: This note is for research and information purposes only and is not personalized investment advice.

AI Infrastructure Multiple Map: Why Samsung Electronics Looks Cheap and Samsung Electro-Mechanics Looks Expensive

Sun, 31 May 2026 09:00:00 +0900

Context This post bridges the Samsung Electronics stock-bonus / memory-supercycle note and the Samsung Electro-Mechanics KRW 138T peer-premium note. The first asked whether Samsung’s DS profit duration is being re-rated. The second asked whether SEMCO deserves Murata / Ibiden-like AI component premiums. This note asks the connecting question: inside one AI infrastructure cycle, why do different layers deserve different multiples? Related hubs: AI HBM Hub, AI PCB & Substrate Hub, and Korea Semiconductor Value Chain Hub.

TL;DR

Inside the same AI infrastructure cycle, the market does not assign the same multiple to every supplier. Multiples are not a simple function of “AI exposure.” They are a function of earnings duration, pricing power, customer lock-in, capacity risk and peak-earnings doubt.

The most dangerous area today is re-rating MLCC / FC-BGA bottleneck stocks as if they were GPU/HBM-style structural monopolies. They are real bottlenecks, but not every bottleneck deserves an NVIDIA-style platform multiple.

The strongest relative-value idea is Samsung Electronics. It has HBM catch-up, memory pricing and foundry optionality, yet the user-provided 2026-05-29/30 public finance-data snapshot shows roughly 7.8x forward P/E and 4.4x P/B. Samsung Electro-Mechanics has the right thesis, but the price already demands several years of successful execution. (Yahoo Finance)

1. The Core Formula

The multiple spread inside AI infrastructure is not about “who is exposed to the same demand cycle.” It is about who can capture excess profit for longer, with less reinvestment, under more certain customer contracts.

Multiple Premium
= pricing power × demand duration × customer lock-in × FCF conversion

Multiple Discount
= capacity capex burden × oversupply risk × peak-earnings doubt × customer concentration

That formula turns GPU, HBM, CPU, MLCC and FC-BGA into different assets, even though they all sit inside the same AI capex cycle.

Layer	What Lifts the Multiple	What Caps the Multiple	Key Question
GPU / platform	CUDA, rack-scale systems, software lock-in, asset-light structure	custom ASICs, China controls, hyperscaler bargaining power	Is the customer buying time-to-market or just a chip?
HBM / memory	HBM4, sold-out supply, LTAs, rising HBM content per accelerator	memory peak-earnings doubt, capex, supplier diversification	Is this a cycle profit or a contracted franchise?
CPU	AI server attach, orchestration demand	substitutability, ARM / custom CPU, lower scarcity	Primary choke point or necessary helper?
FC-BGA	large high-layer substrates, qualification, long lead times	capex-heavy, depreciation, ABF oversupply memory	Is capex underwritten by LTAs / prepayments?
MLCC / Si-Cap	power-integrity bottleneck, high-reliability parts, low customer price resistance	generic MLCC cycle, low AI revenue purity, competition	Shipment blocker or commodity passive?

2. GPU: Why It Gets the Highest Multiple

GPU is not just a semiconductor component. NVIDIA controls GPUs, networking, rack-scale systems, CUDA, the software stack and reference architectures. Customers are not merely buying chips. They are buying time to build an AI factory.

NVIDIA FY2027 Q1 revenue was $81.6B, Data Center revenue was $75.2B, and Q2 revenue guidance was $91.0B. With that growth rate and margin structure, the company’s large market cap alone is not enough to call it a bubble. (NVIDIA Investor Relations)

The GPU multiple is about:

Variable	Interpretation
P	GPU / rack / system-level ASP can be defended or raised
Q	AI datacenter capex maps directly into accelerator demand
C	fabless structure and platform premium push much of the capex burden into the supply chain

The Red Team matters. Hyperscaler custom ASICs, power / land / cooling constraints, China export controls and customer bargaining power are all real risks. Broadcom’s AI revenue is also scaling fast: Q1 AI revenue was $8.4B and Q2 AI semiconductor revenue guidance was $10.7B. (Broadcom Inc.)

Still, the GPU platform is the highest-quality layer in AI infrastructure. The reason is not “GPU.” The reason is control over the AI factory operating system.

3. HBM: Hot Earnings, Peak-Earnings Discount

HBM is the memory-bandwidth bottleneck that determines GPU and custom ASIC performance. Without HBM, GPUs cannot ship properly. Yet HBM companies such as SK hynix and Micron screen at much lower forward P/Es than NVIDIA or Broadcom.

That is not because demand is weak. It is because the market still discounts HBM earnings as potential memory-cycle peak earnings, not fully as structural franchise earnings.

Variable	Interpretation
P	HBM ASP carries a strong premium to commodity DRAM
Q	HBM content per GPU / ASIC, stack count and HBM4 / HBM4E transitions drive growth
C	TSV, stacking, packaging, yield and wafer allocation create heavy execution cost

The key is the LTA. Traditional DRAM is exposed to spot prices and inventory cycles. HBM increasingly relies on customer qualification, allocation, long-term agreements and fixed volume / pricing structures. The stronger the LTA, the more the market can treat HBM as a contracted high-value memory franchise rather than commodity DRAM.

But P/B must be checked alongside P/E. In the input snapshot, SK hynix and Micron screen at low forward P/Es, but their P/B ratios already look structurally re-rated. “Cheap” and “peak-earnings doubt” coexist. (Yahoo Finance, Yahoo Finance)

4. CPU: Essential, But Not the Primary Choke Point

CPUs are necessary in AI servers. Host CPUs, orchestration, storage/network control, preprocessing and inference serving all need them. AMD’s Q1 data-center revenue was $5.8B, up 57% year over year. (Advanced Micro Devices, Inc.)

But the primary AI infrastructure bottlenecks are more clearly GPU, HBM, networking and advanced packaging. CPUs are necessary, but substitutable. Customers can choose across x86, ARM, hyperscaler internal CPUs, NVIDIA Vera CPU, Intel Xeon and AMD EPYC.

So CPU multiples require discipline.

Company	Interpretation
AMD	Server CPU share gain is real, but a high forward P/E requires both EPYC share gains and Instinct GPU ramp success
Intel	The turnaround option exists, but foundry, process or AI accelerator execution must be proven first

CPU benefits from the AI cycle, but a standalone CPU thesis has weaker alpha than GPU, HBM, FC-BGA or high-value power-integrity parts.

5. FC-BGA and MLCC: Right Theme, But Bottleneck Is Not Monopoly

The reason Samsung Electro-Mechanics, Murata and Ibiden matter is clear. As AI accelerators, server CPUs and networking ASICs grow larger, demand rises for large high-layer FC-BGA and high-reliability MLCCs. AI servers run higher power spikes, lower voltage margins and tighter power-integrity requirements.

Samsung Electro-Mechanics reported Q1 2026 revenue of KRW 3.209T and operating profit of KRW 280.6B, citing higher supply of AI-server MLCC and AI accelerator / server-CPU FC-BGA. (Samsung Electro-Mechanics)

SEMCO also signed a roughly KRW 1.5T silicon capacitor supply contract with a global large-scale company for January 1, 2027 to December 31, 2028. (Samsung Electro-Mechanics)

Those are strong facts. The problem is price.

Layer	Why It Is a Bottleneck	Why It Is Not GPU/HBM Multiple by Default
FC-BGA	larger chips raise area, layer count and signal-integrity difficulty	capex-heavy, depreciation burden, ABF oversupply memory
MLCC	tiny part, but power failure can delay full server shipments	mixed with generic MLCC cycle, lower AI revenue purity
Si-Cap	die-near PDN component inside AI GPU / HBM packages	customer, margin, yield and repeat-order data mostly undisclosed

ResearchAndMarkets estimates the global FC-BGA market rising from $2.46B in 2026 to $3.74B in 2032, a 7.1% CAGR. That market growth alone does not easily justify 100x P/E for some FC-BGA-related equities. (Research and Markets)

So the conclusion is not “MLCC / FC-BGA is wrong.” The better conclusion is: the direction is right, but the price demands evidence of LTAs, prepayments, margin and AI revenue purity.

6. Samsung Electronics vs Samsung Electro-Mechanics: Same Cycle, Different Price

Samsung Electronics: AI Memory Hierarchy Option, Not Just Memory Cycle

Samsung Electronics is interesting not only because of HBM. It combines HBM4E catch-up, DDR5, SOCAMM2, eSSD / KV-cache storage and Samsung Foundry optionality. As argued in the NVIDIA Vera Rubin inference-stack note, inference is becoming a memory hierarchy problem across HBM, SRAM, eSSD, KV-cache and networking.

The input snapshot shows Samsung Electronics at roughly 7.8x forward P/E and 4.4x P/B. Its pure HBM exposure is lower than SK hynix’s, but the valuation gap and HBM catch-up option make the risk/reward more attractive. (Yahoo Finance)

Key checkpoints:

Checkpoint	Meaning
HBM4E qualification	condition for closing the SK hynix valuation gap
DDR5 / server DRAM pricing	duration of the broader memory supercycle
eSSD / KV-cache	exposure to the AI inference memory hierarchy
Samsung Foundry	whether the custom ASIC option remains alive

Samsung Electro-Mechanics: Correct Thesis, Price First

SEMCO is one of the few Korean component companies with MLCC + FC-BGA + Si-Cap at once. As covered in the SEMCO Si-Cap and EMIB-T note and the KRW 138T peer-premium note, it can be reclassified as an AI package power-integrity supplier.

But reclassification potential and current stock attractiveness are different things. The input snapshot puts SEMCO around 195x TTM P/E, 100x+ forward P/E and roughly 16x P/B. That price requires not a “good company” but a company with repeat orders, high margins and a visible 2027-2028 earnings step-up.

What matters now is evidence, not story:

Evidence Needed	Why It Matters
additional Si-Cap customers	distinguishes one big contract from platformization
Si-Cap margin	tests the high-value thesis
AI-server MLCC ASP	separates AI MLCC from commodity MLCC
FC-BGA utilization / OPM	tests whether capex-heavy risk is absorbed
LTAs / prepayments	confirms capex payback visibility

7. Practical View

Name	View	Reason
Samsung Electronics	Under / Buy-now candidate	low multiple versus HBM catch-up, memory pricing and foundry option
NVIDIA	Fair to relatively under	forward P/E is not extreme versus growth, but position sizing matters
SK hynix	Fundamental under, tactical wait	strong HBM alpha, but P/B and recent rally create entry risk
Micron	Fair to under, volatile	HBM / DRAM upside and P/B re-rating coexist
Broadcom	Fair to slightly over	strong AI ASIC visibility, already rich valuation
AMD	Over	CPU share gain alone struggles to defend the multiple
Intel	Over / avoid	turnaround expectations have run ahead of execution proof
Samsung Electro-Mechanics	Over / avoid chasing	thesis right, but repeat orders and high margins are already priced
Murata / Ibiden	Over	AI passive / substrate bottlenecks are real, but multiples already price monopoly-like outcomes

The conclusion:

The biggest mistake in this cycle is assigning a platform multiple to every component supplier just because it touches the NVIDIA supply chain.

GPU, HBM, MLCC and FC-BGA all matter. Their margin-capture structures differ. At current prices, Samsung Electronics and NVIDIA still have logical buy cases, while SK hynix and Micron look structurally cheap but timing-sensitive. Samsung Electro-Mechanics, Murata and Ibiden may be excellent companies, but the stocks may not be excellent at current prices.

8. Evidence Classification

[Fact]

NVIDIA FY2027 Q1 revenue was $81.6B, Data Center revenue was $75.2B, and Q2 revenue guidance was $91.0B. (NVIDIA Investor Relations)
Broadcom Q1 AI revenue was $8.4B, up 106% year over year, and Q2 AI semiconductor revenue guidance was $10.7B. (Broadcom Inc.)
AMD Q1 2026 data-center revenue was $5.8B, up 57% year over year. (Advanced Micro Devices, Inc.)
Samsung Electro-Mechanics Q1 2026 revenue was KRW 3.209T and operating profit was KRW 280.6B, with AI-server MLCC and AI accelerator / server-CPU FC-BGA cited as drivers. (Samsung Electro-Mechanics)
SEMCO signed an approximately KRW 1.5T silicon capacitor supply contract for January 1, 2027 to December 31, 2028. (Samsung Electro-Mechanics)

[Inference]

NVIDIA deserves a higher multiple than normal semiconductor component suppliers because it is closer to an AI factory platform owner.
Low forward P/Es in HBM reflect peak-memory-cycle doubt more than weak demand.
Samsung Electronics has lower HBM purity than SK hynix but better relative risk/reward because of valuation gap plus HBM catch-up optionality.
MLCC and FC-BGA bottlenecks have value, but without CUDA-like platform lock-in, 100x P/E requires repeat orders and margin proof.

[Speculation]

If HBM LTAs become fixed-price or prepayment-backed through 2027, SK hynix and Micron can deserve structurally higher memory multiples.
If Samsung Electronics qualifies HBM4E quickly, Samsung may become the cleaner 2H26 alpha than SK hynix.
If SEMCO’s Si-Cap business expands into second and third AI ASIC customers, part of today’s valuation burden may be explained after the fact.

[Blocked]

Vera Rubin MLCC / FC-BGA BOM, Murata versus SEMCO supply share, customer-specific LTAs / prepayments and FC-BGA pricing escalators are not public.
SEMCO Si-Cap ASP, units, yield, OPM, customer name and take-or-pay terms are not public.
Multiples in this post come from the 2026-05-29/30 public-data snapshot in the input, not real-time market data.

How This Blog Is Made: Introducing Thesis OS, Our Open-Source Research Operating System

Sat, 30 May 2026 11:00:00 +0900

🔗 Go to repo: github.com/youngseongshin/thesis-investment-os — the open-source system that runs this blog’s research

Today’s post is a little different from the usual. It isn’t about a stock — it’s about how the posts on this blog actually get made. Let me pull back the curtain for a moment.

What it takes to produce a single post

The posts on Korea Invest Insights are not improvised by a person staring at a blank screen. Behind them runs a small operating system called the Thesis Investment OS. The name sounds grand, but the idea is simple.

Make investment judgment visible, testable, and honest about its own track record.

It is not an automated trading bot, not a signal-selling service, and not an “AI that picks stocks for you.” It is a framework that gathers fragmented market information into a thesis — and lets you go back later and check whether that thesis turned out right or wrong.

The structure breaks into three roles. Think of them as three people on one team.

1. Alpha — the one who gathers the evidence

Alpha is the role that collects and verifies facts.

Quantitative data: prices, volume, flows, fundamentals, filings
Qualitative data: news, filings, earnings transcripts, community signals
Narrowing down candidates with screeners, then layering on context to surface names worth watching

What Alpha produces is evidence records, market snapshots, intraday alerts, screener candidates, and research packets. In short, it is the one who honestly stacks up “what happened." It does not judge yet. It only gathers the raw material.

2. Lattice — the one who builds judgment from evidence

The name Lattice comes from Charlie Munger’s idea of a “latticework of mental models” — a mind built from many interlocking frameworks.

Its role is to take the material Alpha gathered and turn it into an actual investment decision.

Registering a thesis and organizing it into a decision card
Running a devil’s advocate review that deliberately argues the other side
Recording predictions in a prediction ledger, then revisiting them later to see if they held up

The structure you read on the blog — “here’s the conclusion,” “this is a fact and this is speculation” — comes straight from Lattice. The point is to make a call, but leave it in a form you can grade later.

3. Arki — the one who keeps the system running

Arki is the least visible role, and perhaps the most important. It is the one that keeps the whole system healthy.

Defining the schemas that hold the data and the vault layout that stores it
Managing recurring jobs and running health checks
Keeping migration logs and governing the permissions and rules of each role

If the system were a house, Arki is the one making sure the electricity, water and heating keep running while Alpha and Lattice do their work. It is not glamorous, but without Arki the other two wouldn’t last long.

What these three roles have produced — real examples

This is abstract in words, so here are two recent posts that came through this system.

Dell Q1 earnings and the Korea AI-server margin read-through — Alpha gathered Dell’s earnings numbers, and Lattice connected them into the Korean semiconductor and server value chain to build a view.
Marvell Q1 FY2027 results and the Korea semiconductor read-through — same flow: starting from Marvell’s custom-silicon numbers and carrying them into a Korea read-through.

Both posts separate “this is a Fact, this is an Inference, this is Speculation.” That habit is exactly the structure Lattice enforces, and the facts holding it up are the ones Alpha gathered.

Why publish this at all

When you do research long enough, the scariest thing is “not remembering what you said before." Good-looking theses are plentiful; going back to check whether they were actually right is tedious and uncomfortable. So most analysis is written once and forgotten.

Thesis OS deliberately builds that discomfort into the system. Every judgment gets evidence attached, every prediction gets logged, and everything gets graded later. Not because it is perfect, but because it is built so that when it’s wrong, you can see it.

The system is designed to run locally. You can try it with the bundled sample data — no API keys, broker logins, or paid feeds required. The license is MIT, and it needs Python 3.10 or newer.

And the three channels this system publishes through are exactly these: the blog (Korea Invest Insights) you’re reading now, Telegram @korea_invest_insights, and Substack.

Come take a look

The point of this post isn’t to brag — it’s an invitation. If you’ve ever wondered how to make investment research more honest and more testable, take a peek.

You don’t have to read all the code. Even skimming the README should give you a feel for “ah, so this is how these blog posts get made.”

👉 github.com/youngseongshin/thesis-investment-os

A star is welcome, but just browsing the structure is fine too. There is only one reason I opened the curtain: so you can see for yourself where and how this blog’s judgments come from.

Disclaimer: For research and information purposes only. Not personalized investment advice. The open-source system described is a research tool; readers are responsible for their own investment decisions and outcomes.

ADR Hits a Six-Year Low: Will the Megacap Chip Squeeze Tighten, or Is It Time for the Laggards?

Sat, 30 May 2026 10:30:00 +0900

This note follows up on Korea ADR at 67: The Index Holds, So Why Are Stocks Weak?, Korea Foreign Investor Flow Analysis, and Smart Money Returns to the KOSDAQ. Where the earlier pieces looked separately at market breadth (ADR), foreign flows into megacaps, and KOSDAQ liquidity, this one ties all three together to answer the practical question: “Will the concentration tighten, or will the laggards rebound?"

TL;DR

On May 29, 2026 the KOSPI set a fresh record at 8,476pt (+3.55%), but the same day’s ADR fell to roughly 52% — the lowest since March 2020, a six-year trough. A record-high index paired with nine of ten stocks falling is an extraordinarily narrow market.
The stars of this squeeze are Samsung Electronics and SK Hynix (the “megacap chip duo,” or what Korean traders call “Jeon-Nik”). The starting point isn’t a simple theme but a genuine memory earnings cycle — yet single-stock leverage ETFs piled a wave of speculative flow on top of it.
The core conclusion is singular. Don’t try to predict the unwind — confirm it with signals. The default right now is “concentration persists,” and laggards should be added only when ADR, turnover, foreign flows, and earnings estimates all turn together.
And the most important distinction: “riding the concentration” and “riding the concentration product (leverage ETFs)” are two completely different things.

1. What Is Actually Happening in the Korean Market

Start with the facts. ADR doesn’t measure how high the index is — it measures the breadth inside the market.

ADR = Advancing issues / Declining issues × 100

When ADR is 100, advancers and decliners are balanced. Well below 100 means “regardless of the index, most stocks are falling.” The roughly 52% reading on May 29 is a level not seen since the 2020 pandemic crash.

Item	As of May 29
KOSPI	8,476.15pt, +3.55%, all-time high
ADR	About 52%, a six-year low
Market breadth	A breadth collapse close to “nine of ten stocks falling”

The previous note showed ADR at 67; in the interval, the market’s internals narrowed further. In other words, the record-high index is not the result of a broadly strong market, but of a handful of megacaps doing the pulling.

Two Flows That Amplified the Squeeze

Two distinct flows magnified the concentration.

First, single-stock leverage ETFs. On May 27 alone, total turnover in Samsung Electronics and SK Hynix single-stock leverage ETFs was about 10.4 trillion won, against a combined market cap of roughly 4.9 trillion won. Dividing turnover by market cap gives a daily turnover ratio of about 207%. That signals short-term, event-driven flow rather than normal long-term capital. The combined turnover of the four KODEX and TIGER Samsung Electronics/SK Hynix leverage products alone came to about 9.42 trillion won.

Second, the MSCI rebalancing. MSCI applied its May review at the May 29 close, and domestic brokers estimated 1.2 to 1.4 trillion won of passive inflows from higher weightings for SK Hynix and Samsung Electronics. Money that simply tracks the index was structurally forced to buy more of the duo.

In short, today’s concentration is the product of real earnings + passive flows + leverage flows all aligning in one direction.

2. Two Interpretations

There are two broad ways to read this.

Interpretation	One-line summary	Rationale
A. The squeeze tightens further	Strong leaders run for a long time	Memory earnings cycle, passive and leverage flows, MSCI weight increases
B. It’s the laggards’ chance to rebound	A market this narrow snaps back	ADR at a six-year low; extreme breadth collapses don’t last

Both are partly right. A is correct in that “the strongest earnings momentum still sits with the duo,” and B is correct in that “extreme ADR lows have statistically been followed by rebounds.”

The real issue is what to do when both are true.

3. The Answer: Don’t Predict — Confirm

Bottom line first: no one can know in advance when the concentration unwinds. Strong leaders always run longer than expected. So abandoning the leaders too early on the reasoning that “rotation into laggards is coming soon” usually costs you.

Instead, judge the unwind not by “has the duo rolled over,” but by “has the rest of the market, ex-duo, gained the power to rise on its own?" Put plainly, these are the five things to watch:

Ex-duo relative strength: Does the KOSPI excluding the duo start to outperform the duo itself?
Broadening advancers: With the index holding, does the number of advancing stocks grow and does ADR recover above 70?
Broadening foreign flows: Do foreign net buys stop clustering in the duo and spread to other names?
Leverage ETF cooldown: Does leverage-ETF turnover cool while the underlying Samsung and SK Hynix shares hold firm?
Laggard estimate upgrades: Are the earnings estimates (EPS) of the neglected names being revised higher?

When three or four of these light up together, that’s when you can acknowledge a rotation. There’s no need to bet ahead of time with only one or two flashing.

The Most Important Distinction: “Riding the Squeeze” vs. “Riding the Leverage ETF”

Here’s a trap worth flagging. “I’ll ride the concentration in the duo” and “I’ll buy the duo’s leverage ETF” are entirely different statements.

A single-stock leverage ETF is not a stock — it’s a leverage product that resets daily. You’re buying not SK Hynix itself but “twice SK Hynix’s daily return,” and the longer you hold, the more the two become completely different assets.

These products carry negative compounding (volatility decay). Even before listing, Korea’s Financial Supervisory Service warned that “if a price repeatedly rises and falls, cumulative returns can diverge sharply from the underlying.” On the managers’ own simulations, holding a high-volatility name at 2x leverage for about three months often produces a loss of typically 8% to 12% on the leverage ETF even when the underlying is flat (0%).

The key is this: even if the squeeze doesn’t unwind and the duo merely trades sideways, the leverage ETF loses over time. On top of that, with no diversification the daily limit is a wide ±60%, and on June 28 the FSS opened an investigation into possible wash trades and overheating. In other words, it’s a product whose exit you can’t easily control.

So if you want to ride the concentration, the right approach is the underlying (cash) shares, not chasing the leverage ETF. Riding negative compounding — a “publicly disclosed inefficiency” — isn’t alpha; it’s just leverage.

4. Investment Idea Examples (Observation Points, Not Recommendations)

The below are not stock recommendations but examples showing “what conditions make a name a candidate.” This isn’t a call to buy now, but a map of where to look first when the five confirmation signals above light up.

Example 1 — The Next Bottleneck After GPU/HBM: Power, Substrates, Passives

As AI servers grow more complex, demand rises not only for GPUs and HBM but also for parts that stabilize power, substrates, and passive components. MLCCs, FC-BGA, and high-layer-count boards (MLB) are the prime examples. That said, some names here have already run a long way. One large component stock rose about 2.56x between late April and late May. So the point is not “what looks cheap” but “what is seeing fresh upward earnings revisions."

Example 2 — Traces of Foreigners’ Pre-Positioning

There’s one intriguing signal. On May 29, as retail dumped en masse and memory melted up, foreigners quietly net-bought some deeply beaten-down materials, parts, and equipment names. Wonik IPS, Leeno Industrial, and EO Technics were among them. Because domestic institutions mostly sold alongside retail, a “foreigners buying, domestic institutions selling” divergence emerged. Whether that divergence resolves with domestic institutions returning is the second confirmation point.

Example 3 — Where Does the Money the Leverage ETFs Bleed Go?

The most contrarian view is this: when everyone is buying the duo’s leverage ETFs, watch where the money those products bleed via negative compounding will flow. When leverage holders bleed in sideways and choppy ranges, that money ultimately has to flow back into more efficient exposure — index products, cash equities, and neglected quality names. This is a bet on direction rather than timing, which makes it relatively safe.

Common condition: all of the examples above need more than just “oversold.” A name only becomes a candidate when earnings aren’t impaired, estimates are being revised higher, and turnover and foreign flows are attaching for the first time — all at once.

5. The June Calendar Is the Real Variable for Rotation

The biggest enemy of a laggard rebound is, surprisingly, rates and liquidity. June is packed with heavyweight events.

Event	Date	Why it matters
9th nationwide local elections	June 3	Short-term political uncertainty
US May CPI	June 10	Whether inflation reignites
FOMC	June 16–17	Chair Kevin Warsh’s first meeting
SpaceX IPO (target)	June 12	Assumed to be a major liquidity-absorbing event

US April PCE inflation came in at +3.8% year over year, so the inflation burden is still not low. If June CPI accelerates further or the new Fed chair is read as hawkish, rates could rise and rotation into high-valuation laggards (especially high-PER component names) could be delayed.

And keep one statistical trap in mind. The claim that “the win rate of rebounds after extreme ADR lows is nearly 100%” omits that the sample is small and that most past cases occurred during rate-cutting regimes. We may instead be in the opposite regime, with inflation stirring again. You can’t mechanically transplant past win rates onto the present.

6. Wrap-Up — Fund Manager’s Comment

Back to the question: “Will the megacap chip squeeze tighten, or is it the laggards’ chance to rebound?” The honest answer is “both are possible, and right now we’re in the stage of waiting for the signal that separates them."

The two riskiest choices are:

Abandoning the leaders too early on the back of a low ADR alone — handing over strong earnings momentum at a discount.
Chasing an already-overheated single-stock leverage ETF too late — being dragged along with no exit by negative compounding and ±60% volatility.

So the most rational stance right now boils down to this:

Hold the duo via the underlying (cash) shares, do not chase the leverage ETF, and add laggards only after confirming that ADR, turnover, foreign flows, and earnings estimates all turn together.

Unwinding a concentration isn’t something you predict — it’s something you confirm. Not missing the moment that first confirmation signal lights up, using the five-point checklist above — that’s the most practical preparation you can make right now.

The market figures in this article are based on news and disclosures between May 27 and 29, 2026, and the named stocks are examples to illustrate the analytical flow, not investment recommendations. All actual investment decisions and their consequences rest with the investor.

AI Token Futures and Cost Per Token: It's Now a Cost Race, Not a Performance Race

Sat, 30 May 2026 09:00:00 +0900

📚 Context for follow-up posts This is a follow-up to Dell’s Earnings Surprise and Korean Semiconductors and Marvell Q1 FY2027 and Korean Semiconductors. Where the first two posts argued that “the AI bottleneck migrates from the GPU down to memory, interconnect, substrate and power — and who earns the margin at that bottleneck,” this post goes one step further. Once AI usage itself gets a market price, the battleground shifts from “a faster chip” to “a cheaper token.” Related hubs: AI HBM Hub, Korea Semiconductor Value Chain Hub, AI PCB & Substrate Hub.

TL;DR

Until now, AI investing has been a fight over “who builds the more powerful chip.” But a quiet signal has just switched on. AI usage itself is starting to get a market price.

According to Reuters, China’s Shanghai Futures Exchange (SHFE) is early-designing futures linked to AI token prices. In the U.S., CME Group (with Silicon Data) and ICE (with Ornn) announced plans for GPU compute futures. None of these are listed, confirmed products yet. But the direction is clear.

Once AI compute and tokens have a “market price,” the industry’s axis shifts from a performance race to a cost race.

It is the same as when oil got a futures price: refining, shipping and power generation all lined up by “cost per barrel.” Once AI’s price per token becomes visible, every participant asks the same question: “What does it cost me to produce one token?"

The answer is not generic “AI stocks.” It is the bottleneck parts, chips and platforms that actually lower the customer’s cost per token. The priority order:

Priority	What to Buy	Key Names	View
1	Custom ASIC / AI networking	Broadcom, Marvell, TSMC	Most direct to the token-cost stack. But it’s a consensus long — valuation discipline needed, customer-concentration risk
2	HBM + eSSD + KV-cache storage	Samsung Electronics, SK Hynix, Micron	Memory bandwidth and cache management are the inference bottleneck. eSSD/KV-cache cut repeated compute and ease HBM load
3	Advanced packaging / FC-BGA / MLCC / silicon capacitor	Samsung Electro-Mechanics, Daeduck Electronics, Murata, Ibiden	Power & signal integrity → effective token throughput. “Right direction, wrong price” risk
4	Compute pricing rail	CME, ICE	Financialization itself. Option value, pre-launch, liquidity uncertain. Least-crowded 2nd-order idea
5	Nvidia / AMD full-stack	Nvidia, AMD	Still core. But the thesis becomes “highest tokens-per-watt & AI-factory economics,” not “GPU shortage”
6	Power equipment	HD Hyundai Electric, LS Electric	Not an outright short. Power demand keeps rising (IEA: datacenter power ~945TWh by 2030). Use only as a relative-value hedge against an over-stretched power-scarcity premium

The Korea bottom line first: Samsung Electronics is the most balanced practical pick. HBM4E catch-up + DDR5 + SOCAMM2 + PCIe Gen6 eSSD + KV-cache storage give it the broadest exposure across the token-cost stack. SK Hynix is the best-quality name but crowded. Samsung Electro-Mechanics has high thesis purity but price matters. Daeduck Electronics is a second-order FC-BGA beta candidate.

1. What Is Happening: The Financialization of AI Usage

First, separate the facts.

[Fact] Per Reuters, SHFE is early-designing futures linked to AI token prices. The contract specification, the definition of the underlying (which token), the settlement method, and the approval/launch timeline are all unconfirmed.
[Fact] CME Group announced plans to launch compute futures with Silicon Data; ICE announced plans for GPU compute futures contracts with Ornn.

What matters here is not “futures are already listed.” They are not. What matters is that the industry has started trying to put a price on AI usage.

An analogy: imagine using electricity without any standard price of “X per kilowatt-hour.” The moment a power exchange opens and prices get quoted, every factory starts computing “how much electricity do we use per unit of product?” AI is now standing at exactly that threshold.

When the price becomes visible, the cost becomes visible. When the cost becomes visible, whoever lowers it wins.

2. Why This Changes the Rules: Performance Race → Cost Race

For years the AI race was about “benchmark scores” — who runs a bigger model on a faster chip. Investing followed the same logic: “buy the company that makes the most powerful GPU.”

Once usage gets a price, the question changes.

Cost per token, in the simplest form:

Cost per token = total compute cost / tokens processed

You lower it by shrinking the numerator (compute cost) or growing the denominator (tokens processed). Whoever extracts more tokens from the same electricity and the same capital wins. So two metrics become the new battleground:

Tokens per watt = tokens/sec / power (W)
Tokens per dollar = tokens / total cost ($)

Per Reuters, TSMC sees chip-design focus shifting from raw performance to power efficiency because of AI power demand — meaning tokens per watt becomes a key metric.

In investment language, this is clear:

Not “buy AI stocks,” but “buy the companies that lower the customer’s cost per token."

3. The Kitchen Analogy: The Expensive Chef Is Sitting Idle

Let’s make it more concrete.

A GPU is a very expensive chef. Their hands are fast, but they can only cook if the ingredients arrive on time. The conveyor belt that brings the ingredients is the memory (HBM).

What happens if the belt is slow? The expensive chef stands around waiting for ingredients. The wage clock keeps running. No tokens come out. Cost per token spikes.

So the real bottleneck is often not “hire another chef” (= a more powerful GPU). It is more effective to speed up the belt and pre-stack the frequently used ingredients next to the chef (= memory bandwidth + cache).

KV-cache storage is exactly the act of “pre-stacking frequently used ingredients nearby.” It avoids redoing the same calculation, saving the expensive chef’s time. eSSD makes that pantry cheap and large. This also reduces the load placed on HBM.

This analogy is the investment map for this post. Money flows to wherever the cost per token goes down.

4. Priority 1: Custom ASIC / AI Networking — Broadcom, Marvell, TSMC

The most direct global stack for lowering token cost is custom ASIC and AI networking.

When a hyperscaler uses a chip tailored to its own workload (a custom ASIC), it does the same job more cheaply and with less power than a generic GPU. AI networking is what ties those chips into a cluster. Both lift tokens per watt directly.

Company	Confirmed Facts	View
Broadcom	Q1 FY2026 AI revenue $8.4B (+106% YoY); Q2 AI semiconductor revenue guided ~$10.7B	The core of custom ASIC / networking. Consensus long — valuation discipline needed
Marvell	FY2027 Q1 revenue $2.418B; per Reuters, targets custom-chip revenue >$10B by FY2029	Custom-silicon growth path. Customer concentration is the key variable
TSMC	Per Reuters, AI power demand is shifting chip-design focus from performance to power efficiency → “tokens per watt” a key metric	The foundry every custom ASIC / GPU ultimately passes through. The floor of the token-cost stack

A caution, though. This is already the consensus long the market loves most. A great company and a great price are not the same thing. And custom ASIC revenue is concentrated in a few large customers, so a single customer’s order swings can move the numbers sharply.

5. Priority 2: HBM + eSSD + KV-cache Storage — Samsung Electronics, SK Hynix, Micron

In inference, the real bottleneck is not raw compute but memory bandwidth and cache management. It is exactly the chef-belt analogy from Section 3.

HBM is the ultra-fast belt right next to the chef.
eSSD and KV-cache storage are the pantry where frequently used ingredients are pre-stacked. They avoid repeating the same computation, save expensive GPU time, and ease the load on HBM.

Samsung Electronics’ 1Q26 results materials cited demand for DDR5, SOCAMM2, PCIe Gen6 eSSD, and KV-cache storage. That lineup maps precisely onto the token-cost thesis.

Company	Token-Cost Exposure	View
Samsung Electronics	HBM4E catch-up + DDR5 + SOCAMM2 + PCIe Gen6 eSSD + KV-cache storage	Broadest exposure. Spans multiple layers of the token-cost stack
SK Hynix	Best-quality pure HBM exposure	The best company itself, but the most crowded position in the market
Micron	One of the three global HBM/memory leaders	The U.S. comparable. HBM/DRAM cycle exposure

For a deeper discussion of Samsung Electronics, see Samsung Electronics Stock-Bonus Buyback Analysis and the Samsung Electronics Deep Dive; for SK Hynix, see the SK Hynix Deep Dive.

If forced to pick one, the most balanced Korean exposure to the token-cost thesis is Samsung Electronics. SK Hynix is cleaner as pure HBM beta, but across KV-cache, eSSD and SOCAMM2 Samsung spans more layers — and the market crowding is lighter on Samsung than on SK Hynix.

6. Priority 3: Advanced Packaging / FC-BGA / MLCC / Silicon Capacitor — Samsung Electro-Mechanics, Daeduck

No matter how good a chip is, if the power wobbles and the signal breaks, effective token throughput falls. Advanced packaging, substrates and power-stabilization components are what protect power integrity and signal integrity.

FC-BGA is the high-performance substrate that connects the chip to the board.
MLCCs and silicon capacitors dampen transient voltage swings inside the GPU/HBM package.

A solid layer here lets the expensive chef run at full speed, producing more tokens. So while these parts are indirect, they clearly affect cost per token.

Company	Exposure	View
Samsung Electro-Mechanics	FC-BGA + high-end MLCC + silicon capacitor bundle	High thesis purity. But valuation is already stretched — a “right direction, wrong price” spot
Daeduck Electronics	Second-order FC-BGA beta candidate	The next seat after SEMCO. Direct qualification / volume confirmation is a precondition
Murata, Ibiden	Japanese MLCC / packaging comparables	Tracked together as global peers

For detailed SEMCO analysis, see AI Server Passive-Component Bottleneck: A SEMCO Technical Explainer.

The direction is right. But the price is rich.

The token-cost thesis does point to packaging and substrates, but Samsung Electro-Mechanics has already priced much of that in. So in Korea, a second-order FC-BGA beta like Daeduck Electronics can be the “same direction, less stretched price” alternative — provided direct qualification and volume are confirmed first.

7. Priority 4: Compute Pricing Rail — CME, ICE

The starting point of this post, and its least-crowded second-order idea, is the compute pricing rail.

If the financialization of AI usage truly takes hold, the exchanges where that trading happens benefit structurally — just as exchanges enjoyed steady fee flows once oil futures took hold.

Company	Exposure	View
CME	Plans to launch compute futures with Silicon Data	An option on financialization itself. Pre-launch, liquidity uncertain
ICE	Plans for GPU compute futures contracts with Ornn	Similarly a pre-launch option value

But this is option value, not confirmed revenue. Whether the contracts actually list, and whether volume builds, is unknown. So treat it as a “side-branch of AI financialization” — the smallest-weight, second-order idea.

8. Priority 5: Nvidia / AMD Full-Stack — The Thesis Changes

Nvidia and AMD remain core. They are not dropping out. But the thesis changes.

The old thesis was “GPUs are scarce, so whoever makes them wins.” The token-cost-era thesis is different.

Not “GPU shortage,” but “who provides the full stack with the highest tokens-per-watt and the best AI-factory economics."

Nvidia is strong not simply because its chips are fast, but because it bundles GPUs, networking and software to lower the cost per token of the entire AI factory. In this frame, the key question is not “how fast is the next chip” but “how cheaply does the next system produce tokens.”

9. Priority 6: Power Equipment — A Relative-Value Hedge, Not a Short

Here is a common misconception worth addressing. “If token-cost efficiency matters, power demand falls, so just short the power-equipment names, right?”

No. Power demand itself keeps rising.

The IEA forecasts global datacenter electricity consumption reaching ~945TWh by 2030 (roughly 2x versus 2024).

That is no basis for an outright short on power equipment. Simply shorting names like HD Hyundai Electric or LS Electric is dangerous.

There is only one way to use power equipment here: a relative-value hedge. Use it only, and sparingly, when all three of the following hold at once:

Semiconductor EPS estimates are being revised up, while
Power-equipment EPS estimates stall, and
Power-equipment valuations sit at historic highs.

In other words, it is meaningful only as a relative-value hedge against an “over-stretched power-scarcity premium.” Not an outright short.

10. Discipline: “Financialization = Buy Everything” Is Greed

This thesis is attractive. But its size and the investment decision are separate. Three things must be kept apart.

[Fact]: CME-Silicon Data and ICE-Ornn compute-futures plans were announced. Broadcom’s, Marvell’s and TSMC’s numbers were reported. The IEA’s 945TWh forecast is public.
[Inference]: Once usage gets a price, the industry shifts from a performance race to a cost-per-token race. Then the bottlenecks that lower token cost (custom ASIC, memory, packaging) are structurally favored. This is a reasonable inference from the facts.
[Speculation]: The SHFE futures’ contract spec, underlying-token definition, settlement method and approval timeline are all unconfirmed. Which Korean component maker supplies which customer, and at what revenue share, is also undisclosed.

In particular, the SHFE AI token futures should be treated not as “an already-listed, confirmed product” but as an early signal and option value. And specific customer/share claims — “Samsung supplies this customer’s KV-cache storage,” “Daeduck holds X% of this FC-BGA” — are not verifiable from public sources.

Even a sound thesis ends as a theme unless it is validated in the numbers.

The conclusion is not “AI is being financialized → buy all the related stocks.” It is to selectively pick the bottlenecks that actually lower the customer’s cost per token — at a price you can defend.

11. Next Checkpoints

Checkpoint	Strong Signal	Weak Signal
AI usage financialization	SHFE / CME / ICE actually lists and builds volume on at least one	Stalls in design phase, liquidity falls short
Custom ASIC	Broadcom / Marvell custom-revenue visibility rises	Single-customer order delays / concentration risk surfaces
Memory / KV-cache	Samsung eSSD / SOCAMM2 / KV-cache volume and pricing confirmed	HBM competition intensifies, pricing weakens
Packaging / substrate	SEMCO package margin confirmed; Daeduck qualification / volume	Growth decelerates while valuation stays high
Tokens per watt	TSMC / Nvidia foreground power-efficiency metrics	Only performance headlines repeated
Power equipment	Semis EPS↑ + power EPS stall + power valuation at historic high all met	Any one of the three breaks → hold off on the hedge

Final Interpretation

AI token futures and GPU compute futures are still early. Nothing is listed; no volume has built. But the direction is clear.

The moment AI usage gets a market price, the battleground shifts from “a faster chip” to “a cheaper token.”

From this vantage point, what to buy is not generic AI stocks but the bottlenecks that actually lower the customer’s cost per token. Globally, custom ASIC / networking (Broadcom, Marvell, TSMC) is most direct, and memory / KV-cache storage solves the inference bottleneck. In Korea, the most balanced practical pick is Samsung Electronics — its HBM4E catch-up, DDR5, SOCAMM2, PCIe Gen6 eSSD and KV-cache storage span the token-cost stack most broadly. SK Hynix is the best quality but crowded; Samsung Electro-Mechanics has high thesis purity but a rich price; Daeduck Electronics is a second-order FC-BGA beta candidate. Power equipment is not a short — only a relative-value hedge used when the conditions line up.

The conclusion is discipline. The more AI is financialized, the more “at what price” matters as much as “what to buy.”

Fact / Inference / Speculation / Blocked

[Fact]

Per Reuters, SHFE is early-designing AI-token-price-linked futures. The contract spec, underlying-token definition, settlement and approval timeline are unconfirmed.
CME Group announced plans to launch compute futures with Silicon Data; ICE announced plans for GPU compute futures contracts with Ornn.
Broadcom Q1 FY2026 AI revenue $8.4B (+106% YoY); Q2 AI semiconductor revenue guided ~$10.7B.
Marvell FY2027 Q1 revenue $2.418B. Per Reuters, Marvell targets custom-chip revenue >$10B by FY2029.
Per Reuters, TSMC sees chip-design focus shifting from performance to power efficiency due to AI power demand, making “tokens per watt” a key metric.
The IEA forecasts global datacenter electricity reaching ~945TWh by 2030 (~2x versus 2024).
Samsung Electronics’ 1Q26 results materials cited DDR5, SOCAMM2, PCIe Gen6 eSSD, and KV-cache storage demand.

[Inference]

Once AI usage gets a price, the industry shifts from a performance race to a cost-per-token race.
The bottlenecks that lower cost per token (custom ASIC / AI networking → memory / KV-cache → packaging / substrate) are structurally favored.
In Korea, Samsung Electronics has the most balanced exposure across the token-cost stack; SK Hynix is the best quality but heavily crowded.
Because power demand keeps rising, an outright short of power equipment is inappropriate; use it only as a relative-value hedge when the conditions line up.

[Speculation]

Whether the SHFE AI token futures actually list and trade, and how the underlying token would be defined, is uncertain.
The launch timing and liquidity of the CME / ICE compute futures are not confirmed.
Which global customers specific Korean component makers (Samsung Electronics, Samsung Electro-Mechanics, Daeduck Electronics) supply, and at what revenue share, is undisclosed.

[Blocked]

The SHFE futures’ contract spec, settlement method and approval timeline.
Real-time consensus EPS revision rates and NTM valuations for the Korean names.
Korean suppliers’ customer-level share and revenue mix in the custom ASIC / memory / packaging supply chains.

Disclaimer: For research and information purposes only. Not investment advice. Names cited are for analytical illustration; readers should perform their own due diligence and consult licensed advisors before any investment decision.

Dell's Earnings Surprise and Korean Semiconductors: Pricing Power Sits With Components, Not Server Assembly

Fri, 29 May 2026 22:00:00 +0900

📚 Context for follow-up posts This is a follow-up to Marvell Q1 FY2027 and Korean Semiconductors. Where the Marvell post argued that “the AI bottleneck migrates from the GPU down to interconnect, substrate, and power,” this Dell post re-checks “who actually earns the money at that bottleneck” — through margin numbers. Related hubs: AI HBM Hub, Korea Semiconductor Value Chain Hub.

TL;DR

Dell reported FY2027 Q1 results on May 28, 2026 (U.S. time), and by any measure the numbers were a large earnings surprise. Revenue grew 88% in a single year, and adjusted EPS jumped 214%. The next day (May 29) the stock surged roughly 33%.

But buried in the same release was one uncomfortable number. No matter how many AI servers Dell sold, gross margin actually fell from 21.6% a year ago to 18.1%. Management’s own stated AI-server operating-margin target is mid-single-digit. In other words, Dell sells an enormous volume of product, but the margin on that product is thin.

That single line is the entire point of this post.

A company that assembles AI servers can post big revenue but thin margins. The real pricing power and the fat margin sit upstream — in memory and high-value components.

Dell itself said demand exceeds supply and that the primary constraint is memory. The moment the company building the servers says “what’s holding me back is memory,” it becomes clear where investors should look.

The Korea-translated priority order:

Priority	Korea Read-Through	Key Names	View
1	Memory pricing power (HBM, server DRAM, eSSD)	Samsung Electronics, SK Hynix	Most direct. SK Hynix is the best company but richly priced; Samsung has HBM4E / eSSD / foundry base-die catch-up optionality
2	FC-BGA, high-end MLCC, silicon capacitor	Samsung Electro-Mechanics	Direct second-order beneficiary, but P/E already in the 200s — “right direction, wrong price” risk
3	Datacenter power (second-order)	HD Hyundai Electric, LS Electric	Not a direct beneficiary of Dell’s print itself; a side-branch of the larger AI-power-demand trend
4	Avoid	Low-margin server assembly / set makers	The very spot Dell just illustrated: big revenue, thin margin

Dell itself had a strong quarter, but the source’s house view was HOLD, with a $475 12-month target. The discipline is not “AI servers are good → buy everything,” but to selectively look upstream, where margin and pricing power flow.

1. Dell’s Print: Revenue Explodes, Margin Compresses

First, the numbers as reported. (Dell)

Item	FY2027 Q1	Commentary
Revenue	$43.8B	+88% YoY; beat consensus (~$35.4–35.7B) by over $8B
GAAP diluted EPS	$5.24	+282% YoY
Non-GAAP diluted EPS	$4.86	+214% YoY; ~+$1.9 vs consensus (~$2.94–2.96)
ISG (servers, networking, storage)	$29.0B	+181% YoY
AI-optimized server revenue	$16.1B	+757% YoY
Traditional server/networking	$8.5B	+92%
Storage	$4.3B	+8%
CSG (PC)	$14.6B	+17% (Commercial $13.0B, +18%)
Operating cash flow	$4.1B	Adjusted FCF $3.165B, shareholder return $2.1B

Orders and guidance were strong too.

AI orders $24.4B; AI backlog $51.3B; over 5,000 AI customers.
Q2 revenue guidance $44–45B; Non-GAAP EPS midpoint $4.80.
FY2027 revenue $165–169B; Non-GAAP EPS midpoint $17.90.
FY2027 AI-server outlook ~$60B (+144%).

Up to this point it looks flawless. But the same release contains one decisive line.

Non-GAAP gross margin fell from 21.6% to 18.1%.

That means AI-server revenue grew more than sevenfold while the company-wide margin rate actually declined. Management put the AI-server operating-margin target at mid-single-digit. In other words, AI servers carry enormous revenue volume but a lower profitability profile than PCs or traditional servers.

2. Why the Margin Is Thin: The Value Sits in the Components

The reason is simple. The most expensive part of an AI server’s cost is not something Dell makes. It is the components: GPUs, HBM, server DRAM, eSSD, CPUs. Dell buys these, puts them in a chassis, connects the cabling, validates the system, and ships it to the customer.

As components get more expensive, revenue grows accordingly. But the margin on those expensive components is captured by the component makers. The assembler sees inflated revenue and thin margin. Dell’s print this quarter laid that structure bare.

The most important clue in management’s call was not the margin figure but the statement that demand exceeds supply, and the explanation that the primary constraint is memory. DRAM, NAND, CPU, HDD, and leading-node allocation were all cited as bottlenecks in the Q&A.

Translated into investor language:

It’s not that they can’t build the boxes — it’s that they can’t get enough of the key components. Whoever holds the scarce part sets the price. And the scarce part is memory.

If demand exceeds supply and the side that controls supply is memory, then pricing power clearly sits with memory. That is the starting point for the Korea read-through.

3. Korea Priority 1: Memory — Samsung Electronics and SK Hynix

The companies that sell the most of the very memory Dell called constrained are Samsung Electronics and SK Hynix.

Their Q1 results prove the structure directly.

Company	1Q26 Revenue	1Q26 Operating Profit	Note
Samsung Electronics	KRW 133.9T	KRW 57.2T	DS (semiconductor) division revenue KRW 81.7T, OP KRW 53.7T
SK Hynix	KRW 52.58T	KRW 37.61T	72% operating margin. HBM + high-capacity server DRAM + eSSD

SK Hynix’s 72% operating margin is no ordinary number. Set it next to Dell’s mid-single-digit AI-server operating-margin target, and you can see at a glance who captures the margin inside the same AI server. The side assembling the server is in the single digits; the side making the memory that goes inside it is in the 70s.

By name, the call differs.

SK Hynix is the strongest company in itself. It holds the core of AI memory: clean HBM exposure, high-capacity server DRAM, and eSSD. But as of May 29 its price is ~KRW 2,333,000, P/E ~22.1x, near its 52-week high. A great company, but the price is already elevated.
Samsung Electronics, as of May 29, is ~KRW 317,000, P/E ~25.4x, near its 52-week high (KRW 323,000). It is not as clean an HBM beta as SK Hynix, but it carries more catch-up optionality. It shipped the industry’s first HBM4E 12-high samples (in Q2), and eSSD plus a foundry-based base die also sit inside the same AI-memory stack.

For deeper discussion of Samsung Electronics, see Samsung Electronics Stock-Bonus Buyback Analysis and the Samsung Electronics Deep Dive.

The point: the moment Dell admitted that memory is the constraint, the pricing power of the two Korean companies holding that constraint structurally strengthens.

4. Korea Priority 2: High-Value Components — Samsung Electro-Mechanics

After memory, the next place pricing power flows is package substrates and power-stabilization components. In Korea, that seat is Samsung Electro-Mechanics.

Its Q1 results:

Item	1Q26	Note
Revenue	KRW 3.2091T
Operating profit	KRW 280.6B
Package-division revenue	KRW 725.0B	+45% YoY

The company directly cited strong demand for high-value MLCCs for AI servers and datacenters, and FC-BGA for AI/server applications. This dovetails precisely with the structure Dell revealed: as AI servers grow, demand for power-stabilization components and high-performance substrates around GPU/HBM packages grows alongside them.

For detailed analysis of Samsung Electro-Mechanics, see:

But here, price must be flagged. As of May 29, Samsung Electro-Mechanics is ~KRW 2,127,000, P/E ~227x. That is a very high number. It means the market has already priced in big expectations.

The direction is right. But the price is expensive.

Dell’s print did confirm that high-value components are positive, but Samsung Electro-Mechanics has already reflected much of that “good” in its price. So between memory (Priority 1) and Samsung Electro-Mechanics (Priority 2), there is a key difference: even the same beneficiary has a very different entry price.

5. Korea Priority 3: Datacenter Power (a Side Branch)

A lower-priority area is datacenter power. HD Hyundai Electric, as of May 29, is ~KRW 1,053,000, P/E ~48x.

The logic that more AI servers means more datacenter power demand, benefiting transformer and power-distribution makers, is sound. But this is not a direct result of Dell’s Q1 print — it is a side branch of the broader AI-power-demand trend. The constraint Dell explicitly named was not power; it was memory.

So power belongs in a second-order bucket: “related, but not the core read-through of this print.”

6. The Spot to Avoid: Low-Margin Assembly

The most important lesson from Dell’s print may lie less in what to buy than in what to avoid.

The fact that Dell sold more than seven times the AI servers and still saw its margin rate fall lays bare the structural limit of low-margin server assembly and set businesses. The revenue looks dazzling, but the money goes to the component makers.

So buying assembly/set-stage names indiscriminately just because they carry the “AI server beneficiary” label is dangerous. Even within the same AI-server revenue growth, the side with pricing power (memory, high-value components) and the side without it (assembly) are completely different investments.

7. Discipline: “AI Servers Are Good = Buy Everything” Is an Overreach

This print is clearly a strong positive. But the size of the positive and the investment call are separate things.

Three layers must be kept distinct.

Fact: Dell revenue +88%, EPS +214%, AI servers +757%, yet gross margin 21.6% → 18.1%. Memory is the primary constraint. These are numbers the company reported.
Inference: Margin and pricing power flow toward memory and high-value components rather than the assembler. In Korea the order is Samsung Electronics / SK Hynix → Samsung Electro-Mechanics → power. This is a judgment reasonably drawn from the facts.
Speculation: How much HBM4E a specific company supplies to Dell, and what share each maker holds in Dell’s servers, is not disclosed. This is not confirmed fact.

In particular, specific customer/share claims such as “Samsung Electronics supplies HBM4E to Dell” or “Samsung Electro-Mechanics holds X% of Dell server FC-BGA” are not confirmed in public disclosures. Such content should be classified as speculation or unverifiable, not fact.

Even a sound thesis ends as a theme unless it is validated in the numbers.

This post’s conclusion is not “AI servers are good, so buy all the related names.” It is: look upstream, where margin and pricing power flow — and select with the entry price in mind.

8. Next Checkpoints

Checkpoint	Strong Signal	Weak Signal
Dell AI-server margin	Mid-single-digit target achieved or raised	Margin falls further or misses target
Dell memory constraint	Memory shortage persists, sustaining pricing strength	Memory supply eases, prices soften
Samsung Electronics	HBM4E mass production / customer qualification, eSSD volume / pricing	Intensifying HBM competition, price declines
SK Hynix	HBM / server DRAM margins hold	2027 supply-overhang signals
Samsung Electro-Mechanics	Sustained package-revenue growth, confirmed margin	Growth decelerates while P/E stays high

Final Interpretation

Dell FY2027 Q1 is not a signal to “buy all AI servers” for Korean semiconductors. The more precise message is:

As AI-server revenue grows, margin and pricing power move up, away from the assembler and toward memory and high-value components.

From this vantage point, the most direct Korean names are memory — Samsung Electronics and SK Hynix. SK Hynix is the strongest company in itself but already priced high; Samsung Electronics carries HBM4E / eSSD / foundry base-die catch-up optionality. Samsung Electro-Mechanics is a direct second-order beneficiary through FC-BGA, high-end MLCC, and silicon capacitor, but at a P/E in the 200s it sits in a “right direction, wrong price” spot. Datacenter power is a side branch; low-margin assembly is the spot to avoid.

The conclusion is discipline. The stronger the positive, the more it matters not just what you buy, but at what price.

Fact / Inference / Speculation / Blocked

[Fact]

Dell FY2027 Q1 revenue was $43.8B (+88% YoY); Non-GAAP diluted EPS was $4.86 (+214%). (Dell)
ISG revenue $29.0B (+181%); AI-optimized server revenue $16.1B (+757%). (Dell)
Non-GAAP gross margin fell from 21.6% to 18.1%; AI-server operating-margin target is mid-single-digit. Memory (DRAM/NAND) was cited as the primary constraint. (Dell)
AI orders $24.4B; AI backlog $51.3B; FY2027 AI-server outlook ~$60B (+144%). (Dell)
Samsung Electronics 1Q26 revenue KRW 133.9T, OP KRW 57.2T; DS-division revenue KRW 81.7T, OP KRW 53.7T.
SK Hynix 1Q26 revenue KRW 52.58T, OP KRW 37.61T, 72% operating margin.
Samsung Electro-Mechanics 1Q26 revenue KRW 3.2091T, OP KRW 280.6B, package-division revenue KRW 725.0B (+45% YoY).

[Inference]

As AI-server revenue grows, margin and pricing power flow toward memory and high-value components rather than the assembler (Dell).
A reasonable Korea read-through priority is Samsung Electronics / SK Hynix (memory) → Samsung Electro-Mechanics (FC-BGA / MLCC / SiCap) → power (second-order).
SK Hynix is the strongest company in itself but carries a high valuation; Samsung Electronics has more catch-up optionality.

[Speculation]

Whether Samsung Electronics directly supplies HBM4E into Dell AI servers, and at what share, is not disclosed.
Samsung Electro-Mechanics’ share in Dell server FC-BGA / MLCC cannot be specified from public disclosures.

[Blocked]

Supplier identity and share by component within Dell’s AI-server BOM.
Samsung Electronics’ and Samsung Electro-Mechanics’ direct revenue exposure to Dell.
Real-time consensus EPS revision rates and NTM valuations for the Korean names.

Sources

Disclaimer: For research and information purposes only. Not personalized investment advice; this is public-information-based analysis. Names cited are for analytical illustration; readers should perform their own due diligence and consult licensed advisors before any investment decision.

Korea Quality Re-Rating Watch 2026-05-29: VM, Gigavis, DB Hi-Tek

Fri, 29 May 2026 16:30:00 +0900

Macro Dashboard

Indicator	Level	5-Day Δ	Signal
KOSPI	8,476.1	+8.0%	▲ Large-cap bid
KOSDAQ	1,074.8	−7.4%	▼ Weak
USD/KRW	1,506	−0.5%	→ Stable
VIX	15.8	−4.8%	● Calm
US 10Y	4.46%	−10 bps	→ Stable
Brent	$91.0	−8.6%	▼ Soft

Regime: Korea = Bear | US = Bull. The 15.4pp gap between KOSPI and KOSDAQ over five sessions is the headline divergence. US breadth (50MA: 57.2%) stays constructive while Korea’s remains thin (50MA: 22.7%, 200MA: 37.5%). Brent’s pullback reduces one inflation tail risk; the dollar is steady near 1,506.

Market Wrap

Today’s session looked strong at the index level and weak everywhere else. KOSPI held near 8,476, but only 439 of 2,487 tracked stocks advanced against 1,846 decliners. The ADR proxy printed 23.8 — deeply below neutral — and the average stock fell 1.22% on ₩27.9 trillion in total turnover. This was not a broad rally. It was a handful of large-cap moves lifting the headline number.

The dominant theme was a sharp re-rating in consumer electronics and IT services. LG Electronics (066570.KS, +29.93%) posted one of the largest single-day moves in the KOSPI on extraordinary ₩2.55 trillion volume; LG Innotek (011070.KS, +28.57%) followed on ₩1.88 trillion. Samsung SDS (018260.KS, +20.32%) and NAVER (035420.KS, +14.15%) also surged, with institutions stepping in behind both. Hyundai Mobis (012330.KS, +11.95%) and Samsung Electro-Mechanics (009150.KS, +15.04%) extended strength into the auto-component and electronics supply chain.

Institutional flow was decisive in the large caps. Samsung Electronics (005930.KS, +5.84%) attracted ₩1.67 trillion in institutional buying even as foreigners sold. Hyundai Motor (005380.KS, +6.79%) and NAVER drew simultaneous foreign and institutional buying — the cleaner co-buy signal on the day. In contrast, SK Hynix (000660.KS) absorbed ₩1.03 trillion in foreign net selling despite a modest +1.92% gain, with retail partially absorbing the supply.

Weakness was broad and indiscriminate outside the large-cap names. Display components, biotech, and smaller materials stocks dominated the decline list, with several shedding 12–21%. Sector rotation rather than macro panic — but breadth at 23.8 ADR leaves no ambiguity about the Bear-regime reading for Korea at the stock level.

Today’s Quality Re-Rating Candidates

Source: KR Meta Screener (2026-05-29) | Screener intersection of 101-ticker universe

The meta screener ranks candidates by combining quality fundamentals, institutional flow, cycle re-rating, earnings improvement, and post-earnings drift. Names clearing 3+ screeners are the preferred research candidates today.

Candidate Table

Rank	Ticker	Name	Meta Score	Screeners	ROE	OP YoY	Δ OPM	PER
1	089970.KQ	VM	106.0	5 — QC · SM · CR · SME · PEAD	14.8%	+387%	+29.3pp	30x
2	420770.KQ	Gigavis	97.9	4 — QC · SM · CR · SME	7.3%	+777%	+29.9pp	116x
3	000990.KS	DB Hi-Tek	88.1	5 — QC · SM · CR · SME · PEAD	11.8%	+45%	+3.0pp	28x
4	356860.KQ	TLB	77.2	4 — QC · SM · CR · SME	16.0%	+665%	+8.2pp	42x
5	009540.KS	HD Korea Shipbuilding	64.1	4 — SM · CR · SME · PEAD	16.3%	+172%	+7.4pp	12x
6	007810.KS	Korea Circuit	60.9	3 — SM · CR · SME	10.7%	+262%	+5.9pp	46x
7	353200.KS	Daedeok Electronics	47.3	3 — SM · CR · SME	5.3%	+336%	+3.4pp	100x

QC = Quality Compounder · SM = Smart Money Quality · CR = Cycle Rerating · SME = Smart Money Earnings

Top 3 in Focus

VM (089970.KQ) — Semiconductor Equipment, KOSDAQ VM manufactures deposition and etch equipment for semiconductor fabs. It is the only name today topping both the five-screener overlap list and the meta-screener ranking simultaneously. The numbers are striking: operating profit +387% YoY, margin expansion of +29.3pp, and a PEAD score that ranks #1 among 29 drift candidates. Foreign and quality-institutional buyers absorbed roughly ₩24.1 billion in retail selling over five sessions. The Q1 2026 quarterly report filed on DART (May 15) provides the most recent financial baseline. Key question: whether the margin step-up is structural or a one-quarter outsized project mix.

Gigavis (420770.KQ) — Automated Optical Inspection, KOSDAQ Gigavis builds AOI (automated optical inspection) systems used in PCB and panel quality control — a critical link in the electronics supply chain. Four screeners cleared; PEAD is the only framework it misses, likely because the drift window does not yet show a confirmed price reaction. Revenue doubled (+101% YoY) and operating profit expanded +777%, with a +29.9pp margin jump. Smart Money Quality ranks it #1 among nine names in its universe. The elevated PER (116x) reflects growth expectations already priced in; the next earnings disclosure on DART and consensus revision trajectory are the key monitors. An insider ownership filing was made on May 22.

DB Hi-Tek (000990.KS) — Analog/Power Foundry, KOSPI DB Hi-Tek is Korea’s largest dedicated analog and power semiconductor foundry. It cleared all five screeners — matching VM on coverage breadth — but at a more moderate inflection profile: ROE 11.8%, OP +45% YoY, margin +3.0pp. The stock fell -4.53% today alongside broad sector weakness, yet foreign buyers net-bought against quality-institutional selling — a flow split worth watching. Two items require verification before this candidate advances: (1) a DART filing today for a subsidiary operational suspension event and (2) the risk classification note from the Kiwoom surface overlay (quality institutional flow negative, short interest context). The 5-day foreign + quality-institutional net buy of ₩146.0 billion remains the supporting evidence.

Data sourced from Research OS local DB as of 2026-05-29. This post is systematic screener output for market analysis purposes only and does not constitute investment advice.

Samsung Electro-Mechanics at ₩138T Market Cap: An AI Components Play at Hyundai's Weight Class — Can It Command a Premium Over Murata and Ibiden?

Thu, 28 May 2026 18:10:00 +0900

📚 Samsung Electro-Mechanics Series Samsung Electro-Mechanics: Hybrid Challenger / ₩1.5T Silicon Capacitor Contract / Market Cap Crosses ₩100T / AI Server Passive Component Bottleneck / Samsung Electro-Mechanics Si-Cap and Intel EMIB-T Related Hubs: AI Substrate & PCB Hub / Korea Semiconductor Value Chain Hub

TL;DR

Samsung Electro-Mechanics is no longer a components stock you buy because it is cheap. At the May 28, 2026 closing price of ₩1,849,000, the common-share market capitalization stands at approximately ₩138.1 trillion. That puts it in roughly the same weight class as Hyundai Motor on a common-share basis and at 94% of Murata’s market cap of approximately ₩146.2 trillion. Ibiden, at approximately ₩51.8 trillion, is no longer a meaningful comparison point in size. Figures were cross-checked against the Research OS local database, yfinance closing prices as of May 28, 2026, and a JPY/KRW rate of 9.405.

Three conclusions follow.

There is a legitimate premium thesis for Samsung Electro-Mechanics. The company is no longer simply an MLCC supplier; it can be reclassified as a qualified merchant supplier of power-integrity components inside AI packages.
But it is already receiving a peer premium. On a P/B, P/S, and market cap basis, it is difficult to argue that Samsung Electro-Mechanics is undervalued relative to Murata or Ibiden.
A reasonable near-term ceiling is ₩150–165T, with a bull-case ceiling of ₩180–205T. Anything above ₩220T approaches momentum overshoot territory absent a 2027 consensus operating profit north of ₩4T, an additional ₩1T-class silicon capacitor win, or confirmation of a second hyperscaler customer.

The practical stance is: hold existing positions, abandon any prior plans to trim, do not chase at current levels. The ₩2.0–2.2M range is a first ceiling confirmation zone; ₩2.4–2.7M requires 2027 operating profit of ₩4T and repeat silicon capacitor orders to enter market consensus.

1. Where Samsung Electro-Mechanics Stands Now

1.1 Market Cap Comparison

All figures are based on the May 28, 2026 close or same-day snapshot.

Company	Price	Market Cap	KRW Equivalent	Key Interpretation
Samsung Electro-Mechanics	₩1,849,000	~₩138.1T	₩138.1T	Reached common-share weight class of Hyundai Motor
Murata	¥8,538	~¥15.54T	~₩146.2T	Global passive component leader
Ibiden	¥19,740	~¥5.51T	~₩51.8T	AI/HPC package substrate scarcity asset

Samsung Electro-Mechanics has closed to within 94% of Murata’s market cap. That matters. The market is no longer pricing Samsung Electro-Mechanics as a Korean electronic components stock; it has begun pricing it as a hybrid of a Murata-class AI passive platform and an Ibiden-class package scarcity asset.

The Hyundai Motor comparison carries the same message. The exact market cap arithmetic varies depending on whether preferred shares are included (note: the comparison here is on a common-share basis), but the implication is clear: Samsung Electro-Mechanics has entered the tier of Korea’s flagship large-cap manufacturers. From this level, the question shifts from “is there a theme attached?” to “can it attract the capital allocation flows reserved for Korea’s heavyweight industrials?"

1.2 Multiple Comparison

Surface-level multiples are already aggressive.

Company	PER	Forward PER	P/B	Interpretation
Samsung Electro-Mechanics	~175x	~113x	~14.2x	Difficult to justify on 2026 earnings alone
Murata	~67x	~36x	~5.7x	Passive leader premium
Ibiden	~92x	~64x	~10.0x	Advanced substrate scarcity premium

On a P/B basis, Samsung Electro-Mechanics already trades above both Murata and Ibiden. Saying “Samsung Electro-Mechanics is cheap relative to Murata or Ibiden” at today’s price is not accurate. The market, however, is looking at 2027–2028, not 2026.

If one adopts a Macquarie-style bull case of ₩4T operating profit for 2027, after-tax net income of roughly ₩3.0–3.2T becomes plausible. On that basis, the current ₩138T market cap implies approximately 43–46x forward 2027E net income. Still not cheap — but if Samsung Electro-Mechanics is reclassified as an AI package power-integrity platform, calling it an outright bubble is also hard to sustain.

2. Who Else Holds Silicon Capacitor References Outside Samsung Electro-Mechanics

To assess the premium thesis, one must first map the competitive reference landscape. Silicon capacitor references break into three tiers:

Product line existence
AI/HPC reference design or customer validation
Mass production, long-term supply contract, or confirmed customer program

Key axes based on publicly available information:

Company	Role	AI/HPC Reference Assessment	Confidence	Comparison to Samsung Electro-Mechanics
Murata	Merchant Si-Cap supplier	Murata silicon capacitors are listed in OCP Chiplets’ HPC reference PDN design; high-density Si-Cap products for mobile/HPC are publicly disclosed	High	The most direct incumbent. Samsung Electro-Mechanics must differentiate through large AI contract visibility, not claimed technical exclusivity. (OCP, Murata)
TSMC	Packaging platform (embedded)	CoWoS-L supports HPC package power management via LSI and embedded deep trench capacitors (eDTC)	High	Not a direct component peer but a platform-native alternative axis. May constrain the TAM ceiling for Samsung Electro-Mechanics. (TSMC 3DFabric)
Intel Foundry / EMIB	Packaging platform (embedded)	EMIB-M incorporates MIM capacitors within the bridge; EMIB-T adds TSVs for enhanced power delivery	Medium–High	Intel is a packaging platform provider, not a component supplier. EMIB-T proliferation validates the Si-Cap demand direction but does not yet confirm specific customer linkage. (Intel)
Empower / ADI	IVR + Si-Cap power delivery	Empower ECAP is an embedded silicon capacitor for AI/HPC processor package-level integration; ADI announced a $1.5B acquisition of Empower	Medium–High	Not a standalone capacitor but an IVR + Si-Cap + vertical power delivery system solution — the most threatening system-level competitive axis. (Empower, ADI)
AP Memory	S-SiCap / interposer Si-Cap	Announced S-SiCap Gen4 and interposer-integrated Si-Cap for AI server/HPC; claims customer packaging and reliability validation	Medium	A Taiwan-based niche player. Small in scale, but the high-density interposer-integrated direction cannot be dismissed. (AP Memory)
ROHM / TDK / Kyocera AVX / Yageo, etc.	Adjacent passive component suppliers	MLCC, polymer/tantalum, RF/optical Si-Cap for AI server applications	Low for in-package AI Si-Cap	Relevant for AI server passive components broadly, but not strong peers for in-package AI accelerator Si-Cap

In summary, Samsung Electro-Mechanics’ direct peer group is Murata, Empower/ADI, and AP Memory. TSMC and Intel represent “packaging platforms that embed or substitute silicon capacitors” — competitive threats, but simultaneously corroborating references for the broader AI packaging trend.

3. Why Samsung Electro-Mechanics Can Command a Premium

3.1 Reclassification from MLCC to In-Package PDN Component

Conventional MLCC is fundamentally a passive component mounted on a PCB. Silicon capacitors, by contrast, sit inside or immediately adjacent to GPU, HBM, and AI accelerator packages to suppress voltage transients and high-frequency noise.

Samsung Electro-Mechanics has secured a silicon capacitor supply contract of approximately ₩1.5 trillion, covering the period from January 1, 2027 through December 31, 2028. The company states that the product is embedded within high-performance semiconductor packages — such as those housing AI server GPUs and HBM — to stabilize power delivery. (Samsung Electro-Mechanics)

This distinction matters. Board-level MLCC is relatively substitutable. In-package components, however, are co-validated during ASIC and package design. Once designed in, customers face meaningful barriers to switching. If the market begins to view Samsung Electro-Mechanics as an AI packaging supply chain vendor rather than a cyclical passive components stock, a multiple premium becomes defensible.

3.2 Scarcity: A New Merchant Supplier Breaking into the Murata–TSMC Duopoly

Samsung Electro-Mechanics’ official disclosures note that the silicon capacitor market has historically been dominated by a small number of players due to high technical barriers and rigorous customer qualification requirements. The significance of the current contract lies less in “Samsung Electro-Mechanics has built the technology” and more in the fact that a new qualified merchant supplier has passed AI/HPC customer validation. (Samsung Electro-Mechanics)

For AI hyperscalers and ASIC developers, relying solely on Murata, TSMC’s captive CoWoS, or Intel’s captive EMIB may be insufficient for supply security. Samsung Electro-Mechanics entering as a validated merchant supplier aligns with customers’ second-source requirements.

3.3 A Potential Path from Marvell to a Long-Term Global Hyperscaler Contract

The Elec reported that Samsung Electro-Mechanics previously supplied silicon capacitors for AI accelerators to U.S. fabless company Marvell, and has since signed a large-scale contract with a separate global major customer. The customer name is undisclosed under NDA; industry sources characterize it as a large North American technology company. (The Elec)

The customer identity cannot be confirmed from public disclosures. But the investment-relevant observation is the visible trajectory: sample supply → AI fabless reference → long-term contract with a major global customer. If that trajectory holds, Samsung Electro-Mechanics’ Si-Cap business can be modeled as an earnings contributor rather than an option.

3.4 A Bundle Position Spanning FC-BGA, MLCC, and Si-Cap Simultaneously

According to Samsung Electro-Mechanics’ product materials, its silicon capacitors are available in land-side, top-side, and embedded configurations, with capacitance, thickness, size, and pad design customizable to customer requirements. (Samsung Electro-Mechanics)

This creates a differentiation argument relative to Murata. Samsung Electro-Mechanics holds MLCC and package substrate businesses simultaneously. In AI servers, where GPUs and CPUs draw large currents at low voltages, embedded and land-side MLCC and Si-Cap directly beneath and within packages are critical for reducing loop inductance.

The company’s own technical materials confirm that AI servers use substantially more MLCC than conventional servers, and that embedded and land-side MLCC configurations are evolving to minimize loop inductance. (Samsung Electro-Mechanics)

The premium thesis for Samsung Electro-Mechanics is therefore not solely about Si-Cap revenue in isolation.

Si-Cap → Entry into AI package interiors → Co-adoption of MLCC / embedded MLCC / land-side MLCC → Expanded customer touchpoints for FC-BGA and package substrates

If this chain is confirmed, Samsung Electro-Mechanics could earn an AI package solution multiple that exceeds even Murata’s passive component multiple.

4. Conditions Under Which the Premium Could Become Excessive

First, margins are undisclosed. Si-Cap is likely a high-value product, but actual gross margins, yields, depreciation, and customer-level ASPs have not been made public. Arbitrarily assuming high margins introduces meaningful risk.

Second, customer concentration risk is significant. The ₩1.5T contract is large, but it may represent high dependence on a single customer and a single program. If no additional customers materialize, the market may reassess it as a large one-off win.

Third, TSMC/Intel captive capacitor solutions could erode the merchant Si-Cap addressable market. TSMC CoWoS-L integrates eDTC within the packaging platform, and Intel EMIB-M/T continues to strengthen bridge-level capacitors and TSV-based power delivery. If customers conclude that platform-native solutions are sufficient, the TAM for external Si-Cap suppliers may be constrained. (TSMC 3DFabric, Intel)

Fourth, Murata’s incumbent reference position remains strong. Murata appears in OCP Chiplets as part of an HPC reference PDN design and maintains an accumulated technology base in high-density, low-ESL, ultra-thin silicon capacitors. For Samsung Electro-Mechanics to command a higher premium than Murata, what is needed is not evidence of being a Murata substitute but evidence of being “a second major supply axis alongside Murata.” (OCP)

5. Estimating the Market Cap Ceiling

5.1 Peer Parity Method

Scenario	Market Cap	Implied Share Price	Interpretation
Current	₩138T	₩1,850,000	Already at Hyundai Motor common-share weight class; 94% of Murata
Murata parity	₩146T	~₩1,960,000	First peer parity level
Murata +10% premium	₩161T	~₩2,160,000	Reasonable near-term ceiling
Murata +20% premium	₩175T	~₩2,350,000	Aggressive premium
₩2.5M sell-side high target	₩187T	₩2,500,000	Sell-side high target / momentum overshoot ceiling

On this framework, the near-term rational ceiling is ₩150–165T. Extending for additional contract expectations and momentum opens the range to ₩180–205T, but that zone already requires 2027 operating profit of ₩3.5–4.0T and repeat Si-Cap orders.

5.2 Required Earnings Back-Calculation

Market Cap	Net Income Required at 40x	Net Income Required at 50x
₩138T	₩3.45T	₩2.76T
₩160T	₩4.00T	₩3.20T
₩187T	₩4.68T	₩3.74T
₩205T	₩5.13T	₩4.10T
₩220T	₩5.50T	₩4.40T

Samsung Electro-Mechanics’ 2025 net income was approximately ₩0.73T. Even at current levels, the market cap already heavily front-runs a step-change in 2027–2028 earnings. Sustaining a market cap above ₩187T requires net income visibility of ₩3.7T or better, or operating profit guidance approaching ₩4T.

5.3 Summary: Market Cap Range

Zone	Market Cap	Implied Share Price	Assessment
Current	₩138T	₩1,850,000	Already in elevated valuation territory
Near-term rational ceiling	₩150–165T	₩2,000,000–2,210,000	Murata parity to +10% premium
Bull-case ceiling	₩180–205T	₩2,410,000–2,740,000	Requires 2027 OP of ₩3.5–4.0T and EPS upgrades across Si-Cap, FC-BGA, and MLCC
Stretch / Overheating	₩220T+	₩2,950,000+	Difficult to defend without an additional ₩1T-class contract win or 2027 OP of ₩4T entering market consensus

6. Practical Positioning

The current stance is as follows.

Maintain existing holdings
Abandon any prior plans to reduce positions
Do not chase at current prices
₩2.0–2.2M is the first ceiling confirmation zone
₩2.4–2.7M requires 2027 OP of ₩4T entering broad consensus

Samsung Electro-Mechanics can still move higher from here. But from this point, the game is no longer “it’s cheap.” It is holding through the waiting period while confirming that 2027–2028 earnings actually step up.

Conditions That Would Justify Additional Premium

The large Si-Cap contracts for 2027–2028 translate into actual revenue and earnings
Customer base expands beyond Marvell to two or more additional AI ASIC customers
Si-Cap adoption drives co-adoption revenue across MLCC, embedded MLCC, and FC-BGA
Merchant Si-Cap demand increases as EMIB-T/CoWoS alternative packaging proliferates
Yield, margin, and capacity data confirm “manufacturing at scale” rather than “sample-stage technology”

Invalidation Conditions

Si-Cap revenue recognition for 2027 is delayed
Si-Cap customer base remains a single customer
Murata, Empower, and AP Memory enter the same customer base as dual-source suppliers, driving rapid price compression
TSMC CoWoS-L/R eDTC becomes the standard for high-end AI packages, limiting the addressable area for external Si-Cap adoption
Si-Cap operating margins are confirmed to be below the company-wide average

7. Final Portfolio Manager Comment

The premium thesis for Samsung Electro-Mechanics is real. But that thesis rests not on “uniquely proprietary technology globally” but on being a publicly listed component supplier that was first to publicly demonstrate large-customer qualification at scale.

Murata is the incumbent. Empower/ADI is pursuing system-level power delivery. AP Memory is pushing high-density interposer capacitors. TSMC is building platform-native eDTC. Each of these represents a credible threat. For Samsung Electro-Mechanics to continue commanding a premium above Murata and Ibiden, what matters from here is not news flow but repeat orders and disclosed margins.

The final conclusion:

Samsung Electro-Mechanics cannot definitively claim absolute technology leadership over established passive component leaders such as Murata. However, as the first new merchant supplier to secure a large-scale, long-term contract for in-package AI server silicon capacitors, there is a sufficient basis for a multiple above its historical MLCC cyclical trading range. That said, today’s price already reflects a substantial portion of that thesis.

Evidence Classification

[Fact]

Samsung Electro-Mechanics closed at ₩1,849,000 on May 28, 2026; common-share market cap is approximately ₩138.1T. Source: Research OS local database and yfinance cross-check.
Murata closed at ¥8,538 on May 28, 2026; market cap approximately ¥15.54T, equivalent to approximately ₩146.2T at JPY/KRW 9.405. (Yahoo Finance, Yahoo Finance FX)
Ibiden closed at ¥19,740 on May 28, 2026; market cap approximately ¥5.51T, equivalent to approximately ₩51.8T. (Yahoo Finance)
Samsung Electro-Mechanics has signed a silicon capacitor supply contract of approximately ₩1.5T covering 2027–2028. (Samsung Electro-Mechanics)
Murata silicon capacitors are listed in OCP Chiplets’ HPC reference PDN design. (OCP)
TSMC CoWoS-L supports HPC package power management via eDTC. (TSMC 3DFabric)
ADI announced the acquisition of Empower Semiconductor for $1.5 billion. (ADI)
AP Memory announced S-SiCap and interposer-integrated Si-Cap for AI server/HPC applications. (AP Memory)

[Inference]

Samsung Electro-Mechanics’ current share price front-runs not just a single ₩1.5T contract but the prospect of a repeatable AI package power-integrity platform.
The Samsung Electro-Mechanics premium derives from customer qualification and large contract visibility, not claimed technical monopoly.
A premium over Murata may be sustainable due to Samsung Electro-Mechanics’ higher earnings sensitivity to AI server incremental exposure relative to its total business mix.
TSMC eDTC and Intel EMIB captive capacitors may constrain the TAM ceiling for Samsung Electro-Mechanics’ merchant Si-Cap business.

[Speculation]

Claims that Samsung Electro-Mechanics’ contract customer is a specific hyperscaler or AI ASIC company cannot be confirmed from public disclosures alone.
The assumption that Si-Cap operating margins significantly exceed the company-wide average has not yet been verified.
The assertion that Samsung Electro-Mechanics deserves a structurally higher long-term multiple than Murata is premature before additional customer wins are confirmed.

[Blocked]

Samsung Electro-Mechanics’ Si-Cap ASP, unit volumes, yield, cost structure, and operating margin.
Customer name, end-chip identity, and package platform name.
Whether Samsung Electro-Mechanics is the sole supplier.
Actual AI server customer-level market share for Murata, Empower, and AP Memory.
Hyundai Motor market cap comparisons vary depending on whether preferred shares are included and which total share count definition is applied by data providers; this article uses common-share basis only.

Samsung Electro-Mechanics Silicon Capacitors and Intel EMIB-T: The Component That Moved Inside AI Package Power Delivery

Thu, 28 May 2026 17:20:00 +0900

📚 Samsung Electro-Mechanics Series ₩1.5 Trillion Silicon Capacitor Contract / Understanding MLCC and Silicon Capacitors / AI Server Passive Component Bottleneck / Samsung Electro-Mechanics Breaks ₩100 Trillion Market Cap / Marvell Q1 and Korea Semiconductor Read-Through Related Hubs: AI Substrate & PCB Hub / Korea Semiconductor Value Chain Hub

One-page Korean infographic: AI packaging competition is expanding from a CoWoS-vs-EMIB-T platform battle into a broader ecosystem contest across substrates, bridges, capacitors, and EDA. Samsung Electro-Mechanics’ key position sits in FC-BGA, silicon capacitors, and the optional bridge/interposer ecosystem.

TL;DR

The essence of this story is that Samsung Electro-Mechanics has begun moving inside AI semiconductors.

Traditional MLCCs are capacitive components mounted across smartphones, automobiles, and server boards. But the latest packages housing AI accelerators and HBM experience current transients too large for board-level MLCCs to handle alone. Clamping the voltage fluctuations that arise directly beside the die requires placing fast-discharge energy storage components as close to the chip as possible. Samsung Electro-Mechanics’ silicon capacitors are designed precisely for this role.

Intel EMIB-T is a useful technical frame for understanding this shift. Intel describes EMIB as a 2.5D packaging technology using a small silicon bridge embedded in the substrate to connect logic-to-logic and logic-to-HBM dies, with EMIB-T adding TSVs to the bridge to reinforce power delivery and signal routing. As AI chips grow larger and HBM stacks multiply, power delivery networks inside packages become increasingly complex—making ultra-low-ESL components like silicon capacitors an inevitable priority. (Intel)

Customer identity, however, requires caution. What is officially confirmed is that Samsung Electro-Mechanics signed a silicon capacitor supply agreement worth approximately ₩1.5 trillion with a large global company, covering January 1, 2027 through December 31, 2028. Claims connecting this contract to Google TPU v8e, MediaTek, or Intel EMIB-T adoption come from industry media and sell-side analysts—not from official statements by Samsung Electro-Mechanics, Google, or Intel. (Samsung Electro-Mechanics)

The bottom line: this news has the potential to reshape Samsung Electro-Mechanics’ valuation story. The company has historically been viewed as an electronic components name driven by MLCC cycles, smartphone camera modules, and package substrate cycles. But if Si-Cap enters mass production from 2027 and the customer base expands to EMIB-T or other AI ASIC packages, Samsung Electro-Mechanics could be re-rated as a core supplier of AI infrastructure package power delivery components. Conversely, if the ₩1.5 trillion contract remains a single-customer, single-project event, or if customer ramp-up delays or dual-sourcing intensifies, current expectations could deflate quickly.

1. Why This Is Not an MLCC Story

Power delivery challenges in AI packages are not simply about high power consumption. The problem is that voltage droops when current demand changes abruptly. GPUs, AI ASICs, and HBM stacks demand large current bursts over extremely short time intervals. Voltage droop shrinks clock margins, increases the probability of errors, and forces conservative performance guardbands.

High-performance packages therefore place power stabilization components as close to the die as possible. Board-level MLCCs remain essential, but inside or immediately adjacent to AI GPU and HBM packages, thinner components with lower ESL and ESR are required.

Samsung Electro-Mechanics describes its silicon capacitors as follows: dielectric layers and internal electrodes are stacked on a silicon wafer; wafer grinding enables thinning to below 100 µm; and low-ESL characteristics are advantageous for power stabilization. Application configurations include Land-side, Top-side, and Embedded types—meaning the component is not simply a board-level part but one that can be placed inside the package design itself. (Samsung Electro-Mechanics)

This is the crux of the Samsung Electro-Mechanics investment thesis. The company has already accumulated deep expertise in high-temperature, high-voltage, and ultra-high-capacitance MLCC technology, and in FC-BGA it is moving toward high-layer-count, large-format substrates for AI accelerators and server CPUs. Adding Si-Cap to this portfolio extends the company’s “AI server power stabilization component” coverage from board-level MLCCs all the way into package-internal power delivery.

2. Why EMIB-T Matters

Intel EMIB replaces a full silicon interposer with a small silicon bridge embedded in the substrate, connecting multiple dies in a 2.5D packaging configuration. Intel describes EMIB as its solution for logic-to-logic and logic-to-HBM interconnection, with volume manufacturing dating back to 2017. (Intel)

The evolution to EMIB-T is more significant. Intel states that EMIB-M adds MIM capacitors inside the bridge to reinforce power delivery, while EMIB-T goes further by adding TSVs to address the vertical power delivery demands of growing HBM stacks. In other words, the bridge is evolving from a pure signal interconnect into a structure that also carries power delivery infrastructure. (Intel)

Synopsys characterizes EMIB-T in the same direction: TSV-based power delivery, backside bumping, denser MIM capacitors, and high-speed protocol routing. The motivation is that larger compute dies, next-generation HBM, and expanding AI accelerator clusters must simultaneously satisfy signal integrity and power delivery requirements. (Synopsys)

The investment point here is not “Intel EMIB will unconditionally succeed.” A more precise statement is:

As AI packages scale, the power delivery network inside the package—the PDN—becomes more sophisticated. EMIB-T is a concrete example of that trend, and silicon capacitors are a component category rising within it.

3. What Is Confirmed vs. What Remains an Estimate

Item	Judgment	Comment
Samsung Electro-Mechanics ₩1.5 trillion silicon capacitor contract	Fact	Samsung Electro-Mechanics disclosed a supply agreement of approximately ₩1.5 trillion with a large global company, covering January 1, 2027 through December 31, 2028. (Samsung Electro-Mechanics)
Si-Cap’s role in power stabilization inside AI server GPU and HBM packages	Fact	Confirmed in Samsung Electro-Mechanics’ official press release and product page. (Samsung Electro-Mechanics, 2)
EMIB as Intel’s 2.5D packaging for logic-to-HBM interconnection	Fact	Confirmed in Intel’s official packaging page and EMIB technology brief. (Intel, 4)
EMIB-T adding TSVs to enhance power delivery and signal routing	Fact	Intel and Synopsys descriptions are consistent. (Intel, Synopsys)
Ibiden’s capacity expansion for high-performance IC package substrates for AI and high-performance servers	Fact	Ibiden disclosed a ¥500 billion investment plan for FY2026–FY2028, with phased ramp-up and mass production from FY2027. (Ibiden)
Claims that Samsung Electro-Mechanics’ counterparty is Google and that the product will be used in TPU v8e	Unclear	Plausible as a market estimate, but not officially confirmed by Samsung Electro-Mechanics, Google, or Intel.
Claims that Ibiden’s investment is exclusively for EMIB-T and largely funded by a specific hyperscaler	Unclear	Ibiden’s official materials describe expansion for high-performance IC package substrates but do not identify the customer or confirm EMIB-T exclusivity.

The distinction between confirmed fact and market estimate must be maintained rigorously. The quality of this contract is high, but labeling it “confirmed Google win” immediately undermines credibility. The safer framing is:

What is officially confirmed is Samsung Electro-Mechanics’ Si-Cap supply agreement with a large global customer. The market is interpreting this in the context of Intel EMIB-T, Google TPU product lines, and broader AI ASIC package PDN advancement.

4. The Samsung Electro-Mechanics Thesis: MLCC + FC-BGA + Si-Cap

For Samsung Electro-Mechanics, this is not simply a story about one new component being added. Three business axes are converging into a single narrative.

Axis	Existing Position	Extension in AI Packaging
MLCC	Power stabilization for boards, servers, and automotive applications	Premium MLCC demand in AI servers and networking equipment
FC-BGA	High-layer-count, large-format substrates for server CPUs and AI accelerators	Die area and layer count climb as GPU, ASIC, and network ASIC complexity grows
Si-Cap	New high-value passive component	Die-near PDN component inside GPU, HBM, and AI ASIC packages

This combination matters. Samsung Electro-Mechanics does not design or manufacture chips. But for chips to function, power must be delivered stably, signals must travel without loss, and multiple dies and HBM stacks must be integrated inside a single package. Samsung Electro-Mechanics’ products sit in exactly that underlying infrastructure.

Si-Cap’s advantage is not just price. Package-internal components carry high customer qualification barriers, and once a component is designed in, switching costs are substantial—requiring re-validation of electrical characteristics, thermal behavior, reliability, and assembly yield for the specific package. The first large design win therefore carries reference value that exceeds the revenue it represents.

5. Value Chain: Who Captures What

Value Chain	Benefit / Impact	Investment Interpretation
Intel Foundry / Advanced Packaging	EMIB-T emerges as an alternative and complement to CoWoS bottlenecks	Even without catching TSMC on leading-edge logic, securing advanced packaging slots gives Intel a customer touchpoint for AI ASICs.
Ibiden and other high-performance substrate suppliers	Capacity expansion for IC package substrates serving AI and high-performance servers	Ibiden’s ¥500 billion plan signals that the substrate bottleneck is structural. However, EMIB-T-specific customer allocations remain unconfirmed. (Ibiden)
Samsung Electro-Mechanics	₩1.5 trillion Si-Cap contract officially confirmed	Customer identity undisclosed, but this is a strong signal that AI package-internal power stabilization components have entered commercial scale. (Samsung Electro-Mechanics)
Murata	Established leader in silicon capacitors and IPD	Murata holds an existing portfolio of 3D silicon capacitors and custom IPDs. The competitive and dual-sourcing dynamic with Samsung Electro-Mechanics warrants close monitoring. (Murata)
Synopsys and the EDA / package design ecosystem	Beneficiary of rising EMIB-T design complexity	Simultaneously validating chiplets, UCIe, HBM, PDN, thermal, and signal integrity raises the value of 3D-IC design flows. (Synopsys)
TSMC CoWoS	Competitive but near-term substitution is limited	EMIB-T is more likely to serve as a second path for AI ASIC customers amid CoWoS supply constraints than to immediately displace CoWoS.

6. Technical Moat: The Hard Parts of Silicon Capacitors

First, achieving ultra-low ESL and ESR. In AI packages, what matters is not just the capacitor’s intrinsic performance but the length of the current loop from capacitor to die. Placing components below the package, above the die, or embedded within the substrate to shorten current paths is the design challenge.

Second, co-qualification with the package. Silicon capacitors are not sold as standalone generic parts; capacitance value, thickness, pad layout, placement, and reliability requirements are all defined relative to the PDN design of a specific AI ASIC package. This is why Samsung Electro-Mechanics emphasizes customer qualification and technical barriers to entry. (Samsung Electro-Mechanics)

Third, yield and supply stability. Silicon capacitors are wafer-based components. The manufacturing philosophy differs fundamentally from traditional MLCC, and yield management is required across thinning, grinding, dicing, and package assembly.

Fourth, customer lock-in. Once a particular silicon capacitor is designed into an AI package PDN, replacing it with an alternative supplier requires re-validating electrical characteristics, thermal performance, reliability, and assembly yield from scratch. Early design-win incumbents enjoy a structural advantage.

7. Investment Judgment: Good News and Good Price Are Different Things

Samsung Electro-Mechanics’ direction is correct. Si-Cap addresses a power integrity bottleneck in AI packages, and the company has secured a meaningful first large-scale reference. But the equity market has already moved quickly to price this story.

The appropriate investment posture at this point is therefore Wait / Watchlist. The rationale is straightforward.

First, the ₩1.5 trillion contract, while substantial, is recognized over 2027–2028. That single contract alone is insufficient to justify a full re-rating of Samsung Electro-Mechanics’ market capitalization.

Second, the real alpha lies in repeatability. If this contract is limited to a single customer and a single project, re-rating momentum will be difficult to sustain. Conversely, if a second or third AI ASIC customer or a follow-on generation contract is confirmed during 2027, the narrative shifts from “component win” to “platform supplier.”

Third, margins have not been disclosed. Si-Cap is a premium component, but actual operating margins will depend on initial yield, customer pricing terms, process costs, and inspection costs.

The monitoring checklist:

Checkpoint	Why It Matters
Si-Cap revenue recognition begins in 2027	First evidence that the contract translates into the income statement
Customer diversification	Determines whether this is a repeatable platform or a single-project event
Placement configuration	Whether Top-side, Land-side, or Embedded affects ASP and lock-in intensity
Yield and capacity	Thin wafer-based products make initial yield the primary margin driver
MLCC and FC-BGA co-growth	Confirms whether the full AI power integrity portfolio is expanding, not just Si-Cap alone

8. Red Team: Where the Thesis Could Be Wrong

First, customer identity overreach. Linking this contract to Google TPU v8e, MediaTek, or Intel EMIB-T is a compelling narrative for investors, but none of it is officially confirmed. Google’s public TPU page describes the direction of its eighth-generation TPU but does not confirm Samsung Electro-Mechanics Si-Cap content or EMIB-T adoption. (Google Cloud)

Second, EMIB-T yield and schedule risk. EMIB-T must simultaneously improve power delivery and signal routing, making production ramp-up technically demanding. Even if the technology direction is sound, delayed customer production ramp means delayed component revenue.

Third, competitive risk. Murata is an established incumbent with an existing portfolio of silicon capacitors and IPDs. Samsung Electro-Mechanics winning the first large contract does not automatically lock in long-term market share. (Murata)

Fourth, TSMC CoWoS resilience. EMIB-T gaining traction does not mean CoWoS disappears. Real customers will evaluate yield, cost, HBM procurement, package turnaround time, and long-term capacity commitments across multiple options simultaneously.

9. Conclusion

This news illustrates a broader shift in AI semiconductor investment: the focus is expanding from front-end miniaturization to back-end power and signal integrity. EMIB-T is Intel’s packaging card to press into the CoWoS supply bottleneck; silicon capacitors are a component category newly rising within that dynamic.

The most realistic interpretation:

Conviction	Judgment
High	The importance of die-near and package-embedded power stabilization components in AI packages will continue to grow.
Medium	Silicon capacitor suppliers such as Samsung Electro-Mechanics and Murata can establish a new growth axis in the high-performance packaging component market from 2027 onward.
Low–Medium	The chain of evidence connecting Google TPU v8e or Amazon AI ASICs adopting Intel EMIB-T at scale, with a specific Korean supplier’s silicon capacitors inside, remains insufficiently confirmed.

In plain terms: building a great AI chip is no longer enough on its own. Integrating the chip and HBM into a single package—and delivering power to that package without voltage instability—now determines real-world performance. Silicon capacitors are the premium component that suppresses that instability directly beside the die.

If this story develops as anticipated, competition in AI packaging will expand beyond TSMC CoWoS vs. Intel EMIB-T to encompass the entire substrate, bridge, capacitor, and EDA co-design ecosystem that lives inside those packages. Samsung Electro-Mechanics has begun to move inside that ecosystem.

Evidence Classification

[Fact]

Samsung Electro-Mechanics signed a silicon capacitor supply agreement of approximately ₩1.5 trillion with a large global company, covering January 1, 2027 through December 31, 2028. (Samsung Electro-Mechanics)
Samsung Electro-Mechanics describes its silicon capacitors as power stabilization components placed inside high-performance semiconductor packages, including AI server GPUs and HBM. (Samsung Electro-Mechanics)
Samsung Electro-Mechanics silicon capacitors support thinning below 100 µm, low ESL characteristics, and Land-side, Top-side, and Embedded placement configurations. (Samsung Electro-Mechanics)
Intel EMIB is a 2.5D packaging technology for logic-to-logic and logic-to-HBM interconnection; EMIB-T adds TSVs to the bridge. (Intel)
Ibiden plans to invest ¥500 billion in FY2026–FY2028 to expand production capacity for high-performance IC package substrates serving AI and high-performance servers, with phased ramp-up and mass production beginning in FY2027. (Ibiden)

[Inference]

Samsung Electro-Mechanics’ core alpha lies less in Si-Cap unit revenue than in securing a reference position in AI package power integrity.
Silicon capacitors are best understood not as a wholesale replacement for MLCCs but as a premium layer added to handle the near-die, high-frequency domain that MLCCs cannot adequately reach.
The competition between EMIB-T and CoWoS intensifies the complexity of package-internal PDN, substrate, bridge, and EDA co-design rather than merely reprising a front-end process node race.

[Speculation]

Claims that Samsung Electro-Mechanics’ product will be adopted in Google TPU v8e or specific Intel EMIB-T packages have not been officially confirmed.
Whether Samsung Electro-Mechanics’ Si-Cap business will expand into repeat contracts beyond 2028 requires confirmation of additional customers and product generations.
Si-Cap margins may structurally exceed those of MLCC and FC-BGA, but actual yield outcomes and pricing terms are undisclosed.

[Blocked]

Identity of the contracting counterparty.
Per-product ASP, volume, cost structure, and yield.
Exact placement location within the final chip and package.
Whether the contract includes take-or-pay provisions or cancellation clauses.
Whether repeat contracts will be established beyond 2028.

Korea Quality Re-Rating Watch 2026-05-28: VM, SamCNS and Protek — Five Screeners

Thu, 28 May 2026 16:30:00 +0900

Macro Dashboard

Indicator	Level	5-Day Change	Signal
KOSPI	8,185.3	+4.7%	Bullish
KOSDAQ	1,104.4	-0.1%	Flat
USD/KRW	1,502.6	-0.1%	Stable
VIX	16.8	+0.3%	Calm
US 10Y	4.48%	-10 bps	Stable
Brent	$94.9	-8.4%	Falling

Regime: KR Bull / US Bull. Stance: active expansion. KOSPI’s five-day gain of 4.7% stands in contrast to KOSDAQ’s flat print — a 4.9 percentage-point large-cap divergence that has been building all week.

Market Wrap

KOSPI’s headline at 8,185 tells a cleaner story than the market actually delivered on May 28. Under the surface, breadth was poor: of 2,490 listed names, only 553 advanced against 1,709 decliners. The ADR proxy sat at 32.4, average stock change was -0.22%, and total turnover came in at 20.6 trillion KRW — a session of selective rotation, not broad participation.

The clearest flow story of the day was a rotation within large caps, out of semiconductor heavyweights and into components. Foreigners sold SK Hynix (000660.KS) to the tune of 1.89 trillion KRW and dumped Samsung Electronics (005930.KS) for another 1.56 trillion, but simultaneously bought Samsung Electro-Mechanics (009150.KS, +13.44%) and Samsung SDI (006400.KS, +7.30%) in size. LG Innotek (011070.KS, +8.62%) drew combined foreign and institutional interest. The read: overseas money is not leaving Korea’s tech complex — it is repositioning within it, shifting exposure from memory to component and battery names.

Institutions partially offset the foreign sell-down in SK Hynix, ranking it as their top net buy of the session. NAVER (035420.KS, +3.12%) drew joint institutional and foreign interest. On the other side, Korean institutions were decisive sellers of Samsung Electronics, Hanmi Semiconductor (042700.KS, -8.31%), LS Electric (010120.KS, -5.74%), and Jusung Engineering (036930.KS, -9.17%).

Defense and shipbuilding names pulled back: Hanwha Ocean (042660.KS, -7.93%), HD Hyundai Heavy Industries (329180.KS, -5.38%). The 8.4% five-day decline in Brent crude is a broadly supportive input cost signal for Korean manufacturers, but today it appeared to weigh on energy-adjacent names more than it lifted margins stories.

The session’s notable movers were concentrated in small-cap, high-beta names rather than quality compounders — a reminder that, even within a Bull regime, the current breadth environment (only 25.3% of stocks above their 50-day MA, 39.9% above the 200-day) demands careful stock selection over index-level positioning.

Today’s Quality Re-Rating Candidates

Four names swept all five screeners simultaneously today — Quality Compounder, Smart Money Quality, Cycle Rerating, Smart Money Earnings, and PEAD — a rare clean sweep that concentrates the research queue sharply.

Top Candidates Table

Rank	Ticker	Name	Meta Score	Screeners Hit	ROE	OP YoY	Margin Δ	Key Note
1	089970.KQ	VM (브이엠)	102.9	5/5	14.8%	+387%	+29.3pp	Smart money absorbing retail supply
2	252990.KQ	SamCNS (샘씨엔에스)	91.3	5/5	9.7%	+183%	+9.1pp	F+QI net selling — caution on flow
3	053610.KQ	Protek (프로텍)	89.9	5/5	13.7%	+244%	+12.2pp	F+QI net selling — caution on flow
4	039030.KS	Iotechnics (이오테크닉스)	79.5	5/5	8.3%	+159%	+11.5pp	Short interest 19.8% — monitor
5	420770.KQ	Gigavis (기가비스)	73.5	4/5	7.3%	+777%	+29.9pp	Missing Smart Money Quality leg
6	095610.KQ	TES (테스)	67.9	3/5	14.5%	+50%	+0.4pp	Supply contract filed May 21
7	000660.KS	SK Hynix	55.7	2/5	35.6%	+101%	+13.1pp	Treasury share disposal; heavy outflow
8	064850.KQ	FnGuide (에프앤가이드)	54.9	3/5	15.1%	+41%	+5.9pp	Consensus z-score +1.19

Top Three in Detail

#1 — VM (089970.KQ, 브이엠) is a semiconductor and display process equipment maker. Today it scores highest on both the meta screener (102.9) and the PEAD composite (+2.06, #1 in Tier A), reflecting operating income growth of 387% YoY on revenue that more than doubled (+106%). The margin expansion of 29.3 percentage points in a single year is the strongest in the candidate queue. The critical flow read: over the past five sessions, foreign and institutional buyers absorbed 22.75 billion KRW net while retail distributed 12.92 billion — the direction of flow is aligned with the quality signal. Two recent DART filings, neither flagged as risk. Next check: order backlog visibility and whether the operating leverage can sustain or whether Q1 was a peak beat.

#2 — SamCNS (252990.KQ, 샘씨엔에스) produces semiconductor packaging materials. Operating income grew 183% YoY, net income expanded 358%, and margin improved 9.1 percentage points. Smart Money Quality ranks it #1 and Smart Money Earnings ranks it #1 — the earnings-flow combination is exceptional. However, the flow nuance matters here: net foreign and institutional selling over five sessions totals -11.16 billion KRW, with retail stepping in to absorb (+3.34B). This is the reverse of VM’s flow structure. An IR event held May 21 (DART filing confirmed) may clarify the demand picture. The caution is real but does not disqualify the fundamental case; it demands that any entry price reflects the flow headwind.

#3 — Protek (053610.KQ, 프로텍) is a semiconductor bonding and packaging equipment supplier. Operating income +244% YoY, margin +12.2pp, and the name holds the second-highest Cycle Rerating rank (#2) in the full screener. Like SamCNS, the five-session flow shows net foreign and institutional selling (-9.04B), partially offset by retail (+3.29B). Three DART filings are on record, all classified neutral. The company sits at PER 15.9x with ROE 13.7% — relative to the quality of its earnings acceleration, that is inexpensive compared to peers at 30-80x in the same screener cohort.

Framework note: Any name appearing in five screeners simultaneously is rare by construction — today’s four-name clean sweep (VM, SamCNS, Protek, Iotechnics) is a concentrated signal, not a routine output. None of these mentions constitute a buy recommendation. They are research candidates that pass quality, flow, earnings cycle, and post-earnings drift filters as of May 28, 2026. Verify forward guidance, supply chain exposure, and valuation headroom before forming a view.

NVIDIA's Post-GTC 2026 Inference Stack: Why LPX and CMX Moved Ahead of CPX

Thu, 28 May 2026 15:30:00 +0900

📚 NVIDIA·Vera Rubin follow-up series NVIDIA Q1 FY27 and Korea AI Infrastructure / Vera Rubin VR200 BOM Cost Check / AI-RAN and the Korea Supply Chain / Marvell Q1 FY2027 and Korean Semiconductors

📚 Samsung Electronics · Korea semiconductor linked reads Samsung Electronics PER 15x Re-rating Thesis / Samsung Foundry Customer List / AI HBM Hub / Semiconductor Value Chain Hub

Beginner TL;DR

CPX is the accelerator that reads a long prompt for the first time. LPX is the accelerator that rapidly pulls out one token at a time.

Post-GTC 2026, NVIDIA’s message has shifted: the bigger bottleneck is not the one-time prefill read, but the continuously repeated decode and KV cache management. That is why the strategic focus moved from CPX as a standalone chip to an inference system built around Rubin GPU + LPX + CMX + Dynamo.

Conclusion: LPX and CMX moved to the front because the monetizable bottleneck in AI inference has shifted from “reading long context once” to “generating every token quickly and cheaply reusing that memory."

From a Korean semiconductor perspective, this should not be read as “weakening HBM demand.” It should be interpreted as an inference infrastructure value chain that spans HBM + high-speed networking + storage tiers + substrate/packaging + power/cooling.

0. The Simplest Analogy: AI Inference Is “Read, Then Speak”

When you send a question to an AI, two stages happen internally.

Stage	Plain-Language Explanation	Technical Term	Bottleneck
Stage 1	The AI reads the question and supporting material first	Prefill / Context phase	Ability to read long documents quickly
Stage 2	The AI speaks the answer one word at a time	Decode / Generation phase	Ability to generate one token at a time, quickly and stably
Memory	Notes taken while reading	KV cache	Where those notes are stored and how fast they can be retrieved

CPX was primarily aimed at Stage 1: reading long context for the first time. The Rubin CPX NVIDIA unveiled in 2025 was a GPU targeting 1M+ token context workloads — very long context processing — with 30 PFLOPS NVFP4 compute, 128 GB GDDR7, and 3× attention acceleration. (NVIDIA Developer)

LPX and CMX, by contrast, target Stage 2: answer generation and Memory: KV cache management, respectively. Per NVIDIA’s LPX documentation, the Rubin GPU handles prefill and decode attention, while LPX accelerates latency-sensitive operations such as FFN/MoE inside the decode loop. (NVIDIA Developer) CMX creates a dedicated KV cache tier between GPU HBM and conventional storage. (NVIDIA Developer)

0.1 Glossary

GTC 2026

GTC is NVIDIA’s most important developer and customer event for announcing its AI infrastructure strategy. The central message of GTC 2026 was an expansion from “a GPU company for AI training” to “a company selling the full AI factory operating system and hardware stack."

At GTC 2026, NVIDIA unveiled Dynamo 1.0 and described it as the distributed operating system for the AI factory — software that orchestrates GPUs, memory, and storage at the cluster level. (NVIDIA Newsroom)

Inference

Inference is the process by which an AI actually generates a response. AI infrastructure used to revolve around “GPUs for training a model.” But as real-world usage of products like ChatGPT, Claude, coding agents, and search agents has exploded, inference cost and latency now matter more than training.

This shift is significant for investors. Training is a CAPEX driven by a small number of large AI labs. Inference is an OPEX that accrues daily with every user request. That means cost-per-token, response latency, and power efficiency flow directly into service margins.

Token

A token is the smallest unit an AI reads or writes. In English it may be a word or word fragment; in Korean it may be a character, a word, or part of a word. When users feed in long reports, codebases, or video transcripts, token counts surge.

Revenue and cost in AI services ultimately converge on one question: “How many tokens can be processed, how cheaply, and how fast?”

Prefill / Context Phase

Prefill is the stage where the AI reads the question and any attached material for the first time. For example, if you submit a 200-page report and ask for a summary, the model first reads the entire document to build its internal state. That process is prefill.

CPX was originally aimed at this domain. NVIDIA’s own CPX description draws the distinction: the context phase is compute-bound, while the generation phase is memory bandwidth-bound. (NVIDIA Developer)

In plain language:

CPX = a dedicated reading-assist device that speeds up the initial read of long material.

Decode / Generation Phase

Decode is the stage where the AI generates its answer one token at a time. The AI does not produce a full response all at once. Even a short reply is constructed internally as a sequence of small pieces.

The key point is that decode is repetitive work. Long answers, coding agents, reasoning models, and tool-use agents continuously generate thousands to tens of thousands of tokens. Users experience this stage’s speed directly.

NVIDIA’s LPX documentation states that in interactive inference, time-to-first-token, tokens/sec per user, and tail latency are the core metrics. (NVIDIA Developer)

In plain language:

LPX = an accelerator that helps the AI speak quickly and without interruption, one word at a time.

KV Cache

KV cache is the intermediate memory where the AI stores what it has already read, so it does not need to recompute it.

Imagine a 30-minute conversation with an AI. If the model had to re-read and recompute the entire prior conversation from scratch with every new response, costs would explode. Instead, the model stores what it has already processed as a KV cache.

NVIDIA’s CMX documentation calls KV cache “inference context” and notes that in agentic systems it is reused like the model’s long-term memory. (NVIDIA Developer)

In plain language:

KV cache = the working memo where the AI remembers “what we talked about earlier.”

HBM

HBM is the ultra-high-speed memory bonded directly to the GPU. It is the most critical memory in AI training and inference — expensive, supply-constrained, and difficult to package.

There is an important misconception to dispel. LPX and CMX should not be seen as replacing HBM. LPX is an SRAM-based low-latency decode assist device; CMX is a storage tier for KV cache. HBM remains the core memory of the Rubin GPU.

The precise framing is:

LPX and CMX do not eliminate HBM demand. They offload certain bottlenecks that were crowding HBM, allowing HBM to focus on higher-value computation.

SRAM

SRAM is very fast but small-capacity memory. It can be accessed far more quickly than HBM on a GPU, but is difficult to produce in large volumes.

SRAM is the heart of LPX. Per NVIDIA’s official LPX materials, the Groq 3 LPX rack offers 256 chips, 128 GB total SRAM, 40 PB/s on-chip SRAM bandwidth, and 640 TB/s scale-up bandwidth. (NVIDIA Developer)

A quick sanity check:

Item	Formula	Value
LPX chip count	32 trays × 8 chips	256 chips
SRAM capacity	32 trays × 4 GB	128 GB

In other words, LPX is not a large-capacity memory device. It is closer to a device that uses small, very fast SRAM to reduce decode latency.

CPX

CPX is best understood as Context Processing X — a GPU specialized for processing long context. Its original rationale was clear: when an AI first reads a long codebase, long video, long report, or long research document, prefill cost grows. CPX was designed to accelerate that “first read.”

However, post-GTC 2026 public messaging put LPX and CMX more prominently forward than CPX. Tom’s Hardware noted that the Rubin CPX was absent from GTC 2026 keynote slides while Groq 3 LPU/LPX racks appeared, interpreting this as a signal that NVIDIA is more focused on the LPU side than on CPX — though it stopped short of concluding CPX was fully cancelled, noting it could remain as an off-roadmap product for certain customers. (Tom’s Hardware)

In summary:

CPX = an accelerator specialized for reading long material the first time. Its priority in current messaging appears to have declined, though a full cancellation is difficult to confirm.

LPX / LPU

LPX is a low-latency inference rack that attaches the Groq LPU to NVIDIA’s Vera Rubin platform. LPU stands for Language Processing Unit — a processor specialized for language model inference.

LPX has three core strengths:

Strength	Plain-Language Explanation
Low latency	Responses arrive quickly without stutter
Predictable execution	Variance in per-user response latency is reduced
SRAM-based high-speed processing	Small but extremely fast memory handles decode operations

NVIDIA describes LPX as designed to work in tandem with the Rubin GPU: the Rubin GPU handles prefill and decode attention, while LPX takes on latency-sensitive FFN/MoE operations during decode. (NVIDIA Developer)

In plain language:

Rubin GPU = the heavy-duty general-purpose engine. LPX = a high-speed auxiliary engine that boosts token generation throughput.

CMX

CMX is a dedicated memory/storage tier for KV cache.

In the agentic AI era, conversations, code, tool calls, search results, and task histories grow long — and KV cache grows with them. Keeping all KV cache in GPU HBM is too expensive. Offloading it to conventional SSDs or object storage is too slow. CMX sits in between.

Location	Plain-Language Analogy	Characteristics
GPU HBM	Notes on your desk	Fastest but expensive
CMX	The cabinet right beside you	Quite fast and much larger
Conventional storage	The warehouse	Large but slow

NVIDIA describes CMX as a G3.5 tier between GPU HBM and conventional storage, enabling KV cache to be shared and reused across pods to reduce the bottleneck for long-context and agentic inference. (NVIDIA Developer)

In plain language:

CMX = a dedicated cache warehouse that stores the AI’s conversation memory cheaply and quickly.

Dynamo

Dynamo is the traffic control system for NVIDIA’s inference infrastructure. No matter how good the GPU, LPX, CMX, KV cache, and storage are, efficiency collapses without knowing where to route a request, which cache to reuse, and which GPU already holds the relevant memory.

Dynamo orchestrates all of that. NVIDIA states that Dynamo 1.0 splits inference work between GPUs and lower-cost storage, and can route requests to the GPU that already holds the relevant KV cache for agentic AI and long prompts. (NVIDIA Newsroom)

In plain language:

Dynamo = the AI factory operating system that decides “send this request to that GPU — it already has the relevant memory.”

0.2 Why LPX and CMX Rather Than CPX

CPX Is Good at “Reading Once”; LPX Is Good at “Talking Continuously”

CPX targets prefill. It is useful when reading a long document for the first time.

But the bottleneck users actually feel in live AI services is often in decode. If a response comes back slowly, if a coding agent takes a long time to move to the next file edit, or if a voice assistant cuts out, users notice immediately.

NVIDIA’s LPX documentation itself states that workloads are shifting more toward decode due to longer reasoning outputs, prefix caching, and longer context. (NVIDIA Developer)

The logic for putting LPX ahead of CPX is therefore simple:

The speed of generating every token repeatedly is more directly tied to service quality and billing than the speed of reading something once.

In Agentic AI, KV Cache Becomes a Core Asset

A conventional chatbot answers one question and stops. An agent is different. A coding agent, for example, cycles through reading code, generating a patch, running tests, reading errors, revising, retesting, and reporting results.

Throughout that process, it must continuously remember prior context. KV cache is no longer a temporary scratchpad — it becomes the agent’s working memory.

NVIDIA’s CMX documentation notes that as long-context and agentic workflows grow, KV cache capacity requirements grow proportionally, and the ability to reuse and store that cache is essential to both performance and efficiency. (NVIDIA Developer)

That is why CMX matters:

The bottleneck in the agentic AI era is not just “computation” — it is “where you store the memory and how fast you can retrieve it.”

LPX Directly Improves User-Perceived Performance

In AI services, average throughput is not the only thing that matters. Users care about whether their answer arrives right now.

LPX reduces latency and jitter through SRAM, compiler-orchestrated execution, and deterministic execution. NVIDIA notes that the LPU’s deterministic execution helps maintain stable time-to-first-token and per-token latency. (NVIDIA Developer)

In plain language:

A GPU-only setup is like a large restaurant with high total cooking capacity but variable per-table wait times. LPX is a dedicated express lane for VIP orders.

CMX Keeps Expensive GPUs from Sitting Idle

GPUs are the most expensive component. If a GPU is waiting for data, money is leaking.

CMX keeps KV cache near the GPU and prefetches it, reducing GPU idle time and redundant recomputation. NVIDIA states that CMX targets up to 5× improvement in tokens-per-second and 5× power efficiency compared to traditional storage. (NVIDIA Developer)

Translated into investor language:

CMX is infrastructure that raises GPU CAPEX utilization and lowers cost-per-token.

For NVIDIA, “The Full System” Is a Stronger Lock-In Than “A Single Chip”

CPX is essentially an individual accelerator. LPX, CMX, and Dynamo together are how NVIDIA controls the entire AI factory.

Layer	Role
Rubin GPU	Large-scale compute, prefill, attention, training/inference general purpose
LPX / LPU	Low-latency decode acceleration
CMX	Dedicated KV cache context memory tier
Spectrum-X / NVLink	Data movement
BlueField-4 DPU	Storage/network I/O offload
Dynamo	Request routing, KV cache movement, GPU/memory orchestration

This structure is powerful because customers are not simply buying a GPU. They are optimizing their entire inference service operation the NVIDIA way.

0.3 CPX, LPX, and CMX in One Sentence Each

Term	One-Sentence Description	Analogy	Core Bottleneck
CPX	Accelerator for reading long context the first time	A speed-reading device for thick books	Prefill
LPX	Accelerator for generating one token at a time, quickly	A speech engine that delivers words fast and consistently	Decode latency
CMX	KV cache tier for storing and reusing the AI’s prior memory	A dedicated cabinet for conversation notes	KV cache capacity / movement
Dynamo	Operating system that schedules and routes all inference work	Air traffic control tower	GPU utilization / routing
Rubin GPU	Large-scale general-purpose AI engine	Main engine	Training, prefill, attention, general inference

0.4 Correcting the Most Important Misconceptions

Misconception 1: “LPX Replaces HBM”

Inaccurate. LPX does not replace HBM. It complements the decode latency domain where HBM-based GPUs are less efficient.

HBM remains the core memory of the Rubin GPU. LPX handles small, fast tasks using SRAM. CMX partially extends KV cache storage beyond HBM.

The accurate framing is:

LPX and CMX do not eliminate HBM demand. They concentrate HBM on higher-value computation and distribute surrounding bottlenecks elsewhere.

Misconception 2: “CPX Has Become Irrelevant”

This is uncertain. The unresolved question is whether CPX was fully cancelled or merely deprioritized on the public roadmap.

The verification paths are NVIDIA’s latest official roadmap, GTC 2026 keynote slides, the next earnings call, or customer system announcements. Based on current public reporting, CPX was absent from GTC 2026 keynote slides and roadmap presentations while LPX was prominent — but the possibility of CPX remaining as an off-roadmap product for specific customers has not been ruled out. (Tom’s Hardware)

The conservative framing is:

Rather than “CPX fully cancelled,” the safer statement is “the strategic front-line message has moved to LPX and CMX.”

Misconception 3: “A Better GPU Is All You Need for Inference”

Not anymore. Inference bottlenecks are a combined function of GPU compute, HBM, SRAM, KV cache, storage, networking, and routing software.

That is precisely why NVIDIA’s emphasis post-GTC 2026 is not the standalone GPU but the AI factory stack.

0.5 Key Takeaway for Non-Technical Readers

The old AI infrastructure race was “who has the most powerful GPU?” Training was the focus: building large models required massive GPUs and HBM.

Now the battlefield is shifting. As AI moves into live services, billions to trillions of tokens must be processed every day. Users leave if responses are slow. Companies see margins erode if cost-per-token is too high. AI agents must continuously remember long conversations and task histories.

In this environment, a single fast GPU is no longer sufficient.

First, the AI must read long material for the first time — a task for CPX or the Rubin GPU. Second, the AI must generate answers quickly, one token at a time — the domain of LPX. Third, the AI must store prior conversation and task state without recomputing it — the domain of CMX. Fourth, someone must decide which GPU receives which request and which KV cache gets reused — the domain of Dynamo.

The essence of NVIDIA’s post-GTC 2026 strategy is therefore:

A shift from “sell more GPUs” to “design and operate the entire AI inference factory.”

In this light, CPX may be a fine product, but it is too narrow to anchor a strategy. CPX excels at a specific interval — processing long context the first time. LPX and CMX, by contrast, target the repetitive, expensive bottlenecks of real AI services: decode latency and KV cache reuse.

For investors, this shift matters. The beneficiaries of AI inference infrastructure can no longer be explained by HBM alone. HBM remains critical, but the analysis must simultaneously cover high-speed networking, DPUs, SSDs/storage, CXL/memory tiers, substrates, packaging, power, and cooling. The AI inference stack is transitioning from a “GPU-centric single bottleneck” to a “composite bottleneck across memory, networking, storage, and software."

The conclusion of this piece is simple. The directional thesis is correct. But the language must be precise. Post-GTC 2026, NVIDIA’s inference strategy is not about selling a bigger Vera Rubin GPU — it has moved to a heterogeneous AI factory combining Vera Rubin GPU/CPU, Groq 3 LPX/LPU, BlueField-4 STX·CMX, Spectrum-X/SPX, and Dynamo. CPX was originally a GPU targeting long-context prefill/context phase, but GTC 2026’s official front-line message brought forward the LPX·STX·CMX combination. However, there is still insufficient public language to conclude that NVIDIA has officially cancelled CPX.

Key Summary

[Fact] At GTC 2026, NVIDIA presented the Vera Rubin platform as a rack-scale AI factory comprising a Vera Rubin NVL72 GPU rack, Vera CPU rack, Groq 3 LPX inference accelerator rack, BlueField-4 STX storage rack, and Spectrum-6 SPX Ethernet rack. (NVIDIA Newsroom)
[Fact] CPX was introduced in 2025 as a GPU targeting 1M+ token long-context processing and context phase acceleration, with 128 GB GDDR7, 30 PFLOPS NVFP4, and attention acceleration as its core messages. (NVIDIA Developer)
[Fact] LPX is a different product with a different character. NVIDIA states that LPX handles latency-sensitive FFN/MoE execution and speculative decoding draft generation inside the decode loop, while the Rubin GPU continues to handle prefill, decode attention, and verification. (NVIDIA Developer)
[Inference] LPX is therefore not HBM-bearish. Rubin GPU/HBM handles large memory and attention; LPU/SRAM complements the low-latency decode path. CMX/STX creates the KV cache storage tier on top of that.
Korea read-through is widest for Samsung Electronics. HBM4·SOCAMM2, Groq LPU foundry, and PCIe Gen6 eSSD/KV-cache are bundled within a single company. SK하이닉스 has a cleaner HBM beta, but the incremental alpha in this piece is not “HBM alone” — it is the full memory hierarchy.

1. Verdict on Each Hypothesis

Claim	Verdict	Comment
Post-GTC 2026, NVIDIA’s front-line strategy shifted to a combination of Vera Rubin GPU + Groq 3 LPX/LPU + storage/networking	Largely correct	The official Vera Rubin platform announcement bundled GPU, CPU, LPX, STX, and SPX into one platform. Closer to pod/rack orchestration than “GPU-only scaling.”
CPX was a product targeting the prefill/context bottleneck	Correct	NVIDIA’s CPX documentation explicitly targeted long-context context phase and 1M+ token workloads.
The monetizable bottleneck is decode rather than prefill	Conditionally correct	For coding agents, voice assistants, and multi-turn agentic workflows, decode latency and tail latency are more closely tied to billing and perceived quality. However, for long-document ingestion, full codebase analysis, and batch summarization, prefill/context remains a high-value bottleneck.
NVIDIA may have prioritized LPX over CPX	Strong inference	LPX·STX·SPX moved to the front in GTC 2026 messaging and CPX stepped back from the main stage. However, “official CPX cancellation” remains [Blocked].
CPX’s role was absorbed into the Vera Rubin + LPX + CMX/STX combination	Partially correct	Parts of CPX’s context role are distributed across the Rubin GPU and the CMX/STX KV cache tier, while the low-latency decode that CPX could not address directly is now handled by LPX. It is not a 1:1 replacement.
Groq LPX/LPU complements the decode weakness of HBM GPUs rather than replacing HBM	Correct	The Rubin GPU is an HBM-based large-memory/attention engine; the LPU is an SRAM-based low-latency token engine.
Samsung Electronics entered as a Groq LPU manufacturing partner	Largely correct	Samsung Semiconductor stated that Jensen Huang mentioned Samsung’s Groq LPU manufacturing role at GTC 2026. Specific LP30 volumes, margins, and yields have not been disclosed. (Samsung Semiconductor)

2. Vera Rubin Is Not a Single GPU — It Is a POD-Scale AI Factory

NVIDIA’s official message is not “the Vera Rubin GPU is fast.” The more important development is that NVIDIA decomposed the AI factory into five rack-scale systems and reassembled them.

System	Role	Investment Implication
Vera Rubin NVL72 GPU rack	Pretraining, post-training, prefill, decode attention, verification	HBM4 and GPU compute remain the mainstream.
Vera CPU rack	CPU orchestration for agentic AI workloads, coherent memory, host-side scheduling	CPUs are revalued as an orchestration layer within the AI rack.
Groq 3 LPX inference accelerator rack	Low-latency decode FFN/MoE, draft generation, deterministic token path	An attempt to price the tail latency of premium interactive inference.
BlueField-4 STX / CMX storage rack	KV cache storage, context memory tier, cache reuse	A structure to move KV cache costs — which had been crowding GPU HBM — down to pod-level storage.
Spectrum-6 SPX / Spectrum-X fabric	Deterministic fabric among GPU, LPU, storage, and DPU	Rack utilization and data movement become the bottleneck, not just the chip.

NVIDIA’s moat in this structure is not GPU FLOPS alone. NVIDIA is attempting to capture the full token economics — prefill cost, decode latency, KV cache reuse, networking jitter, watts/token, rack utilization — in a single integrated offering. This shift should be read not as “CPX dropped out” but as “NVIDIA decomposed inference into finer-grained components and created a monetization point at each layer.”

3. Role Decomposition: CPX, LPX, and CMX

Treating CPX and LPX as the same category of “inference chip” creates confusion. They target different bottlenecks.

Dimension	CPX	LPX/LPU	CMX/STX
Basic character	GDDR7-based context GPU	SRAM-based low-latency decode accelerator	BlueField-4-based context memory / KV cache storage tier
Core bottleneck	Long-context prefill, context phase, attention-heavy input processing	FFN/MoE and pointwise operations inside the decode loop, speculative decoding draft generation	KV cache storage, movement, and reuse as multi-turn and long-context workloads grow
Key resources	128 GB GDDR7, 30 PFLOPS NVFP4	256 LPUs, 128 GB SRAM, 40 PB/s SRAM bandwidth (per LPX rack)	Flash/storage + DPU + Spectrum-X + DOCA/Dynamo
Relationship to GPU/HBM	Context-dedicated GPU alongside the Rubin GPU	SRAM decode tier complementing Rubin GPU/HBM	External context memory tier complementing GPU HBM
Investment interpretation	The solution to “context is too long”	The solution to “interactive token latency is where the money is”	The solution to “KV cache is eating into HBM”

NVIDIA’s LPX technical blog draws the division fairly clearly. The Rubin GPU handles long-context prefill, decode attention, and high-concurrency inference. LPX handles latency-sensitive token generation, FFN/MoE expert execution, and the draft path of speculative decoding. LPX is therefore not a chip that kills HBM; it is an auxiliary engine that accepts the small-batch, low-latency decode path that HBM-based GPUs handle less efficiently. (NVIDIA Developer)

4. Strengths and Limits of “Decode Is the Monetizable Bottleneck”

This statement is more than half right. Decode is close to the monetization bottleneck in particular for the following workloads:

Agentic coding assistants
Multi-agent workflows
Voice interaction
Real-time translation
Enterprise copilots with high tool-calling loop frequency
Premium AI services demanding long reasoning outputs

Users do not directly feel prefill throughput. But time-to-first-token, tokens/sec/user, and tail latency are immediately felt. And in agentic workflows, a single-call delay compounds across dozens of model calls. This is why LPX moved to the front.

However, it is an overstatement to say decode is always more valuable than prefill. In long-context RAG, full codebase analysis, large-document processing, video understanding, and batch summarization, the cost of ingesting the input context and building the KV cache remains significant. CPX existed because that bottleneck was real. The more precise post-GTC 2026 framing is:

For high-value interactive and agentic inference, decode latency has emerged as the monetization bottleneck, and NVIDIA is addressing it with LPX/LPU. However, in long-context AI, prefill and KV cache movement remain core bottlenecks — addressed by the Rubin GPU and the CMX/STX storage/networking tier.

5. Investment Read-Through

NVIDIA: From GPU Company to Token Factory OS

NVIDIA’s long-run logic is less about “a faster GPU” and more about “more inference attachment.” When an LPX, BlueField-4, Spectrum-6, CMX, and Dynamo attach to every Vera Rubin rack alongside the GPU, NVIDIA collects a toll at more layers of the AI factory.

The bull case is clear. What customers need is not a chip benchmark — it is token throughput, low tail latency, watts/token, and utilization. NVIDIA is positioning itself as the company that sells all four as a single rack/POD bundle.

The counter-argument also exists. Google TPU, hyperscaler custom ASICs, and specialist inference chips such as Cerebras are all trying to reduce the NVIDIA tax. And if LPX and CMX fail to demonstrate the TCO improvement claimed by the vendor in production workloads, the attachment narrative weakens. NVDA’s next checkpoint is therefore not just GPU revenue — it is the LPX·CMX·Spectrum attach rate per Rubin rack.

Samsung Electronics: From Memory Cycle Play to Inference Memory Hierarchy Supplier

Samsung Electronics has the widest Korea-listed exposure to this shift, for four reasons:

HBM4/HBM4E: The large-memory tier of the Vera Rubin GPU.
SOCAMM2: Vera CPU and AI server memory architecture.
Groq LPU foundry: AI logic manufacturing option for the SRAM decode tier inside LPX.
PCIe Gen6 eSSD / KV cache storage: The context memory tier that CMX/STX opens.

Samsung Semiconductor disclosed that at GTC 2026 it showcased HBM4E, HBM5 architecture, SOCAMM2, and PM1763 PCIe 6.0 SSD, and that Jensen Huang mentioned Samsung’s Groq LPU manufacturing role. (Samsung Semiconductor) Samsung Electronics’ 1Q26 earnings materials also referenced HBM4 and SOCAMM2 mass product sales for the Vera Rubin platform and PCIe Gen6 SSD development. (Samsung Newsroom)

The Samsung Electronics thesis is therefore too narrow if framed only as “HBM laggard catching up.” The broader statement is:

Samsung Electronics may be reclassified as an inference memory hierarchy supplier with simultaneous exposure across HBM4, SOCAMM2, LPU foundry, and KV cache SSD within the NVIDIA inference stack.

This thesis demands evidence, however. LPU yield and margin, HBM4E customer acceptance, SOCAMM2 shipment volume, and actual KV cache attachment for PCIe Gen6 eSSD all need to be confirmed.

SK하이닉스 · Micron: HBM Winners, but Incremental Alpha Is Narrow Here

LPX is not HBM-bearish. It keeps Rubin GPU/HBM at the center of the premium inference stack while delegating only low-latency decode to SRAM LPUs. The HBM thesis for SK하이닉스 and Micron is therefore not impaired.

The new alpha generated in this piece, however, is not “HBM is good.” That is already consensus. The new alpha is the addition of an SRAM decode tier, a KV cache storage tier, and a rack networking tier above and below HBM. SK하이닉스 has the cleaner pure HBM beta, but Samsung Electronics has the wider architectural read-through.

삼성전기: Not the Primary Subject, but Power Integrity Read-Through Holds

The LPX/CMX architecture is not just a story about adding more GPUs. As the number of chip types inside a rack increases and low-latency paths run in parallel with high-bandwidth memory paths, the importance of power integrity, high-speed substrates, SiCap/MLCC, and FC-BGA is maintained.

삼성전기 is not the protagonist of this piece. But just as the Marvell earnings confirmed the custom XPU·optical·scale-up networking thesis, NVIDIA’s heterogeneous AI factory signals that the “small bottlenecks next to the GPU” in the Korean components supply chain may continue to command premium pricing.

6. Checklist

Maintaining this thesis requires verifying the following items in order:

Checkpoint	Significance
Groq 3 LPX H2 2026 availability	Confirms whether LPX moves from slide to actual deployment
Samsung LPU mass production	Whether Samsung Foundry secures an AI inference logic reference win
HBM4E sample/customer acceptance	Samsung Memory penetration rate into the next Vera Rubin platform
SOCAMM2 shipment continuity	Whether the CPU/agentic AI memory architecture converts to revenue
PCIe Gen6 eSSD and CMX/STX adoption	Whether the KV cache storage tier translates to real sales
CPX follow-on roadmap	Whether CPX is on hold, a niche product, or set to reappear
Dynamo/AFD production benchmark	Whether heterogeneous decode actually lowers TCO

Final Judgment

The user’s thesis is directionally correct. Post-GTC 2026, NVIDIA is decomposing inference from GPU-only scaling into HBM GPU + SRAM LPU + KV cache storage + high-speed networking + orchestration software.

The safest language is:

Post-GTC 2026, NVIDIA’s inference strategy has expanded from homogeneous scaling centered on the Vera Rubin GPU to a heterogeneous AI factory combining Vera Rubin NVL72 + Groq 3 LPX/LPU + BlueField-4 STX/CMX + Spectrum-X/SPX. The existing Rubin CPX was a GDDR7-based context GPU targeting the long-context prefill/context bottleneck, but GTC 2026’s official platform messaging placed the LPX and KV cache storage/networking tier at the front. However, whether CPX has been officially cancelled cannot be concluded from public materials alone.

The more important investment statement is this:

LPX is not an HBM replacement. The HBM GPU continues to handle large models, large context, attention, and verification. LPX complements the small-batch, low-latency decode path where the GPU is less efficient. This change is therefore not HBM-bearish — it is a signal that the AI inference memory hierarchy has become more complex.

Evidence Classification Appendix

[Fact]

The NVIDIA Vera Rubin platform comprises Vera Rubin NVL72, Vera CPU, Groq 3 LPX, BlueField-4 STX, and Spectrum-6 SPX racks. (NVIDIA Newsroom)
CPX was announced in 2025 as a GPU for 1M+ token context workloads. (NVIDIA Developer)
LPX was presented as a rack-scale inference accelerator with 256 LPUs, 128 GB SRAM, 40 PB/s on-chip SRAM bandwidth, and 640 TB/s scale-up bandwidth. (NVIDIA Developer)
The Groq-NVIDIA deal is a non-exclusive inference technology licensing agreement; Groq continues to operate independently. (Groq)
Samsung showcased HBM4E, SOCAMM2, and PM1763 PCIe 6.0 SSD at GTC 2026 and was mentioned as a Groq LPU manufacturing partner. (Samsung Semiconductor)

[Inference]

The interpretation that LPX·CMX/STX·SPX were prioritized over CPX in GTC 2026 front-line messaging is well-supported.
LPX is more likely to complement the utilization and premium inference economics of Rubin GPU/HBM than to replace HBM demand.
Samsung Electronics’ investment thesis lies in the combination of HBM4 + SOCAMM2 + LPU foundry + eSSD/KV cache rather than in HBM4 alone.

[Speculation]

The claim that CPX has been fully cancelled or formally absorbed into LPX+CMX has not been confirmed.
Groq LP30/LPU-specific volumes, ASP, wafer allocation, and gross margins cannot be verified from public materials alone.
Whether LPX and CMX will reproduce NVIDIA’s vendor-claimed perf/W, revenue uplift, and TPS improvements in production workloads remains unverified.

[Blocked]

Official cancellation status of CPX.
Groq 3 LPX rack ASP and per-customer order quantities.
Samsung Foundry LPU yield, wafer price, and margin contribution.
Per-customer KV cache storage attachment for CMX/STX and actual TCO improvement.

This piece should be used for research and commentary purposes only and does not constitute investment advice. Product roadmaps, yields, customer adoption, pricing, and revenue recognition are subject to change even after public disclosures and company announcements.

Marvell Q1 FY2027 and Korean Semiconductors: The Bottleneck Is Interconnect, Substrate, and Power — Not Just HBM

Thu, 28 May 2026 10:20:00 +0900

📚 Context for follow-up posts This is a follow-up to Marvell & Broadcom Pre-Earnings: Checking Korea AI Semiconductor Bottlenecks. The preview asked whether the single-bet on HBM was broadening into AI ASICs, networking, and power stabilization — this post revisits that question against Marvell’s Q1 FY2027 results. Related hubs: AI HBM Hub, AI Substrate & PCB Hub, Korea Semiconductor Value Chain Hub.

TL;DR

The key takeaway from Marvell Q1 FY2027 is not the EPS beat. It is that the company raised its AI data center growth trajectory for FY2027–FY2028, attributing the upside to custom XPUs, optical interconnects, Ethernet switching, DCI, scale-up/scale-across networking, and XPU attach.

Translating this into Korean semiconductors, the answer is not simply “buy more HBM.” HBM remains the primary thesis, but the incremental alpha confirmed by these results is in the interconnect, substrate, power integrity, and test bottlenecks surrounding the GPU.

Priority order:

Priority	Korea Read-Through	Key Names/Group	View
1	FC-BGA + AI server MLCC + silicon capacitors	Samsung Electro-Mechanics (SEMCO)	Most direct. Already re-rated, so SiCap/FC-BGA margin verification is the next step
2	HBM4, SOCAMM2, eSSD, advanced packaging	Samsung Electronics, SK Hynix	HBM beta intact. New alpha is in memory attach and packaging beyond core HBM
3	Custom ASIC/XPU test sockets & interfaces	ISC, Leeno Industrial, TSE	Structural tailwind, but watchlist until direct revenue is confirmed
4	High-speed PCB/MLB for AI networking	ISU Petasys, Daeduck Electronics, Simtech, etc.	Selective. Not all “AI PCB” exposure carries equal benefit

For Marvell itself, the call is HOLD / Buy watch. Reference price $198.70, 12-month target $225, implying roughly +13% upside. The growth path is strong, but the stock already reflects elevated expectations. For Korean investors, what matters more than Marvell’s own valuation is where the bottlenecks Marvell identified are flowing next.

1. What Actually Matters in the Marvell Print

Marvell reported Q1 FY2027 results after the U.S. close on May 27, 2026. The official figures are as follows. (Marvell)

Item	Q1 FY2027	Commentary
Revenue	$2.418B	+28% YoY, +$18M above guidance midpoint
GAAP EPS	$0.04	Depressed by Celestial AI / XConn acquisition accounting
Non-GAAP EPS	$0.80	Near consensus
GAAP gross margin	52.1%	Improved YoY
Non-GAAP gross margin	58.9%	Held in the high-58% range despite expanding AI mix
Operating cash flow	$638.8M	Record operating cash flow
Q2 revenue guidance	$2.70B ±5%	Range: $2.565B–$2.835B
Q2 non-GAAP EPS guidance	$0.93 ±$0.05	Next-quarter profitability baseline

The print itself was roughly a “modest beat / EPS in-line.” What matters for the stock and for the Korea semiconductor read-through is not a few cents in Q1 — it is the FY2027–FY2028 revenue trajectory management described on the call.

Based on third-party call transcripts, management indicated FY2027 revenue of approximately $11.5B and FY2028 revenue of approximately $16.5B, and raised the FY2027 interconnect growth outlook from roughly 50% to above 70%. These figures come from unofficial transcripts rather than the formal IR filing, but the direction is consistent with the language in the company’s official press release. Marvell’s official release explicitly named the sources of growth as 800G/1.6T optics, 51.2T Ethernet switches, scale-up optical for NPO/CPO, DCI modules, custom XPUs, and XPU-attach. (Marvell, StockAnalysis transcript)

The one-sentence summary:

Marvell confirmed in numbers that AI data center capex is shifting from buying GPUs to building cluster interconnect infrastructure.

2. The Core Story Is Interconnect Architecture, Not SOCAMM

The most dangerous misreading of this earnings call for Korean investors is to reduce it to a “SOCAMM theme.” SOCAMM matters, but it is not the center of gravity here.

The actual hierarchy is as follows:

Marvell Growth Pillar	Why It Matters	Korea Semiconductor Translation
Custom XPU / custom ASIC	Signal that hyperscalers are expanding proprietary AI silicon beyond NVIDIA GPUs	HBM customer diversification, package substrates, test sockets
Optical interconnect	800G/1.6T, DCI, scale-up optics are the binding constraint for AI cluster expansion	High-speed PCB/optical is selective; SEMCO FC-BGA and power stabilization are structural
Ethernet switching	51.2T, 100T, 200T roadmap means rising dollar content in AI networking silicon	FC-BGA for network ASICs, high-speed boards, inspection/test
XPU attach	CXL, NIC, memory attach, inference KV cache exposure	Samsung Electronics SOCAMM2, eSSD, server DRAM; memory IP options such as OpenEdges
NVLink Fusion	Custom silicon coexisting inside the NVIDIA ecosystem	Supply chain expansion rather than an NVIDIA-vs-ASIC binary

The NVIDIA-Marvell announcement reinforces the same direction. NVIDIA invested $2 billion in Marvell; Marvell supplies NVLink Fusion-compatible custom XPUs and scale-up networking. The two companies also collaborate on silicon photonics and AI-RAN. (Marvell NVIDIA)

The implication is straightforward:

An AI data center is not a GPU box. It is a system in which GPUs, custom XPUs, HBM, optics, switches, retimers, CXL, NICs, FC-BGA substrates, and MLCCs move as a single organism.

Korean semiconductor investing must therefore shift the frame from “should we hold just the large-cap memory names” to “which bottlenecks below memory are translating into numbers.”

3. Korea Read-Through Priority 1: Samsung Electro-Mechanics (SEMCO)

The cleanest Korean translation of the Marvell print is Samsung Electro-Mechanics. Three reasons:

First, every growth pillar Marvell highlighted — custom XPUs, Ethernet switches, optical interconnects, XPU attach — requires high-speed, high-density packaging and substrates. That maps directly to FC-BGA demand.

Second, AI servers draw large instantaneous currents at low voltages. Suppressing voltage fluctuations around GPU, HBM, and XPU packages requires power stabilization components: MLCCs and silicon capacitors.

Third, Samsung Electro-Mechanics already holds both layers. In its Q1 2026 earnings, the company reported Package Solution revenue of KRW 725.0 billion, up 45% YoY and 12% QoQ, citing expanding supply of high-value substrates for AI accelerators, server CPUs, and networking applications. (Samsung Electro-Mechanics Q1)

Additionally, Samsung Electro-Mechanics signed a silicon capacitor supply contract worth KRW 1.5570 trillion with a large global company. The contract runs from January 1, 2027 through December 31, 2028. The component is described as improving power supply stability inside AI server GPU and HBM packages. (Samsung Electro-Mechanics SiCap)

The investment thesis for Samsung Electro-Mechanics has now shifted:

Old Frame	New Frame
Smartphone MLCC & camera module cyclical	AI package power integrity + FC-BGA platform
Mobile demand recovery	AI accelerator, server CPU, network ASIC demand
Commodity MLCC cycle	High-capacitance, low-ESR, ultra-thin, high-reliability MLCC/SiCap mix
Comparison vs. Ibiden or Murata individually	Hybrid comparison: MLCC + FC-BGA + SiCap

This conclusion is not the same as “buy at any price.” Samsung Electro-Mechanics has already seen meaningful re-rating on the silicon capacitor contract and the AI infrastructure narrative. The next confirmation variables are therefore not price momentum but gross margin, Package Solution OPM, SiCap production yield, and further customer diversification.

4. Samsung Electronics & SK Hynix: HBM Beta Intact, New Alpha Is at the Periphery

The Marvell print is not negative for HBM — if anything, the opposite is true. Even as custom XPUs and scale-up networking grow, memory dollar content per cluster does not shrink. As AI models evolve toward agentic AI, reasoning, and mixture-of-experts architectures, data movement and memory requirements grow larger, not smaller.

From an investment standpoint, however, “HBM is positive” is already the consensus thesis. What the Marvell call added incrementally:

The HBM customer base is broadening from a single NVIDIA GPU structure toward hyperscaler custom XPUs.
XPU attach connects CXL, NIC, memory attach, and KV cache, extending influence to server DRAM, SOCAMM, and eSSD.
As AI clusters scale, the value of the packaging, substrates, and power integrity that link memory and compute rises alongside them.

Samsung Electronics carries optionality across multiple dimensions here. HBM4, HBM4E, SOCAMM2, PM1763 SSD, foundry, and advanced packaging all sit within the same AI infrastructure stack. At NVIDIA GTC 2026, Samsung presented HBM4E, SOCAMM2, and the PM1763 SSD as part of a collaborative AI infrastructure product family with NVIDIA. (Samsung Semiconductor)

SK Hynix remains the purest HBM beta. Looking solely at the Marvell print, however, the new incremental alpha is larger in Samsung Electro-Mechanics, test sockets, and high-speed substrates growing alongside HBM than in SK Hynix itself. SK Hynix is the primary beneficiary, but it is also where the most market attention is already concentrated.

5. Test Sockets: The Quiet Beneficiary of Custom ASIC Proliferation

The importance of custom revenue in the Marvell call is not about chip volume alone. The key is SKU count and qualification cycles.

In a world where AI accelerators are standardized around a single NVIDIA GPU, test component demand is relatively predictable. Conversely, as hyperscaler-specific custom XPUs, XPU attach, CXL, NICs, switch ASICs, and DPUs proliferate, test conditions and socket designs fragment further.

In this scenario, test sockets and interface components can see three simultaneous tailwinds:

Variable	Direction	Reason
Volume	Increasing	More custom ASIC, network ASIC, and memory attach SKUs
ASP	Rising	Higher pin counts, faster signal speeds, higher power test complexity
Replacement cycle	Structural	Generational transitions and per-customer qualification repeats

The Korean names on the watchlist are ISC, Leeno Industrial, and TSE. Conviction should be held lower here, however. Whether Korean test socket companies are directly embedded in Marvell’s or a specific hyperscaler’s custom XPU chain cannot be determined from public disclosures alone. The current assessment is therefore “possible beneficiary” — not “confirmed customer mapping."

The metrics to track in quarterly results: AI/HPC logic revenue, new customer count, high-value socket mix, and OPM defense.

6. Generic PCB Is Not a Blanket Buy

The Marvell print is positive for AI networking and optical interconnects. But the inference that “AI networking is good → all PCBs are good” is dangerous.

The benefit concentrates in companies that satisfy the following conditions:

Ability to process low-loss materials for high-speed signal integrity.
Exposure to high-layer-count MLB or high-value package substrates.
Qualified supplier status with AI server or network equipment customers.
Confirmed ASP, layer count, and area expansion — not just volume growth.

More GPUs and XPUs do not translate into proportionally more substrate units. In an architecture where multiple chips integrate more densely onto a single high-performance package and board, what matters is substrate area, layer count, material difficulty, and yield — not unit shipment volume.

Grouping ISU Petasys, Daeduck Electronics, Simtech, TLB, and Korea Circuit into one undifferentiated basket is therefore imprecise. The actual read-through from Marvell’s results is closer to: “be selective — identify names with high-layer-count substrates for networking and materials qualified for high-speed signal environments.”

7. MRVL Valuation: A Great Company and a Great Entry Price Are Not the Same Thing

The standalone view on Marvell is HOLD / Buy watch.

Item	Detail
Reference price	$198.70, regular-session close 2026-05-27
12-month price target	$225
Implied upside	~+13.2%
Valuation framework	FY2028E EV/Sales
Core view	Growth path raised, but valuation already elevated

The base-case target price calculation:

Target price = (FY2028E Revenue × Target EV/Sales − Net Debt) ÷ Diluted shares

Assumptions: FY2028E revenue $16.5B, target EV/Sales 12.5×, net debt ~$1.117B, diluted shares 915M. This produces a target of approximately $224, rounded to $225.

For Marvell to re-rate toward Broadcom-level multiples, three things are required:

Custom silicon revenue must be confirmed as a recurring program — not a single-customer event.
Non-GAAP gross margin must hold in the 58–59% range even as interconnect and switching grow.
Supply-chain prepayments must convert into actual revenue ramp and free cash flow.

In short: Marvell has become a great company. Whether it is a great entry price is a separate question.

8. Next Checkpoints

Checkpoint	Strong Signal	Weak Signal
Q2 FY2027 revenue	At or above $2.835B (guidance upper end)	Below $2.70B midpoint
Data Center revenue	High-teens or better sequential growth	Sequential growth deceleration
Non-GAAP GM	59.25%+ or upper-end defense	Below 58.25%
Interconnect	FY2027 +70%+ outlook maintained or raised	800G/1.6T ramp slowing
Custom XPU	FY2028 2×+ growth, FY2029 $10B+ visibility	Per-customer ramp delays
Scale-up switching	Tier-1 customer volume production confirmation	Many engagements but no revenue
Korea read-through	SEMCO Package Solution, SiCap, FC-BGA numbers confirmed	Theme strong but margins/orders absent

Per-name Korea confirmation variables:

Name/Group	What to Track
Samsung Electro-Mechanics	Package Solution growth rate, AI network FC-BGA revenue, SiCap yield/margin, additional long-term contracts
Samsung Electronics	HBM4 customer qualification, SOCAMM2 actual shipments, eSSD pricing/volume, foundry/packaging customers
SK Hynix	HBM4 ramp, customer diversification, 2027 supply overhang risk
ISC, Leeno Industrial, TSE	AI logic/test socket revenue, new customers, high-value mix, OPM
PCB/MLB	AI network customer qualification, ASP expansion, low-loss materials and higher layer counts

9. Invalidation Conditions

The conditions under which this thesis weakens are clear:

Marvell Q2 revenue comes in below the $2.70B midpoint and Data Center growth decelerates.
Non-GAAP gross margin falls below 58.25%, confirming that custom/interconnect mix is diluting margins.
The FY2028 $16.5B revenue outlook is revised lower.
Custom XPU and XPU attach are tied up in specific customer schedule delays.
Samsung Electro-Mechanics SiCap is confirmed as low-margin revenue, or FC-BGA growth decelerates.
AI/HPC logic revenue growth fails to appear in test socket company results.
HBM lead times shorten and 2027 supply overhang signals emerge.

Final Interpretation

Marvell Q1 FY2027 is not a signal to “just buy HBM” for Korean semiconductors. The more precise message is:

As AI clusters scale, the bottleneck migrates down from the GPU toward interconnects, substrates, power stabilization, and test.

From this vantage point, the most direct Korean name is Samsung Electro-Mechanics. SEMCO’s MLCC, FC-BGA, and silicon capacitor businesses all sit inside the same AI package bottleneck. Samsung Electronics and SK Hynix remain central as the core HBM plays, but the incremental alpha that Marvell’s results newly surfaced is larger at the HBM periphery. ISC, Leeno Industrial, and TSE are second-order beneficiaries of the custom ASIC buildout — but watchlist is the right posture until direct revenue is confirmed.

The conclusion is not to buy indiscriminately. Even a sound thesis ends as a theme unless it is validated in the numbers. After this quarter, the variables to watch in Korean semiconductors are not price but Package Solution revenue, SiCap margin, AI logic test socket revenue, and high-speed substrate ASP.

Fact / Inference / Speculation / Blocked

[Fact]

Marvell Q1 FY2027 revenue was $2.418B, +28% YoY; Q2 revenue guidance is $2.70B ±5%. (Marvell)
Marvell Q1 non-GAAP gross margin was 58.9%; Q2 non-GAAP gross margin guidance is 58.25–59.25%. (Marvell)
Marvell named 800G/1.6T scale-out optics, 51.2T Ethernet switches, scale-up optical, DCI modules, custom XPUs, and XPU-attach as growth drivers. (Marvell)
NVIDIA invested $2 billion in Marvell; Marvell announced it will supply NVLink Fusion-compatible custom XPUs and scale-up networking. (Marvell NVIDIA)
Samsung Electro-Mechanics Q1 2026 Package Solution revenue was KRW 725.0 billion, up 45% YoY and 12% QoQ. (Samsung Electro-Mechanics Q1)
Samsung Electro-Mechanics signed a silicon capacitor supply contract worth KRW 1.5570 trillion; contract period is January 1, 2027 through December 31, 2028. (Samsung Electro-Mechanics SiCap)

[Inference]

It is reasonable to map Marvell’s growth axes to Korean names in the following order: SEMCO FC-BGA/MLCC/SiCap, Samsung Electronics and SK Hynix memory attach, test sockets, and high-speed substrates.
HBM remains the primary thesis, but the incremental alpha newly surfaced by these results is larger in the interconnect, packaging, power integrity, and test layers than in the large-cap memory names.
Samsung Electro-Mechanics’ re-rating should be interpreted as a transition to an AI infrastructure passive/substrate platform — not a recovery in conventional MLCC demand.

[Speculation]

Whether Samsung Electro-Mechanics’ SiCap customer is linked to a specific North American hyperscaler or AI accelerator supply chain is speculated by the market, but the counterparty has not been disclosed publicly.
Whether domestic Korean test socket companies are directly embedded in Marvell’s or any hyperscaler’s custom XPU program cannot be determined from public disclosures.
AI-RAN could create long-term opportunities in Korean RF and network semiconductors, but it is premature to treat it as a near-term earnings catalyst in 2026.

[Blocked]

Direct Korean supplier status in Marvell’s custom XPU, optical, and switch programs by company.
Customer identity, per-product margin, and monthly ramp pace for Samsung Electro-Mechanics’ SiCap contract.
AI logic revenue breakdown by customer for ISC, Leeno Industrial, and TSE.
Real-time NTM PER, EV/EBITDA, and consensus EPS revision rates for Korean PCB and substrate names as of current 2026 data.

Sources

Korea Quality Re-Rating Watch 2026-05-27: SK Square, Gigavis & HD Hyundai Heavy

Wed, 27 May 2026 16:30:00 +0900

Macro Dashboard

Indicator	Level	5-Day Δ	Signal
KOSPI	8,228.7	+14.2%	Bull
KOSDAQ	1,133.1	+7.3%	Bull
USD/KRW	1,498	−0.1%	Stable
VIX	16.9	+0.7%	Calm
Brent	$94.3	−8.1%	Declining
US 10Y	4.49%	−0.08pp	Easing

Both Korea and US regimes are confirmed Bull. KOSPI’s five-day advance of +14.2% reflects a sustained large-cap recovery; KOSDAQ’s +7.3% points to broadening participation in growth names. VIX at 16.9 is calm. Brent’s −8.1% five-day slide reduces cost-push pressure across Korean industrials and manufacturers. The overall stance is aggressive expansion.

Market Wrap (most recent session data: May 22)

The May 22 session was best described as selective risk-on with notable stock-level dispersion rather than a uniform broad rally. KOSDAQ surged over +4%, outpacing KOSPI’s modest +0.5%, as money rotated into smaller-cap growth stories across pharma/biotech, power/energy, and semiconductor component makers. KOSDAQ breadth reached 36.0% of names above their 50-day MA with 99 names clearing the full quality screen — improving, but still a stock-picker’s market rather than a rising-tide session.

The single strongest performer was Samsung Electro-Mechanics (009150.KS), up +11.3% on the day and +32.7% over five sessions. A Morgan Stanley note circulating across financial channels raised its forecast for high-performance ABF substrate shortage rates from 15% to 22% by 2030, directly reinforcing the investment thesis for both advanced PCB substrate and semiconductor packaging component names. Daeduck Electronics (353200.KS) rose +3.4% on the same thematic read, though foreign and institutional flows split — a divergence worth watching before treating it as confirmed smart-money entry.

On the other side of the ledger, foreign investors sold roughly −1.01 trillion won in Samsung Electronics (005930.KS) on the day, accompanied by program selling of approximately −998 billion won. This came despite Mirae Asset naming both Samsung Electronics and SK Hynix (000660.KS) as H2 re-rating candidates in its latest top-picks note. The simultaneous sell pressure indicated the market needed more confirmation before committing to the memory re-entry narrative — a reminder that broker upgrades and actual capital allocation are different things.

Sector flow summary: biotech/pharma, power infrastructure, and semiconductor equipment components were the strong axes. Large-cap memory semiconductors, telecoms, and prior-session surge names faced profit-taking. Discovery-level names flagged by the daily screener included Jusung Engineering (036930.KS, +20.95%, foreign and institutional co-buy) and Symtec (036070.KS, +5.02%, ABF/PCB axis co-buy with both foreign and institutional participation).

Today’s Quality Re-Rating Candidates (screener data: May 27)

The meta-screener ranks SK Square (402340.KS) first with a score of 84.8 — a meaningful gap above the field. The raw screener-overlap table, by contrast, places HD Hyundai Heavy Industries (329180.KS), EO Technics (039030.KQ), and Hanwha Ocean (042660.KS) first on four screener hits each. The meta-screener order is preferred here: HD Heavy and Hanwha Ocean both miss the Quality Compounder layer, carry debt ratios of 180% and 226% respectively, and have elevated short interest. The meta framework penalizes that profile, pulling their composite scores below names with smaller but structurally cleaner businesses.

Rank	Ticker	Name	Meta Score	Overlaps	Screeners	Key Metrics
1	402340.KS	SK Square	84.8	3	QC · SMQ · SME	ROE 31.9%, OP YoY +124.8%
2	420770.KQ	Gigavis	73.0	3	QC · CR · PEAD	OP YoY +777%, margin +29.9pp
3	064850.KQ	FnGuide	66.1	3	QC · CR · PEAD	ROE 15.1%, OPM 30.1%, consensus z +0.98
4	329180.KS	HD Hyundai Heavy	64.8	4	SMQ · CR · SME · PEAD	OP YoY +189%, 5d F+QI +₩164bn
5	089970.KQ	VM Company	62.3	3	QC · CR · PEAD	OP YoY +387%, margin +29.3pp
6	042660.KS	Hanwha Ocean	62.0	4	SMQ · CR · SME · PEAD	OP YoY +391%, 5d F+QI +₩196bn
7	092460.KS	Halla IMS	59.2	2	QC · CR · PEAD	ROE 13.8%, OPM 16.7%
8	000660.KS	SK Hynix	57.8	2	QC · CR	ROE 35.6%, OPM 48.6%, 5d F+QI −₩2,152bn
9	039030.KQ	EO Technics	56.5	4	SMQ · CR · SME · PEAD	margin +11.5pp, 5d F+QI +₩61bn
10	082920.KQ	Vitzrocell	55.2	2	QC · CR · PEAD	ROE 17.2%, OPM 28.5%, supply contract filed May 26

QC = Quality Compounder · SMQ = Smart Money Quality · CR = Cycle Rerating · SME = Smart Money Earnings

SK Square (402340.KS) — Meta #1, Score 84.8

SK Square is the investment holding entity for SK Telecom’s technology assets, with SK Hynix as the primary NAV driver. The quality case is unusually strong for a holding company: ROE of 31.9%, operating profit up +124.8% YoY, and a debt ratio of just 9.1%. What elevates it today is the flow structure — retail investors distributed approximately −194.9 billion won over five days while foreign capital and quality institutional accounts absorbed that supply (+11.5 billion won net). A share buyback completion report was disclosed via DART on May 13 (a concrete return-of-capital catalyst). This combination of quality fundamentals, clean balance sheet, active buyback, and smart-money absorption is the highest-priority configuration in the framework. Next check: Does the retail-to-institutional rotation persist, and does the NAV discount to SK Hynix’s recent rally narrow on volume?

Gigavis (420770.KQ) — Meta #2, Score 73.0

Gigavis manufactures automated optical inspection equipment used in semiconductor and display processes — a direct beneficiary of rising advanced packaging and HBM capex cycles. The earnings data is striking: operating profit up +777% YoY, revenue more than doubled (+100.6%), and margin expanded 29.9 percentage points. The PEAD composite score of +2.20 is the highest in the entire Tier A PEAD universe for this session, suggesting the market has not yet fully re-priced the earnings inflection. It clears Quality Compounder (#5), Cycle Rerating (#12), and PEAD (#1 rank). The 52-week high gap stands at 13.2%, leaving meaningful drift room. Next check: Volume confirmation as the stock approaches prior highs, and whether the margin expansion is durable into the next quarter.

HD Hyundai Heavy Industries (329180.KS) — Meta #4, Score 64.8

Korea’s largest shipbuilder hits four screeners on the strength of operating profit up +188.9% YoY and a margin expansion of +6.7pp. Cumulative five-day foreign and quality institutional buying totals +164.1 billion won as retail investors supplied −99.7 billion won — a classic smart-money absorption pattern. Four official DART catalyst filings (including a preliminary earnings disclosure) landed in the 21-day lookback window. This name fits a momentum/earnings re-rating frame rather than a pure quality compounder angle. Key caution: Short interest at 17.9% is the highest among the top-ten candidates, and the debt structure is leveraged. Position sizing discipline is warranted.

Screener data sourced from KR Meta Screener, KR Quality Compounder, KR Smart Money Quality, KR Cycle Rerating, KR Smart Money Earnings, and KR PEAD engines as of 2026-05-27. Market session data reflects 2026-05-22 close, the most recent available at time of writing. This post is market analysis only and does not constitute investment advice or reflect any specific portfolio holdings.

Korea ADR at 67: Why the Index Can Hold While Most Stocks Are Weak

Wed, 27 May 2026 16:25:00 +0900

This note follows the complex risk-off and recovery-trigger framework, KOSPI foreign ownership versus Samsung and SK Hynix, Korea foreign-investor flow analysis and the National Growth Fund / KOSDAQ smart-money map. Those pieces looked at macro gates, KOSPI mega-cap flows and KOSDAQ policy capital. This one asks a simpler question: how healthy is the tape beneath the index?

TL;DR

Korea’s 20-day ADR fell from 113.1 to 67.3 over the last month. KOSPI fell from 116.3 to 68.5 and KOSDAQ from 111.6 to 66.7. The average stock is already in a correction even if the index still holds.
This is not broad risk-on. It is narrow leadership concentrated in AI infrastructure bottlenecks, MLCC / FC-BGA / SOCAMM / back-end semis, and selected shipbuilding / defense / power names.
The next trade is not chasing first-line leaders. It is watching for ADR recovery + rising turnover + first foreign / institutional flow into second-line candidates. The local screen points to HPSP, SFA Semicon, Hana Micron, Dongjin Semichem and KMW.

Source: Research OS local DB prices_daily + universe.csv. Data cut is the May 26, 2026 close. May 27 intraday data is not included.

1. What ADR Tells Us

ADR measures market breadth, not index level.

The formula is simple.

Daily ADR = advancing stocks / declining stocks × 100
20D ADR = sum of advancing stocks over 20 sessions / sum of declining stocks over 20 sessions × 100
Advance ratio = advancing stocks / total stocks × 100

An ADR of 100 means advancing and declining stocks are balanced. Below 80, declining stocks are clearly dominating. If ADR sits in the 60s or 70s while the index remains firm, the market is usually being carried by a few large caps or a narrow set of themes.

That is Korea now.

Market	2026-04-16 20D ADR	2026-05-26 20D ADR	Change	Read
KOSPI	116.3	68.5	-47.8p	Large-cap breadth has weakened
KOSDAQ	111.6	66.7	-44.9p	Small/mid-cap breadth has cooled sharply
All Korea	113.1	67.3	-45.8p	The average stock is already weak

The important point is that this is not only a KOSDAQ problem. KOSPI’s 20-day ADR is also down to 68.5. So the right read is not “KOSDAQ is weak but KOSPI is healthy.” It is this:

Korean market breadth has narrowed, and surviving capital is compressed into AI infrastructure, shipbuilding, defense and a few other leadership pockets.

2. Why the Late-May Bounce Is Not Enough Yet

On May 20, only 333 stocks rose while 2,082 fell. Daily ADR was 16.0 and the advance ratio was 13.5%. That was close to a short-term capitulation session.

The May 21-22 rebound was strong.

Date	Advancers	Decliners	Daily ADR	20D ADR	Advance Ratio	Read
2026-05-20	333	2,082	16.0	58.3	13.5%	Short-term capitulation
2026-05-21	1,695	720	235.5	61.9	68.6%	Technical rebound
2026-05-22	2,097	283	742.2	67.9	86.1%	Strong rebound
2026-05-26	749	1,660	45.1	67.3	30.2%	Breadth weakened again

The problem is May 26. After the strong rebound, decliners rose again to 1,660 and the advance ratio fell back to 30.2%.

So the current state is not confirmed breadth expansion. It is a bottoming attempt in breadth that is still being tested.

That links directly to the earlier macro risk-off framework. The condition there was that oil, long rates, the dollar, KRW, Chinese credit and foreign flows needed to stabilize together. The ADR data says that recovery has not yet spread through the market.

3. Current Regime: Narrow Leadership

The current Korean equity regime is Narrow Leadership / Selective Risk-On.

Item	Read
Market breadth	Weak. All-market 20D ADR is 67.3
Leadership	Very strong in AI infrastructure and selected shipbuilding / defense
Trading difficulty	High. Stock selection matters more than index direction
New exposure	Prefer pullbacks and second-line names over chasing first-line leaders
Portfolio stance	Hold relative-strength leaders, replace weak positions, avoid using all cash before breadth recovers

The mistake is to say “the index is holding, so the whole market is fine.” With ADR in the 60s, most stocks are not fine. The opposite mistake is to say “breadth is bad, so avoid everything.” That is also wrong, because capital is still very active in a narrow set of sectors.

The surviving groups are:

AI infrastructure bottlenecks: Samsung Electro-Mechanics, Jeju Semiconductor, Daeduck Electronics, Simmtech, Haesung DS, Hana Micron, HPSP
Memory mega-caps: SK Hynix, Samsung Electronics
Shipbuilding / defense / nuclear-SMR: HD Hyundai Heavy Industries, Hanwha Ocean, Doosan Fuel Cell
Power / optical / network: selected power cable, RF and optical names

The market is weak, but leadership is not. That is not a contradiction. When breadth collapses, capital often crowds even harder into the few themes that still work.

4. Where Leadership Actually Was

The last-month leadership list is compressed into AI infrastructure and shipbuilding / defense.

Returns are %, turnover and flows are KRW 100 million units.

Stock	1M	5D	Avg Turnover	Foreign 1M	Institution 1M	Retail 1M	Read
Jeju Semiconductor	+173.6	+28.4	3,692.8	+1,440.8	+595.8	-1,962.1	LPDDR second-order discovery. Hot
Samsung Electro-Mechanics	+146.0	+52.5	8,986.4	-9,036.0	+4,953.0	+3,922.2	MLCC + FC-BGA leader. Institution-led
Daeduck Electronics	+81.4	+20.5	1,198.1	+524.4	+827.0	-954.7	FC-BGA / MLB core name
Simmtech	+74.5	+32.3	822.6	+678.3	+1,376.0	-2,070.4	SOCAMM / substrate core
Haesung DS	+72.7	+19.2	297.3	+132.7	+656.4	-53.9	Heat spreader / substrate option
Hana Micron	+48.3	+5.0	1,200.6	+2,172.5	+158.3	+135.8	Foreign-led back-end recovery
HD Hyundai Heavy Industries	+51.0	+21.1	3,642.0	-5,344.5	+7,038.2	-1,987.9	Institution-led shipbuilding + nuclear option
Hanwha Ocean	+2.1	+16.3	2,369.4	+1,472.3	+1,393.3	+6,409.7	5D flow improvement, still a laggard

The key is that not all leadership has the same flow quality.

Samsung Electro-Mechanics rose 146.0% over one month, but foreigners sold KRW 903.6 billion. Institutions and retail absorbed the supply. That fits the SEMCO KRW 100T market-cap note: the AI passive-component re-rating is real, but chasing efficiency has fallen.

Hana Micron, by contrast, rose 48.3% with KRW 217.25 billion of one-month foreign net buying. That is why it looks more like a second-line back-end expansion candidate than an already over-owned first-line leader.

HD Hyundai Heavy Industries was sold by foreigners but bought heavily by institutions. That connects to the HD Hyundai Heavy Industries SMR option analysis: the shipbuilding / engine / SMR story is alive, but the rally is institution-led and price location matters.

5. Is KOSDAQ Weak, Or Selectively Coming Back?

KOSDAQ’s 20D ADR is 66.7. On the surface, that is weak. But it does not mean the whole KOSDAQ should be avoided.

The earlier KOSDAQ smart-money and Pearl Abyss rebound note argued that flows can turn before prices. The ADR data narrows that frame.

What matters is not buying KOSDAQ broadly. It is finding second-line names where turnover is just starting to accelerate, foreign/institutional flows are positive, and the 20-day moving-average extension is still manageable.

The local screen highlights:

Rank	Stock	Theme	5D	20D	5D Avg Turnover	Turnover Accel.	20D MA Gap	fi5	Read
1	HPSP	Semi equipment / AI infra	+12.7	+3.4	1,989.2	1.28x	+2.9	+509.3	Cleanest second-line candidate
2	SFA Semicon	Back-end	+19.9	-2.5	3,020.5	3.41x	+7.4	+259.1	Back-end expansion candidate
3	Hana Micron	Back-end	-1.2	+16.6	1,386.3	1.18x	+3.8	+129.4	Pullback candidate
4	Dongjin Semichem	Materials	+5.4	-4.0	642.5	1.15x	+3.1	+447.4	Materials-flow recovery candidate
5	KMW	RF / AI-RAN	+11.7	+20.8	259.0	1.30x	n/a	+226.8	AI-RAN event-confirmation candidate

fi5 is five-day foreign plus institutional net buying. Some local flow fields can be missing or incomplete, so Kiwoom / KRX flow validation is required before stock-level execution.

6. Breadth Expansion Triggers

This is not yet a broad-market buy zone. At an all-market ADR of 67.3, decliners still dominate.

The confirmation checklist is:

Trigger	Threshold	Meaning
20D ADR recovers to 80	All-market ADR above 80	Decliner dominance is easing
20D ADR recovers to 100	All-market ADR above 100	Advancers and decliners are balanced
Daily advance ratio above 55%	2-3 consecutive sessions	The rebound is not a one-day bounce
KOSDAQ turnover rises	Turnover up with more advancers	Small/mid-cap breadth can broaden
Foreign selling is absorbed	FX stable and index holds/rises	More absorption than true risk-off

By sector, the sequence to watch is:

Sector	Signal	Meaning
AI infra second-line	HPSP, SFA Semicon, Hana Micron, Dongjin Semichem turnover rising	Internal expansion within semis
Optical / RF / AI-RAN	KMW, RFHIC, Oi Solution flow turning	Marvell / NVIDIA AI-RAN linkage
FC-BGA / MLB	Daeduck, ISU Petasys, Korea Circuit re-accelerating	Custom ASIC / AI networking confirmation
Test sockets / back-end	ISC, Leeno, TSE, Doosan Tesna turnover rising	SOCAMM / ASIC test-infra expansion
Shipbuilding / defense second-line	Flows rotate into laggards during leader pullbacks	Rotation within an existing leadership theme

This links back to the Marvell / Broadcom Korea AI bottleneck preview. If the market is rotating from a single HBM trade into custom ASICs, AI networking, optical links and power integrity, the broad ADR can stay weak while those lower-stack bottlenecks keep attracting turnover.

7. Practical Read

The current Korean tape can be summarized in two lines:

The broad market is weak.
Leadership is not dead.

The action plan must reflect both.

Action	Condition	Targets
Hold existing leaders	Relative strength remains intact despite weak ADR	Samsung Electro-Mechanics, Daeduck and other AI infra names
Avoid chasing first-line leaders	20D ADR below 80	Hot first-line names
Watch second-line candidates	Turnover acceleration + positive fi5 + manageable MA gap	HPSP, SFA Semicon, Hana Micron
Replace weak positions	Holdings underperform the market and flows are weak	Non-leadership / non-core positions
Manage cash	Portfolio concentration is high before breadth recovers	Do not fully deploy cash while breadth is still weak

Chasing first-line leaders is inefficient here. Samsung Electro-Mechanics, Jeju Semiconductor and Simmtech have already moved sharply. But avoiding all equities because breadth is weak would miss the narrow leadership regime.

The cleaner stance is: hold leaders, watch second-line names, and require ADR above 80 before broadening exposure.

8. Conclusion

Korea is not a dead market. But it is not a broad market either.

As of the May 26, 2026 close, the all-market 20-day ADR is 67.3. KOSPI is 68.5 and KOSDAQ is 66.7. That is a poor backdrop for buying the average stock. At the same time, Samsung Electro-Mechanics, Jeju Semiconductor, Daeduck Electronics, Simmtech, Hana Micron and HD Hyundai Heavy Industries still show strong leadership.

So the regime is narrow leadership, not broad risk-on.

The three checks are:

Does all-market ADR recover above 80?
Do KOSDAQ turnover and the number of advancers rise together?
Do foreign and institutional flows begin to appear in AI-infra second-line names?

Until those conditions improve, chasing first-line leaders is less attractive than observing second-line candidates. If breadth recovers, the market can broaden. If it does not, only a few leaders will survive. The right stance is neither optimism nor pessimism. It is to accept that market breadth is narrow and follow where real turnover and quality flow are appearing.

Appendix. Evidence Classification

[Fact]

As of May 26, 2026, the all-market 20D ADR was 67.3.
On April 16, 2026, the all-market 20D ADR was 113.1.
On May 26, 2026, 749 Korean stocks advanced, 1,660 declined, and daily ADR was 45.1.
The last-month leadership list includes Samsung Electro-Mechanics, Jeju Semiconductor, Daeduck Electronics, Simmtech, Haesung DS, Hana Micron and HD Hyundai Heavy Industries.
HPSP, SFA Semicon, Hana Micron and Dongjin Semichem screened relatively well on turnover and flow conditions.

[Inference]

The current market is narrow leadership, not broad risk-on.
More important than simple foreign buying is whether a sector can hold price and attract turnover despite foreign selling.
Within AI infrastructure, rotation can move from SOCAMM / LPDDR into FC-BGA / MLB, back-end, optical and RF.
Second-line turnover and flow turns have better expected value than chasing first-line leaders after large moves.

[Speculation]

Marvell earnings may strengthen custom ASIC, optical and AI-RAN themes more than SOCAMM alone.
Broadcom earnings may re-ignite AI networking, FC-BGA / ABF and high-speed MLB names.
If ADR recovers above 80, the probability of second-line expansion rises.

[Blocked]

May 27, 2026 closing ADR is not included in this note.
Some five-day flow fields may be missing or incomplete in the local DB.
AI-RAN candidates such as RFHIC and Oi Solution require separate Kiwoom / KRX flow validation before firm stock-level judgment.

Data source: Research OS local DB prices_daily, universe.csv, korea_adr_recent_20260526.csv, korea_leaders_20260415_20260526.csv, second_line_theme_flow_candidates_20260527.csv. Data cut: May 26, 2026 close. This is research commentary, not investment advice.

HD Hyundai Heavy Industries’ SMR Option: TerraPower Natrium, Shipbuilding Earnings and Engine Re-Rating

Wed, 27 May 2026 14:40:00 +0900

Related reading: Hanwha Ocean Deep Dive · Samsung Heavy Industries Deep Dive · Hanwha Engine: From Marine Engines to Data-Center Power · Northern Sea Route and Korean Shipbuilding Beneficiaries · Korea Daily Market Hub

TL;DR

HD Hyundai Heavy Industries has real SMR exposure. It has been selected as a preferred manufacturer for TerraPower Natrium Reactor Enclosure System, or RES, components. That is a higher-quality option than a generic nuclear theme.
But SMR is not a 2026-2027 EPS driver yet. Contract value, margin, delivery schedule and revenue recognition remain undisclosed.
The current numbers are coming from shipbuilding and engines. 1Q26 sales were KRW 5.91 trillion, operating profit was KRW 905.4 billion and net profit was KRW 773.8 billion. Research OS local DB shows a 26 May 2026 close of KRW 745,000 and 2026E consensus P/E around 24x.
The stance is Wait. The trend and institutional flow are strong, but the stock already prices in a lot of shipbuilding, engine, defense and SMR re-rating. Better entry points are a KRW 705,000-720,000 pullback or a KRW 733,000-745,000 retest.
The relative-value proxy may be HD Korea Shipbuilding & Offshore Engineering. It owns 69.2% of HD Hyundai Heavy Industries, giving group-level exposure with a potential valuation cushion, although holdco discount and exchangeable-bond overhang matter.

1. The Core Question

This note asks whether HD Hyundai Heavy Industries (329180.KS) should still be treated as a cyclical shipbuilder, or whether the market is correctly reclassifying it as a hybrid of shipbuilding, engines and power infrastructure, defense, and SMR manufacturing supply chain.

This is a mixed TYPE A/B analysis.

TYPE A: stock-specific price, flow, valuation and entry conditions.
TYPE B: the industry frame around TerraPower Natrium, SMRs, data-center power and naval shipbuilding.

The conclusion is straightforward.

The SMR option is real. But at the current price, it is not yet enough as a standalone Buy-now argument.

The investment stack should be separated like this:

Layer	Thesis	Earnings Timing	Current Read
Core shipbuilding	High-priced backlog and better merchant/offshore/naval margins	Now	Main EPS driver
Engines / power	Marine engines plus HiMSEN for data-center power needs	2026-2028	Fastest re-rating driver
Defense / MASGA	Naval ships, MRO, U.S. shipbuilding cooperation and nuclear-submarine optionality	Medium term	Event premium
SMR	TerraPower Natrium RES preferred manufacturer	After 2028-2030	Long-duration multiple option

2. Fact Check: What Is Actually Confirmed?

2.1 TerraPower Natrium preferred manufacturer status is confirmed

On 21 May 2026, HD Hyundai announced that it signed a Natrium Reactor Supply Framework Agreement with TerraPower. The key point is that HD Hyundai Heavy Industries was selected as a preferred manufacturer for Natrium Reactor Enclosure System, or RES, components. HD Hyundai also said the companies plan to move from FOAK experience toward NOAK serial manufacturing. (HD Hyundai)

TerraPower confirmed the same point in its own release, naming HD Hyundai Heavy Industries as the preferred manufacturer for Natrium RES components and describing a broader collaboration with HD Hyundai and Hyundai E&C covering design, manufacturing, supply chain, construction, commercial structure and multi-unit delivery. (TerraPower)

This is more advanced than a generic MOU. However, it is not yet a disclosed binding purchase order with contract value, margin and delivery schedule.

2.2 Meta’s eight-unit agreement opens Q, but not yet HHI revenue

TerraPower and Meta announced an agreement in January 2026 to develop up to eight Natrium reactor and energy storage system plants. TerraPower said this could provide up to 2.8 GW of baseload energy, or up to 4 GW with output boost, with initial units as early as 2032. (TerraPower)

That matters because SMR manufacturing value comes from fleet deployment, not a one-off demonstration plant.

But Meta’s eight units do not automatically mean eight units of HD Hyundai Heavy Industries revenue. The exact component scope, unit allocation, price and margin remain [Blocked].

2.3 Regulatory risk is lower, not gone

The U.S. Department of Energy said the NRC issued a construction permit for TerraPower’s Kemmerer Natrium project. The same DOE note also says TerraPower must submit a separate operating license application before it can operate the facility. The project is expected to be completed in 2030. (U.S. DOE)

The investment read is:

Question	Answer
Is HHI in TerraPower’s SMR supply chain?	Yes. That is a fact.
Does SMR materially lift 2026-2027 earnings?	Not yet proven.
Does Meta’s eight-unit agreement equal eight HHI units?	No. It creates a scenario, not confirmed revenue.
Is Natrium commercialization risk gone?	No. Operating license, fuel, timing and cost risks remain.

3. Current Earnings: Strong Even Without SMR

The core business is already strong before assigning any SMR value.

Yonhap reported the following 1Q26 results for HD Hyundai Heavy Industries. (Yonhap)

Item	1Q26	YoY
Sales	KRW 5.91tn	+54.8%
Operating profit	KRW 905.4bn	+108.8%
Net profit	KRW 773.8bn	+172.3%

Research OS local DB confirms the same basic earnings picture:

Verified Item	Value	Source / Date
Close	KRW 745,000	Research OS `prices_daily`, 2026-05-26
Intraday high	KRW 763,000	Research OS `prices_daily`, 2026-05-26
1Q26 revenue	KRW 5.9163tn	Research OS `kr_fundamentals_daily`, Naver collected 2026-05-26
1Q26 operating profit	KRW 905.4bn	Research OS `kr_fundamentals_daily`, Naver collected 2026-05-26
1Q26 net profit	KRW 773.8bn	Research OS `kr_fundamentals_daily`, Naver collected 2026-05-26
1Q26 EPS	KRW 7,372	Research OS `kr_fundamentals_daily`, Naver collected 2026-05-26
2026E revenue consensus	KRW 24.8651tn	Research OS `consensus_daily`, Naver, 2026-05-26
2026E operating profit consensus	KRW 3.8385tn	Research OS `consensus_daily`, Naver, 2026-05-26
2026E net profit consensus	KRW 2.9616tn	Research OS `consensus_daily`, Naver, 2026-05-26
2026E EPS	KRW 28,131	Research OS `consensus_daily`, Naver, 2026-05-26
2026E P/E	24.17x	Research OS `consensus_daily`, Naver, 2026-05-26

The valuation is the point.

Dividing the KRW 745,000 close by 2026E EPS of KRW 28,131 gives roughly 26.5x. The local consensus table shows 24.17x because the underlying price snapshot differs. Either way, the message is the same:

This is not a cheap cyclical shipbuilder. The stock already embeds high-margin shipbuilding, engines, defense and some SMR optionality.

4. Value Chain: What Bottleneck Does HHI Actually Own?

In SMR, HHI does not own the reactor design IP. TerraPower is the developer. HHI’s role is nuclear-grade heavy fabrication.

The chain looks like this:

AI data-center power demand
 ↓
Big Tech demand for firm clean power
 ↓
TerraPower Natrium / GE-Hitachi technology
 ↓
NRC / DOE regulatory and policy support
 ↓
Hyundai E&C EPC + HD Hyundai manufacturing supply chain
 ↓
HHI precision fabrication of RES components
 ↓
FOAK delivery → possible NOAK serial production

The bottleneck is not just shipyard capacity.

nuclear quality assurance
large-scale welding and precision fabrication
delivery reliability
repeatable production-process design
managing rework and quality costs in nuclear components

SMR economics depend less on the phrase “small reactor” and more on whether serial production can reduce cost. If HHI moves from preferred manufacturer to serial production vendor, the option becomes a repeat revenue pool rather than a one-off headline.

But there are no numbers yet. That is why this is an option, not EPS.

5. Idea 1: TerraPower Natrium Serial Production Option

Classification: Idiosyncratic Alpha

HHI’s SMR upside is more specific than a generic nuclear theme. It has entered the TerraPower Natrium RES supply chain and may move from FOAK work toward NOAK serial production.

P x Q x C

Variable	Content	Investment Meaning
P, price	RES component price and nuclear QA premium	Undisclosed, but potentially higher-value than ordinary marine equipment
Q, quantity	Kemmerer FOAK, Meta initial two units, up to eight, additional customers	Fleet deployment expands option value
C, cost	Nuclear QA, dedicated capacity, skilled labor, delay/rework	Margin remains uncertain

Market Mispricing

The upside case is that preferred manufacturer status becomes repeat production status. In that scenario, HHI can be reclassified as strategic manufacturing capacity, not just a shipbuilding cycle stock.

The downside to the narrative is treating the current framework as already-booked revenue. Without contract value, it should not be loaded into 2026-2027 EPS.

Red Team

Natrium operating-license timing could slip.
HALEU fuel supply, rates, U.S. nuclear policy and FOAK cost overrun remain risks.
If HHI misses cost, quality or delivery expectations on FOAK work, NOAK supply may be split.

6. Idea 2: Engines and Data-Center Power Are Faster Than SMR

Classification: Idiosyncratic Alpha

While the market focuses on SMR, the faster earnings lever is engines.

AI data-center power shortages will not wait for 2030s SMR deployment. Gas engines, backup generation, distributed power and power equipment can move earlier. HHI’s HiMSEN engine business is exposed to that bridge-power layer.

Public broker summaries and the user-supplied data cite 1Q26 engine OPM of 21.1%. If the market starts valuing the engine segment as data-center power infrastructure rather than a shipbuilding component, it can lift the company’s consolidated multiple.

The key question is:

Is the engine segment a marine-cycle component business, or a data-center power-infrastructure business?

If the answer shifts toward the second, HHI can receive a blended multiple closer to power infrastructure plus shipbuilding plus defense.

Required Conditions

Additional data-center HiMSEN orders
Engine OPM around 20%
Capacity expansion without margin damage
High-margin shipbuilding backlog converting at the same time

Failure Conditions

Data-center engine orders prove one-off
Capacity expansion brings pricing pressure
Carbon rules or gas prices limit demand

7. Idea 3: Defense, MASGA and Nuclear-Submarine Premium

Classification: Beta Trade / Event-driven

Defense and U.S. shipbuilding cooperation add another option layer.

Korea’s nuclear-submarine discussions, U.S. naval MRO and shipbuilding cooperation, and MASGA-style policy narratives all push investors to view HHI as strategic manufacturing capacity rather than only merchant-ship exposure.

The issue is timing. Policy events move the stock quickly; revenue conversion takes time.

Required Conditions

Naval and special-purpose vessel order visibility
U.S. MRO / shipbuilding cooperation becomes contractual
Nuclear-submarine budget, prime contractor and reactor/fuel structure become clearer
Benefit split between HHI and Hanwha Ocean is visible

Failure Conditions

Policy headlines fail to become budgets and contracts
Benefits accrue more to competitors
U.S. local production, labor and regulatory constraints reduce profitability

8. Flow: Strong Trend, But Not Yet Foreign-Led

Research OS local DB shows HHI closed at KRW 745,000 on 26 May 2026, with an intraday high of KRW 763,000. Regular-session volume on 26 May was 686,723 shares, about 1.34x the 20-day average.

The key flow point: this is institution-led, not foreign-led.

KRW bn equivalent shown as KRW 100mn units. Source: Research OS investor_flow_daily.

Window	Retail	Foreign	Institution	Financial Inv.	Insurance	Investment Trust	Private Fund	Pension etc.	Other Corp.	Quality Institution
5D	-1,384	-1,404	+2,734	+327	+86	+1,080	+956	+282	+61	+2,405
10D	-1,154	-746	+1,814	-60	+102	+319	+1,072	+394	+63	+1,887
20D	-1,343	-1,699	+2,880	-654	+66	+1,120	+1,146	+1,239	+128	+3,572

The read:

First, retail investors are selling into the move.

Second, foreigners are not yet the lead buyer. They were net sellers over 5D and 10D. One day of mild foreign buying on 26 May does not prove a full turn.

Third, the real buyers are investment trusts, private funds and pensions. The 5D quality-institution buy of KRW 240.5bn is meaningful.

But the tape is not clean:

Item	Value	Read
26 May short-sale value ratio	26.2%	High
5D program net buy	-KRW 180.3bn	Still a drag
26 May program	+KRW 5.0bn	Slight positive turn
Stock-loan balance	26 May DB value is zero, abnormal. 22 May balance of roughly 7.69mn shares is more reliable	Data-quality caution
Foreign ownership	13.59%	As of 26 May 2026

Conclusion:

Institutions are buying, but foreign, program and short-sale pressure has not disappeared. After a large breakout candle, a retest is preferable to chasing.

9. Technical Setup: Minervini Passes, Entry Is Less Clean

The trend itself is strong.

Research OS local DB:

Metric	Value
Close	KRW 745,000
5-day MA	KRW 666,200
20-day MA	KRW 672,350
50-day MA	KRW 579,430
150-day MA	KRW 569,650
200-day MA	KRW 550,318
252-day closing low	KRW 372,000
252-day closing high	KRW 745,000

Minervini Trend Template: 9/9 pass.

Condition	Result
Price > 150DMA	Pass
Price > 200DMA	Pass
150DMA > 200DMA	Pass
200DMA rising	Pass
50DMA > 150DMA	Pass
50DMA > 200DMA	Pass
Price > 50DMA	Pass
Price at least 25% above 52-week low	Pass
Price within 25% of 52-week high	Pass

The stock qualifies. The price is the problem.

It is roughly 10.8% above the 20DMA and 28.6% above the 50DMA. The ideal Minervini-style buy point was the breakout around KRW 733,000. Now it is extended.

Key levels:

Level	Meaning	Action
KRW 763,000	26 May intraday high	Small momentum entry only if it breaks and holds on volume
KRW 745,000	26 May close	Retest zone
KRW 733,000	Prior high / breakout line	Most important first support
KRW 705,000-720,000	Better pullback zone	Attractive if institutional flow persists
KRW 689,000	26 May low	Break below damages the short-term setup

10. Valuation: SMR Alone Does Not Explain the Stock Yet

The current price already embeds high expectations.

Local consensus:

Item	2026E
Revenue	KRW 24.8651tn
Operating profit	KRW 3.8385tn
Net profit	KRW 2.9616tn
EPS	KRW 28,131
P/E	24.17x
ROE	27.63%
Target price consensus	KRW 894,842

The user-supplied recent broker targets range from KRW 900,000 to KRW 970,000, KRW 1,000,000 and KRW 1,170,000. The re-rating reflects 1Q26 earnings, engine re-rating, SMR optionality and defense/MASGA expectations.

SMR Option Sensitivity

The following is [Speculation], not fact. Since the contract value has not been disclosed, this is a reverse-sensitivity exercise.

Assumptions:

Eight Natrium units
76% after-tax conversion
20x P/E
Roughly 104.96mn shares outstanding

Scenario	Assumption	Per-Share Option Value	% of Current Price
Low	8 units x KRW 100bn RES revenue per unit x 10% EBIT margin	~KRW 11,600	~1.6%
Base	8 units x KRW 300bn RES revenue per unit x 15% EBIT margin	~KRW 52,100	~7.0%
High	8 units x KRW 500bn RES revenue per unit x 20% EBIT margin	~KRW 115,900	~15.6%

The point is:

For SMR alone to justify more than 20% upside from here, the RES revenue and margin assumptions need to be large. Public data does not yet support that.

SMR is better understood as a reason HHI might deserve a higher multiple beyond the current cycle, not as an immediate EPS bridge.

11. Preferred Exposure

Stock	Ticker	Stance	One-Line Thesis
HD Hyundai Heavy Industries	329180	Wait	Shipbuilding and engines make EPS; TerraPower Natrium SMR is a long-duration multiple option
HD Korea Shipbuilding & Offshore Engineering	009540	Watchlist / conditional preference	Owns 69.2% of HHI and gives group-level shipbuilding, SMR and MASGA exposure
Hyundai E&C	000720	Watchlist	EPC bridge in TerraPower-HDEC-HDH cooperation, but lower directness and higher EPC execution risk

HD Korea Shipbuilding is interesting because of relative value. Asia Business Daily reported that HD Korea Shipbuilding owns 69.2% of HD Hyundai Heavy Industries. (Asia Business Daily)

Multiplying HHI’s market cap by 69.2% produces a gross stake value far above HD Korea Shipbuilding’s own market cap. This is not arbitrage. Net debt, holdco discount, exchangeable-bond overhang, other subsidiaries and double-listing structure all matter.

Still, it may offer better risk/reward than chasing HHI at a high multiple.

12. Trading Plan

HD Hyundai Heavy Industries: Wait

At the current price, the setup is not simple.

Stock quality: strong
Trend: strong
Institutional flow: strong
Foreign flow: still not strong
Short sale and program pressure: not clean
Valuation: not cheap
SMR numbers: undisclosed

Entry conditions:

Approach	Condition	Read
Trend / flow trade	Retest of KRW 733,000-745,000, or pullback to KRW 705,000-720,000	Enter only if institutional flow holds
SMR-only value entry	Below roughly KRW 590,000, or around 20x 1Q26 run-rate EPS	Margin of safety while SMR contract value is undisclosed
Momentum breakout	Break and hold above KRW 763,000 with volume	Small position only
Do not chase	Gap above KRW 780,000	Risk/reward deteriorates
Short-term invalidation	Break below KRW 689,000	Breakout setup damaged

Catalysts

Catalyst	What To Check
2Q26 earnings	OPM around 15%, OP above KRW 900bn
Additional engine orders	Data-center HiMSEN follow-on contracts, capacity expansion
TerraPower follow-up disclosure	Binding PO, contract value, delivery schedule, margin clues
Meta initial two units	Site selection, financing and offtake structure
Defense / MASGA	Naval MRO, nuclear-submarine budget, prime contractor and benefit split

Invalidation

TerraPower framework does not evolve into binding orders or disclosed value within 12-18 months.
Kemmerer project schedule slips structurally from the 2030 target.
Operating license, HALEU fuel or NRC/policy risks become binding constraints.
HHI faces FOAK cost, quality or delivery issues.
Core shipbuilding OPM falls back below 10%.
Engine OPM falls below 17% and data-center power demand proves one-off.
The stock trades above KRW 1,000,000 without 2027E EPS upgrades or SMR contract disclosure.

13. Final View

HHI’s SMR news matters. TerraPower Natrium RES preferred manufacturer status is a real option. Meta’s eight-unit agreement opens a long-term quantity scenario, and the DOE/NRC construction permit reduces part of the regulatory risk.

But sequence matters.

Shipbuilding and engines make the earnings. SMR raises the multiple.

The current stock price already discounts a meaningful amount of core shipbuilding strength, engine re-rating, defense/MASGA and SMR optionality. Chasing HHI solely for SMR is not attractive from a risk/reward perspective.

Existing holders can follow the trend. New buyers should prefer a KRW 733,000-745,000 retest or a KRW 705,000-720,000 pullback. More conservative investors can wait for TerraPower contract value and 2Q26 margin confirmation.

One-line portfolio view:

HD Hyundai Heavy Industries may be a good company, but this is not yet a clean entry price. SMR is a long-term option; near-term buying should be governed by 2Q26 shipbuilding/engine earnings and institutional-flow retest.

Appendix: Evidence Classification

[Fact]

HHI was selected as preferred manufacturer for TerraPower Natrium RES components. (HD Hyundai)
TerraPower named HHI as preferred manufacturer and described broader multi-unit collaboration with HD Hyundai and Hyundai E&C. (TerraPower)
TerraPower and Meta announced support for up to eight Natrium plants. (TerraPower)
DOE said the Natrium construction permit was issued, while a separate operating license is required before operation. (U.S. DOE)
HHI’s 1Q26 revenue was KRW 5.91tn, operating profit KRW 905.4bn and net profit KRW 773.8bn. (Yonhap)
Research OS local DB shows a 26 May 2026 close of KRW 745,000 and 2026E consensus EPS of KRW 28,131.

[Inference]

SMR is more of a 2030s multiple option than a 2026-2027 EPS driver.
If TerraPower reaches NOAK serial production, HHI’s nuclear-grade fabrication moat can expand.
The current rally is institution-led rather than foreign-led.
HD Korea Shipbuilding may offer a valuation-cushion proxy versus direct HHI chase.

[Speculation]

RES revenue per unit of KRW 100bn-500bn and EBIT margin of 10-20%.
HHI supplying key RES components across all eight Meta-linked Natrium units.
Floating nuclear or SMR-powered vessels becoming commercial shipbuilding markets.
Defense, nuclear-submarine and U.S. shipbuilding cooperation structurally lifting HHI’s multiple.

[Blocked]

TerraPower contract value.
RES component margin.
HHI SMR-dedicated capex.
Actual revenue-recognition year.
NOAK volume allocation.
TerraPower operating-license certainty and timing.
Hyundai E&C EPC profitability and risk-sharing structure.

Did Samsung Signal a Memory Supercycle Through 2028? Reframing the DS Stock-Bonus Buyback

Wed, 27 May 2026 09:30:00 +0900

Samsung Electronics series: Citi’s ₩460,000 target and the memory re-rating · Strike risk vs memory supercycle · Can Samsung deserve 15x PER? · Foreign ownership divergence · AI HBM hub

TL;DR

This issue should be framed around two questions: (1) did Samsung internally acknowledge a memory supercycle through 2028? (2) how much additional treasury-stock buying could the special-bonus structure create?
The answer to the first question is conditional yes. If the hurdle is annual DS operating profit of KRW 200 trillion, it is a strong internal signal on HBM4, server DRAM and eSSD durability through 2028. If it is cumulative 2026-2028 profit of KRW 200 trillion, the signal is much weaker.
The answer to the second question depends heavily on the denominator. The reported KRW 80 trillion is not an official fixed amount. It looks like an estimate based on KRW 1,220 trillion of 2026-2028 consolidated operating-profit forecasts, a 10.5% pool and a 60% after-tax payout assumption.
If the headline net KRW 80 trillion is correct, it would buy roughly 252 million shares at KRW 317,500, or about 4.3% of common shares outstanding. But this is employee compensation stock, not a cancellation-style shareholder-return program.
The event is best read as internal supercycle signaling + HBM talent retention + near-term float absorption + future overhang, not simply as a larger shareholder-return announcement.

1. The Two Questions

The market headline has focused on the phrase “Samsung may buy KRW 80 trillion of shares for special bonuses.” That is too blunt.

The investment question should be reduced to two parts:

Does this mean Samsung internally recognizes a DS and memory supercycle through 2028?
How much additional treasury-stock buying could be created by the special-bonus structure?

The first question is about earnings-cycle durability: how long Samsung sees the HBM4, server DRAM, enterprise SSD and SOCAMM2 demand cycle lasting.

The second question is about flow and capital-policy mechanics: how much stock the company might need to buy, and whether those shares are cancelled or paid to employees.

2. Confirmed Facts and Blocked Items

The confirmed facts are:

Item	Confirmed Detail	Source
1Q26 consolidated results	Revenue KRW 133.9tn, operating profit KRW 57.2tn	Samsung Global Newsroom
1Q26 DS results	DS revenue KRW 81.7tn, DS operating profit KRW 53.7tn	Samsung Global Newsroom
AI memory commentary	HBM4 and SOCAMM2 mass-product sales for NVIDIA Vera Rubin, HBM4E samples, strong server-memory demand	Samsung Global Newsroom
Existing shareholder-return policy	KRW 9.8tn annual regular dividend and 50% of FCF returned for FY2024-2026	Samsung IR
2024-2025 buybacks	Official IR shows KRW 1.8118tn in 2024 and KRW 8.1893tn in 2025, or roughly KRW 10.0tn combined	Samsung IR
Special-bonus negotiation	Tentative agreement reports linking DS special bonuses to business performance and stock payment	Reuters, Hankyoreh

The blocked items are even more important:

Blocked Item	Why It Matters
Whether the 10.5% denominator is consolidated OP, DS OP or adjusted OP	It changes the stock-purchase size materially
Whether the KRW 200tn 2026-2028 condition is annual or cumulative	It determines whether this is truly a 2028 supercycle signal
Whether the assumed 60% after-tax payout ratio is correct	It changes the net purchase amount
Lockup, tradability and disposal schedule after employees receive shares	It determines near-term flow and future overhang
Whether current consensus estimates already include the bonus expense	It determines whether analysts are double-counting or under-counting the cost

The analysis below is therefore a scenario framework, not a confirmed company guidance model.

3. Question 1: Did Samsung Acknowledge a Supercycle Through 2028?

The answer is conditional yes.

Samsung has not officially declared “a supercycle through 2028.” But if the reported bonus hurdle is tied to annual DS operating profit of KRW 200 trillion, it becomes a strong internal signal.

That is because 1Q26 DS operating profit was KRW 53.7 trillion.

1Q26 DS OP KRW 53.7tn x 4 = annualized KRW 214.8tn

Annual KRW 200 trillion is not fantasy. It is close to the current run-rate.

3.1 If the Hurdle Is Annual KRW 200tn, It Matters

If the agreement says DS must earn around KRW 200 trillion each year from 2026 to 2028, the implication is clear.

Samsung may be internally assuming that:

HBM4 and HBM4E qualification and ramp stay broadly on track.
Server DRAM and enterprise SSD demand persists beyond a one-year spike.
NVIDIA Vera Rubin, next-generation GPUs/CPUs and hyperscaler AI capex keep improving the memory mix.
AI and server-memory mix improvement offsets ordinary DRAM supply growth.

In that case, the special-bonus plan is not just a labor settlement.

It is a sign that Samsung designed DS incentives around a high-profit memory environment lasting through 2028.

3.2 If the Hurdle Is Cumulative KRW 200tn, the Signal Is Weak

If the condition is cumulative 2026-2028 DS operating profit of KRW 200 trillion, the interpretation changes completely.

Cumulative KRW 200tn / 3 years = annual average KRW 66.7tn

Compared with the annualized 1Q26 DS run-rate of KRW 214.8 trillion, that is a low hurdle. Under this interpretation, the plan is closer to a compensation formula than a strong supercycle signal.

So the key is annual versus cumulative.

Interpretation	DS Profit Hurdle	Versus 1Q26 Run-Rate	Investment Meaning
Annual KRW 200tn	KRW 200tn each year	Close to annualized KRW 214.8tn	Strong internal signal on a 2026-2028 supercycle
Cumulative 2026-2028 KRW 200tn	Annual average KRW 66.7tn	Far below the current run-rate	Weak as a supercycle signal
Annual KRW 100tn for 2029-2035	Long-term high-profit condition	Above many past cycle levels	Structural profitability claim after the peak cycle

3.3 Official Commentary Already Confirms 2026 Strength

Samsung’s official 1Q26 release is enough to confirm that management sees strong 2026 AI memory demand.

The company reported DS operating profit of KRW 53.7 trillion and explained that memory results were driven by AI demand, limited supply, industry-wide price increases and mass-product sales of HBM4 and SOCAMM2. It also said it plans to provide HBM4E samples in Q2 and expects server-memory demand to remain strong in the second half due to hyperscaler and enterprise AI/LLM adoption.

But that official release is not a 2028 guidance statement. The 2028 signal can only be inferred from the special-bonus formula.

The clean conclusion:

Samsung did not officially declare a supercycle through 2028. But if the bonus hurdle is annual DS operating profit of KRW 200 trillion, the plan can be read as a strong internal signal that Samsung sees the high-profit memory cycle lasting through 2026-2028.

4. Question 2: How Much More Stock Could Samsung Buy?

The second question is numerical.

The reported KRW 80 trillion figure appears to come from this bridge:

Estimated net treasury-stock purchase
= 2026-2028 operating-profit forecast x 10.5% x after-tax payout ratio

4.1 Headline Case: Consolidated Operating Profit

Using the operating-profit forecasts cited in Korean coverage:

Item	2026E	2027E	2028E	Total
Consolidated OP forecast	KRW 352tn	KRW 444tn	KRW 424tn	KRW 1,220tn
10.5% pool	KRW 37.0tn	KRW 46.6tn	KRW 44.5tn	KRW 128.1tn
60% after-tax assumption	KRW 22.2tn	KRW 28.0tn	KRW 26.7tn	KRW 76.9tn

So the “KRW 130tn gross / KRW 80tn net” headline is broadly consistent with the math.

But this only holds if the denominator is consolidated operating profit. If the final agreement uses DS operating profit, adjusted operating profit or pre-bonus operating profit, the number changes.

4.2 What Does Net KRW 80tn Mean in Shares?

Dividing headline net KRW 80 trillion by the cited May 27 price of KRW 317,500:

KRW 80tn / KRW 317,500 = roughly 252mn shares

Against roughly 5.846bn common shares, that is about 4.3%.

Assumption	Net Purchase Amount	Shares at KRW 317,500	% of Common Shares
Headline net	KRW 80.0tn	~252mn	~4.3%
Consolidated OP x 10.5% x 60%	KRW 76.9tn	~242mn	~4.1%
DS annual KRW 200tn x 3 years x 10.5% x 60%	KRW 37.8tn	~119mn	~2.0%
Cumulative DS KRW 200tn x 10.5% x 60%	KRW 12.6tn	~40mn	~0.7%

This is the important table.

KRW 80 trillion is closer to an upper-end scenario. Depending on the denominator, the actual incremental stock purchase could be closer to the low-teens trillion won.

4.3 Compared With Existing Buybacks

Samsung’s official IR page shows roughly KRW 10.0 trillion of buybacks over 2024-2025: KRW 1.8118 trillion in 2024 and KRW 8.1893 trillion in 2025. (Samsung IR)

If special-bonus stock purchases reach the headline net KRW 76.9-80.0 trillion, they would be about 7.7-8.0x the official 2024-2025 buyback amount.

Comparison	Amount
Official 2024-2025 buybacks	~KRW 10.0tn
Headline net special-bonus purchase estimate	~KRW 76.9-80.0tn
Multiple	~7.7-8.0x

That is meaningful for flow.

But the critical point is that the shares are not being cancelled. They are being paid to employees.

5. Shareholder Return or Employee Compensation?

The precise answer is employee compensation.

During the purchase phase, the company buying shares can absorb float and support near-term flow. But those shares are later transferred to employees, not retired.

The economics have three stages:

Stage	Effect	Shareholder Read
Company buys shares in the market	Float absorption and buying flow	Near-term positive
Shares are paid to employees	Treasury stock becomes individually held stock	No cancellation effect
Lockup expires or tradable portion is sold	Some shares can return to market	Future overhang

A cancellation-style buyback permanently reduces shares outstanding and increases EPS. Employee-compensation stock does not. It can eventually return to the market, and voting rights revive once treasury shares are transferred to individuals.

That does not make the event bad. It simply means the objective is different:

Retain HBM-critical employees.
Link DS performance to employee compensation.
Give employees exposure to Samsung’s share price.
Reduce immediate cash-bonus pressure.
Reduce labor risk.

This is not primarily a shareholder-return event. It is an HBM execution and talent-retention event.

6. Why Stock Compensation Matters for HBM

HBM competitiveness is not only about capex.

Samsung’s HBM4 and HBM4E recovery depends on:

customer qualification
TSV process stability
packaging yield
base-die and foundry integration
thermal and power characteristics
NVIDIA and custom-ASIC qualification
mix shift across enterprise SSD, SOCAMM2 and DDR5

The bottleneck is people as much as equipment. Stock compensation can be read as a long-duration incentive designed to retain DS talent through 2028.

This links the two questions:

Samsung expects high DS profits through 2028
 ↓
It allocates part of that profit to employees in stock
 ↓
Employees are exposed to Samsung's share price and DS execution
 ↓
Samsung gets HBM talent retention and lower labor risk

For investors, the structure is two-sided.

Opportunity	Risk
HBM execution recovery, talent retention, reduced labor risk	Higher compensation cost, labor receives part of cycle upside, future overhang
Near-term purchase flow	Not EPS-accretive like cancellation
Internal supercycle signal	Signal weakens if the KRW 200tn hurdle is cumulative rather than annual

7. Final Answer to the Two Questions

Question 1. Did Samsung Internally Recognize a Supercycle Through 2028?

Not as an official declaration. But it can be a strong internal signal under the annual-profit interpretation.

If the hurdle is annual DS operating profit of KRW 200tn: close to yes. 1Q26 DS annualized run-rate was KRW 214.8tn, and embedding that kind of target in compensation through 2028 signals confidence in a high-profit memory cycle.
If the hurdle is cumulative 2026-2028 DS operating profit of KRW 200tn: closer to no. Annual average KRW 66.7tn is far below the current run-rate, making it more of a compensation threshold than a supercycle signal.

The final agreement text is the key.

Question 2. How Much Additional Stock Could Samsung Buy?

Headline case: KRW 76.9-80.0tn, roughly 242-252mn shares, or about 4.1-4.3% of common shares.

But the range is wide:

Scenario	Estimated Stock Purchase	% of Common Shares
Cumulative DS KRW 200tn basis	~KRW 12.6tn	~0.7%
DS annual KRW 200tn x 3 years	~KRW 37.8tn	~2.0%
Consolidated OP forecast KRW 1,220tn	~KRW 76.9tn	~4.1%
Headline net KRW 80tn	~KRW 80.0tn	~4.3%

The most important sentence:

These are not cancellation shares. They are employee-compensation shares. The purchase phase can be positive for flow, but the shares can become future overhang after they are paid to employees.

So the core of the event is not “more shareholder return.”

It is how long Samsung sees the DS/HBM supercycle lasting, and how much of that upside it is willing to use to retain key HBM talent in stock.

Checklist

The next checks are about the formula and the actual execution path.

Checkpoint	What To Confirm
Final agreement text	Is KRW 200tn annual or cumulative?
Bonus denominator	Consolidated OP, DS OP or adjusted OP?
After-tax payout	Actual net payout ratio and withholding method
Purchase mechanics	Market-purchase pace, execution window and shareholder approval
Post-payment overhang	Immediate tradable portion, lockup period and unlock schedule

Until those are known, KRW 80 trillion should be treated as an upper-end scenario, not a fixed number.

Evidence Classification

[Fact]

Samsung reported 1Q26 consolidated revenue of KRW 133.9tn, operating profit of KRW 57.2tn, DS revenue of KRW 81.7tn and DS operating profit of KRW 53.7tn. (Samsung Global Newsroom)
Samsung cited HBM4 and SOCAMM2 mass-product sales for NVIDIA Vera Rubin, HBM4E samples and strong server-memory demand in the second half. (Samsung Global Newsroom)
Samsung’s FY2024-2026 shareholder-return policy includes KRW 9.8tn of annual regular dividends and 50% of FCF returned to shareholders. (Samsung IR)
Samsung’s official IR page shows roughly KRW 10.0tn of 2024-2025 share purchases. (Samsung IR)

[Inference]

The reported gross KRW 130tn and net KRW 80tn appear to be based on KRW 1,220tn of 2026-2028 consolidated operating-profit estimates, a 10.5% pool and an after-tax payout assumption.
If the hurdle is annual DS operating profit of KRW 200tn, the plan can be read as an internal signal that Samsung expects a high-profit memory cycle through 2028.
The stock-bonus purchase can support near-term flow, but it should not be valued like a cancellation-style shareholder-return program.

[Speculation]

Samsung likely chose stock compensation to combine HBM talent retention with share-price-linked incentives.
If HBM4/HBM4E qualification and yield recovery are confirmed, the compensation cost could generate high ROI.

[Blocked]

The exact denominator of the 10.5% bonus pool.
Whether the KRW 200tn 2026-2028 condition is annual or cumulative.
Whether the 60% after-tax payout assumption is accurate.
The actual buyback schedule and daily market-purchase pace.
Employee lockup and sale schedule after shares are paid.

Korea Quality Re-Rating Watch 2026-05-26: GigaVis, VM, Micro Contact Sol — Triple-Screen Compounders Lead the Bull

Tue, 26 May 2026 16:30:00 +0900

Macro Dashboard

Indicator	Level	5-Day Change	Signal
KOSPI	8,047.5	+10.7%	Bullish
KOSDAQ	1,172.5	+8.1%	Bullish
USD/KRW	1,502	−0.4%	Stable
VIX	16.6	−4.6%	Risk-On
US 10Y	4.56%	−0.11pp	Easing
Brent	$95.7	−8.9%	Deflationary Tailwind

Regime verdict: KR Bull / US Bull — both markets in confirmed uptrend. Breadth is building (>50MA: 35%, >200MA: 46%), and the VIX collapse below 17 removes a key overhang. The KR Discovery signal is on Day 19 of confirmation. Stance: aggressive expansion.

Market Wrap

Korea’s session on May 22 had a split personality. KOSPI ground higher in a narrow +0.5% band while KOSDAQ surged +4%, driven by a rotation out of large-cap semiconductors and into biotech, energy, and parts-and-materials names.

The day’s defining theme was breadth expansion over index performance. Screener passes jumped to 99 names, the widest count in several sessions. That breadth was real — but selective. The gain was concentrated in pharmaceuticals and biotech, power/energy equipment, and semiconductor materials and components. Large-cap semiconductor names were a drag, with Samsung Electronics (005930.KS) seeing foreign net selling of over ₩1 trillion and program selling approaching ₩1 trillion.

Samsung Electro-Mechanics (009150.KS) was the standout price performer across the analysis universe, rising +11.3% over the session and +32.7% over five days, reaching a relative strength of 98.9. The catalyst: a Morgan Stanley note circulating through institutional channels upgrading their 2030 ABF substrate shortage forecast from 15% to 22%. The report directly supported the ABF/packaging substrate thesis for names like Samsung Electro-Mechanics and Daeduck Electronics (353200.KQ), though institutional profit-taking accompanied the move.

Flow signals told a nuanced story. Foreign investors were net buyers in niche semiconductor equipment and ABF names while selling large-caps. Institutions rotated out of single-day momentum plays but were net buyers in several materials names. The retail-versus-institution divergence was sharp: in names like Daeduck Electronics, institutions accumulated (+₩12.7B) while foreigners sold (−₩7.4B) — a mixed signal requiring a second-day confirmation.

Jusung Engineering (036930.KQ), which was not part of the core screener set at session time, spiked +21% with combined foreign and institutional buying (“CoBuy”), flagging it as a new discovery candidate. Simtech (036710.KQ) also appeared in the ABF/PCB cluster with moderate institutional support. The broader message: the market is rewarding materials and parts exposure, not just the large-cap chip names.

Today’s Quality Re-Rating Candidates

The Meta Screener ran across 118 tickers and produced a clear top tier. The framework ranks on quality fundamentals, institutional/foreign flow, and re-rating or earnings catalysts — then cross-checks against official DART filings for catalyst confirmation and risk flags.

Top Candidates Table

Rank	Ticker	Name	Meta Score	Overlap	Screeners Hit	Key Metrics
1	420770.KQ	기가비스 (GigaVis)	97.3	4	Quality, Smart Money Quality, Cycle Rerating, Smart Money Earnings	ROE 7.3%, OP YoY +777%, OPM Δ+29.9pp
2	089970.KQ	브이엠 (VM)	94.8	5	Quality, Smart Money Quality, Cycle Rerating, Smart Money Earnings, PEAD	ROE 14.8%, OP YoY +387%, OPM Δ+29.3pp
3	098120.KQ	마이크로컨텍솔 (Micro Contact Sol)	77.8	4	Quality, Smart Money Quality, Smart Money Earnings, PEAD	ROE 23.1%, OP YoY +74%, 5d F+QI +₩8.0B
4	082920.KS	비츠로셀 (Vitzrocell)	71.3	3	Quality, Smart Money Quality, Smart Money Earnings	ROE 17.2%, OP YoY +33%, DART catalyst 1 / risk 1
5	042660.KS	한화오션 (Hanwha Ocean)	69.9	4	Smart Money Quality, Cycle Rerating, Smart Money Earnings, PEAD	OP YoY +391%, OPM Δ+6.9pp, 5d F+QI +₩137.2B
6	058470.KS	리노공업 (Lino Industrial)	64.7	4	Quality, Smart Money Quality, Smart Money Earnings, PEAD	ROE 20.8%, OPM 47.5%, short interest 39.9% ⚠
7	053610.KQ	프로텍 (Protek)	59.9	3	Quality, Cycle Rerating, PEAD	ROE 13.7%, OP YoY +245%, consensus z +1.12
8	329180.KS	HD현대중공업 (HD Hyundai Heavy)	57.5	4	Smart Money Quality, Cycle Rerating, Smart Money Earnings, PEAD	OP YoY +189%, 5d F+QI +₩100.3B
9	147830.KQ	제룡산업 (Jearyong Industrial)	56.8	3	Quality, Cycle Rerating, PEAD	ROE 13.9%, OP YoY +159%, OPM Δ+13.9pp
10	000660.KS	SK하이닉스 (SK Hynix)	56.1	2	Quality, Cycle Rerating	ROE 35.6%, OP YoY +101%, DART buyback catalyst

Deep Dive: Top 3

GigaVis (420770.KQ) — Score 97.3, 4-screener overlap

GigaVis makes inspection equipment for display and semiconductor manufacturing. It clears all four core screeners: Quality Compounder, Smart Money Quality, Cycle Rerating, and Smart Money Earnings. Operating profit grew +777% year-on-year, and operating margin expanded 29.9 percentage points — the kind of inflection that the Cycle Rerating screen is built to catch. The smart money signal is clean: foreigners and quality institutions bought a net ₩7.3B over five days, absorbing ₩3.75B in retail supply. Consensus estimates have been revised upward. There are no DART risk flags; the three recent filings are officer ownership disclosures and a quarterly report — both neutral. What to check next: whether the premium valuation (PER 117x) is justified by continued order momentum, and whether institutional accumulation continues into next week.

VM (089970.KQ) — Score 94.8, 5-screener overlap

VM is the only name in the current universe hitting all five screeners simultaneously. The company produces thin-film deposition equipment for semiconductor fabs. Revenue nearly doubled year-on-year (+105%), operating profit grew +387%, and margin expanded 29.3 percentage points. The PEAD score of +1.67 is the second-highest across all Tier A names, suggesting the post-earnings drift window is still open. Institutional and foreign net buying totaled +₩33.6B over five days, against ₩23.1B of retail supply absorbed. The stock is trading at its 52-week high. Risks: the company has meaningful debt (debt ratio 40%), and valuation at PER 34x prices in continued execution. One DART filing — a Q1 quarterly report — is neutral.

Micro Contact Sol (098120.KQ) — Score 77.8, 4-screener overlap

Micro Contact Sol manufactures test sockets and probe pins used in semiconductor and display testing. It ranks #4 in the Quality Compounder screen with ROE of 23% and near-zero leverage (debt ratio 19%). Revenue grew +44% and operating profit +74%, with an amended Q1 filing confirming the numbers. Smart money flow is positive: ₩8.0B net foreign and institutional buying over five days against ₩5.0B retail supply. The PEAD signal is mild (+0.18), placing it in Tier A but near the lower edge — the drift window exists but is narrower than for VM or GigaVis. Next check: whether the Q1 filing amendment changes any key line items and whether institutional accumulation extends past the May 22 session.

Screener data as of 2026-05-26 16:22 KST. Market close data from 2026-05-22 session; macro regime updated 2026-05-26. All figures sourced from KR_Crawler DB and official DART filings. This post is market analysis, not investment advice.

Korea Invest Insights Thesis Check: What 207 Korean Articles Say About Our Edge

Tue, 26 May 2026 12:30:00 +0900

This is a meta follow-up to the Korea Daily Market Hub and Why Korea Part 5. The goal is not to claim portfolio performance. It is to ask which investment theses published on Korea Invest Insights lined up with subsequent price action, and where the process needs correction.

TL;DR

We collected all 207 Korean-language /ko/post/ articles from the Korea Invest Insights sitemap and mapped them to 593 article-stock pairs.
Using primary mentions only, 527 pairs had calculable performance. The win rate was 54.1%, +10% outcomes were 29.4%, +20% outcomes were 21.1%, and -10% outcomes were 24.3%.
The strongest cluster was AI infrastructure, semiconductors, HBM and substrates. Across 305 primary pairs, the average return was +15.5%, the median was +7.8%, and the win rate was 67.9%.
A separate thesis map compresses the corpus into 14 investment theses plus 53 operating logs. Roughly 39% of non-operating posts sat inside a broad Korea growth / AI-semiconductor risk-on cluster. That was a source of performance, but also a correlation risk.
The weakest areas were gaming, parts of biotech and medtech, K-beauty / consumer, listed VC proxies and some idiosyncratic deep dives. These cases often had good narratives but weaker price bridges.
The core lesson: KII’s edge was strongest when theme diffusion + flows + bottleneck logic aligned. It was weaker when product narratives, binary events or low-liquidity themes were stretched into structural equity theses too early.

This is not investment advice and not an actual portfolio return. It does not include position sizing, execution price, stops, taxes, transaction costs, publication-time effects or realized trading decisions.

What Was Measured

The purpose was to test research signals, not writing quality. The question was: which types of thesis were followed by positive price action?

Item	Basis
Article source	Korean sitemap, Korean index.json, LLM guide
Articles	207
Article-stock pairs	593
Successful performance calculations	557
Primary-mention successful calculations	527
Primary-mention ticker count	166
Price source	Research OS local DB, KR prices_daily + Kiwoom open fallback, US_Crawler prices_daily
Latest price date	2026-05-22
Entry price	Publication-date open, or next trading-day open if the publication date was a market holiday
Latest price	2026-05-22 close

Primary mentions were based on tag_code, tag_name and tag_alias. Title or description aliases were excluded from the core interpretation because they create more false positives.

Overall Scorecard

Metric	Result
Positive-return share	54.1%
+10% or better	29.4%
+20% or better	21.1%
-10% or worse	24.3%
Average return	+7.9%
Median return	+0.6%
25th percentile	-9.3%
75th percentile	+13.9%

The result is constructive but not clean. The mean is positive, but the median is only slightly above zero. That means a relatively small number of large winners pulled up the average. The research process produced right-tail outcomes, but it also carried meaningful left-tail risk.

Thesis Map: The Bigger Risk Is Correlation

The performance workbook answers one question: what went up after publication? The thesis map answers a different question: what were we repeatedly exposed to?

On that lens, 206 posts compress into 14 investment theses and 53 operating logs. The count differs by one from the 207-article sitemap performance run because the two exercises used slightly different grouping rules.

Item	Value	Read
Thesis-map posts	206	Comparable to, but not identical with, the 207-article performance dataset
Investment theses	14	Large idea buckets: stocks, sectors, macro, policy and process
Operating logs	53	Daily Wraps, weeklies and screeners; roughly 26% of the map
Korea growth / AI-semi risk-on cluster	59	Pearl Abyss 25, AI infra 18, Samsung Electro-Mechanics 8, Samsung Electronics 8
Share of non-operating posts	Roughly 39%	59 divided by roughly 150 non-operating posts

Pearl Abyss is not a semiconductor company. The reason it appears in this correlation bucket is that the stock behaved more like a Korea growth / domestic risk-on exposure than a defensive standalone asset. In other words, the map is not a sector taxonomy. It is an exposure map.

The mid-review by thesis looks like this:

Thesis	Posts	Residual Alpha	Mid-Review
Pearl Abyss platform / franchise re-rating	25	Low	Deep tracking, but -28.1% price outcome. The market still wants a 2027 earnings bridge and next-title visibility.
Samsung Electro-Mechanics AI package components	8	Low	Thesis strongly validated by price. The remaining question is execution toward a 2028 OP path, not first-time reclassification.
AI semiconductor infrastructure beyond HBM	18	Medium	Best-performing cluster, but still exposed to an AI CapEx peak-out.
Samsung Electronics memory-cycle to AI-platform reclassification	8	Medium	Re-rating logic remains plausible, but foundry and memory-cycle risks must stay inside the thesis.
Why Korea / Korea discount compression	5	Medium	Structural frame worked, but the actual market was heavily concentrated in Samsung and SK Hynix.
KOSDAQ policy / smart money	8	Medium	Policy direction looks right; execution delays and fund-deployment risk need tracking.
Pamicell perception-change series	4	High	Clean recognition-gap thesis, but still needs direct evidence for the Doosan Electronics BG proxy logic.
Korean AI / sovereign AI / listed VC proxies	6	High	Potential information edge, but the bridge from private-market exposure to listed-stock re-rating is slow.
Humanoid robotics valuation skepticism	4	Medium	One of the few clusters with built-in valuation skepticism. That tone should be used more broadly.
Operating logs	53	System	Trust infrastructure rather than a single alpha claim; weak-side and short/risk screeners are still underbuilt.

Combining the thesis map with the performance table produces a more balanced conclusion. The good outcomes were not random. But many of the good outcomes were also not independent. A large share of the winners sat in the same AI-memory / AI-infrastructure / Korea growth factor complex.

What Worked

First-mention winners were concentrated in AI infrastructure and semiconductor bottlenecks.

Stock	First Mention	Return From First-Mention Open	Read
Samsung Electro-Mechanics	2026-04-09	+161.7%	AI MLCC, FC-BGA, silicon capacitors, power-integrity re-rating
Jeju Semiconductor	2026-05-13	+121.0%	Memory / SoCAMM / earnings-surprise diffusion
Opticore	2026-05-09	+100.0%	High-beta optical / CPO theme
SK Square	2026-04-14	+96.8%	SK Hynix NAV and holding-company discount compression
TES	2026-04-09	+89.7%	Semiconductor equipment, PEAD, smart money
SK Hynix	2026-04-14	+77.8%	HBM leader and AI-memory earnings leverage
Daeduck Electronics	2026-04-20	+61.7%	AI substrates and FC-BGA diffusion
Marvell	2026-04-10	+58.8%	Custom AI silicon and interconnect bottleneck

The common pattern was:

A large theme existed: AI infrastructure, HBM, substrates, power integrity, networking, optics.
Flows confirmed the story: foreigners, institutions or smart-money screens showed demand.
The logic was bottleneck-based: these were not optional products, but components needed for the system to work.
The theme moved down the value chain: from Samsung Electronics and SK Hynix into Samsung Electro-Mechanics, Daeduck, Simtech, TES and Jeju Semiconductor.

The Samsung Electro-Mechanics case is the clearest example. The thesis began with AI substrates and MLCCs, then expanded into silicon capacitors, Murata comparisons and AI-server passive-component bottlenecks. That sequence worked. But it also means the easy part of the thesis may already be priced.

What Did Not Work

The weaker cases were more idiosyncratic.

Stock	First Mention	Return From First-Mention Open	Read
Pearl Abyss	2026-04-04	-28.1%	Gap between game data and equity re-rating
GigaLane	2026-04-17	-27.5%	Low-liquidity neglected-stock volatility
Next Biomedical	2026-04-14	-26.3%	Medtech event risk and limited earnings bridge
OE Solutions	2026-05-09	-25.1%	CPO theme proximity and valuation risk
Openedges Technology	2026-04-25	-23.9%	Long-duration IP thesis versus short-term multiple compression
Hyundai Rotem	2026-05-01	-23.7%	Late entry into a strong defense / rail theme
DSC Investment	2026-04-29	-23.2%	Listed VC exposure not yet translating into equity demand

These cases had different stories, but similar failure modes. Product or industry narratives were often better than the near-term equity bridge. Some themes were directionally right but not directly monetized by the listed stock. Several names were small or high-multiple stocks with weak liquidity, making them more vulnerable when risk appetite softened.

Pearl Abyss is the most important negative case. The game data did not collapse. The issue was that the market wanted a bridge from Crimson Desert success to repeatable earnings: DLC, the next title, capital allocation and the 2027 earnings path. Product data can support a hold thesis. It does not automatically create a stock re-rating.

Cluster Results

Keyword-based clusters sharpen the lesson.

Cluster	Pairs	Tickers	Average	Median	Win Rate	+10% or Better	-10% or Worse
AI infra / semis / HBM / substrates	305	92	+15.5%	+7.8%	67.9%	40.7%	18.7%
Smart money / PEAD / quality	138	59	+11.1%	+4.6%	60.1%	41.3%	23.9%
Power / nuclear / energy	67	33	+9.1%	+2.4%	56.7%	35.8%	31.3%
Robotics / physical AI	31	19	+7.9%	+2.1%	58.1%	29.0%	22.6%
Biotech / healthcare	34	29	+2.7%	-4.2%	35.3%	26.5%	35.3%
Gaming / Pearl Abyss	49	15	+2.0%	-13.9%	34.7%	20.4%	57.1%
Listed VC / venture proxies	18	13	-5.5%	-4.6%	22.2%	11.1%	38.9%
K-beauty / consumer	22	11	-8.7%	-8.5%	22.7%	4.5%	31.8%

The clearest edge was in the AI infrastructure value chain. The second-best signal was the screener process: PEAD, quality and smart-money intersection notes were more robust than pure narrative writeups.

Power and energy were positive but volatile. K-beauty, listed VC, gaming and parts of biotech were weaker because the bridge from story to earnings, flows or timing was less direct.

Process Lessons

1. Separate good thesis from good entry price

A correct thesis can become a poor entry after the market prices it. Samsung Electro-Mechanics is now a valuation question, not just a thesis question. Pearl Abyss is the reverse: product evidence may be strong, but the equity needs a clearer earnings bridge.

2. Make primary tags stricter

Performance attribution depends heavily on tag quality. Comparison names, risk examples and passing mentions should not become primary tags. Core thesis names should always have ticker tags.

3. Separate article types

A Daily Wrap, a screener note, a deep dive and an explainer are not the same product.

Article Type	Correct Performance Lens
Daily Wrap / Screener	Short-term flow and earnings-drift validation
Deep Dive	One-to-three-month thesis validation
Hub / Explainer	Discovery, SEO/GEO and reader navigation value

4. Update failures faster

Losers should not simply be called wrong. They need classification: broken thesis, delayed thesis, overpaid entry, weak liquidity, or missing catalyst. Pearl Abyss, Openedges, K-beauty and listed VC proxies deserve follow-up reviews under that framework.

Current Judgment

The interim read is straightforward:

KII has a research edge, but it is not evenly distributed across all article types or sectors.

The edge is strongest where AI infrastructure bottlenecks, flows and earnings drift overlap. It is weaker where the analysis relies on product quality, binary events, long-duration narratives or listed proxies for private-market exposure.

The thesis map adds one more point: the best-performing posts were not independent exposures. AI memory, AI infrastructure, Samsung Electro-Mechanics, Samsung Electronics, KOSDAQ growth and parts of Pearl Abyss were all tied to Korea risk-on conditions. The next review should therefore track not just returns, but factor concentration and thesis correlation.

The operating change is therefore:

Keep AI infrastructure and semiconductor bottlenecks as core coverage.
Keep Daily Wrap and screener-driven notes, but tighten tagging and tracking.
Apply more price discipline to gaming, biotech, consumer and listed VC stories.
Repeat this performance review monthly, not as a scorecard for bragging, but as a feedback loop for the research process.
Track correlation risk separately. Many posts can still be one large position if they share the same factor exposure.

Data Notes

Facts: 207 Korean articles, 593 article-stock pairs, 557 successful performance calculations, 527 successful primary-mention calculations, 166 primary-mention tickers. Latest price date: 2026-05-22.

Inference: KII’s strongest historical signals came from AI infrastructure, semiconductors, power and smart-money screening. The weaker areas were more idiosyncratic and less directly tied to earnings or flows.

Limitations: This is not actual portfolio performance. It excludes sizing, execution, costs, taxes, stops, publication-time effects and post-publication judgment changes. Posts published after the latest price date are blocked from performance calculation.

Sources

Korea Invest Insights Korean sitemap
Korea Invest Insights Korean index.json
Korea Invest Insights LLM guide
Research OS local DB: KR prices_daily + Kiwoom open fallback, US_Crawler prices_daily
Internal outputs: kii_article_performance_workbook.xlsx, ticker_summary_primary_mentions.csv, article_stock_performance.csv, summary_report.md, generated 2026-05-26 KST

For research and information purposes only. Not investment advice. Past performance does not guarantee future results.

KOSPI Foreign Ownership Is High, But Samsung and SK Hynix Are at 2026 Lows

Tue, 26 May 2026 10:30:00 +0900

This is a follow-up to Korea Foreign Investor Flow Analysis. That note showed that most 2026 foreign selling was concentrated in Samsung Electronics and SK Hynix. This note asks the next question: how much ownership-based selling pressure is left?

TL;DR

KOSPI market-cap-weighted foreign ownership is still high. On the user’s benchmark series, it rose from 36.02% at the start of 2026 to 38.56% as of May 22.
Yet the two stocks actually being sold are already near 2026 lows. Research OS local DB shows Samsung Electronics foreign ownership falling from 52.37% to 48.32%, and SK Hynix from 53.81% to 51.62%.
That divergence exists because KOSPI ownership is value-weighted. Foreigners can sell shares, but if their remaining Samsung/SK Hynix positions surge in price, their market-cap-weighted KOSPI ownership can still rise.
The practical trigger is now flow speed: first stabilization needs KOSPI foreign ownership below roughly 38.2% plus five-day average foreign net selling below about KRW 1tn per day.

The Core Data

Item	Start of Year	Current	Change	Read
KOSPI market-cap-weighted foreign ownership	36.02%	38.56%	+2.54pp	Foreign-owned market value remains high
Samsung Electronics foreign ownership	52.37%	48.32%	-4.05pp	2026 low; 2022+ local DB low
SK Hynix foreign ownership	53.81%	51.62%	-2.19pp	2026 low, but not a long-term trough
Samsung Electro-Mechanics	37.63%	38.21%	+0.58pp	Less ownership pressure
Daeduck Electronics	15.96%	17.23%	+1.27pp	Looks more like accumulation

Yonhap Infomax reported that Samsung Electronics’ 48.32% foreign ownership was the lowest since September 2013, while SK Hynix’s 51.62% was the lowest since May 2023. It also reported YTD foreign net selling of KRW 50.2027tn in Samsung and KRW 34.2279tn in SK Hynix. (Yonhap Infomax)

Samsung’s historical anchor is much higher. In May 2019, Korea JoongAng Daily reported foreign ownership at 57.33%, above the prior 2001 record of 57.30%. (Korea JoongAng Daily)

Why The Aggregate Still Looks High

The KOSPI foreign ownership ratio is a market-value measure, not simply a share-count measure. Seoul Economic Daily explained the gap clearly: foreign selling pulled down the ratio, but price appreciation in foreign-held shares added even more to the ownership value. (Seoul Economic Daily)

That means the aggregate KOSPI ratio says foreign portfolio value is still heavy, while Samsung/SK Hynix share-count ownership says the two key positions have already been reduced materially.

Flow Is Still Not Calm

The problem is speed. From May 18 to May 22, foreigners net-sold KRW 14.4477tn in KOSPI. SK Hynix and Samsung represented roughly 73% of that one-week selling. From May 7 to May 22, foreigners net-sold KRW 46.3383tn, with Samsung and SK Hynix together accounting for 82.9%. (Daum / Energy Economy)

Research OS local all-listed-stock proxy points in the same direction:

Period	Total Foreign Net Buy	Samsung	SK Hynix	Two-Stock Share
May 18-22	-KRW 13.16tn	-KRW 5.28tn	-KRW 5.35tn	80.7%
May 7-22	-KRW 43.31tn	-KRW 18.94tn	-KRW 19.59tn	89.0%
Jan 2-May 22	-KRW 89.20tn	-KRW 50.41tn	-KRW 34.39tn	95.1%

So the ownership level is getting closer to a clearing point, but the flow is still too fast to call the selloff over.

Practical Framework

Stage	Conditions	Interpretation
Selling pressure remains	KOSPI foreign ownership above 38.5% and five-day average selling above KRW 2tn/day	Rebalancing still active
First stabilization	Below 38.2% and five-day average selling below KRW 1tn/day	Selling speed has slowed
Meaningful stabilization	Below 38.0% and Samsung/SK Hynix ownership stops falling	Mechanical selling largely absorbed
Re-buying setup	37.5-38.0% stable, foreign net buying for 3-5 sessions, KRW stabilizes	Korea beta can be re-entered

Investment Read

For Samsung Electronics, the foreign share-count ratio already looks low. Below 48% would look like a late-stage reduction signal. But the stock still needs foreign selling speed to slow before new money gets a cleaner entry.

For SK Hynix, 51-52% is a lower-neutral zone, not a full trough. A move toward 50.5-51.0% with price resilience would make the risk/reward more comfortable.

The final line is simple: KOSPI-level foreign ownership still argues for rebalancing pressure, but Samsung and SK Hynix ownership already argues that the selling is closer to exhaustion than the headline flow suggests. This is not a buy-now signal. It is a wait-for-flow-to-slow signal.

Disclaimer: For research and information purposes only. Not investment advice.

AI Server Passive-Component Bottlenecks: Why Tiny Power-Stability Parts Now Matter

Tue, 26 May 2026 00:00:00 +0000

Follow-up to the Samsung Electro-Mechanics series. See the pieces on SEMCO at KRW 100T, the KRW 1.5T silicon-capacitor contract, MLCC vs silicon capacitors, and AI chip-design bottlenecks. Related hubs: AI HBM, AI PCB/substrate, and Korea semiconductor value chain.

TL;DR

The AI server passive-component bottleneck is not a GPU shortage. It is the need for more and higher-spec parts that buffer, filter, and stabilize the power that GPUs consume.
Think of an AI server as a high-performance engine. MLCCs, silicon capacitors, and inductors are the fuel-pressure regulators, shock absorbers, and filters.
NVIDIA says a DGX GB200 rack consumes roughly 120kW. Lenovo’s GB300 NVL72 guide cites 135kW TDP and up to 155kW peak per rack. (NVIDIA, Lenovo)
The investable bottleneck is not generic MLCC. It is high-capacitance, low-ESR, low-noise, low-profile AI server components.
Samsung Electro-Mechanics matters because it can connect MLCC + FC-BGA + silicon capacitors in one AI package power-integrity stack.

One-Sentence Explanation

AI server passive-component bottlenecks mean GPUs now draw such large and fast-changing currents that the small parts stabilizing voltage near the chip become a performance bottleneck.

GPU performance is usually explained through GPUs, HBM, and networking. That is true, but those chips need a stable power base. When GPU load changes, current can swing rapidly. If voltage droops or noise spikes, the system can lose performance or stability.

What Passive Components Do

Component	Simple analogy	Role in an AI server
MLCC / capacitor	Water tank, shock absorber	Buffers current and absorbs voltage noise near chips and boards
Silicon capacitor	Tiny emergency battery next to the GPU	Stabilizes power inside or very near GPU/HBM packages
Inductor	Current inertia device	Smooths current in voltage conversion
Resistor / filter / ferrite	Speed limiter, noise filter	Reduces noise on high-speed signal and power lines
VRM-side parts	Power kitchen for the GPU	Converts 48V/12V rails into sub-1V GPU voltage

TDK describes the data-center power chain as UPS → PSU → IBC → VRM → CPU/GPU voltage, with efficiency, ripple, heat tolerance, and long-term reliability required at every step. (TDK)

Why AI Servers Changed the Bottleneck

GPUs and CPUs can run below 1V, while current can change by tens to hundreds of amperes depending on load. Samsung Electro-Mechanics says high-capacitance MLCCs near the GPU must act as current buffers for stable operation. (Samsung Electro-Mechanics)

The bigger issue is peak power, not only average power. NVIDIA’s GB300 NVL72 PSU with energy storage is designed to smooth AI workload power spikes and reduce peak grid demand by up to 30%. (NVIDIA Developer)

That logic travels down the stack. Rack-level power smoothing matters, but chip-near smoothing matters too. The closer the component is to the GPU/HBM package, the more valuable it becomes.

MLCC vs Silicon Capacitor

Item	Location	Role	Analogy
MLCC	Many places on board and near chips	Broad power stabilization	Fire extinguishers placed across the building
Silicon capacitor	Inside or right next to GPU/HBM package	Ultra-close transient suppression	Fuel-pressure control next to the engine cylinder

Samsung Electro-Mechanics announced a roughly KRW 1.5 trillion silicon-capacitor supply contract for 2027-2028. The company says the product improves power stability inside high-performance AI server semiconductor packages and has far lower ESL/ESR than conventional MLCCs. (Samsung Electro-Mechanics)

Samsung Electro-Mechanics Read-Through

The SEMCO thesis is not “MLCC prices go up.” It is:

AI rack power rises
 → transient power swings become harder to manage
 → GPU/HBM packages need better power integrity
 → high-end MLCC, FC-BGA, and silicon capacitors gain strategic value

That is why SEMCO’s rerating is tied to AI package power integrity, not only the legacy smartphone component cycle.

The KPIs to watch:

AI server MLCC ASP and customer qualification
Silicon-capacitor revenue recognition from 2027
Additional silicon-capacitor customers or platforms
AI server and networking FC-BGA growth
Group operating margin moving toward the high-teens or 20% path

Bottom Line

Passive components used to look like cheap background parts. In AI servers, they are becoming part of the performance stack. The GPU is the engine, HBM is the high-speed fuel tank, the substrate is the road, and MLCC/silicon capacitors keep the engine from stuttering.

That is the technical foundation behind the Samsung Electro-Mechanics, Murata, and TDK rerating thesis. It does not make every price attractive, but it explains why the category now deserves a different lens.

Evidence Map

Facts: NVIDIA GB200 rack power, Lenovo GB300 rack power, TDK power-chain architecture, Samsung Electro-Mechanics AI MLCC and silicon-capacitor disclosures.

Inference: The bottleneck is a combination of higher quantity, higher specification, closer placement, and harder qualification.

Blocked: SEMCO’s exact silicon-capacitor customer, product margins, and platform-by-platform backlog are not public.

Samsung Electro-Mechanics at KRW 100T: Can It Overtake Murata and Hyundai Motor?

Tue, 26 May 2026 00:00:00 +0000

Follow-up in the Samsung Electro-Mechanics series. See the earlier pieces on AI infrastructure rerating, the KRW 1.5 trillion silicon-capacitor contract, and MLCC vs silicon capacitors. Related hubs: Korea semiconductor value chain, AI PCB/substrate, and AI HBM.

TL;DR

A Hyundai Motor market-cap overtake is now mathematically visible. At KRW 1,340,000 and 74.69 million shares, Samsung Electro-Mechanics is worth about KRW 100.1 trillion. To exceed Hyundai Motor common stock at roughly KRW 134.1 trillion, SEMCO would need about KRW 1.80 million per share, or +34% from the reference price. A KRW 150 trillion benchmark requires about KRW 2.01 million, or +49.9%. (Samsung Electro-Mechanics, Hyundai)
Murata is proof that the ceiling exists. Murata guides FY3/2027 revenue of JPY 1.96 trillion, operating profit of JPY 380 billion, and 19.4% operating margin. It explicitly cites server capacitors and power modules as growth drivers. (Murata)
But SEMCO is not cheap here. Q1 2026 normalized operating profit annualizes to roughly KRW 1.41 trillion, making the current market cap about 71x normalized operating profit. The stock is already discounting 2027-2028 AI component earnings.
A durable KRW 134-150 trillion market cap needs a 2028 operating-profit path of roughly KRW 3.8-4.5 trillion or more. The silicon-capacitor contract matters less as one revenue line and more as proof that SEMCO has entered the GPU/HBM package power-integrity socket. (Samsung Electro-Mechanics)
My stance is Watchlist / Wait. The long-term thesis is stronger, but the stock now needs earnings revisions, not just narrative.

1. The Math

Samsung Electro-Mechanics’ official IR page shows KRW 1,340,000 per share and 74,693,696 listed shares as of the reference date. That implies:

SEMCO market cap = KRW 1,340,000 × 74,693,696 = KRW 100.09 trillion

Hyundai Motor common shares at KRW 655,000 and 204,757,766 common shares imply roughly KRW 134.1 trillion. (Hyundai)

Benchmark	Target market cap	Required SEMCO share price	Upside from KRW 1.34m
Hyundai Motor common stock	KRW 134.1T	KRW 1.796m	+34.0%
Murata rough KRW market cap	around KRW 138T	KRW 1.849m	+38.0%
KRW 150T round number	KRW 150.0T	KRW 2.008m	+49.9%

The point: a temporary overtake is no longer fantasy. The harder question is whether it can be sustained.

2. What Murata Proves

Murata is the most useful global benchmark because it shows how the market prices AI server passive components. Its FY3/2027 guidance is:

Item	FY3/2026	FY3/2027 guidance	Change
Revenue	JPY 1.8309T	JPY 1.9600T	+7.1%
Operating profit	JPY 281.8B	JPY 380.0B	+34.8%
Net income	JPY 233.9B	JPY 293.0B	+25.3%
Operating margin	15.4%	19.4%	+4.0pp
Capacitors revenue	JPY 936.4B	JPY 1.0617T	+13.4%
Capacitors mix	51.1%	54.2%	+3.1pp

Murata is already being priced as an AI infrastructure passive-component bottleneck, not just a cyclical electronics supplier.

But that does not automatically mean Samsung Electro-Mechanics deserves the same multiple. Murata has higher capacitor purity and a more proven margin profile. SEMCO has a broader but messier portfolio: MLCC, FC-BGA, optical modules, and now silicon capacitors.

3. SEMCO’s New Frame

SEMCO’s Q1 2026 revenue was KRW 3.209 trillion, operating profit was KRW 280.6 billion, and operating margin was 8.7%. Adding back KRW 71.4 billion of one-off severance cost, normalized operating profit is about KRW 352.0 billion, or an 11.0% margin. (Samsung Electro-Mechanics)

Division	Q1 2026 revenue	YoY	Read-through
Components	KRW 1.409T	+16%	AI server, power, and networking equipment revenue growth
Package Solution	KRW 725B	+45%	AI/server FC-BGA growth and new big-tech networking substrates
Optics Solution	KRW 1.076T	+5%	Camera modules; still a margin-dilution risk

The KRW 1.5 trillion silicon-capacitor contract for 2027-2028 is the real classification event. The product goes inside high-performance semiconductor packages such as AI server GPUs and HBM, stabilizing power supply. (Samsung Electro-Mechanics)

Even if annualized revenue contribution is only around KRW 0.75 trillion, the strategic value is larger: SEMCO has passed qualification for a package-internal power-integrity socket.

4. The Valuation Bridge

Use a simple formula:

Justified market cap = Revenue × Operating margin × after-tax conversion × P/E

Assume 75% after-tax conversion from operating profit to net income.

Scenario	Revenue	OPM	OP	Net income proxy	P/E	Justified market cap	Overtake?
Bear	KRW 15T	12%	KRW 1.80T	KRW 1.35T	30x	KRW 40.5T	No
Base-upside	KRW 18T	18%	KRW 3.24T	KRW 2.43T	40x	KRW 97.2T	Around current
Bull / Murata path	KRW 22T	22%	KRW 4.84T	KRW 3.63T	40x	KRW 145.2T	Hyundai common overtake
Extreme	KRW 24T	25%	KRW 6.00T	KRW 4.50T	35x	KRW 157.5T	KRW 150T+ possible

The message is blunt: a durable Hyundai/Murata-level market cap requires KRW 20T-plus revenue, 20%-plus OPM, and a 35-40x earnings multiple.

5. Investment View

Samsung Electro-Mechanics is now a stronger long-term AI infrastructure component thesis. But at KRW 100 trillion, the stock is no longer a simple “cheap Korean parts supplier” idea.

Entry conditions

Q2-Q3 normalized group OPM at 12-13% or higher
Components plus Package revenue growth above 25%
AI server MLCC ASP or shortage showing up in earnings
FC-BGA revenue from AI/server/networking customers accelerating
Additional silicon-capacitor customers or platform wins
Consensus beginning to price a 2028 operating-profit path above KRW 4 trillion

Invalidation

AI MLCC ASP rolls over with general MLCC pricing
Package Solution growth falls below 20%
The silicon-capacitor contract remains one customer and one program
Group OPM stays near 12% into 2027
Optics keeps diluting the margin bridge
Murata, TDK, Yageo, or others ease AI passive-component tightness earlier than expected
Hyperscaler AI capex enters a digestion phase

Bottom Line

Samsung Electro-Mechanics at KRW 100 trillion is a symbol. It says the market is no longer treating the company as a smartphone component vendor. It is starting to price SEMCO as an AI server power-integrity and package-substrate bottleneck.

Could it overtake Hyundai Motor? Yes. Could it sustain a KRW 150 trillion valuation? Only if the 2028 operating-profit path moves toward KRW 4 trillion-plus.

Murata opens the ceiling, but it also raises the standard. From here, SEMCO needs earnings revisions to defend the narrative.

Pearl Abyss Long-Tail Check: Crimson Desert Holds, Shorts Fade, Foreigners Absorb

Mon, 25 May 2026 21:20:00 +0900

Pearl Abyss × Crimson Desert series Neowiz and CD Projekt comparables / KOSDAQ smart money and Pearl Abyss / Flow bottom test / Patch 1.07 weekend data / Pearl Abyss hub

The Pearl Abyss debate has moved from “did Crimson Desert sell?” to “how long can the sales tail last?” The May 25 data is clear enough: the game data supports the long tail, short pressure is easing, and foreign investors are absorbing supply. The stock has not fully priced that combination yet.

Key Takeaways

Crimson Desert is not a collapsing game. As of May 25, 20:20 KST, Steam sales ranks were global #18, Korea #5, US #13, and China #17. For a day-68 premium game, that is upper-tail persistence.
CCU is decaying normally, but purchase-conversion proxies improved. Average CCU for the weekend-plus-holiday window fell 12.8% vs the prior comparable window, but new reviews rose 5.2%, reviews/day rose 4.7%, and new-review positivity improved to 89.0%.
Patch 1.08 improved quality signals. Since May 22, 17:20 KST, there were 1,318 new reviews with 90.0% positivity. The sequence is patch, review recovery, then US/global rank defense.
Short pressure is easing. Short-sale balance fell from roughly KRW 131.2B in early April to KRW 62.1B as of May 20. Short-sale value ratio dropped from 32.1% on May 13 to 2.7% on May 22.
Foreign investors are absorbing. From April 1 to May 22, foreigners bought KRW 56.6B net, individuals bought KRW 19.2B, and institutions sold KRW 76.1B. Over the latest five trading days, foreigners bought KRW 15.1B while individuals sold KRW 14.9B.
Conclusion: Hold thesis strengthened. More aggressive buying still needs official 6M units, DLC/expansion roadmap, DokeV visibility, or a capital-return message.

1. Latest Snapshot

Metric	Latest
CCU	38,254
Today peak CCU	60,687
Cumulative positive reviews	130,176
Cumulative negative reviews	24,665
Total reviews	154,841
Cumulative positive rate	84.07%
Global sales rank	#18
Korea	#5
US	#13
China	#17

Estimated units are now close to the next milestone.

Anchor	Estimated units	Remaining to 6M
Total-review anchor	5.926M	about 74k
Positive-review anchor	5.940M	about 60k

The practical estimate is roughly 5.93M units. An actual 6M crossing in late May to early June is plausible, although the official announcement may lag.

2. Game Data: The Sales Tail Is Holding

Region	Current read	Investment read-through
Korea	Improved vs early May, around #4-5	Core fandom and patch response remain intact
US	Recovered to #12-13 after Patch 1.08	New-purchase conversion is still alive in the key AAA market
China	Holding #15-20	Steam China demand remains sticky
Global	Re-entered #15-18	Upper long-tail defended

The US recovery matters most. In a tough AAA market, a return to the low-teens rank is not just existing-user retention. It implies that new buyers are still converting.

Weekend Plus Holiday Comparison

Metric	Prior comparable window	This weekend plus holiday	Change
Average CCU	48,705	42,475	-12.8%
Peak CCU	71,867	60,687	-15.6%
New reviews	1,053	1,108	+5.2%
New-review positivity	85.2%	89.0%	improved
Reviews/day	371.6	389.2	+4.7%
Global average rank	#19.9	#19.0	improved
Best global rank	#16	#14	improved
Korea average rank	#6.6	#4.8	improved
US average rank	#18.6	#17.0	improved
China average rank	#16.4	#17.1	stable

CCU decay is normal for a two-month-old premium game. The important question is whether purchase conversion decays with it. This window says no: CCU fell, but reviews and ranks improved.

3. Reviews: New Buyers Still React Well

Daily reviews are off the launch peak, but the recent month has held a 250-500/day range. The recent 7-day positive rate recovered to the 85-90% zone. Since Patch 1.08:

Metric	Since Patch 1.08
Window	May 22 17:20 to May 25 20:20
Average CCU	41,278
Peak CCU	60,687
New reviews	1,318
New-review positivity	90.0%
Global average rank	#20.2
Korea average rank	#5.0
US average rank	#19.0
China average rank	#17.4

This is not the pattern of a game only being held up by remaining users. New reviews continue, review quality improved, and US/global ranks recovered.

4. Short-Sale Data: Pressure Is Easing

Metric	Peak / prior	Latest
Short-sale balance	about KRW 131.2B in early April	KRW 62.1B as of May 20
Short-sale value ratio	32.1% on May 13	2.7% on May 22

Short selling spiked around the May 13 earnings/event window, but both trading intensity and balance have since moved lower. KRX balance data is delayed, so the latest balance read is May 20, not May 22. Still, the direction is easing pressure.

5. Flow: Foreigners Are Absorbing Supply

Investor	Net buying, Apr 1-May 22
Foreigners	+KRW 56.6B
Individuals	+KRW 19.2B
Institutions	-KRW 76.1B

Investor	Latest 5 trading days
Foreigners	+KRW 15.1B
Individuals	-KRW 14.9B
Institutions	-KRW 0.7B

Since April, foreigners and individuals bought while institutions sold. Recently, individuals have been selling and foreigners have absorbed that supply. This is not a pure retail accumulation setup; it is foreign absorption plus institutional selling digestion.

6. Integrated Read

Layer	Current signal	Read
Sales ranks	Korea, US, and global recovered; China holds Top 20	Upper long-tail defended
Reviews	New reviews hold; post-1.08 positivity at 90%	New-buyer response improved
Shorts	Balance down, value ratio collapsed	Short pressure easing
Flow	Foreign net buying, institutional selling	Repricing delayed, not thesis broken

Pearl Abyss has improving game data and easing short pressure, but repricing is still delayed by institutional selling and the market’s “one-off hit” suspicion.

7. Investment Read

Stance: Hold / add only after triggers.

The data supports holding because the game is holding, new-buyer reactions improved, short pressure eased, and foreigners are absorbing supply. But adding capital requires official catalysts.

Add trigger	Meaning
Official 6M units	Next proof of durable sales
DLC / expansion roadmap	One-off package sale becomes repeatable IP revenue
DokeV visibility	Validates the engine and AAA pipeline thesis
Capital-return message	Converts CCP sale proceeds and 2026 cash flow into shareholder value

Failure signal	Meaning
Global rank falls outside #25-30 after the holiday	Upper long-tail breaks
US rank falls back to #25-30	Key AAA market conversion weakens
Daily reviews fall below 200 and positivity slips to low 80s	New inflow and quality trust weaken together
Short ratio rises again and foreign buying stops	Price pressure resumes

Next checkpoint is Tuesday after the holiday effect fades. If global remains inside #20, US stays #15-20, and Korea keeps Top 5, the long-tail thesis strengthens again.

Data source: Research OS local DB, Steam sales-rank and review tracking, KRX short-sale data, and investor-flow data. Unless otherwise stated, timestamps are May 25, 2026 20:20 KST or May 22, 2026 market-close data.

Disclaimer: For research and information purposes only. Not investment advice.

Korea Quality Re-Rating Watch 2026-05-25: VM, RFHIC and MicroContact Sol

Mon, 25 May 2026 16:30:00 +0900

Macro Dashboard

Indicator	Level	5-Day Δ	Signal
KOSPI	7,847.7	+4.4%	📈 Bull
KOSDAQ	1,161.1	+4.5%	📈 Bull
VIX	16.6	−7.9%	🟢 Stable
US 10Y	4.56%	−0.07pp	→ Neutral
USD/KRW	1,514	+1.4%	⚠️ KRW soft
Brent	$100.2	−9.9%	📉 Falling

Regime: Korea Bull / US Neutral → Operative stance: Selective KR expansion

Korea and the US are diverging at the regime level. Breadth metrics on the KR side — 36% above the 50-day MA, 48% above the 200-day — are not euphoric, but sufficient to support active stock selection. KRW weakness (+1.4% in five sessions) adds currency drag for USD-based investors; the Brent slide is a mild demand-signal flag worth watching for Korea’s energy-exposed industrials.

Market Wrap

Close briefing data below is from the most recent available session (2026-05-22). Macro regime data is current as of 2026-05-25.

The May 22 session ran as selective risk-on with stock-level dispersion rather than a clean index sweep. KOSPI held near +0.5% intraday while KOSDAQ pushed toward +4% — a gap that signals the real rotation was inside the small and mid-cap growth universe, not in large-cap index names.

Sector leadership was meaningfully broader than the prior week’s semiconductor-dominated move:

Pharma and biotech led the tape with the widest breadth gains of any sector on the day
Power equipment and energy infrastructure names remained active, building on the multi-week theme around grid upgrade and transformer demand
Advanced packaging and ABF substrates extended the prior week’s move. A Morgan Stanley note circulating in market channels raised its 2030 ABF substrate supply-shortage forecast to a 22% deficit (from 15%), reinforcing the demand narrative for Korean PCB and substrate makers. Samsung Electro-Mechanics (009150.KS) posted the session’s strongest large/mid-cap combination — a 1-day gain of +11.3%, 5-day +32.7%, RS percentile 98.9 — with foreign inflows despite institutional profit-taking. Daedeok Electronics (000240.KS) added +3.3% with institutional buying against light foreign selling, a cross-current that signals the stock is in a hand-change phase rather than a clean accumulation

Weak spots: Samsung Electronics (005930.KS) drew ₩1.01 trillion in foreign net selling and nearly ₩1 trillion in program trade outflows on the day — this despite a solid 5-day recovery of +8.1%. The large-cap memory trade appears to be in a confirmation-seeking mode after a sharp bounce: buyers want to see the foreign supply absorbed before re-engaging. Telecom names were flat to down with dual foreign/institutional selling; the sector is functioning as a defensive placeholder rather than a conviction holding in a broad risk-on environment.

Screener universe breadth expanded to 99 passing names under Bull regime conditions — a meaningful acceleration from prior weeks. The signal: stock-picking conditions are improving, but at 36% above the 50-day MA, the market is not yet in a breadth thrust that would argue for aggressive position-building. Quality rotation over momentum chasing remains the operative frame.

Today’s Quality Re-Rating Candidates

The 2026-05-25 meta-screener processed 137 tickers across five frameworks: Quality Compounder (QC), Smart Money Quality (SMQ), Cycle Rerating (CR), Smart Money Earnings (SME), and Post-Earnings Drift (PEAD). Names appearing in three or more screeners are the primary candidates; the five-screener sweep by VM (브이엠) is exceptional.

Candidate Table

Rank	Ticker	Name	Meta Score	Overlap	Screeners	Key Metrics
1	089970.KQ	VM (브이엠)	94.7	5	QC + SMQ + CR + SME + PEAD	ROE 14.8%, OP YoY +387%, Δmargin +29.3pp, RS 97
2	098120.KQ	MicroContact Sol (마이크로컨텍솔)	77.2	4	QC + SMQ + SME + PEAD	ROE 23.1%, OP YoY +73.5%, RS 88
3	218410.KQ	RFHIC	73.9	4	QC + SMQ + SME + PEAD	OP YoY +1,919%, RS 98, short 7.2% ⚠️
4	039030.KQ	EO Technics (이오테크닉스)	64.2	4	QC + SMQ + SME + PEAD	OP YoY +159%, RS 92, short 19.6% ⚠️
5	095610.KQ	TES (테스)	73.3	3	QC + SMQ + SME	ROE 14.5%, OP YoY +50.3%, new contract catalyst
6	356860.KQ	TLB (티엘비)	60.7	3	SMQ + CR + SME	OP YoY +665%, Δmargin +8.2pp, RS 94
7	147830.KQ	Jerong Industries (제룡산업)	56.5	3	QC + CR + PEAD	ROE 13.9%, OP YoY +158.5%, Δmargin +13.9pp
8	322000.KS	HD Hyundai Energy Solutions	56.2	3	QC + CR + PEAD	OP YoY +1,077%, RS 96, Δmargin +7.5pp
9	000660.KS	SK Hynix	64.6	2	QC + CR	ROE 35.6%, consensus z-score +1.11, RS 98
10	053610.KQ	Protec (프로텍)	57.3	2	QC + PEAD	ROE 13.7%, OP YoY +244.5%, consensus z-score +1.10

QC = Quality Compounder | SMQ = Smart Money Quality | CR = Cycle Rerating | SME = Smart Money Earnings | PEAD = Post-Earnings Drift

Top 3 in Focus

1. VM (089970.KQ) — Meta Score 94.7 | Five-Screener Sweep

VM makes semiconductor process equipment — primarily diffusion furnaces and CVD systems — for memory and logic fabs. It is the only name in the 137-ticker universe to clear all five screener layers simultaneously, making it the most structurally supported article candidate in today’s output.

Quality layer: The business has executed a sharp earnings re-rating. Operating profit grew +387% year-on-year with margin expansion of +29.3 percentage points — the second-highest margin improvement figure in today’s entire Cycle Rerating screener. ROE of 14.75% is solid for a Korean capital equipment name at this stage of the cycle. The Quality Compounder score of 0.887 (ranked #9 in a 142-stock field) confirms the fundamentals pass the durability test, not just the momentum test.

Money flow layer: Over the last five sessions, foreign buyers and quality institutional accounts absorbed ₩33 billion while retail sold ₩21.5 billion net — a clean transfer-of-ownership pattern. Smart Money Quality ranked VM at #5; Smart Money Earnings ranked it #8.

Re-rating / timing layer: PEAD score of +2.30 is the highest in Tier A today, with the stock up +19% in five days and sitting at its 52-week high (0% gap). The Q1 2026 DART filing (May 15) carries no risk flags. Key question for further research: whether the post-earnings drift window still has momentum left after the recent five-day run, or whether a consolidation period is required before the next leg.

2. MicroContact Sol (마이크로컨텍솔, 098120.KQ) — Meta Score 77.2 | Four-Screener Hit

MicroContact Sol manufactures test sockets and precision contact components used in semiconductor device testing — a niche but structurally important segment that benefits directly from rising test complexity in advanced-node chips and HBM stacks.

Quality layer: ROE of 23.1% is among the higher values in today’s screener output, and one of the cleaner profitability profiles in the electronic components segment. Operating profit grew +73.5% year-on-year with a 3.0pp margin increase, suggesting genuine operating leverage rather than revenue-only growth. Quality Compounder rank #7 of 142.

Money flow layer: Foreign and quality institutional buyers accumulated ₩7.5 billion against ₩7.4 billion of retail supply in the last five sessions — a tight, nearly one-for-one absorption. Smart Money Quality ranked the name #1 in the universe today; Smart Money Earnings ranked it #3.

What to check next: Two DART filings stand out — a Q1 2026 amendment filed May 22 (alongside the original May 14 report). The amendment scope should be reviewed before sizing conviction; amendments can be administrative corrections or they can flag material restatement. The PEAD score of −0.21 (ranked #24) is not a positive drift signal — meaning the post-earnings price reaction has been muted. The bull case rests more on quality and flow than on price momentum, which may suit patient accumulation.

3. RFHIC (218410.KQ) — Meta Score 73.9 | Four-Screener Hit | Short Interest Flag

RFHIC designs and manufactures GaN-on-SiC RF power amplifier components — the core active element in 5G macro base stations and defense/radar electronics. This is a structurally differentiated materials position: GaN-on-SiC offers superior power density and efficiency relative to GaAs alternatives, and RFHIC holds meaningful share in the Korean supply chain for telecom OEMs.

Quality and earnings layer: The operating income growth figure of +1,919% year-on-year looks dramatic in isolation; it reflects a return from near-zero profitability in 2025 to a 16.6% operating margin, driven by a 5G infrastructure build-out cycle and improving defense electronics volumes. Revenue growth of +61.7% confirms the demand recovery is genuine, not just base-effect arithmetic. The RS percentile of 98 places RFHIC among the top relative-strength names in the Korean universe; the five-day price gain of +7.1% is still within the pre-blow-off range.

Money flow layer: Foreign and institutional buyers added ₩13.9 billion over five sessions against ₩16.1 billion of retail supply — directionally consistent with institutional conviction building. Smart Money Earnings ranked RFHIC #2 today; PEAD score of +1.72 (Tier A, ranked #2) confirms post-earnings drift momentum is still active.

Caution: Short interest of 7.2% of float with increasing stock lending (대차잔고 expanding) is a risk that cannot be ignored. Financial investment accounts are also net sellers. RFHIC screens at the top of the quality and money flow layers, but the short overhang introduces gap risk on negative catalysts. Preferred approach: treat as a monitoring candidate and require the borrow trend to stabilize or reverse before adding meaningfully to sector exposure here.

All screener data is from 2026-05-25. Close briefing flow data is from 2026-05-22. Screener signals are research candidates, not buy or sell recommendations. DART official filings are the primary catalyst and risk verification source.

Korea Foreign Investor Flows: 95% of the 2026 Net Selling Came From Samsung Electronics and SK hynix

Sun, 24 May 2026 23:55:00 +0900

This is a follow-up to Why Korea Part 5: Korea’s Stock Market Is Now Near Global No. 6. That post asked why Korea’s market cap has exploded. This one asks a more practical question: who was buying, and who was selling?

TL;DR

Foreign investors did not simply abandon Korea. They sharply reduced Samsung Electronics and SK hynix, the two AI-memory mega-caps.
Research OS local DB shows foreign net selling of KRW 89.2T across 2,730 Korean stocks from Jan. 2 to May 22, 2026. Samsung Electronics contributed KRW -50.4T and SK hynix KRW -34.4T. Together: KRW -84.8T, or roughly 95% of total net selling.
Samsung Electronics rose 127.6% YTD while foreigners sold KRW 50.4T and foreign ownership fell from 52.37% to 48.39%.
SK hynix rose 186.7% YTD while foreigners sold KRW 34.4T. This looks less like thesis abandonment and more like distribution after an extreme HBM-led rally.
The portfolio conclusion is: hold core memory exposure, avoid chasing, and watch whether foreign buying rotates into non-memory sectors with real earnings.

1. The Main Decomposition

Segment	Foreign net buying
Korea total, 2,730 stocks	KRW -89.2T
Samsung Electronics	KRW -50.4T
SK hynix	KRW -34.4T
Ex-Samsung Electronics	KRW -38.8T
Ex-Samsung Electronics + SK hynix	KRW -4.4T

The key line is the last one. Excluding Samsung Electronics and SK hynix, the foreign outflow shrinks to KRW -4.4T. The better description is therefore:

Foreign investors were not primarily selling Korea. They were cutting Korea’s AI-memory mega-cap exposure.

Seoul Economic Daily similarly reported that from May 7 to May 22, SK hynix and Samsung Electronics accounted for more than 83% of foreign net selling on the KOSPI. (Seoul Economic Daily)

2. Samsung Electronics: Strong Price, Foreign Distribution

Item	Value
Period	2026-01-02 to 2026-05-22
Price	KRW 128,500 → KRW 292,500
YTD return	+127.6%
Intraday high	KRW 300,500, May 22
Foreign net buying	KRW -50.4T
Institutional net buying	KRW +6.3T
Retail net buying	KRW +29.2T
Foreign ownership	52.37% → 48.39%

Samsung Electronics remains a core Korean equity. But this was not a foreign-led rally. Foreigners sold through most of the move, while domestic liquidity absorbed the supply.

The most important interval was May 7-May 22. Samsung rose another 7.7%, while foreigners sold KRW 18.9T. That is a sign of a strong market, but also a sign of heavy distribution near the top. Yonhap also reported that on May 7 Samsung reached the “270,000 won” level despite large foreign selling in both Samsung Electronics and SK hynix. (Yonhap)

3. SK hynix: Even Stronger HBM Leadership, Same Distribution Pattern

Item	Value
Period	2026-01-02 to 2026-05-22
Price	KRW 677,000 → KRW 1,941,000
YTD return	+186.7%
Intraday high	KRW 1,995,000, May 15
Foreign net buying	KRW -34.4T
Institutional net buying	KRW +9.2T
Retail net buying	KRW +24.1T
Foreign ownership	53.81% → 51.66%

SK hynix is the purer HBM beta, which explains why it outperformed Samsung. Foreign selling here does not mean the company is weak. It means a very successful position is being reduced.

The fundamentals remain powerful. SK hynix reported KRW 52.5763T in 1Q26 revenue, KRW 37.6103T in operating profit, and a 72% operating margin, citing strong AI infrastructure demand and high-value memory products. (SK hynix / PR Newswire)

4. What Foreigners Bought Instead

Foreign buying was much more diversified than foreign selling.

Rank	Stock	Foreign net buying
1	Doosan Enerbility	KRW +2.15T
2	Celltrion	KRW +1.42T
3	Samsung SDI	KRW +1.25T
4	APR	KRW +1.15T
5	Hanwha Ocean	KRW +0.76T
6	Doosan Robotics	KRW +0.65T
7	SK Inc.	KRW +0.63T
8	Hyundai Engineering & Construction	KRW +0.60T
9	Doosan	KRW +0.56T
10	POSCO Holdings	KRW +0.52T

The rotation map is clear: energy infrastructure, biotech, batteries, K-beauty, shipbuilding/defense and robotics received selective inflows. Opinion News described a similar rotation from semiconductor profit-taking toward Doosan Enerbility, Celltrion, Samsung SDI, APR and Hanwha Ocean. (Opinion News)

5. Investment Read-Through

Asset or basket	View	Rationale
Samsung Electronics	Hold core, pause additions	Strong price action, but foreign structural rebuying is not yet visible
SK hynix	Watch, no chase	HBM leadership is real, but foreign selling is extreme
Celltrion	Buy on pullback candidate	Foreign buying plus lower correlation to AI hardware
Samsung SDI	Conditional buy	Strong foreign accumulation, but battery-cycle recovery still needs proof
Hanwha Ocean	Watchlist	Good shipbuilding/defense setup, but recent foreign flow cooled
Doosan Enerbility	Hold/watch	Strong YTD foreign buying, weaker recent flow
Doosan Robotics / Padu / Seojin System	Avoid chasing	Strong momentum, but price risk is high

Bottom Line

The right sentence is not “foreigners sold Korea.”

It is:

Foreigners sold Korea’s AI-memory mega-caps while selectively rotating into energy infrastructure, biotech, batteries, K-beauty, shipbuilding/defense and robotics.

That does not kill the Korea thesis. But it changes the trading discipline. Samsung Electronics and SK hynix are still core holdings, not fresh-cash priority buys until foreign buying returns or the stocks digest the distribution.

Data source: Research OS local DB, prices_daily, investor_flow_daily, foreign_ownership_daily, through 2026-05-22. Flow values are estimated from local daily investor-flow data and may differ from final exchange totals.

Disclaimer: For research and information purposes only. Not investment advice.

Why Korea Part 5: Korea’s Stock Market Is Now Near Global No. 6 — Buy Signal or Warning Light?

Sun, 24 May 2026 21:20:00 +0900

Why Korea Part 5. Part 1 covered semiconductor substrates, Part 2 covered K-beauty, Part 3 covered Samsung Electronics and SK hynix earnings leverage, and Part 4 covered the $6.7B Korea ETF inflow and value-trap debate. This article connects that thesis to the new headline: Korea’s equity market is now appearing near global No. 6 by market cap. The dedicated flow follow-up is Korea Foreign Investor Flow Analysis.

TL;DR

The direction is real. WFE data show KRX domestic market capitalization at $2.67T in December 2025, up 71.4% in USD terms versus the prior year. CEIC shows Korea at $4.03T in February 2026.
The Wikipedia-style $4.89T / global No. 6 / 190% of GDP snapshot is plausible as a recent market snapshot, but the GDP ratio needs methodology checking before it should be treated as a hard input.
Our Research OS local DB proxy puts KRX-listed market capitalization at roughly KRW 6,851T on May 22, 2026. Samsung Electronics and SK hynix together account for roughly KRW 3,093T, or 45.1% of that proxy.
The investment conclusion is not “buy Korea indiscriminately.” It is: verify AI-memory earnings durability, then look for the next bottleneck where the re-rating can broaden.

1. Start With the Data

Different sources use different scopes: WFE and CEIC focus on domestic listed companies; local KRX reports often include KOSPI, KOSDAQ and KONEX in won terms; public ranking tables may combine multiple sources.

Item	Value	Interpretation
WFE KRX domestic market cap, Dec. 2025	$2.67T	+71.4% YoY in USD terms
CEIC Korea market cap, Feb. 2026	$4.03T	Monthly all-time high in CEIC’s series
KRX-reported Korean market cap, Apr. 27, 2026	KRW 6,031.97T	First break above KRW 6,000T
Public ranking table	$4.89T	Global No. 6, behind India and ahead of Taiwan
KOSPI close, May 22, 2026	7,847.71	Historic high zone
Research OS local DB proxy, May 22, 2026	KRW 6,851T	Proxy derived from foreign-held shares and ownership ratio

The $4.89T snapshot is not absurd. It is about 83.2% above the WFE December 2025 number and about 21.4% above the CEIC February 2026 number. Given the strength of Korean equities in March-May, the direction is credible.

2. Be Careful With the GDP Ratio

The public table shows Korea at 190% of GDP. That number implies a GDP denominator of roughly $2.57T.

CEIC, however, puts Korea’s market-cap-to-GDP ratio at 149.6% for year-end 2025. It can certainly move higher in 2026 as prices rise, but the 190% figure should not be used without checking the market-cap scope, GDP denominator, FX rate and date.

Use the hierarchy this way:

Fact: Korea market cap rose sharply from WFE’s $2.67T in Dec. 2025 to CEIC’s $4.03T in Feb. 2026.
Likely: A late-May snapshot around the high-$4T range is directionally consistent.
Blocked: 190% of GDP requires source-methodology validation.

3. The Real Driver Is AI Memory

Korea’s market-cap expansion is mostly an AI-memory story. Samsung Electronics and SK hynix are the core.

Variable	Change	Meaning
Price	HBM and DRAM pricing power	Higher revenue per bit
Quantity	AI server, GPU and custom ASIC capex	More HBM, DDR5, LPDDR and SOCAMM demand
Cost	Fixed-cost operating leverage	Price gains flow hard into operating profit

Our local proxy shows Samsung Electronics plus SK hynix at KRW 3,093T, or 45.1% of the covered market-cap proxy. That is the heart of the story.

So yes, Korea has re-rated. But no, this is not yet a balanced full-market re-rating. It is still a concentrated large-cap semiconductor rally.

4. Global No. 6 Is Not a Buy Button

Market-cap rankings are late indicators. By the time a country enters the global top-six table, the first easy money has usually been made.

The headline hides three issues:

It is backward-looking. Rankings change after prices move.
It hides concentration. Korea looks broad, but Samsung and SK hynix dominate the incremental value.
It creates overvaluation debates. A high market-cap-to-GDP ratio can mean structural re-rating, but it can also mean a crowded position near peak earnings.

The correct response to “Korea is global No. 6” is not automatic buying. It is opening the checklist.

5. The Checklist

The next phase depends on five questions:

Are Samsung Electronics and SK hynix 2026-2027 EPS estimates still being revised up?
Are HBM and DRAM contract prices still firm?
Are foreign flows spreading beyond mega-cap semiconductors?
Is KOSDAQ relative strength improving for multiple sessions?
Is Value-Up turning into real dividends, buyback cancellation and ROE improvement?

If these hold, global No. 6 can become the middle stage of a structural Korea re-rating.

If they break, the headline becomes a late-cycle warning light.

Investment View

Area	View	Reason
Broad KOSPI chase	Wait	The global-rank headline is late-cycle evidence, not fresh alpha
Samsung Electronics	Watchlist	HBM recovery, DRAM and foundry optionality need execution proof
SK hynix	Watchlist / buy-on-pullback candidate	Strongest pure HBM exposure, but much is already priced
Samsung Electro-Mechanics	Second-order bottleneck candidate	FC-BGA, MLCC and silicon capacitor exposure to AI package power integrity
Power infrastructure	Selective watchlist	AI data centers and semiconductor clusters need grid capacity
KOSDAQ quality names	Conditional buy candidates	Only after breadth and smart-money flow confirm broadening

Conclusion

Korea’s market-cap surge is real. WFE, CEIC, KRX reports and our local Research OS proxy all point in the same direction. Korea is no longer a quiet cheap market. AI memory, Value-Up reform and global capital rotation have moved it up the global ranking table.

But the investment conclusion is colder:

Global No. 6 is not the reason to buy. The reason to buy is whether the forces that got Korea there can persist and broaden.

The next alpha is not broad Korea beta. It is the next bottleneck: AI memory, package substrates, testing, power infrastructure, silicon capacitors, and select KOSDAQ names that can turn the Korea re-rating into earnings.

Evidence

WFE: KRX domestic market capitalization was $2.67T in Dec. 2025, up 71.4% YoY in USD terms. (WFE)
CEIC: Korea market capitalization was $4.03T in Feb. 2026. (CEIC)
KRX/Asia Business Daily: Korea market cap broke KRW 6,031.97T on Apr. 27, 2026. (Asia Business Daily)
Seoul Economic Daily: KOSPI closed at 7,847.71 on May 22, 2026. (Seoul Economic Daily)
CEIC: Korea market cap to GDP was 149.6% in 2025. (CEIC)

After NVIDIA, the AI Semiconductor Bottleneck Is Data Movement, HBM, FC-BGA and Power Integrity

Sun, 24 May 2026 19:20:00 +0900

Related series NVIDIA Q1 FY27 and Korean AI infrastructure / Marvell and Broadcom earnings preview / Samsung Electro-Mechanics KRW 1.5T silicon-capacitor contract / MLCC and silicon capacitors explained / Samsung Electronics re-rating thesis / AI PCB and substrate hub

TL;DR

The most important of the three videos is Dwarkesh Patel’s interview with Reiner Pope. It does not start with market headlines. It starts with what actually happens inside a chip: where the electricity flows, where the data sits, and how often the chip has to move that data.¹

The key point is simple. AI performance is not only about FLOPS. The real bottleneck is where the data comes from, where it is stored, and how short the path is between memory and compute. In that frame, NVIDIA remains central, but the investment read-through spreads into HBM, package substrates, FC-BGA, high-layer PCBs, Ethernet, optical links, power-stability components and testing.

For Korean investors, the translation is clear. Samsung Electronics and SK Hynix are the memory core. Samsung Electro-Mechanics is the FC-BGA and silicon-capacitor power-integrity node. Daeduck, Isu Petasys, Simmtech, Korea Circuit and TLB are substrate and PCB spread candidates. This is not a market to chase a Monday gap. The key is whether turnover and foreign / institutional flow hold into the afternoon.

1. Why the Reiner Pope Interview Matters

A common mistake in AI semiconductor investing is to read chip performance as “more FLOPS equals a better chip.” Reiner Pope’s explanation breaks that frame from the bottom up.

Much of AI computation is repeated matrix multiplication: multiply many numbers, add them, and do it again. Making the arithmetic unit faster matters, but at chip level the bigger question is often where the numbers come from.

An AI chip has several layers of storage and movement.

Location	Plain-English description	Investment translation
Registers and SRAM	The workbench inside the chip	Very fast, but area is expensive
HBM	The high-speed warehouse next to the GPU	Bandwidth bottleneck; SK Hynix and Samsung Electronics
Package substrate / interposer	The floor connecting chips and memory	FC-BGA, ABF, advanced substrates
Server board and network	Roads inside and outside the rack	High-layer PCB, Ethernet, optical links
Data-center power	Electricity for the whole system	Transformers, distribution, cooling, total operating cost

If the arithmetic unit waits for data, the chip is underused. So the real AI-chip question is not just “can we add more compute?” It is “can we move less data, keep it closer, and feed the chip without wasting power?"

That is why HBM, FC-BGA, silicon capacitors and high-speed PCBs belong in the same discussion. They all solve the same physical problem: keep AI chips fed with data and stable power.

2. Why Low-Precision Compute Leads to Substrates and Power

Reiner Pope’s point on low precision is not simply that FP8 or FP4 doubles speed. Lower precision also changes area, wire count and power. Fewer bits mean smaller circuits, less switching and lower energy per operation.

That matters for investors because lower precision is not only an NVIDIA GPU feature. If more compute gets packed into the same power envelope, the whole system must evolve.

Technology shift	System requirement	Korean supply-chain link
FP8 and FP4 adoption	More compute inside the same power budget	HBM4, server DRAM, SOCAMM
Tensor Core / systolic-array style designs	Less data movement inside the chip	HBM and package interconnect
Larger GPUs and ASICs	Larger die and package formats	FC-BGA, ABF, advanced substrates
Rack-scale expansion	More chip-to-chip and rack-to-rack bandwidth	High-layer PCB, Ethernet, optical links
Higher power density	Faster response to current swings and voltage noise	MLCC, silicon capacitors

So AI semiconductor value does not end at the GPU vendor. NVIDIA anchors the system. Marvell and Broadcom sit in custom AI chips, connectivity, Ethernet and optical links. Korean memory, substrate and power-stability components attach underneath.

3. NVIDIA’s Numbers Are Strong. The Market Question Has Changed

NVIDIA reported Q1 FY27 revenue of $81.6 billion, Data Center revenue of $75.2 billion, and Q2 revenue guidance of $91.0 billion ±2%. On official numbers, AI infrastructure demand has not rolled over.²

But the market is no longer asking only “is NVIDIA good?” That is already known. The new questions are:

Does this capex return as revenue and free cash flow for cloud customers?
Can data-center power and cooling bottlenecks be solved?
Can model companies keep paying high token bills?
Will AI backlash and regulation slow deployment?

The All-In episode belongs in this bucket. Anthropic, Karpathy, SpaceX, NVIDIA earnings and AI regulation are all part of one larger issue: AI is not over, but it now has to prove capital efficiency.³

In market language:

2023-2025 frame	2026 and beyond frame
GPUs are scarce	Rack-level data movement and power are scarce
HBM is scarce	HBM + substrates + power integrity + networking are scarce together
AI models improve	AI usage must convert into revenue and cash flow
Buy NVIDIA	Track the bottlenecks underneath NVIDIA

4. Marvell and Broadcom: Connectivity Becomes the Next Test

Marvell and Broadcom are the next earnings events to watch through this frame. Neither is simply an “NVIDIA competitor.” As AI data centers scale, both attach to connectivity, switching, optical signals and custom AI silicon.

For Marvell, the core question is whether custom AI silicon and optical connectivity are accelerating into actual revenue. For Broadcom, the core question is whether AI ASICs and AI Ethernet networking are scaling together. If both companies sound strong, the Korean read-through should broaden beyond pure HBM.

U.S. signal	Korean read-through
Custom AI chip demand	HBM, package substrates, testing
AI Ethernet and switch growth	Isu Petasys, high-speed PCB, low-loss materials
Optical links and silicon photonics	Selective optical exposure; indirect substrate and power benefits
Rack-scale expansion	Power equipment, cooling, data-center operating cost
Larger packages	Samsung Electro-Mechanics FC-BGA, Daeduck, Korea Circuit

The right translation is not “Broadcom / Marvell good means all Korean semis are good.” It is: AI-chip growth benefits the bottleneck suppliers that feed, connect, stabilize and test those chips.

5. What the AT&S News Confirms

On May 21, 2026, AT&S announced a capacity expansion for high-end IC substrates used in AI, at its Chongqing site in China. The investment is in the high-double-digit million-euro range and is backed by long-term customer agreements. AT&S expects a high-double-digit million-euro positive EBIT effect in fiscal 2026/27.⁴

The key point is not the customer name. The key point is that AI substrate capacity is now being expanded under long-term agreements.

AT&S also framed glass-core substrates in April as a next-generation foundation for AI, high-performance computing, high-speed communications and photonics. As packages become larger and more complex, dimensional stability, signal quality, power efficiency and data movement become limiting factors.⁵

This lines up directly with the Reiner Pope interview. If the AI bottleneck is data movement, advanced substrates and packages are no longer passive parts. They become performance infrastructure.

6. Korean Stock-Group Translation

6.1 Samsung Electronics and SK Hynix: The Memory Core

Reducing data-movement cost requires HBM and server memory. Whether the accelerator is a GPU or a custom ASIC, high-performance AI silicon consumes memory bandwidth.

SK Hynix is the cleanest HBM exposure. Samsung Electronics is a broader option: HBM recovery, HBM4, server DRAM, SOCAMM, eSSD and foundry optionality. But Samsung still needs execution proof: customer qualification, yield and visible AI-silicon wins.

6.2 Samsung Electro-Mechanics: FC-BGA plus Silicon Capacitors

Samsung Electro-Mechanics is one of the clearest Korean second-order bottlenecks in this frame.

First, FC-BGA is the package substrate that connects high-performance chips to the board. Larger GPUs, CPUs, custom AI chips and switch ASICs all need advanced substrates.

Second, silicon capacitors stabilize power inside the AI GPU / HBM package. Samsung Electro-Mechanics announced a roughly KRW 1.5 trillion silicon-capacitor supply contract with a large global customer in May 2026, and described the product as a component that improves power stability inside high-performance semiconductor packages such as AI-server GPUs and HBM.⁶

The point is not “more MLCC content.” The point is that Samsung Electro-Mechanics could be reclassified as a substrate company with in-package power-integrity components.

6.3 Daeduck, Isu Petasys, Simmtech, Korea Circuit and TLB

These names should not be lumped together too loosely.

Group	What to watch	Risk
Daeduck Electronics	FC-BGA, package substrates, AI chip substrates	Customer proof, yield, utilization
Isu Petasys	High-layer networking PCB	Flow confirmation after a strong rally
Simmtech / TLB	Memory modules, SoCAMM, server PCB	AI revenue mix and margin proof
Korea Circuit	SoCAMM and FC-BGA optionality	Qualification and actual revenue timing

The phrase “AI server exposure” is not enough. The evidence investors need is direct supply, ASP uplift, long-term agreements and margin resilience.

7. What 20VC Shows About Capital-Market Temperature

The 20VC episode discussed Anthropic, Karpathy joining Anthropic, Cerebras, SpaceX and AI token costs.⁷ This is less about semiconductor physics and more about capital-market temperature.

The positive signal is that investors are still willing to fund AI infrastructure and AI hardware stories. Model-lab financing, AI hardware IPO appetite and mega-tech IPO expectations remain alive.

The negative signal is concentration risk. Too much capital is gathering around a small number of mega AI and infrastructure companies. Private AI model spending eventually has to pass through public-market scrutiny, big-tech capex budgets or real revenue.

So the 20VC takeaway is not “buy private AI exposure.” It is: AI risk appetite is alive, but the market will increasingly demand monetization and cash-flow proof.

8. Practical Checklist

For the Korean market next week, the question is not just whether Samsung Electronics and SK Hynix rise. The better signal is whether money moves down the stack.

Order	Checkpoint	Meaning
1	Samsung Electronics and SK Hynix hold on volume	Memory core confirmed
2	Samsung Electro-Mechanics and Daeduck hold after the open	FC-BGA and power-integrity re-rating
3	Isu Petasys, Simmtech, TLB and Korea Circuit participate	PCB and SoCAMM spread
4	Foreign and institutional flow holds into the afternoon	Real money, not just a theme trade
5	Power equipment and data-center infrastructure join	Full AI infrastructure bottleneck spread

Entry discipline matters.

Condition	Read
Gap-up at the open with weak turnover	Do not chase
Afternoon hold with foreign / institutional buying	Broaden watchlist
PCB names rally while memory megas weaken	More likely thematic
Memory → substrates → power-equipment sequence	Bottleneck spread signal
“AI” label without revenue or orders	Avoid

9. Conclusion

The three videos combine into one sentence:

The AI trade is not over. It is moving down the stack.

2023-2025 was the GPU and HBM phase. From 2026 onward, the next layers are data movement, packaging, FC-BGA, Ethernet and optical links, power integrity, testing and data-center operating cost. NVIDIA remains the center. But the investor question is now: “which Korean bottleneck turns NVIDIA’s revenue growth into its own earnings?”

The current order is:

Memory core: Samsung Electronics, SK Hynix
Package and power integrity: Samsung Electro-Mechanics, Daeduck Electronics
High-speed PCB and modules: Isu Petasys, Simmtech, Korea Circuit, TLB
Testing and sockets: ISC, LEENO Industrial, TSE
Power and data-center operations: power equipment and cooling infrastructure

The most important investment sentence is this: AI performance competition is a data-movement-cost competition, and that cost translates into HBM, packaging, substrates, networking and power.

If that spread is confirmed by turnover and foreign / institutional flow, FC-BGA and high-speed PCB should be treated as AI infrastructure bottlenecks, not just another theme.

Evidence Classification

[Fact]

Dwarkesh Patel’s Reiner Pope interview explains AI chip design through MAC operations, data movement, low-precision arithmetic, Tensor Core / systolic-array structures and total operating cost.¹
NVIDIA reported Q1 FY27 revenue of $81.6 billion, Data Center revenue of $75.2 billion, and Q2 revenue guidance of $91.0 billion ±2%.²
AT&S announced on May 21, 2026 that it would expand AI high-end IC substrate capacity based on long-term customer agreements.⁴
Samsung Electro-Mechanics announced a roughly KRW 1.5 trillion silicon-capacitor supply contract and described silicon capacitors as power-stability components for AI-server GPU and HBM packages.⁶

[Inference]

The AI chip bottleneck is moving from arithmetic units into data movement and power integrity.
NVIDIA’s strong results may flow through to Korea not only through HBM, but also through FC-BGA, high-speed PCB, silicon capacitors and test sockets.
Marvell and Broadcom earnings are key checks for whether “custom AI chips + connectivity bottlenecks” are turning into real numbers.

[Blocked]

AT&S’s key customer name and exact product mix.
Whether specific Korean substrate and PCB companies directly supply NVIDIA, Marvell or Broadcom programs.
Samsung Electro-Mechanics’ silicon-capacitor customer, product margin and exact package location.
Official confirmation of several private-company valuation and funding numbers discussed in All-In and 20VC.

Dwarkesh Patel, “Chip design from the bottom up / Reiner Pope,” YouTube, 2026-05-22. https://www.youtube.com/watch?v=oIk3R-sMX5o ↩︎ ↩︎
NVIDIA Investor Relations, “NVIDIA Announces Financial Results for First Quarter Fiscal 2027,” 2026-05-20. https://nvidianews.nvidia.com/news/nvidia-announces-financial-results-for-first-quarter-fiscal-2027 ↩︎ ↩︎
All-In Podcast, “SpaceX’s $2T case, Nvidia’s shock selloff, America turns on AI,” YouTube, 2026-05-22. https://www.youtube.com/watch?v=HGbA6ze0_3M ↩︎
I-Connect007, “AT&S Expands Capacity for AI Substrates,” 2026-05-21. https://iconnect007.com/article/150109/ats-expands-capacity-for-ai-substrates/150106/pcb ↩︎ ↩︎
AT&S, “AT&S advances glass core substrates for AI, high-performance computing and photonics,” 2026-04-22. https://ats.net/en/press/ats-advances-glass-core-substrates-for-ai-high-performance-computing-and-photonics/ ↩︎
Samsung Electro-Mechanics, “Samsung Electro-Mechanics Signs 1.5 Trillion KRW Silicon Capacitor Supply Contract with Global Large-Scale Company,” 2026-05-20. https://samsungsem.com/global/newsroom/news/view.do?id=10310 ↩︎ ↩︎
20VC, “Andrej Karpathy joins Anthropic / Cerebras / SpaceX,” YouTube, 2026-05-21. https://www.youtube.com/watch?v=z94zlbVn048 ↩︎

National Growth Fund and KOSDAQ Smart Money: Where the ₩150T Policy Capital Actually Flows

Sun, 24 May 2026 16:20:00 +0900

📚 KOSDAQ follow-up series
KOSDAQ Smart Money Returns / KOSDAQ 2026 Policy Triggers / Korea Daily Market Hub

The previous KOSDAQ flow note argued that price and flow were diverging. KOSPI large caps had dominated the tape, but recent foreign, program and trust-account flow into KOSDAQ was quietly improving. This post is the policy-capital follow-up: if money is starting to look back at KOSDAQ, which policy channels and industrial bottlenecks can actually absorb it?

The short answer: Korea’s National Growth Fund is not a promise that the government will buy the KOSDAQ index. It is better understood as a tool for lowering the cost of capital for strategic industries and unlocking delayed scale-up or capex decisions. The investable question is therefore not “which stocks have the policy label?” but which bottlenecks can turn policy capital into orders, capacity, revenue and earnings?

Executive Summary

Separate the numbers by layer. The full National Growth Fund is a five-year program of more than ₩150T, combining ₩75T from the Advanced Strategic Industry Fund and at least ₩75T from private, pension, financial-institution and public money. Financial Services Commission
The 2026 deployment plan is ₩30T+. It consists of ₩3T direct investment, ₩7T indirect investment, ₩10T infrastructure financing and ₩10T ultra-low-interest loans. The simple sum is 3 + 7 + 10 + 10 = ₩30T. Financial Services Commission
The ₩50T+ package announced in April 2026 is a separate advanced-industry ecosystem layer. It is centered on ₩35T of indirect investment and ₩15T+ of direct investment. Financial Services Commission
The first-round ₩6.6T figure is not the total size of the first seven mega-projects. It is the amount of support approved in January-March 2026: Shinan Ui offshore wind ₩3.4T, Ulsan next-generation battery materials ₩0.1T, Pyeongtaek AI semiconductor production base ₩2.5T, and Rebellions capital participation ₩0.6T. Financial Services Commission
For KOSDAQ, the indirect-investment structure matters most. The FSC explicitly discusses support for Pre-IPO companies, early KOSDAQ-listed companies, M&A, restructuring, KOSDAQ funds and regional funds. Financial Services Commission
The best public-market ideas are bottleneck suppliers, not generic policy-theme stocks. Power infrastructure, AI semiconductor design services, OLED equipment, packaging, substrates, testing and late-stage biotech are more important than the headline names.

In one sentence:

The National Growth Fund is not an index buyer for KOSDAQ; it is a capital channel that narrows the search for the next KOSDAQ alpha candidates by funding selected growth companies and industrial bottlenecks.

1. Fact Check: Do Not Mix ₩150T, ₩50T and ₩6.6T

The broad direction is correct. Korea is using policy capital to support AI, semiconductors, biotech, defense, robotics, hydrogen, batteries, displays, future mobility and related supply chains. But the numbers need to be separated.

Item	Correct Reading	Investor Caveat
Full National Growth Fund	More than ₩150T over five years	This is the full policy frame, not the amount buying listed stocks in one year.
2026 support plan	₩30T+	Mix of direct investment, indirect investment, infrastructure financing and ultra-low-interest loans.
Advanced-industry ecosystem package	₩50T+	A five-year investment-capital layer, not the entire fund.
First-round ₩6.6T	January-March 2026 approvals	Not the total value of the first seven mega-projects.
Second-round roughly ₩10T	Press-reported estimate	The FSC release does not break out every second-round project allocation in detail.

According to the FSC, the full National Growth Fund is more than ₩150T over five years. The government expanded the original ₩100T plan to at least ₩150T, combining ₩75T from the Advanced Strategic Industry Fund and at least ₩75T from private, pension, financial-institution and public money. Financial Services Commission

The 2026 plan is separate. The FSC states that the 2026 support plan is ₩30T+: ₩3T in direct investment, ₩7T in indirect investment, ₩10T in infrastructure financing and ₩10T in ultra-low-interest loans. Financial Services Commission

The advanced-industry ecosystem package announced on April 14, 2026 is another layer. Yonhap described it as more than ₩50T over five years, split between ₩35T of indirect investment and ₩15T of direct investment. Yonhap News Agency The FSC release also says the ₩15T direct-investment channel will be used for large, long-term strategic support, while indirect investment includes Pre-IPO, early KOSDAQ-listed companies, M&A, restructuring, KOSDAQ funds and regional funds. Financial Services Commission

The most common mistake is the ₩6.6T first-round figure. That is not the total size of the first seven mega-projects. It is the amount of support approved in January-March 2026:

Shinan Ui ₩3.4T + Ulsan battery materials ₩0.1T + Pyeongtaek AI semiconductors ₩2.5T + Rebellions ₩0.6T = ₩6.6T

So the right starting question is not “where does the government immediately inject ₩50T?” The right question is which funding channel unlocks which bottleneck, and how quickly?

2. Why This Is a Follow-Up to KOSDAQ Smart Money

The prior KOSDAQ smart-money note focused on a divergence. Over the past two months, KOSPI large caps dominated price performance. But over the latest 5-20 trading days, KOSDAQ foreign, program and trust-account flows improved more than the headline price action suggested.

The National Growth Fund adds the policy-capital layer. The key is not index-level buying. The fund’s indirect-investment design points to a more specific KOSDAQ transmission channel.

Policy Structure	KOSDAQ Transmission
Pre-IPO support	Expands the pipeline of later-stage companies that can list or raise follow-on capital.
Early KOSDAQ-listed company support	Reduces the funding gap after IPO.
KOSDAQ fund creation	Creates follow-on demand for selected growth stocks.
M&A and restructuring funds	Supports consolidation and business-model repair.
AI and semiconductor ecosystem funds	Flows into design, IP, testing, equipment and materials.
Regional funds	Supports local deep-tech and manufacturing-linked KOSDAQ candidates.

This structure does not lift the entire KOSDAQ market mechanically. It repairs the capital chain for selected companies. For VC investors, that means a stronger bridge from Pre-IPO financing to post-listing liquidity. For public-equity investors, it means the focus should be on companies where policy capital becomes actual orders, expansion, loans, guarantees or investment disclosures.

The core question is:

Where in KOSDAQ can policy capital translate into real operating numbers?

3. Translating the 13 Mega-Projects Into Public-Market Language

The first seven projects and the second six projects add up to 13 mega-projects. The list is broad, so investors need to translate it into value-chain bottlenecks.

Project Cluster	Investment Translation	Public-Market Watch Area
K-Nvidia, Sovereign AI	Domestic AI semiconductors, models and data centers	AI chip design, DSP, servers, data-center power
National AI Computing Center, Saemangeum advanced belt	Computing infrastructure and regional data centers	Power, cooling, networking, security
Pyeongtaek AI semiconductor production base	HBM, foundry and advanced process capex	Samsung Electronics, equipment, materials, gases, power
Semiconductor cluster energy infrastructure	Grid, transformers and distribution bottlenecks	Electrical equipment, cables, ESS, switchboards
OLED super-gap	Panel capex and equipment orders	OLED deposition, laser, inspection and logistics equipment
Biotech and vaccines	Late-stage clinical and production financing	Phase 3 biotech, CDMO, vaccine facilities
Future mobility and defense	Unmanned systems and defense supply chains	Avionics, sensors, power systems, materials

The investability differs by layer. Many core AI semiconductor startups are private. Samsung Electronics is too large for policy capital alone to determine the stock. Biotech has clinical risk. Defense and power equipment have already been strongly re-rated.

The public-market alpha is most likely where policy capital either unlocks a previously delayed investment decision or reinforces an already visible demand bottleneck.

4. Idea 1: AI Data-Center Power Infrastructure

The highest-quality public-market idea is power infrastructure. Sovereign AI and the National AI Computing Center need electricity before they need more acronyms. Data centers consume power, and semiconductor fabs cannot run without stable energy supply.

Policy materials include energy infrastructure as a core axis. The second-round projects include renewable-energy infrastructure, and Yonhap described that project as building stable power supply for data centers through solar and wind generation. Yonhap News Agency

LS ELECTRIC is the easiest listed proxy to understand. The company reported first-quarter 2026 consolidated revenue of ₩1.3766T and operating profit of ₩126.6B, while order backlog at the end of the quarter was reported at ₩5.6T. Media coverage attributed the strength to AI data-center investment, semiconductor capex and renewable-energy-driven power-infrastructure demand. Chosun Biz

But investors should be careful. Many power-equipment stocks have already been re-rated. Even if the National Growth Fund strengthens the story, the market may already price in a large part of the global grid and AI data-center cycle. The real indicators are order backlog, data-center revenue mix and margin durability, not the policy label.

Item	View
Investment type	Structural growth, but valuation discipline required
Key indicators	Backlog, transformers, switchboards, data-center power projects
Entry rule	Prefer pullbacks after earnings confirmation over headline chasing
Invalidation	Backlog slowdown, margin decline, delayed data-center power projects

5. Idea 2: AI Semiconductor Design and Implementation Bottlenecks

K-Nvidia and Sovereign AI are powerful headlines, but listed access is limited. Rebellions, FuriosaAI, DEEPX and other domestic AI chip names are largely private. So where can public investors look?

The answer is the implementation bottleneck. Even when AI chip startups receive capital, they still need design services, verification, foundry interface, packaging and mass-production transition. Design houses and ASIC service firms sit in this workflow.

ADTechnology is the clearest listed watchlist name. On its IR page, as of May 22, 2026, the company showed a share price of ₩42,600, PER of 153.24x and 13,462,007 shares outstanding. The simple market-cap calculation is ₩42,600 × 13,462,007 = roughly ₩573.5B. ADTechnology

Arm describes ADTechnology as South Korea’s largest ASIC design house and Arm’s first Total Design Partner in Korea. Arm also says ADTechnology provides ASIC design services across mobile, automotive and HPC markets. Arm

The logic is attractive, but the price is not easy. A 153x PER means the market already expects substantial growth. Policy capital flowing into AI semiconductors is not enough by itself. Investors need evidence:

New AI/HPC design wins
More tape-outs
Transition from design work to mass-production revenue
Improved credibility of Samsung Foundry leading-edge processes
Repeatable revenue rather than one-off project fees

The conclusion is Watchlist. ADTechnology is one of the more interesting listed proxies for domestic AI semiconductor ecosystem expansion, but actual order and production evidence should come before chasing.

6. Idea 3: OLED Super-Gap and the Equipment Cycle

For KOSDAQ, OLED equipment is one of the cleanest policy-linked areas. OLED super-gap is part of the second-round mega-projects, with policy support aimed at large-scale facility investment in high-value OLED. Financial Services Commission

Even if policy capital flows to large panel makers, the public-market alpha can be clearer in equipment suppliers. Panel companies bear demand, depreciation, utilization and Chinese competition risk. Equipment makers see the investment decision translate into orders and revenue recognition.

Watchlist areas include Sunic System, AP System, inspection equipment, laser equipment and logistics equipment. But the key is not the OLED label. It is actual equipment orders.

Evidence to Watch	Why It Matters
Panel-company investment approval	First sign that policy finance becomes capex
Equipment order disclosures	Converts theme into revenue evidence
Down payments and backlog	Shows that orders are real
Delivery and revenue timing	Equipment earnings can be lumpy
Customer diversification	Reduces dependence on a single capex cycle

This is where the policy story fits KOSDAQ especially well. Many of the relevant names are mid-cap equipment companies, not mega-cap index stocks. If policy capital accelerates capex, earnings sensitivity can be meaningful. But investors should avoid chasing names that have already run on the policy label. OLED equipment should be bought only where order evidence and valuation still leave room.

7. Idea 4: Late-Stage Biotech Extends Runway, But Does Not Change Drug Efficacy

The second-round mega-projects include next-generation biotech and vaccines. The policy direction supports companies near commercialization, including global Phase 3 clinical programs. This is closely tied to KOSDAQ.

Biotech policy support should be interpreted differently from equipment or power infrastructure. It can lower the cost of Phase 3 funding and reduce dilution. It does not increase drug efficacy, safety or approval probability.

So biotech policy support is best treated as an event-driven theme rather than a blanket sector buy.

Item	What to Check
Clinical stage	Phase 3 entry or progress
Use of funds	Clinical cost, production facility, approval preparation
Regulatory path	Prior consultation with FDA, EMA or MFDS
Partnering	Global co-development or licensing potential
Downside risk	Policy capital does not protect against clinical failure

8. Practical Checklist: Watch Disclosures, Orders and Backlog

To trade or invest around this policy, investors need a checklist. Following headlines alone tends to end in overheated policy-theme names.

Step	Evidence	Meaning
1	Manager selection and sub-fund formation	Policy capital becomes deployable
2	Direct investment, loan or guarantee approval	Capital cost declines for a specific project
3	Expansion, order or equipment-purchase disclosure	Operating translation begins
4	Backlog growth	The story becomes numeric
5	Revenue and operating profit recognition	The thesis is verified in financial statements

For the KOSDAQ follow-up framework, add flow confirmation:

KOSDAQ/KOSPI relative-ratio rebound
Sustained KOSDAQ foreign and program net buying
Retail selling absorbed by foreign and institutional buyers in growth names
Trading-value expansion in policy-linked candidates
Post-disclosure strength that holds into the afternoon session

Policy capital is a favorable backdrop. It becomes an investment opportunity only when price and flow confirm it.

9. Sector and Stock-Group View

Name / Group	Market	View	Core Logic	Caveat
LS ELECTRIC	KOSPI	Hold / watch pullbacks	Power bottleneck for AI data centers and semiconductor capex	Already re-rated; monitor backlog and margins
ADTechnology	KOSDAQ	Watchlist	Listed proxy for K-Nvidia / Sovereign AI implementation demand	High PER; needs AI/HPC order evidence
Sunic System / AP System	KOSDAQ	Watchlist	OLED super-gap capex beneficiaries	Theme until actual equipment orders arrive
Samsung Electronics	KOSPI	Watchlist	Pyeongtaek AI semiconductor base, HBM and foundry optionality	HBM qualification and foundry credibility matter more than policy
Late-stage biotech	KOSDAQ	Selective event-driven	Policy capital can reduce Phase 3 funding gaps	Clinical risk remains
Generic policy-theme names	KOSPI/KOSDAQ	Avoid	Weak evidence of earnings translation	News may mark a local top

Priority ranking:

Power infrastructure: Closest to real-world demand, but valuation matters.
AI semiconductor design services: Scarce listed proxy, but must grow into valuation.
OLED equipment: Strong earnings sensitivity if policy capital unlocks capex.
Late-stage biotech: Event-driven, not a blanket sector call.

10. Conclusion: KOSDAQ Needs Capital Channels, Not Just Policy Labels

The National Growth Fund is positive for KOSDAQ, but not because it buys the index. Its significance is structural.

It reduces funding gaps for Pre-IPO and early-listed companies. It supports M&A and restructuring. It provides long-duration capital to KOSDAQ-friendly sectors such as AI, semiconductors, OLED, biotech and regional deep tech. And if public money pulls in private money, it can improve follow-on financing and post-listing liquidity for selected growth companies.

The KOSDAQ smart-money follow-up is therefore:

Do not buy the whole policy basket. Find the bottleneck companies where policy capital turns into disclosed orders, expansion, investment or financing.

The next four things to wait for:

Fund-manager selection and sub-fund formation from the second quarter onward
Public confirmation of direct investments, loans and guarantees
OLED, AI semiconductor and power-infrastructure order disclosures
Simultaneous improvement in KOSDAQ relative strength and foreign/program flows

KOSDAQ price action is still weak. But flows have started to improve, and the National Growth Fund creates the policy-capital channel those flows can attach to. The next alpha is unlikely to come from the label “policy beneficiary.” It is more likely to come from the narrow places where money must actually pass.

Evidence Classification

[Fact]

The full National Growth Fund is a five-year ₩150T+ framework combining ₩75T from the Advanced Strategic Industry Fund and at least ₩75T from private, pension, financial-institution and public capital. Financial Services Commission
The 2026 support plan is ₩3T direct investment, ₩7T indirect investment, ₩10T infrastructure financing and ₩10T ultra-low-interest loans. Financial Services Commission
The April 2026 advanced-industry ecosystem package is a ₩50T+ layer centered on ₩35T indirect investment and ₩15T direct investment. Financial Services Commission, Yonhap News Agency
The first-round ₩6.6T figure refers to January-March 2026 support approvals. Financial Services Commission
The six second-round projects are biotech/vaccines, OLED, future mobility/defense, sovereign AI, renewable-energy infrastructure and the Saemangeum advanced belt. Yonhap News Agency
ADTechnology’s IR page showed, as of May 22, 2026, a share price of ₩42,600, PER of 153.24x and 13,462,007 shares outstanding. ADTechnology

[Inference]

The best public-market alpha from the National Growth Fund is more likely in bottleneck suppliers than in large direct policy beneficiaries.
KOSDAQ benefits more from Pre-IPO, early-listed company support, KOSDAQ funds, M&A and restructuring capital than from index-level buying.
ADTechnology is not confirmed as a direct policy beneficiary; it is a listed proxy for domestic AI semiconductor ecosystem expansion.

[Speculation]

The final company-level allocation of the second-round projects remains uncertain.
Which listed AI software or semiconductor design firms will convert sovereign-AI policy capital into revenue still requires disclosure evidence.
Whether OLED policy financing accelerates domestic panel capex and equipment orders is still an assumption.

[Blocked]

Exact official allocation by each of the six second-round projects.
Company-level direct investment, loan and guarantee terms.
Confirmed supplier lists for cooling, EPC, water treatment and data-center power management.
Latest full-consensus 12-month forward PER, EV/EBITDA and earnings estimates for Korean individual names.

Pearl Abyss Comps: Between Neowiz's Post-Lies-of-P Discount and CD Projekt's Re-Rating Path

Sat, 23 May 2026 10:40:00 +0900

Pearl Abyss × Crimson Desert Series KOSDAQ smart money and Pearl Abyss rebound setup / Flow-bottom test / 1Q recognized copies and the 2027 cliff / What the DLC comment really means / Pearl Abyss Hub

Pearl Abyss is not priced as if Crimson Desert failed. It is priced as if the market accepts Crimson Desert’s success but still discounts it as a 2026 one-off. The closest domestic comp is Neowiz after Lies of P. The more important structural comp is CD Projekt around the Cyberpunk 2077 recovery and Phantom Liberty. Capcom is not today’s comp; it is the model Pearl Abyss wants to graduate into.

Key Takeaways

The closest Korean comp is Neowiz. Lies of P sold over 1M copies within one month and turned Neowiz profitable in 3Q23, but the market did not immediately grant a global-IP multiple. A good game and a repeatable franchise were priced differently. (Neowiz 3Q23 IR)
The key overseas comp is CD Projekt. Cyberpunk 2077 lost trust at launch, then regained it through patches, Update 2.0 and Phantom Liberty. Phantom Liberty sold over 4.3M copies within two months. (CD Projekt)
Capcom is the target model, not the current comp. Capcom already has globally validated IP, repeat launches and long-tail sales. Monster Hunter: World stands at 22.1M units, or 29.6M including Iceborne Master Edition. (Capcom)
Pearl Abyss has larger scale than Neowiz. Crimson Desert has officially surpassed 5M copies and reached a Steam peak concurrent user count of 276,261. Pearl Abyss’s 1Q26 Crimson Desert revenue was KRW 266.5B. (Pearl Abyss) (1Q26 Earnings Letter)
But the market question is still similar. Can the company earn money in 2027? What is the DLC roadmap? When does DokeV become visible? Will cash flow return to shareholders?

The conclusion: Pearl Abyss sits between a Neowiz-style discount and a CD Projekt-style re-rating path. Crimson Desert’s success is already proven. The next proof points are 6M official sales, DLC/expansion visibility, DokeV visibility and capital allocation.

1. What the Current Price Implies

As of the May 22, 2026 close:

Item	Current State
Share price	KRW 46,050
Change vs May 14 close	-9.0%
Post-earnings position	Near post-1Q low zone
Estimated Crimson Desert unit sales	~5.88M
Gap to 6M	~116K copies
Recent foreign flow	Continued net buying
Short-sale ratio	2.69% on May 22, easing

The important point is this:

The market is not pricing Crimson Desert failure. It is pricing the possibility that Crimson Desert success is a one-time 2026 earnings event.

So the debate has moved from game success to company classification.

Stage 1: Did Crimson Desert sell?
→ Already answered. Official 5M+, estimated ~5.88M.

Stage 2: Does Crimson Desert generate 2026 profit?
→ Already answered. 1Q26 OP above KRW 200B and FY26 OP guidance of KRW 487.6B-572.6B.

Stage 3: Does Crimson Desert become a repeatable IP?
→ Not answered yet. This is what the market is waiting for.

That is why the Neowiz and CD Projekt comps matter.

2. Neowiz: The Korean Discount Case

The most direct domestic comp is Neowiz. Lies of P was a Korean-developed global package action RPG with meaningful commercial success. Neowiz’s 3Q23 presentation says Lies of P sold over 1M copies within one month of official launch, with 93% of sales from overseas markets and 73% from North America and Europe. The company also laid out expansion-pack and sequel plans. (Neowiz 3Q23 IR)

The earnings turn was real.

Item	Neowiz 3Q23
Revenue	KRW 117.48B
Operating profit	KRW 20.25B
2Q23 operating profit	-KRW 4.89B
PC/console revenue	KRW 54.78B
Lies of P initial sales	Over 1M within one month

4Q23 was also respectable. PC/console revenue rose to KRW 61.31B, up 11.9% QoQ, and FY2023 operating profit was KRW 31.73B, up 62.2% YoY. (Neowiz 4Q23 IR)

But the stock did not re-rate.

Point	Date	Close	vs Launch-Day Close
D0	2023-09-19	KRW 34,500	0.0%
D+7	2023-09-26	KRW 25,900	-24.9%
D+30	2023-10-19	KRW 26,050	-24.5%
D+60	2023-11-17	KRW 27,500	-20.3%
D+65	2023-11-23	KRW 27,400	-20.6%

The October 17, 2023 1M-copy announcement produced a short-term bounce, but not a structural re-rating.

The company turned. The stock did not.

That is the lesson. The Korean market often prices “one good game” and “repeatable global IP company” very differently.

3. Pearl Abyss Is Bigger, But the Discount Question Rhymes

Pearl Abyss is larger than Neowiz on the relevant metrics.

Axis	Neowiz / Lies of P	Pearl Abyss / Crimson Desert
Initial sales	Over 1M within one month	Official 5M+, estimated ~5.88M
Launch-reflecting quarter revenue	3Q23 KRW 117.48B	1Q26 KRW 328.1B
Launch-reflecting quarter OP	3Q23 KRW 20.25B	1Q26 KRW 211.7B
New-title revenue	PC/console KRW 54.78B	Crimson Desert KRW 266.5B
Technology argument	Global package-game capability	Proprietary BlackSpace Engine and AAA open-world capability
Next variable	DLC / sequel	DLC / expansion, DokeV, capital allocation

Pearl Abyss has larger unit sales, stronger earnings leverage and a stronger technology-platform argument. Crimson Desert is not just “a good Korean console game.” It is evidence that Pearl Abyss can build and operate a large global AAA open-world title.

But the market’s discount question is similar.

Market Question	Neowiz	Pearl Abyss
Is the hit recognized?	Yes	Yes
Is repeatability recognized?	Not yet	Not yet
Is next-year profit visible?	Weak	Still debated
Is DLC quantified?	Weak	Still weak
Is the next-title schedule visible?	Weak	DokeV proof needed

The risk is a Neowiz-style discount: good game, good reviews, strong launch sales and a profit turn are still not enough unless the market sees next-year earnings and a repeatable franchise path.

4. CD Projekt: The Trust-Recovery and Expansion Re-Rating Case

CD Projekt is the more important structural comp. The direction is different: Cyberpunk 2077 lost trust at launch, then recovered through major patches, Update 2.0 and Phantom Liberty.

CD Projekt’s 3Q23 release says Phantom Liberty launched on September 26, 2023 and sold over 4.3M copies within two months. The company also said Update 2.0 and Phantom Liberty lifted the quality of the Night City experience, while 3Q23 revenue reached PLN 443M and net profit reached PLN 203M. (CD Projekt)

This is the core path: patches → trust recovery → expansion sales → IP durability.

Axis	CD Projekt	Pearl Abyss
Market doubt	Can Cyberpunk recover trust?	Can Pearl Abyss repeat AAA execution after long delays?
Evidence	Patches, Update 2.0, Phantom Liberty	Patch 1.04-1.08 cadence, long-tail sales, China rank
Re-rating trigger	Expansion sales and IP durability	DLC/expansion, DokeV, capital return
What the market waited for	Is recovery real?	Is this a platform company?

The difference matters. CD Projekt was a recovery from a damaged launch. Pearl Abyss is a successful launch whose repeatability is still not trusted.

So Pearl Abyss is not discounted because it is worse. It is discounted because the market is waiting for the next CD Projekt-like proof point: DLC, expansion and next-title visibility.

5. Capcom Is the Target Model

Capcom is not the current comp. It is the target model.

Capcom’s Platinum Titles page shows Monster Hunter: World at 22.1M units, or 29.6M including Monster Hunter World: Iceborne Master Edition. Monster Hunter World: Iceborne itself stands at 16.0M units. (Capcom)

Capcom already has what Pearl Abyss still needs to prove:

Multiple global IPs.
Repeatable engine and production systems.
Base games, expansions, remasters and sequels.
Long-tail sales that recur through cycles.
A market multiple that already reflects that repeatability.

Pearl Abyss’s destination, if the proprietary-engine and AAA-platform thesis works, looks more like Capcom. But the current stock is not there. It is between Neowiz’s discount and CD Projekt’s re-rating bridge.

6. Comp Map

Company	Ticker	Similarity	Why
Neowiz	095660.KQ	High	Global package-game hit followed by repeatability discount
CD Projekt	CDR.WA	High	Patch and expansion path can restore IP trust
Shift Up	462870.KS	Medium	Stellar Blade success plus platform/next-IP expectations
Remedy	REMEDY.HE	Medium	Strong game quality and IP but monetization/repeatability questions
Capcom	9697.T	Low today; target model	Repeat global IP, engine and long-tail sales already proven

Shift Up is directionally relevant because Stellar Blade surpassed 1M units and then built expectations around updates and platform expansion. But Shift Up has Nikke as a cash generator, while Pearl Abyss needs Crimson Desert DLC, DokeV and capital allocation to carry the bridge. (Gematsu)

The flow comparison is important.

Period	Neowiz Foreign	Neowiz Institution	Pearl Abyss Foreign	Pearl Abyss Institution
D+7	+KRW 4.58B	-KRW 0.74B	+KRW 14.98B	-KRW 54.68B
D+14	+KRW 4.88B	-KRW 1.73B	+KRW 68.92B	-KRW 4.63B
D+30	+KRW 1.48B	-KRW 5.54B	+KRW 68.03B	-KRW 58.31B
D+45	+KRW 4.66B	-KRW 5.33B	+KRW 103.37B	-KRW 51.27B
D+60	+KRW 0.96B	-KRW 3.40B	+KRW 71.83B	-KRW 77.83B
D+64	+KRW 1.46B	-KRW 3.33B	+KRW 83.95B	-KRW 77.72B

Neowiz did not have strong foreign sponsorship after launch. Pearl Abyss does. By D+64, foreign investors had bought roughly KRW 83.95B net and foreign ownership had risen by about 3.28 percentage points.

But that flow has not been enough to move the stock because domestic institutional selling and high short-sale turnover absorbed it.

Period	Neowiz Short Sale / Trading Value	Pearl Abyss Short Sale / Trading Value
D+7	0.04%	4.49%
D+14	0.04%	3.58%
D+30	0.06%	5.48%
D+45	0.07%	6.07%
D+60	0.16%	6.92%
D+64	0.16%	6.91%

In short:

Neowiz had a hit but weak foreign sponsorship.
Pearl Abyss has foreign buying, but domestic institutions and short-sale turnover are blocking re-rating.

This means Pearl Abyss’s weakness is not simply about weak game data. It is more a tape where success is visible, but domestic event money, institutions and short sellers are slowing the re-rating.

8. What Investors Should Watch

Whether Pearl Abyss stays in a Neowiz-style discount or moves toward a CD Projekt/Capcom re-rating depends on three proof points.

1. Official 6M Copies

The current local estimate is near 5.88M copies, while the official number is 5M+. An official 6M milestone would help the near-term tape. But it is not enough by itself. It confirms base-game durability, not franchise durability.

2. DLC / Expansion Confirmation

This is the most important variable. DLC or expansion confirmation weakens the “2027 cliff” discount. To matter, the market needs more than direction. It needs timing, content scope, pricing and revenue contribution.

3. DokeV or Next-Pipeline Visibility

DokeV visibility would expand the thesis from “Crimson Desert worked” to “Pearl Abyss has a repeatable AAA production pipeline.” That is the first step toward the Capcom target model.

Capital return from CCP proceeds or treasury-share cancellation would accelerate the re-rating.

Final Read

Pearl Abyss is not priced like a failed game company. It is priced like a company with one successful game that has not yet been accepted as a platform company.

The cleanest analogy is:

Pearl Abyss is a blend of Neowiz’s post-Lies-of-P discount and CD Projekt’s pre-Phantom-Liberty trust-recovery waiting period.

Neowiz is the warning: a good game, profit turn and DLC/sequel references were not enough. CD Projekt is the opportunity: patches and expansions can restore trust and extend the IP multiple.

The next re-rating triggers are clear:

Official 6M sales.
DLC / expansion roadmap.
DokeV development visibility.
CCP proceeds use or shareholder return.
Slower domestic institutional selling and short-sale turnover.

One of these can support a rebound. Two or more together can move Pearl Abyss from a Neowiz-style discount toward a CD Projekt-style re-rating path.

Fact / Inference / Blocked

[Fact]

Neowiz said Lies of P sold over 1M copies within one month, with 93% of sales from overseas markets and 73% from North America and Europe. (Neowiz 3Q23 IR)
Neowiz 3Q23 revenue was KRW 117.48B and operating profit was KRW 20.25B. (Neowiz 3Q23 IR)
Neowiz 4Q23 PC/console revenue was KRW 61.31B and FY2023 operating profit was KRW 31.73B. (Neowiz 4Q23 IR)
CD Projekt said Phantom Liberty sold over 4.3M copies within two months. (CD Projekt)
Capcom lists Monster Hunter: World at 22.1M units and 29.6M including Iceborne Master Edition. (Capcom)
Pearl Abyss announced Crimson Desert surpassed 5M copies worldwide and reached a Steam peak of 276,261 concurrent players. (Pearl Abyss)
Pearl Abyss’s 1Q26 material shows Crimson Desert 1Q revenue of KRW 266.5B and 2026 Crimson Desert guidance of KRW 644.1B-734.8B. (1Q26 Earnings Letter)

[Inference]

Pearl Abyss’s current discount is more about 2027 earnings durability and repeatable IP status than about game failure.
The Neowiz case shows how the Korean market can discount a global package-game hit as one-off until repeatability is quantified.
The CD Projekt case shows how patches and expansion sales can restore IP trust and support re-rating.
Capcom is a target model, not the current comp.

[Blocked]

The 5.88M Crimson Desert unit figure is a local estimate; the official company number is 5M+.
DLC timing, pricing, content scope and revenue contribution are not disclosed.
DokeV launch timing and 2027 earnings contribution are not confirmed.
The flow and short-sale comparison is based on Research OS local data and KRX data, not a public company disclosure.

Marvell and Broadcom Earnings Preview — Korea's AI Bottleneck Check Before the Calls

Sat, 23 May 2026 09:30:00 +0900

Related series: ARM rally and Korea’s next AI bottlenecks / NVIDIA Q1 FY27 and Korean AI infrastructure / Samsung Electro-Mechanics silicon capacitor contract / Samsung Electronics re-rating thesis

Marvell and Broadcom’s next earnings calls are not just U.S. semiconductor events. They are a check on whether the AI infrastructure bottleneck is moving from GPU-only demand into custom ASICs, Ethernet networking, optical links, package substrates, silicon capacitors, high-speed PCBs and test sockets. If the numbers are strong, Korea’s read-through expands from SK hynix and Samsung Electronics HBM into Samsung Electro-Mechanics, FC-BGA, power-integrity components and selected back-end names.

Summary

Broadcom is the larger macro signal. In Q1 FY2026, Broadcom reported revenue of $19.311B, AI revenue of $8.4B, Q2 revenue guidance of $22.0B, and Q2 AI semiconductor revenue guidance of $10.7B. On that guidance, AI semiconductor revenue alone would be 48.6% of total Q2 revenue. Broadcom is no longer just “semis plus VMware.” It is increasingly an AI custom-chip and AI networking company with a VMware cash-flow engine attached. (Broadcom)

Marvell is smaller, but the signal is cleaner. Q4 FY2026 revenue was $2.219B, Q1 FY2027 guidance is $2.4B ±5%, and Q4 data-center revenue was $1.651B, or 74% of revenue. Marvell’s call will test whether custom silicon, electro-optics and CXL/PCIe switching are moving from narrative into revenue acceleration. (Marvell)

For Korean investors, the question is not “Are Marvell and Broadcom AI winners?” The question is whether their comments confirm a broader physical bottleneck: XPU + HBM + Ethernet fabric + optical links + advanced packaging + power integrity.

If that signal is strong, the Korean priority stack becomes:

SK hynix / Samsung Electronics — HBM, server DRAM, LPDDR/SOCAMM.
Samsung Electro-Mechanics — silicon capacitor, FC-BGA, MLCC and package-level power integrity.
High-speed PCB and test sockets — Isu Petasys, Daeduck, Simmtech, TLB, Korea Circuit, ISC, LEENO, TSE, with direct revenue proof required.

The conclusion is simple: Marvell and Broadcom earnings are a test of whether Korea’s AI semiconductor thesis should broaden beyond HBM into ASIC networking, packaging and power-delivery bottlenecks.

1. Dates and Baselines

Item	Marvell	Broadcom
Next earnings	Q1 FY2027, May 27, 2026 at 1:45 p.m. PDT	Q2 FY2026, June 3, 2026 at 5:00 p.m. EDT
Korea time	May 28, 2026 05:45 KST	June 4, 2026 06:00 KST
Latest reported revenue	Q4 FY26 $2.219B	Q1 FY26 $19.311B
Next guidance	Q1 FY27 $2.4B ±5%	Q2 FY26 $22.0B
AI / data-center benchmark	Q4 data-center $1.651B, 74% mix	Q1 AI revenue $8.4B, Q2 AI semiconductor $10.7B guide

Marvell’s Q1 guide implies roughly +8.2% QoQ growth at the midpoint. Broadcom’s Q2 AI semiconductor guide implies +27.4% QoQ from Q1 AI revenue. These are the two numbers to watch first.

2. Broadcom — The Bigger Macro Signal

The key Broadcom question is not only whether AI semiconductor revenue beats $10.7B. The more important question is whether AI networking is rising inside the AI revenue mix.

Broadcom’s AI infrastructure exposure includes custom XPUs, Ethernet switches, NICs, SerDes, DSPs, retimers and optical connectivity. As clusters expand, the bottleneck shifts from individual accelerators to the fabric connecting chips, racks and data centers.

Broadcom checkpoint	Why it matters	Korea read-through
Q2 AI semiconductor revenue above $10.7B	Confirms custom-chip demand strength	HBM, FC-BGA, test, package components
Rising AI networking share	Confirms Ethernet and optical bottlenecks	High-layer PCB, low-loss materials, sockets
Google TPU and long-duration customer visibility	Extends the custom-ASIC cycle	HBM customer diversification, packaging duration
Margin defense	Revenue growth must convert into profit	Supplier pricing power check
Packaging and supply-chain comments	Shows where the physical constraint sits	Samsung Electro-Mechanics, memory, substrates, test

Broadcom’s OpenAI collaboration is also important. The companies announced a 10GW custom AI accelerator collaboration, with deployment expected to begin in the second half of 2026 and run through the end of 2029. The system includes accelerators as well as Ethernet, PCIe and optical connectivity. (Broadcom OpenAI)

3. Marvell — The Cleaner Connectivity Signal

Marvell’s market cap and revenue base are smaller than Broadcom’s, but it is a cleaner signal for the “connectivity bottleneck” thesis.

The call should answer four questions:

Marvell checkpoint	Why it matters	Korea read-through
Q1 FY27 revenue vs $2.4B guide and Q2 guide	Confirms FY27 acceleration	AI ASIC and interconnect demand durability
Custom-silicon design wins	Shows hyperscaler ASIC diversification	HBM, packaging and test demand broadening
Electro-optics / DSP / 1.6T-3.2T roadmap	Confirms data-movement bottleneck	Limited direct Korea exposure; long-term option
XConn / CXL / PCIe	Connects to memory pooling	CXL memory, SOCAMM, server DRAM
NVLink Fusion collaboration	Shows custom ASIC and NVIDIA ecosystems can coexist	HBM diversification, test and package complexity

In April 2026, Marvell announced the Polariton acquisition to strengthen its optical roadmap toward 3.2T and beyond. As AI workloads expand, data-center bandwidth demand rises and optical links become a power-efficiency problem, not only a speed problem. (Marvell Polariton)

Marvell’s XConn acquisition points to PCIe/CXL switching and scale-up fabric. That matters for Korean memory because CXL and memory pooling broaden the hierarchy from “HBM next to GPU” into rack-level memory architecture. (Marvell XConn)

4. Korea Translation — What Comes After HBM?

The first-order Korea frame has been:

AI = NVIDIA GPU = HBM = SK hynix.

That remains true, but it is no longer enough. If Broadcom and Marvell print strong numbers, the frame broadens:

AI = GPU + custom ASIC + HBM + Ethernet + optics + packaging + power integrity + test.

Memory Still Comes First

Custom ASICs may compete with NVIDIA GPUs for some workloads, but they do not remove the memory bottleneck. Google TPUs, Broadcom XPUs, OpenAI accelerators and Marvell custom silicon all need high-bandwidth memory. SK hynix remains the cleaner exposure. Samsung Electronics has a larger option set, but needs HBM4, HBM4E and customer qualification proof.

Samsung Electro-Mechanics Is the Cleanest Non-Memory Bottleneck

Samsung Electro-Mechanics has three relevant exposures: FC-BGA, MLCC and silicon capacitors. The company announced an approximately KRW 1.5T silicon capacitor supply contract for 2027-2028 and described the product as a component used inside high-performance packages such as AI server GPUs and HBM to stabilize power supply. (Samsung Electro-Mechanics)

This is the cleanest Korean non-memory read-through from AI ASIC and AI networking. As chips get larger and more power-dense, power integrity becomes a package-level constraint.

PCB and Test Need Proof

Isu Petasys, Daeduck, Simmtech, TLB and Korea Circuit are plausible high-speed board and module substrate candidates. ISC, LEENO and TSE are plausible test-socket candidates. But this group needs proof: customer qualification, AI-related revenue, ASP uplift and margin preservation.

5. Practical Watchlist

Priority	Layer	Korean names	View
1	HBM / server DRAM / SOCAMM	SK hynix, Samsung Electronics	Core layer. Custom ASICs expand the customer base
2	Silicon capacitor / FC-BGA	Samsung Electro-Mechanics	Clearest non-memory bottleneck with an official large contract
3	High-speed PCB / low-loss materials	Isu Petasys, Daeduck, Simmtech, TLB, Korea Circuit, Doosan Electronics BG	Positive if AI networking grows; direct proof required
4	Test sockets / inspection	ISC, LEENO, TSE, Intekplus	Benefits from ASIC SKU growth and I/O complexity
5	Optical interconnect	Korean optical modules and materials candidates	Directionally right, but direct Korea exposure is less clear

6. What To Listen For

For Broadcom, bullish lines would be:

AI semiconductor revenue tracking above $10.7B.
AI networking share rising.
Better visibility into 2027.
Supply chain and packaging constraints under control.

For Marvell, bullish lines would be:

Q1 FY27 above the $2.4B midpoint and Q2 guidance showing acceleration.
Record custom-silicon bookings or design wins.
Electro-optics, optical DSP and CXL/PCIe switching turning into FY27/FY28 revenue.
Data-center growth and gross margin holding together.

Final Read

Marvell and Broadcom earnings are not a blanket buy signal for Korean semiconductors. They are a bottleneck map.

If Broadcom beats the AI semiconductor guide and talks up AI networking, Korea’s read-through expands beyond HBM into FC-BGA, high-speed PCBs, test sockets and silicon capacitors. If Marvell confirms custom silicon and optical interconnect acceleration, AI servers should be viewed less as GPU boxes and more as memory, connectivity, package and power-delivery systems.

The safest conclusion is still disciplined: SK hynix and Samsung Electronics remain the memory core; Samsung Electro-Mechanics is the clearest non-memory bottleneck; PCB and test-socket names need revenue and margin proof before they deserve full re-rating.

Fact / Inference / Blocked

[Fact]

Marvell’s Q1 FY2027 call is scheduled for May 27, 2026 at 1:45 p.m. PDT. (Marvell)
Marvell’s Q4 FY2026 revenue was $2.219B and Q1 FY2027 guidance is $2.4B ±5%. (Marvell)
Marvell’s Q4 FY2026 data-center revenue was $1.651B, or 74% of revenue. (Marvell)
Broadcom’s Q2 FY2026 earnings call is scheduled for June 3, 2026 at 5:00 p.m. EDT. (Broadcom Events)
Broadcom’s Q1 FY2026 AI revenue was $8.4B and Q2 AI semiconductor guidance is $10.7B. (Broadcom)
Samsung Electro-Mechanics announced an approximately KRW 1.5T silicon capacitor supply contract. (Samsung Electro-Mechanics)

[Inference]

Custom ASIC growth is more likely to broaden HBM demand than reduce it.
For Korea, the more actionable read-through is memory, package substrates, silicon capacitors and test sockets rather than generic optical themes.
Samsung Electro-Mechanics’ silicon capacitor contract is evidence that package-level power integrity is moving into commercial supply.

[Blocked]

Customer-level proof that specific Korean suppliers ship directly into Broadcom or Marvell ASIC programs.
The exact customer, margin and package placement for Samsung Electro-Mechanics’ silicon capacitor contract.
AI customer revenue split for Korean PCB and test-socket suppliers.

This article is for research and commentary only, not investment advice. Earnings dates and guidance are based on company IR materials as of May 23, 2026 KST. Korean supply-chain exposure includes inference because customer-level disclosures are limited.

KOSDAQ Surges 4% as ABF Substrate Stocks Break Out

Fri, 22 May 2026 23:30:00 +0900

KOSDAQ Jumps 4%, ABF Substrate Names Break Out While Samsung Electronics Faces Foreign Selling

South Korea’s equity market split sharply on Friday, May 22, 2026. The KOSPI — South Korea’s large-cap benchmark index — gained a modest 0.5%, but the story was in the secondary market: the KOSDAQ, home to South Korea’s mid-cap growth and tech names, surged roughly 4% on the session. Market breadth expanded meaningfully, with 99 stocks passing momentum and flow screens — a signal that institutional interest is rotating beyond a handful of mega-caps into a broader set of Korean equities.

The divergence was not random noise. It reflects a shift in what the market is pricing: not the AI semiconductor giants, but the infrastructure layer underneath them.

ABF Substrate: Morgan Stanley Raises the Bar

The most significant fundamental catalyst of the session came from a Morgan Stanley research note circulating in Korean market channels. The bank raised its forecast for high-performance ABF (Ajinomoto Build-up Film) substrate shortage rates — a key bottleneck in advanced semiconductor packaging — from 15% to 22% by 2030.

ABF substrate is a critical material used in high-end chip packaging, particularly for AI accelerators and high-bandwidth memory (HBM) applications. Supply has been structurally constrained as chipmakers race to stack more logic and memory in tighter packages. A 22% shortage rate by decade’s end implies sustained pricing power for the handful of companies that dominate global ABF substrate supply.

Two Korean names sit at the center of this theme.

Samsung Electro-Mechanics (009150.KS), South Korea’s largest passive components and substrate manufacturer, gained +11.30% on Friday — bringing its five-day return to +32.67%. Foreign investors were net buyers of ₩17.4 billion during the session. Institutional funds were net sellers, suggesting profit-taking after the sharp run-up, but price action held above the ₩1.219 million breakout level that technicians had flagged as the key reference.

Daedeok Electronics (317400.KS), a specialist in ABF and HDI printed circuit boards for advanced packaging, rose +3.35% with five-day gains of +11.41%. Institutional buying of ₩12.7 billion was notable, though foreign investors trimmed ₩7.4 billion — a split that bears watching heading into next week.

Why does the Morgan Stanley revision matter now? Because the market had already begun pricing a substrate recovery, but the revised shortage estimate provides fresh analytical support for continuing that move. The question is whether the 5-day momentum in these names has gotten ahead of near-term fundamentals, or whether the re-rating has further to run as more sell-side houses update their models.

Samsung Electronics: Foreign Selling Resumes After 5-Day Recovery

Samsung Electronics (005930.KS), South Korea’s largest company by market cap and the world’s top memory chip producer by revenue, gave back -2.34% on Friday after a strong five-day run of +8.13%. The one-day pullback was amplified by foreign investor net selling of ₩1.01 trillion — one of the largest single-session foreign outflows for the stock in recent weeks. Program trading added another ₩998.2 billion of net selling pressure.

The timing is awkward. Mirae Asset, one of Korea’s largest domestic brokerages, included Samsung Electronics and SK Hynix (000660.KS) — South Korea’s second-largest memory chipmaker — in its second-half Top Picks list, citing re-rating potential as memory pricing and AI demand visibility improve. That positive framing from domestic research sits uneasily against Friday’s foreign outflow.

The pattern is familiar: a sharp recovery rally, then a test of whether foreign investors use strength to reduce exposure. At ₩1 trillion of selling in a single session, Friday qualifies as the latter. Until foreign and program selling decelerates, Samsung Electronics is likely range-bound rather than in a clean breakout.

Jusung Engineering and the Semiconductor Equipment Breakout

Friday’s top-performing new entrant in momentum screens was Jusung Engineering (036930.KS), a Korean semiconductor equipment maker focused on deposition and etch tools for advanced logic and memory fabs. The stock surged +20.95% in a session characterized by simultaneous buying from both foreign investors (+₩24.3 billion) and domestic institutions (+₩15.8 billion) — a “co-buy” pattern that screeners flag as a higher-conviction entry signal.

The ₩214,000 level becomes the key reference: it represents where the stock opened the gap, and holding above it in the coming sessions would validate the breakout rather than mark it as a one-day spike. Equipment stocks have underperformed the semiconductor supply chain for most of 2025; if capex guidance from Samsung and SK Hynix for the second half of 2026 holds firm, the equipment cycle laggards have room to catch up.

Simtech (222800.KS), a PCB and ABF substrate specialist, also appeared in upper tiers of momentum screens with a +5.02% gain on co-buy flows (foreign +₩8.6 billion, institutional +₩5.8 billion). The name is worth tracking alongside Daedeok Electronics as the ABF substrate supply-chain thesis widens.

Sector Rotation: Beyond Semiconductors

One of the clearer signals from Friday’s session is that leadership is no longer concentrated in semiconductors. The strongest sectors on the day included:

Pharmaceuticals and biotech: several names saw double-digit gains, extending a recent run that some desks attribute to pipeline updates and improving regulatory sentiment
Power and energy infrastructure: beneficiary of AI data center electricity demand narratives gaining traction in Korean media
Semiconductor materials and components (소부장): the ABF names above, plus smaller equipment and chemical suppliers

Telecom was notably weak. SK Telecom (017670.KS), South Korea’s largest mobile carrier by subscriber count, was flat on the day with net foreign selling of ₩58.6 billion and institutional selling of ₩4 billion. Defensive sectors tend to underperform on days when breadth expands into growth names — and Friday fit that pattern cleanly.

The won/dollar exchange rate at ₩1,516 remains a background constraint. A weaker won increases the cost of dollar-denominated imports and adds translation risk for foreign investors considering Korean equity exposure, which partly explains why even a broadly positive day produced heavy foreign selling in liquid large-caps like Samsung Electronics.

What to Watch Next Week

Five checkpoints define the near-term outlook for Korean equity market participants:

Samsung Electro-Mechanics price defense: Does the ₩1.219 million breakout level hold, and does the stock extend toward ₩1.34 million? Institutional profit-taking is normal after a 32% five-day run; the question is whether foreign buying absorbs it.
Samsung Electronics foreign flow: Does Friday’s ₩1.01 trillion outflow represent a one-day flush, or does it continue into Monday? Two consecutive sessions of heavy foreign selling would shift the near-term bias.
Daedeok Electronics foreign buyer conversion: Institutional accumulation is constructive, but ABF substrate stocks tend to get a second leg only when foreign investors also commit. Watch whether foreign selling of ₩7.4 billion shrinks or reverses.
Jusung Engineering breakout validation: A close above ₩214,000 on Monday — ideally on continued co-buy flows — would confirm the equipment sector is re-entering a leadership window.
KOSDAQ breadth composition: Friday’s 4% KOSDAQ surge was broad, but distinguishing between a bio-led one-day spike and a sustained rotation into semiconductor equipment and growth names matters for whether the trend extends.

The ABF substrate re-rating, if the Morgan Stanley supply shortage estimates prove durable, is the kind of multi-year structural theme that tends to produce several distinct entry windows rather than a single breakout. Friday may have been the first window. Whether it holds depends on whether the fundamental thesis outpaces the technical overextension.

Data sourced from KRX intraday flow data, KR Crawler DB (prices_daily, investor_flow_daily, program_trade_daily) as of 16:05 KST on May 22, 2026. Ticker symbols reference KRX listings. This post is market analysis, not investment advice.

ARM's Rally — The Next AI Bottleneck Is CPU Orchestration, Optical Links and Power Integrity

Fri, 22 May 2026 21:40:00 +0900

Related series: NVIDIA Q1 FY27 and Korean AI infrastructure / Samsung Electro-Mechanics’ KRW 1.5T silicon-capacitor contract / Understanding MLCCs and silicon capacitors / Samsung Electronics rerating thesis

ARM’s rally is not simply a statement that “CPUs matter again.” More precisely, AI servers are being redefined from GPU boxes into systems that combine CPUs, XPUs, HBM, optical links, power-stabilizing components, high-speed substrates and test infrastructure. ARM is the visible price reaction. But at the current valuation, a large part of the five-year success case is already embedded. For Korean investors, the more useful signal is the next bottleneck: Marvell’s custom silicon and optical links, Samsung Electro-Mechanics’ silicon capacitors, SK hynix and Samsung Electronics memory, and the high-speed substrate / test-socket layer.

Key Takeaways

ARM’s rally is a reclassification event: from mobile IP royalty company toward AI data-center CPU platform company. As AI agents run continuously, call tools, move data and coordinate GPUs, ASICs and memory, the CPU becomes the orchestration layer of the AI rack.

The external catalyst was NVIDIA’s Q1 FY27. NVIDIA reported $81.6B in revenue, $75.2B in Data Center revenue, and Q2 guidance of $91.0B ±2%. The market read that as acceleration, not peak-out, and began repricing the layers below GPUs: CPU, memory, interconnect, power and substrates. (NVIDIA)

ARM’s own story is strong. FY26 revenue was $4.92B, royalty revenue $2.61B, licensing revenue $2.31B, and non-GAAP EPS $1.77. The Arm AGI CPU is positioned as a data-center CPU platform with Meta as lead partner, and Arm says FY27-FY28 customer demand is already above $2B. (Arm)

The problem is price. Research OS local DB shows ARM closed at $298.23 on May 21, 2026. Against FY26 non-GAAP EPS of $1.77, that is roughly 168.5x earnings. Using a market capitalization around the low-$300B range and FY26 revenue of $4.92B, the stock trades around the 60x sales zone. The thesis is real; the stock is not cheap.

The practical alpha is therefore not ARM itself, but the bottlenecks ARM’s move points toward: Marvell’s custom silicon and optical fabric, Samsung Electro-Mechanics’ silicon capacitor, SK hynix / Samsung Electronics HBM and DRAM, high-speed substrates and test sockets.

The one-line conclusion: ARM is the signal. The alpha sits in the bottlenecks.

1. What ARM’s Rally Really Means

Calling this a “CPU renaissance” is only half right. The deeper message is that the constraint is moving.

In 2023-2024, the core AI infrastructure bottleneck was GPUs. In 2025, HBM and advanced packaging rose to the front. In 2026 and beyond, as AI racks become larger systems, CPU orchestration, memory movement, optical links, power integrity, substrates and testing all become limiting factors.

GPU scarcity
→ HBM / packaging scarcity
→ CPU orchestration / memory movement / optical interconnect bottlenecks
→ package-level power integrity / high-speed substrate / test bottlenecks

AI agents are not ordinary web requests. They call models, use tools, read documents, query databases and store intermediate outputs. GPUs handle compute, but CPUs coordinate task order, memory management, network I/O, security, error handling and system operation.

This makes the CPU the control plane of the AI rack. If the GPU is the engine, the CPU is the traffic controller. ARM’s rally reflects the market starting to price that role.

Arm’s FY26 results support the narrative. The company reported FY26 revenue of $4.92B and Q4 revenue of $1.49B, with record annual royalty and licensing revenue. Data-center royalties more than doubled year over year. (Arm)

But the right investment question is not “is ARM’s thesis right?” It is “does the current price still leave enough risk-reward?”

2. Arm AGI CPU: From IP Vendor to Silicon Platform

Arm has put the Arm AGI CPU at the center of the story. It describes the product as production silicon for agentic AI data centers, with Meta as lead partner. Arm also says commercial systems are available to order through ASRock, Lenovo, Quanta and Supermicro. (Arm)

The old Arm model was:

IP licensing
→ customers design and manufacture chips
→ Arm receives license fees and royalties

The AGI CPU extends that model:

IP + compute subsystems + production silicon
→ customers deploy the Arm platform faster at data-center scale
→ Arm gains more platform control

This is powerful, but it changes the risk profile. Arm is no longer only a neutral blueprint provider. As it sells its own silicon, customers and licensees may begin to see it as a potential competitor.

That is why regulatory risk matters. Bloomberg has reported that the FTC is looking into whether Arm could restrict or degrade license access while expanding its own chip ambitions. (Bloomberg)

The AGI CPU is a long-term call option, but it also makes Arm heavier, more complex and more exposed to channel conflict.

3. Valuation: Correct Thesis, Expensive Stock

The uncomfortable part of the ARM rally is the math. Directionally, the thesis makes sense. Valuation already discounts a long runway.

Research OS local DB shows ARM closed at $298.23 on May 21, 2026. Against FY26 non-GAAP EPS of $1.77:

FY26 non-GAAP P/E
= 298.23 / 1.77
= about 168.5x

Against FY26 revenue of $4.92B and a market cap in the low-$300B range:

FY26 P/S
= roughly low-$300B / $4.92B
= about mid-60x sales

This is not a price for “good current earnings.” It is a price for a 2030-2031 scenario: higher data-center CPU penetration, AGI CPU revenue, higher royalty rates and Armv9 / Neoverse expansion.

If Arm reaches something like $25B of revenue and $9+ of EPS by FY31, today’s price can be explained. But that is a five-year success case. Investors need to separate a correct story from a good entry.

Our view is Avoid at current price / Wait. The company is not the problem. The entry price is.

4. Better Alpha 1: Marvell and the Connectivity Bottleneck

If ARM represents CPU orchestration, Marvell represents custom compute plus optical connectivity plus switching. This may be the more actionable derivative idea.

Marvell reported FY26 revenue of $8.195B and non-GAAP EPS of $2.84. It expects FY27 revenue to approach $11B, with data center leading growth and interconnect revenue growing more than 50%. (Marvell)

Marvell is not just a networking chip company.

Layer	Meaning
Custom AI silicon	Design partner as hyperscalers build their own XPUs
Optical interconnect	Next bottleneck for intra-rack and inter-rack data movement
PCIe / CXL switching	Connects CPU, GPU, ASIC and memory pools
NVLink Fusion partnership	Strategic position inside heterogeneous NVIDIA-connected AI infrastructure

The Celestial AI acquisition adds a photonic scale-up platform. Marvell has said it expects the business to reach a $500M annualized run-rate by Q4 FY28 and a $1B annualized run-rate by Q4 FY29, if milestones are met. (Marvell Celestial)

The caveat: Marvell has already rallied. Research OS local DB shows MRVL closed at $190.69 on May 21, 2026. At this level, the May 27 earnings report needs to confirm the FY27 revenue path, data-center demand, custom silicon and optical interconnect momentum.

Our view: Wait / Buy on pullback.

5. Better Alpha 2: Samsung Electro-Mechanics and Power Integrity

In Korea, the clearest second-order bottleneck is Samsung Electro-Mechanics.

On May 20, 2026, Samsung Electro-Mechanics announced a silicon-capacitor supply contract worth approximately KRW 1.5T over two years, from January 1, 2027 to December 31, 2028. The company explained that silicon capacitors are installed inside high-performance semiconductor packages such as AI server GPUs and HBM to improve power stability. (Samsung Electro-Mechanics)

This is the same direction as ARM’s rally. As AI racks become systems rather than GPU boxes, CPU, GPU, HBM and ASIC packages become denser and draw power faster. Voltage noise and power fluctuations become a bottleneck.

Samsung Electro-Mechanics’ thesis is not “MLCC demand rises.” It is:

Old SEMCO:
MLCC + camera modules + FC-BGA

New SEMCO:
AI package power-integrity components
+ FC-BGA
+ AI-grade MLCC

A silicon capacitor does not replace all MLCCs. It is a high-performance complement used inside or very close to AI GPU / HBM packages. This can reclassify Samsung Electro-Mechanics from a passive-parts cycle name into an AI package power-integrity supplier.

The risks are clear: the customer is undisclosed, 2026 earnings impact is limited, margins are not disclosed, and the stock has already reacted. Our view remains Watchlist → Buy on pullback.

6. Better Alpha 3: SK hynix and Samsung Electronics Memory

ARM’s rally is positive for Korean memory as well. More CPU orchestration means more CPU-side memory, more data movement to HBM, more LPDDR / DDR / CXL memory pooling, and more server DRAM.

GPU-centric server
→ GPU + CPU + DPU + NIC + switch ASIC + HBM + DRAM
→ higher rack-level memory capacity and bandwidth

SK hynix is the cleanest HBM and server DRAM beneficiary, but it is also a crowded winner. The view is Hold / Pullback Buy.

Samsung Electronics is different. It is the catch-up option: HBM4, SOCAMM2, server DRAM, eSSD, HBM4 base die and foundry optionality. Since Arm AGI CPU manufacturing is tied to TSMC 3nm, one should not overstate direct Samsung Foundry exposure. But AI rack diversification still improves the value of Samsung’s memory and packaging ecosystem.

Company	Role	View
SK hynix	Proven HBM winner	Hold / Pullback Buy
Samsung Electronics	HBM catch-up + broad IDM option	Watchlist / Buy on confirmation

Samsung needs evidence: HBM4 qualification, volume, margins and foundry loss reduction.

7. Better Alpha 4: High-Speed Substrates and Test Sockets

As AI racks evolve from single GPU boards into dense systems, substrates and testing become second-order winners.

High-speed substrates and PCBs. More CPUs, GPUs, ASICs, HBM, NICs and switch ASICs mean more high-speed signaling and more complex board design. Isu Petasys, Daeduck Electronics, TLB, Korea Circuit and Simmtech are candidates. But direct orders, customer qualification, ASP and margin must be verified.

Test sockets. As chips become faster, hotter and more pin-dense, sockets and probe-card requirements rise. ISC, LEENO Industrial and TSE are relevant candidates. This layer can carry higher margins and less capex cyclicality than substrates, but product mix matters.

Layer	Candidates	What to Verify
High-speed PCB	Isu Petasys, Daeduck, TLB, Korea Circuit, Simmtech	AI-server revenue, layer count, ASP, OPM
Test sockets	ISC, LEENO, TSE	ASIC/HBM/CXL customers, product mix, new socket revenue
Package substrate	Samsung Electro-Mechanics, Daeduck, Korea Circuit	FC-BGA utilization, qualification, margins

The rule: do not buy the phrase “AI exposure.” Buy verified customers, orders and margins.

8. Strategy

Name / Domain	View	Reason
ARM	Avoid at current price / Wait	Correct thesis, but very rich valuation
Marvell	Wait / Buy on pullback	Custom silicon + optical + switching bottleneck
Samsung Electro-Mechanics	Watchlist → Buy on pullback	Silicon capacitor can reclassify it as AI package power-integrity supplier
Samsung Electronics	Watchlist	HBM4, SOCAMM, server DRAM, eSSD, foundry option
SK hynix	Hold / Pullback Buy	Cleanest HBM winner, but crowded
High-speed substrates	Watchlist	Beneficiary, but verify customers and margins
Test sockets	Watchlist	Chip-complexity consumables beta

Catalysts

Marvell earnings on May 27, 2026.
Arm AGI CPU supply expansion and conversion from demand to revenue.
NVIDIA Vera Rubin second-half ramp.
Samsung Electro-Mechanics follow-on silicon-capacitor customers or capacity announcements.
Samsung Electronics HBM4 qualification.
Korean substrate / test-socket companies separating AI revenue in quarterly results.

Invalidation

Arm AGI CPU demand fails to convert because of supply, margins or channel conflict.
FTC / regulatory risk damages Arm’s licensing model.
Marvell loses custom design wins to Broadcom or optical revenue ramps too slowly.
Samsung Electro-Mechanics’ silicon-capacitor contract carries weak margins or lacks follow-on customers.
Korean substrate / test names report AI revenue growth without OPM support.

Final Line

ARM’s rally is a correct revaluation signal. AI servers are no longer GPU boxes. They are CPU, XPU, HBM, optical, power-integrity, substrate and test systems.

But buying ARM after the move may confuse the signal with the asset. The better question is: which bottleneck has not yet been fully priced?

Our answer: ARM is the signal. The alpha sits in the bottlenecks.

Evidence Map

[Fact]

NVIDIA Q1 FY27 revenue was $81.6B, Data Center revenue was $75.2B, and Q2 revenue guidance was $91.0B ±2%. (NVIDIA)
Arm FY26 revenue was $4.92B, royalty revenue $2.61B and licensing revenue $2.31B. Arm said AGI CPU customer demand for FY27-FY28 was above $2B. (Arm)
Marvell FY26 revenue was $8.195B, non-GAAP EPS was $2.84, and the company expects FY27 revenue approaching $11B. (Marvell)
Samsung Electro-Mechanics announced a KRW 1.5T silicon-capacitor supply contract on May 20, 2026. (Samsung Electro-Mechanics)
Research OS local DB shows ARM closed at $298.23 and MRVL at $190.69 on May 21, 2026.

[Inference]

ARM’s rally signals a bottleneck shift from GPUs toward CPU orchestration, memory movement, optical interconnect and power integrity.
The better risk-reward may sit in Marvell, Samsung Electro-Mechanics, Korean memory, high-speed substrates and test sockets rather than ARM itself.

[Blocked]

Arm AGI CPU customer-level orders and margins.
Marvell customer-level custom silicon revenue.
Samsung Electro-Mechanics silicon-capacitor customer, ASP, margin and exact package position.
Direct ARM/NVIDIA/Marvell exposure for Korean substrate and test names.

This article is research commentary, not investment advice. US price data is based on Research OS local DB closing prices for May 21, 2026. Company data is based on official NVIDIA, Arm, Marvell and Samsung Electro-Mechanics materials. Scenario views are analyst frameworks and may be wrong. Data cut: May 22, 2026 KST.

Korea Quality Re-Rating Watch 2026-05-22: VM, SamC&S and TLB — Five-Screener Clean Sweep

Fri, 22 May 2026 16:30:00 +0900

Macro Dashboard

Indicator	Level	5-Day Δ	Signal
KOSPI	7,847.7	+4.4%	▲ Bull
KOSDAQ	1,161.1	+4.5%	▲ Bull
VIX	17.0	−4.5%	✓ Stable
US 10Y	4.59%	−0.01pp	→ Flat
USD/KRW	1,517	+1.3%	⚠ KRW weak
Brent	$105.4	−6.0%	▼ Falling
KR Regime	Bull	—	Aggressive expand
US Regime	Bull	—	Supportive

Both Korea and the US are in Bull regime. KOSPI five-day breadth above the 50-day MA stands at 36.0% and above the 200-day MA at 47.8% — constructive but not overheated. The main friction: USD/KRW at 1,517 compresses USD-denominated returns even as Korean prices advance. The tailwind: Brent down 6% over five sessions takes pressure off Korea’s energy-import bill and widens industrial margins.

Market Wrap

Today’s session was selective risk-on, not a full-market breakout. KOSPI held near +0.5% intraday while KOSDAQ surged approximately +4%, with 99 names clearing quality screener filters — the widest breadth count in recent sessions.

Sectors leading. Pharmaceuticals and biotech drove KOSDAQ’s advance, joined by power/energy equipment and semiconductor materials-and-equipment plays. The ABF substrate theme re-ignited after Morgan Stanley reportedly raised its 2030 supply-shortage-rate forecast for high-performance ABF substrates from 15% to 22% — a demand-outlook upgrade that directly lifted PCB and advanced packaging names. Samsung Electro-Mechanics (009150.KS) surged +11.3% on the day (+32.7% over five sessions) on that thesis; Daedeok Electronics (353200.KQ) added +3.3% in the same theme, though institutional and foreign flows diverged between the two. Semiconductor equipment saw a separate breakout: Juseong Engineering (036930.KQ) rallied +20.9% with simultaneous foreign and institutional co-buying.

Sectors lagging. Large-cap semis absorbed profit-taking after the prior session’s rally. Samsung Electronics (005930.KS) recorded net foreign selling of approximately ₩1.01tn and program selling of ₩998bn, capping KOSPI upside. Telecoms and defensives underperformed as capital rotated toward higher-beta growth themes.

Flow picture. Institutional and foreign buying concentrated in mid-cap semiconductor equipment, ABF substrate, and biotech names while reducing in large-cap semis and telecoms. KOSDAQ breadth above the 50-day MA reached 36.0%; quality screener breadth above the 200-day MA hit 47.8% — both consistent with a broadening rather than narrowing rally. That said, USD/KRW above 1,515 and the magnitude of Samsung Electronics’ foreign outflow signal that full risk-on has not been confirmed. Today reads as quality mid-cap accumulation with selective large-cap rotation out.

Today’s Quality Re-Rating Candidates

The May 22 meta screener processed 115 names and produced a sharp hierarchy. Three stocks cleared all five quality-flow screeners simultaneously — Quality Compounder, Smart Money Quality, Cycle Rerating, Smart Money Earnings, and PEAD — a combination that rarely fires on more than one or two names per session.

Top 10 Candidates

#	Ticker	Name	Meta Score	Overlap	Screeners Hit	Key Metrics
1	252990.KQ	SamC&S	97.5	5/5	QC · SMQ · CR · SME · PEAD	ROE 9.7%, OP +183%, OPM +9.1pp, F+QI +₩5.5bn 5d
2	089970.KQ	VM	96.5	5/5	QC · SMQ · CR · SME · PEAD	ROE 14.7%, OP +387%, OPM +29.3pp, F+QI +₩33.0bn 5d
3	356860.KQ	TLB	90.1	4/5	QC · SMQ · CR · SME	ROE 16.0%, OP +665%, OPM +8.2pp, F+QI +₩8.5bn 5d
4	053610.KQ	Protek	72.8	3	QC · CR · PEAD + Consensus↑	ROE 13.7%, OP +245%, OPM +12.2pp
5	095610.KQ	TES	72.2	3	QC · SMQ · SME	ROE 14.5%, OP +50%, F+QI +₩26.1bn 5d, DART supply contract
6	000660.KS	SK Hynix	65.8	2	QC · CR + Consensus↑	ROE 35.6%, OP +101%, OPM +13.1pp ⚠ F+QI sell ₩5.35tn 5d
7	218410.KQ	RFHIC	59.7	4	SMQ · CR · SME · PEAD	OPM +15.3pp, F+QI +₩13.9bn 5d ⚠ short interest 7.2%
8	147830.KS	Jearyong Ind.	58.4	3	QC · CR · PEAD	ROE 13.9%, OP +159%, OPM +13.9pp
9	092460.KQ	Halla IMS	53.5	2	QC · CR	ROE 13.8%, OP +39%, DART earnings pre-disclosure
10	482630.KQ	Samyang NCC	51.7	2	QC · PEAD	ROE 14.7%, OP +65%, IR held May 18

QC = Quality Compounder · SMQ = Smart Money Quality · CR = Cycle Rerating · SME = Smart Money Earnings · F+QI = foreign + qualified institutional net flow

Deeper Look: Top 3

SamC&S (252990.KQ) — Meta Score 97.5

SamC&S is an electronic components manufacturer. The three-layer logic is intact: (1) Quality — operating profit grew +183% YoY with OPM expanding +9.1pp, an early earnings inflection for a business with stable fundamentals; (2) Flow — foreign and qualified institutional investors net-bought ₩5.5bn over five days, absorbing ₩5.4bn of retail selling — a clean smart-money accumulation pattern; (3) Catalyst — the company held an IR session on May 21 (DART-confirmed), indicating management is actively engaging investors around the improved story. Next checks: OPM sustainability into Q2 2026; whether institutional buying persists after the post-IR window closes.

VM (089970.KQ) — Meta Score 96.5

VM manufactures semiconductor equipment. The cycle signal is exceptional: operating margin expanded +29.3pp on +105% revenue growth, producing an operating leverage spread of 281pp. The Cycle Rerating screener ranks VM #1 (score 0.747) — typically a signal that revenue is converting to earnings at an unusually high rate. Foreign and qualified institutional buying of ₩33.0bn over five sessions absorbed ₩21.5bn in retail selling. The stock trades at its 52-week high. Next checks: H2 2026 order backlog disclosures; whether the margin expansion is structural or concentrated in a single large order.

TLB (356860.KQ) — Meta Score 90.1

TLB makes electronic components in the PCB space. Operating profit grew +665% YoY with OPM up +8.2pp; the operating leverage spread of 621pp is among the most extreme cycle re-rating readings in the KR universe today. Smart Money Earnings ranks TLB #1 (score 0.854) and it also ranks #1 in Smart Money Quality — institutional and foreign co-buying is firing simultaneously across both money-flow screens. An IR session on May 15 was DART-confirmed. Today’s Morgan Stanley ABF/substrate supply-upgrade narrative may provide a sector tailwind if TLB’s product mix directly overlaps. Next checks: confirm product exposure to high-performance ABF substrates; two insider trading plan filings logged on DART May 15 are worth monitoring for directional signals.

SK Hynix (000660.KS) — flag. Fundamentals remain strong (ROE 35.6%, OPM +13.1pp, active consensus upgrades), but net foreign and qualified institutional selling of ₩5.35tn over five sessions is being absorbed by retail — the inverse of the accumulation pattern this framework is built to catch. The meta screener scores it 65.8, reflecting this divergence. Watch for flow stabilization before treating it as an actionable re-rating candidate.

Screener data as of 2026-05-22 16:24 KST. This is public market analysis for informational purposes only. All screened names are research candidates requiring independent due diligence.

Smart Money Is Returning to KOSDAQ: Prices Still Look Weak, but Flows Are Moving First

Fri, 22 May 2026 09:55:00 +0900

📚 Related Series Pearl Abyss Flow Bottom Test / Pearl Abyss Hub / May 17 Market Snapshot / Complete KOSDAQ Guide / National Growth Fund and KOSDAQ Policy Capital

The past two months were a decisive KOSPI large-cap win and an extreme KOSDAQ relative slump. On a market-cap-weighted proxy, KOSPI gained +49.8% while KOSDAQ gained +11.4%, leaving KOSDAQ -38.4 percentage points behind KOSPI. But flows are starting to move differently from prices. Over the past five sessions, KOSDAQ recorded +KRW 707.9bn of foreign-plus-institutional buying and +KRW 410.4bn of program buying. KOSPI, over the same period, recorded -KRW 14.058tn of foreign-plus-institutional selling and -KRW 8.157tn of program selling. Price says KOSPI won. Fresh money is starting to look back at KOSDAQ.

Key Takeaways

Two-month price action is a KOSPI victory. On a market-cap-weighted proxy, KOSPI returned +49.8% versus KOSDAQ +11.4%. Over the past 20 sessions, the gap was also extreme: KOSPI +27.0%, KOSDAQ -0.1%.
Equal-weight returns tell a different story. On a two-month equal-weight basis, KOSPI returned +14.5% and KOSDAQ +17.4%. The breadth of the overall KOSDAQ universe has not collapsed. The market was pulled upward by mega-cap AI infrastructure names in KOSPI.
The KOSDAQ/KOSPI relative ratio is near a two-month low. The current relative ratio is 74.1, down -29.5% from the two-month high of 105.1. The 40-session standard score is -1.88σ, which is already an oversold zone in relative-strength terms.
Flows are improving first in KOSDAQ. Over the past five sessions, KOSDAQ foreigners plus institutions bought +KRW 707.9bn, while program flows added +KRW 410.4bn. KOSPI, by contrast, continued to see foreign-plus-institutional and program selling.
The institution mix is different. KOSPI institutional buying is mostly financial-investment flow, which can include index, derivatives, and arbitrage activity. KOSDAQ has smaller but visible investment-trust buying of +KRW 124.9bn.
Pearl Abyss ranks 7th among the top 100 KOSDAQ names by drawdown. It fell -41.3% from its two-month high of KRW 77,400 to KRW 45,450, roughly 19 percentage points deeper than the top-100 median drawdown of -22.0%.
Pearl Abyss looks closer to an investable rebound candidate than a simple oversold bounce. It has 2026E OP of KRW 476.9bn, 2026E PER of 8.0x, 2026E ROE of 36.9%, five-session foreign-plus-institutional buying of +KRW 22.08bn, and individual selling of -KRW 22.33bn.

The conclusion is simple. KOSDAQ has not yet been confirmed as the new market leader. But flows are starting to turn before prices confirm it. Within that setup, Pearl Abyss is one of the most interesting oversold large-cap candidates.

1. Data Basis and Caveats

The numbers in this post are not official KRX index values. They are market-cap-weighted and equal-weighted proxy series built from individual-stock data in the Research OS local database.

Item	Basis
Data	Research OS local DB
As of	2026-05-21
Analysis period	2026-03-21 to 2026-05-21
Trading sessions	42 sessions
Index method	Individual-stock-based market-cap-weighted proxy
Caveat	Not official KRX index data

The purpose is not to replicate the official index perfectly. The purpose is to find the point where relative price strength and flow direction diverge.

2. Price Says KOSPI Large Caps Won

Looking only at price over the past two months, the conclusion is clear: KOSPI large caps won decisively.

Category	KOSPI	KOSDAQ	Interpretation
2M market-cap-weighted return	+49.8%	+11.4%	KOSPI wins decisively
2M equal-weight return	+14.5%	+17.4%	Breadth is not bad for KOSDAQ
Recent 20D return	+27.0%	-0.1%	KOSDAQ was fully sidelined in the past month
Recent 5D return	+4.8%	-0.5%	KOSPI still leads short term
2M average advancer ratio	43.1%	41.5%	Breadth is weak in both markets

The key is the gap between market-cap-weighted and equal-weighted returns. On a market-cap-weighted basis, KOSPI crushes KOSDAQ. But on an equal-weight basis, KOSPI is +14.5% and KOSDAQ is +17.4%.

That means the entire KOSPI universe was not uniformly strong. It means mega-cap AI infrastructure names such as Samsung Electronics, SK Hynix, and Samsung Electro-Mechanics pulled the KOSPI index upward.

By contrast, KOSDAQ looked weak at the index level, but the broader stock universe did not completely break down. The essence of KOSDAQ weakness was not that every stock was worse. It was that market-cap leadership and flow leadership were captured by KOSPI mega caps.

3. KOSDAQ/KOSPI Relative Strength Is Oversold

If we set the KOSDAQ/KOSPI relative ratio to a base of 100, the current position becomes clearer.

Item	Value
Relative ratio at start of 2M period	99.7
2M high	105.1, 2026-03-31
Current	74.1
Versus high	-29.5%
2M change	-25.6%
Versus 20D average	-12.5%
40-session standard score	-1.88σ

KOSDAQ is now at its weakest relative level versus KOSPI in the past two months. A 40-session standard score of -1.88σ is statistically a fairly deep relative-weakness zone.

But that does not automatically mean “buy KOSDAQ now.” Relative-strength oversold conditions show rebound potential, not rebound confirmation.

The current setup is more accurately described this way:

Price: KOSDAQ at a relative low
Flows: early improvement signs
Trend: not yet confirmed
Conclusion: rebound preparation is possible, but price confirmation is still needed

KOSDAQ has entered a zone that looks cheap in relative terms. It is still too early to say the market has started buying it aggressively.

4. Flows Are Looking at KOSDAQ Before Prices Confirm It

The full two-month flow picture is not clean in either market.

Unit: KRW 100mn

Market	Individuals 2M	Foreigners 2M	Institutions 2M	Foreigners+Institutions 2M	Program 2M
KOSPI	+252,628	-555,191	+177,042	-378,149	-293,401
KOSDAQ	+4,765	+19,909	-36,542	-16,632	+6,300

Over two months, foreigners plus institutions were negative in both markets. But the quality is different.

In KOSPI, foreigners and program flows sold heavily, while individuals and financial-investment accounts absorbed the supply. The index was strong, but the quality of the flow was not particularly clean.

In KOSDAQ, institutions sold, but foreigners and program flows were net buyers. Even though KOSDAQ prices were weak, foreigners and program trading did not fully abandon the market.

The more important signal is the recent 20-session and five-session change.

Market	Foreigners+Institutions 20D	Program 20D	Foreigners+Institutions 5D	Program 5D
KOSPI	-323,604	-219,789	-140,578	-81,572
KOSDAQ	+876	+5,182	+7,079	+4,104

This is the core of the post.

Prices are overwhelmingly stronger in KOSPI, but recent flow improvement is better in KOSDAQ.

Over the past five sessions, KOSDAQ recorded +KRW 707.9bn of foreign-plus-institutional buying and +KRW 410.4bn of program buying. KOSPI continued to see foreign-plus-institutional and program selling.

Price is the result of the past. Flow is a hint about the next move. KOSDAQ is weak in price terms, but in flow terms it is starting to look like a market testing a bottom.

5. Institution Breakdown: KOSPI Is Financial Investment, KOSDAQ Shows Investment Trusts

The internal institutional breakdown sharpens the contrast.

Market	Financial Investment 5D	Insurance 5D	Investment Trusts 5D	Private Funds 5D	Pension/National 5D
KOSPI	+45,340	-623	+1,143	+3,299	-2,409
KOSDAQ	+177	+4	+1,249	+205	-690

Most KOSPI institutional buying came from financial-investment accounts. That flow can include index, derivatives, and arbitrage activity. It is risky to read it as long-duration directional capital.

KOSDAQ is smaller in absolute size, but it shows investment-trust buying. That matters when looking for rebound candidates among oversold growth stocks. Investment trusts can be the first capital to touch KOSDAQ mid- and large-cap growth baskets again.

Pensions and insurance accounts have not yet joined. So this is not a confirmed flow turn. It is an early turn. But the interpretation that KOSDAQ has been completely abandoned is wrong.

6. Pearl Abyss: 7th-Largest Drawdown Among Top 100 KOSDAQ Names

Now narrow the lens to Pearl Abyss. Among the top 100 KOSDAQ names, Pearl Abyss ranks 7th by drawdown from its two-month high.

Item	Basis
Data	Research OS local DB
Price date	2026-05-21
Analysis period	2026-03-21 to 2026-05-21
High price basis	Intraday high over the two-month window
KOSDAQ top 100	Market-cap proxy based on foreign-held shares and foreign ownership ratio
Coverage	1,688 KOSDAQ names; drawdowns calculated for top 100

Pearl Abyss’s two-month high was KRW 77,400 on April 1, 2026. Its closing price on May 21, 2026 was KRW 45,450.

Drawdown from high = 45,450 / 77,400 - 1
= -41.3%

The drawdown distribution for the top 100 KOSDAQ names is as follows.

Category	Drawdown from high
Average	-22.3%
Median	-22.0%
Bottom quartile threshold	-29.5%
Bottom decile threshold	-34.3%
Down more than -20%	57 names
Down more than -30%	23 names
Down more than -40%	7 names

At -41.3%, Pearl Abyss sits exactly inside the seven-name group down more than -40%. This is not merely “somewhat oversold.” It is one of the deepest large-cap drawdowns in KOSDAQ.

7. Comparing the Largest Drawdowns

The top 10 drawdowns among the top 100 KOSDAQ names are:

Rank	Name	Market-cap rank	2M high	Current	Drawdown
1	SCD Pharm	7	1,233,000	355,000	-71.2%
2	Orum Therapeutics	82	124,800	65,500	-47.5%
3	Rznomics	63	245,250	130,000	-47.0%
4	Seers Technology	88	63,200	34,300	-45.7%
5	Woori Technology	39	29,500	16,700	-43.4%
6	ABL Bio	11	189,400	109,900	-42.0%
7	Pearl Abyss	38	77,400	45,450	-41.3%
8	LigaChem Biosciences	15	224,000	139,500	-37.7%
9	Sampyo Cement	84	19,690	12,780	-35.1%
10	Curiox Biosystems	81	126,800	82,500	-34.9%

Pearl Abyss has two distinguishing features in this table.

First, it fell alongside biotech-event names, but it is a different animal. SCD Pharm, ABL Bio, LigaChem, Peptro, and HLB are largely clinical, licensing, or regulatory event options. Even when flows improve, valuation is difficult to anchor on earnings.

Second, Pearl Abyss can be discussed with earnings numbers. It has 2026E OP of KRW 476.9bn, 2026E PER of 8.0x, and 2026E ROE of 36.9%. The debate is not “is this an option with no earnings?” The debate is “are the earnings estimates right?”

That is why Pearl Abyss is different within the oversold group. It is not a biotech-event rebound candidate. It can become an investable rebound candidate if earnings and flows confirm together.

8. Pearl Abyss Among the Top 20 Oversold Names

Looking again at the key oversold names through earnings and flows makes Pearl Abyss’s position clearer.

Rank	Name	From high	26E OP	26E PER	26E ROE	5D Foreigners+Institutions	20D Foreigners+Institutions	5D Individuals	5D Program
1	SCD Pharm	-71.2%	450	[Unconfirmed]	[Unconfirmed]	+228.8	+281.4	-42.0	[Unconfirmed]
4	Seers Technology	-45.7%	607	22.6x	86.1%	+114.5	+22.2	-121.7	+172.3
5	Woori Technology	-43.4%	[Unconfirmed]	[Unconfirmed]	[Unconfirmed]	+368.5	+9.6	+6.1	+190.1
6	ABL Bio	-42.0%	-486	Loss-making	-26.4%	+564.0	-1,026.5	-556.5	+779.1
7	Pearl Abyss	-41.3%	4,769	8.0x	36.9%	+220.8	-73.8	-223.3	+235.6
8	LigaChem Biosciences	-37.7%	268	116.1x	8.1%	-860.4	-815.0	+821.1	-149.4
13	Peptro	-33.2%	[Unconfirmed]	[Unconfirmed]	[Unconfirmed]	-263.9	-958.8	-70.3	[Unconfirmed]
15	Oscotec	-32.7%	340	46.5x	16.6%	-76.4	-237.1	+80.7	+7.5
19	HLB	-31.7%	[Unconfirmed]	[Unconfirmed]	[Unconfirmed]	-123.1	-664.5	-10.5	-18.1

Pearl Abyss has the best balance in this table.

The drawdown is deep enough.
Valuation pressure is low at 8.0x 2026E PER.
It has an earnings base with 2026E OP of KRW 476.9bn.
The five-session flow shows a handoff structure: foreigners plus institutions +KRW 22.08bn, individuals -KRW 22.33bn.
But 20-session foreign-plus-institutional flow is still -KRW 7.38bn, so this is not yet confirmed accumulation.

Pearl Abyss is therefore not a stock where the reversal is already complete. It is a stock where we need to verify whether the early reversal is real.

9. Pearl Abyss Detailed Flow

Pearl Abyss’s recent five-session flow looks different from the prior two weeks.

Investor group	5D net buying
Individuals	-KRW 22.33bn
Foreigners	+KRW 20.77bn
Institutions total	+KRW 1.32bn
Financial investment	-KRW 1.61bn
Insurance	+KRW 0.01bn
Investment trusts	+KRW 0.30bn
Private funds	+KRW 2.62bn
Pension/National	+KRW 0.01bn
Other corporations	-KRW 0.04bn
Other foreigners	+KRW 0.29bn

The good part is clear. Individuals sold, while foreigners and private funds absorbed the supply. That is exactly the kind of structure investors want to see first in an oversold bottom test.

But it is not yet a complete turn. Over 20 sessions, foreigners plus institutions are still -KRW 7.38bn. The past five sessions improved, but cumulative flow has not fully turned.

The wording needs to be precise.

The flow turn is not confirmed yet.
But early signs of a flow turn have appeared.

If the May 18 note framed Pearl Abyss as a “flow bottom test,” the May 21 data updates that judgment by one step. The key now is whether the bottom test continues without failing.

10. Comparative Judgment

Pearl Abyss vs Oversold Biotech Names

Pearl Abyss is different from SCD Pharm, ABL Bio, LigaChem, Peptro, and HLB.

Most oversold biotech names are clinical, licensing, or regulatory event options. Even when flows are strong, valuation is hard to anchor on earnings. ABL Bio has strong five-session flow, but 2026E operating profit is -KRW 48.6bn. LigaChem trades at 116.1x 2026E PER and has -KRW 86.04bn of foreign-plus-institutional selling over the past five sessions.

Pearl Abyss is different. It has 2026E OP of KRW 476.9bn, 2026E PER of 8.0x, and 2026E ROE of 36.9%. So the core question is not “can it succeed?” but “how much of the 2026 earnings base can be sustained into 2027?"

Pearl Abyss vs Seers Technology

Seers Technology is the most interesting comparison. It has 2026E OP of KRW 60.7bn, 2026E PER of 22.6x, 2026E ROE of 86.1%, and five-session foreign-plus-institutional buying of +KRW 11.45bn. Its -45.7% drawdown also gives it rebound torque.

But institutions are still weak overall, and foreigners are absorbing most of the supply. Pearl Abyss has +KRW 20.77bn of foreign buying and +KRW 2.62bn from private funds. Its valuation is much lower.

On a risk-adjusted basis, Pearl Abyss > Seers Technology. But both still require additional confirmation of flow quality.

Pearl Abyss vs Woori Technology and Sungho Electronics

Woori Technology and Sungho Electronics have strong flows, but weaker earnings and consensus anchors. Woori Technology has +KRW 36.85bn of five-session foreign-plus-institutional buying, but earnings-based comparison is blocked. Sungho Electronics may have strong short-term institutional buying, but if foreign selling and individual buying are mixed in, investors need to watch for distribution.

Pearl Abyss may have weaker flow intensity than some thematic names. But it is more investable because the drawdown can be explained with earnings numbers and valuation.

11. Practical Judgment

The current market can be summarized as follows.

Item	Judgment
Price	KOSPI large-cap leadership, KOSDAQ at relative lows
Flow	KOSPI price is strong, but flow quality is weak
KOSDAQ	Price is weak, but foreign, program, and investment-trust flows are improving
Meaning	KOSDAQ rebound setup is forming, but price confirmation is needed
Pearl Abyss	Top investable oversold candidate, but flow turn is not confirmed

There are three entry triggers for KOSDAQ.

KOSDAQ/KOSPI relative ratio rebounds for three consecutive sessions.
KOSDAQ foreign-plus-program buying continues.
Oversold KOSDAQ large caps repeatedly show individual selling absorbed by foreigners and institutions.

Pearl Abyss has four separate triggers.

Defense near KRW 45,000.
Continued foreign and private-fund buying.
Recovery above KRW 47,200.
Recovery in 2Q guidance or long-tail/DLC expectations after IR/NDR.

The invalidation conditions are also clear.

Condition	Interpretation
Break below the KRW 44,000 zone	Recent flow turn likely failed
Foreign five-session buying flips back to selling	Low-price re-entry was a one-off defense
KOSDAQ/KOSPI relative ratio fails to rebound	Market-wide rotation is delayed
2Q guidance or long-tail thesis weakens	Fundamental thesis is discounted again

12. Final Line

The past two months were a decisive KOSPI large-cap win and an extreme KOSDAQ relative slump. On a market-cap-weighted proxy, KOSPI gained +49.8% and KOSDAQ +11.4%, leaving KOSDAQ -38.4 percentage points behind. But on an equal-weight basis, KOSPI was +14.5% and KOSDAQ +17.4%. This was not a market where all of KOSPI was strong. It was a market where mega-cap AI infrastructure names pulled the KOSPI index higher.

The KOSDAQ/KOSPI relative ratio is now 74.1, down -29.5% from its two-month high, with a 40-session standard score of -1.88σ. In relative-strength terms, it is already deeply oversold. The more important point is flow. Over the past five sessions, KOSDAQ foreigners plus institutions bought +KRW 707.9bn and program flows added +KRW 410.4bn. KOSPI saw -KRW 14.058tn of foreign-plus-institutional selling and -KRW 8.157tn of program selling over the same period. Price says KOSPI won, but flows are starting to look back at KOSDAQ.

Pearl Abyss is the representative watch name in this setup. It ranks 7th among the top 100 KOSDAQ names by drawdown, falling -41.3% from KRW 77,400 to KRW 45,450. That is roughly 19 percentage points deeper than the top-100 median drawdown of -22.0%. But it has 2026E OP of KRW 476.9bn, PER of 8.0x, and ROE of 36.9%. Recent five-session flow shows individuals -KRW 22.33bn, foreigners +KRW 20.77bn, and institutions +KRW 1.32bn. That is why this is not just a simple oversold bounce. It can be treated as an investable rebound candidate.

Still, the reversal is not confirmed. Twenty-session foreign-plus-institutional flow is still -KRW 7.38bn, and price has not yet confirmed a KOSDAQ relative-strength rebound. This is a phase where flows are starting to return, but prices have not confirmed yet. Pearl Abyss is less a stop-loss candidate than an oversold large-cap where rebound torque can rise if KRW 45,000 holds and foreign/private-fund buying continues.

This article is for research and commentary purposes only and is not investment advice. Market returns, relative ratios, flows, drawdown ranks, and consensus numbers are based on the Research OS local database. The data date is May 21, 2026. The figures are individual-stock-based proxy series, not official KRX index values. The KOSDAQ top 100 is constructed from a market-cap proxy using foreign-held shares and foreign ownership ratio, and may differ from actual market-cap rankings. Pearl Abyss 2026E OP of KRW 476.9bn, 2026E PER of 8.0x, and 2026E ROE of 36.9% are local DB consensus figures and may differ from actual results. [Unconfirmed] marks items that could not be reliably verified on the same basis in the comparison group. The analysis may be wrong. Data as of May 21, 2026 KST.

Understanding MLCCs and Silicon Capacitors — What Samsung Electro-Mechanics' ₩1.5 Trillion Contract Reveals About the AI Package Power Bottleneck

Fri, 22 May 2026 00:00:00 +0000

📚 Related Series Samsung Electro-Mechanics Silicon Capacitor ₩1.5 Trillion Contract / Samsung Electro-Mechanics Hybrid Challenger / Samsung Electro-Mechanics MLCC & FC-BGA Business Deep Dive / Semiconductor Value Chain Hub / AI Substrate & PCB Hub

An MLCC is one type of capacitor — an ultra-compact ceramic component found in virtually every electronic device, functioning as a power stabilizer. A silicon capacitor is placed closer to the semiconductor chip than an MLCC, in some cases inside the package itself, to suppress instantaneous power fluctuations in AI GPUs and HBM. From an investment standpoint, what matters is the signal that “passive components” are graduating from simple commodity parts to AI packaging and power-integrity bottleneck components.

Key Summary

A capacitor stores electrical charge briefly and releases it on demand. For non-engineers, think of it as a water tank, shock absorber, or noise filter for an electrical circuit. An MLCC is the highest-volume ceramic multilayer version of that capacitor. Samsung Electro-Mechanics describes MLCCs as playing the role of a “dam” — holding electricity and releasing it in controlled amounts.¹

A silicon capacitor is also a type of capacitor. Unlike an MLCC, which stacks hundreds of ceramic layers, a silicon capacitor forms dielectric and electrode layers on a silicon wafer. According to Samsung Electro-Mechanics’ product documentation, it can be thinned to under 100 micrometers, embedded inside a package, and benefits from low parasitic inductance that is favorable for power stabilization.²

The essential point is not substitution but a shift in location. MLCCs are mounted primarily on the PCB surface or around the chip. Silicon capacitors can go inside the package, beneath the substrate, or directly adjacent to the die. Because AI GPUs and HBM draw sudden, large current spikes, proximity matters as much as raw capacitance — “how quickly can power be delivered” becomes as important as “how much can be stored.”

Samsung Electro-Mechanics’ May 20, 2026 silicon capacitor supply contract worth approximately ₩1.5 trillion is the first large-scale evidence of this structural shift. The contract runs from January 1, 2027 through December 31, 2028. The company stated that the product is embedded inside high-performance semiconductor packages — including AI server GPUs and HBM — to enhance power-supply stability.³

Investment judgment, however, must remain cold. This week (May 18–22), the three core MLCC names averaged +35.6%, handily outpacing the five core AI substrate names at +14.0%. Samsung Electro-Mechanics gained +32.7% on the week; Samwha Capacitor surged +56.4%. Theme momentum is unmistakable, but the efficiency of fresh capital deployment has diminished sharply.

The bottom line: the significance of silicon capacitors is not MLCC replacement — it is the expansion of the passive component battlefield from the PCB surface into the interior of AI semiconductor packages. Premium product lines at Samsung Electro-Mechanics and Murata may be re-classified as AI infrastructure supply chain assets. But a good industry shift and a good entry price remain entirely separate questions.

1. The Relationship Between Capacitors and MLCCs

A capacitor stores electrical charge and releases it rapidly when needed. When a semiconductor chip demands a sudden surge of current, a nearby capacitor supplies it, preventing voltage from collapsing. Conversely, when voltage spikes, the capacitor absorbs some of it, stabilizing the circuit. This is why capacitors serve the roles of energy storage, voltage smoothing, noise rejection, and power stabilization in electronic circuits.

An MLCC — Multi-Layer Ceramic Capacitor — is an ultra-compact capacitor constructed by stacking hundreds of thin ceramic dielectric layers alternating with metal electrode layers. Samsung Electro-Mechanics references production capability for high-capacitance MLCCs with up to 600 layers, noting that MLCC importance grows alongside the expansion of 5G, consumer electronics, autonomous vehicles, and the Internet of Things.¹

The hierarchy is straightforward:

Capacitor = the entire category of electrical stabilization components
MLCC = a mass-market, multilayer ceramic variant within that category
Silicon Capacitor = a high-performance, silicon-wafer-based variant optimized for ultra-close proximity

Both MLCCs and silicon capacitors are capacitors. They differ in where they are deployed and what performance demands they are designed to meet.

2. Capacitor vs. MLCC vs. Silicon Capacitor

	General Capacitor	MLCC	Silicon Capacitor
Category	Broad family	One type of capacitor	One type of capacitor
Simple analogy	Circuit water tank	Ultra-compact ceramic water tank	Ultra-proximity water tank embedded next to or inside the chip
Key material	Ceramic, aluminum electrolyte, tantalum, film, silicon, etc.	Ceramic dielectric + internal metal electrodes	Silicon wafer-based structure
Primary role	Energy storage, voltage smoothing, noise rejection	High-frequency noise filtering, power stabilization around chips, miniaturization	Suppression of instantaneous power transients in AI GPUs, HBM, and high-performance chips
Typical location	Power stages, boards, modules, motors, inverters	Primarily on the PCB, around chips, around modules	Inside packages, within substrates, immediately adjacent to the die
Advantages	Wide selection across applications	Small, low-cost, mass-producible	Extremely thin; can be placed close to the die; favorable for power stabilization
Limitations	Large performance variance by type	Limited in ultra-high-speed, ultra-proximate package power stabilization	High manufacturing complexity and cost; current addressable market concentrated in high-performance semiconductors
Primary growth vectors	Automotive, power grid, industrial, AI server	Automotive, AI server, smartphone, 5G, IoT	AI GPU, HBM, high-performance computing, high-performance mobile chips

Even a cursory look at TDK’s capacitor product portfolio illustrates the breadth of the category: small MLCCs, high-voltage ceramic capacitors, film capacitors, aluminum electrolytic capacitors, and power capacitors all sit under the same “capacitor” umbrella.⁴ “Capacitors are good” is too broad a statement. In investment terms, what matters is which capacitor, in which location, for which application.

3. Where They Are Used

3.1 General Capacitors

General capacitors appear in virtually every electrical circuit, from consumer electronics to power infrastructure.

Application	Primary Capacitor Type	Role
Smartphone, PC	MLCC, tantalum, polymer	Power stabilization around application processors, memory, and PMIC
Server, AI server	MLCC, polymer, silicon capacitor	Suppression of power transients in CPUs, GPUs, and HBM
Automotive electronics	MLCC, film, aluminum electrolytic, tantalum	Stabilization of ECUs, ADAS, inverters, chargers, and BMS
EV power systems	Film, high-voltage MLCC, aluminum electrolytic	High-voltage DC link, charging, power conversion
Industrial, power grid	Film, aluminum electrolytic	Power factor correction, motor drives, energy storage
Telecom, high frequency	Ceramic, silicon capacitor	High-frequency filtering, impedance matching

3.2 MLCC

The core application for MLCCs is power stabilization around chips. They appear in smartphones, automobiles, servers, network equipment, home appliances, and industrial machinery. As devices shrink, semiconductors switch faster, and the number of sensors and communication modules grows, the number of points requiring voltage stabilization increases proportionally.

MLCC importance in AI servers is also rising. TDK notes that as AI and cloud demand drives higher integration density and performance in data center servers, power density per rack and per server increases — making the performance and footprint of passive components a genuine design constraint.⁴ This is why the MLCC thesis has expanded beyond a simple smartphone-cycle recovery story into AI server power stabilization.

3.3 Silicon Capacitors

The core application for silicon capacitors is power stabilization inside packages. Samsung Electro-Mechanics describes silicon capacitors as ultra-compact, high-performance devices built on silicon wafers, embedded inside high-performance semiconductor packages — including AI server GPUs and HBM — to enhance power-supply stability.³

The company’s product documentation reinforces this: silicon capacitors form dielectric and internal electrodes on silicon, can be thinned to under 100 micrometers through wafer processing, and are well-suited for embedding inside packages. Low parasitic inductance is the key advantage for power stabilization.²

To demystify the terminology:

Parasitic inductance = an interference factor that delays the instantaneous response of current flow
Parasitic resistance = an interference factor that causes energy loss as current passes through
Power integrity = the ability to deliver stable, ripple-free voltage and current to a chip exactly when needed

AI semiconductors draw power in sudden, large spikes. A capacitor far from the die responds too slowly. The solution is to move the capacitor immediately adjacent to the chip — or inside the package itself. The investment case for silicon capacitors is not “higher capacitance” but faster response from a closer location.

4. Key Suppliers and Customer Base

4.1 MLCC Key Suppliers

Region	Key Companies	Strengths
Japan	Murata, TDK, Taiyo Yuden	Ultra-small, high-reliability, automotive-grade MLCC leadership
Korea	Samsung Electro-Mechanics	High-value MLCC for IT, automotive, and AI server; expanding into silicon capacitors
Taiwan / Greater China	Yageo, Walsin	Commodity, industrial, and mid-to-low price passive component portfolio
US / Japan-affiliated	KYOCERA AVX, KEMET/Yageo, Vishay	Specialty capacitors, automotive, industrial, and high-frequency product lines

The MLCC customer base is extremely broad: smartphone OEMs, PC and server manufacturers, automotive OEMs, Tier-1 suppliers, telecom equipment makers, and semiconductor module companies. For MLCC investing, demand mix by end-market and the share of high-value products matter more than exposure to any single customer.

4.2 Silicon Capacitor Key Suppliers

	Key Companies	Positioning
Murata	Leader in silicon capacitors and integrated passive devices	High-performance mobile, telecom, and high-performance computing silicon capacitors
Samsung Electro-Mechanics	Large-scale new entrant	Secured approximately ₩1.5 trillion supply contract for 2027–2028 delivery
KYOCERA AVX et al.	Specialty MOS capacitors	High-frequency, microwave, and specialty industrial product lines
Advanced packaging ecosystem	Captive or co-design options	Linked to interposer and in-package power integrity technology

The customer dimension is critical. MLCCs have a broad, substitutable customer base. Silicon capacitors, by contrast, must be designed into a package from the outset. They require customer qualification, integration into package architecture, and validation of power integrity. Once designed in, they are difficult to replace — the switching cost is high.

Samsung Electro-Mechanics’ contract counterparty is officially disclosed only as a “large global corporation.” It would therefore be incorrect to identify the customer as NVIDIA, AMD, Broadcom, Google, Meta, Microsoft, Amazon, or any other specific company. That remains unconfirmed.

5. What Silicon Capacitors Actually Signify

5.1 Technical Significance — The Power Bottleneck Migrates Inside the Package

As AI GPUs and HBM scale in compute intensity and data movement, instantaneous current demand has grown substantially. The problem is that power arriving from a distance is slowed by parasitic elements in the power traces, substrate, and package — degrading the transient response. The solution is to move capacitors closer to the die.

Conventional MLCCs sit primarily on the PCB or around the package perimeter. Silicon capacitors can go inside the package or immediately adjacent to the die. With capacitors of equivalent capacitance, placement determines performance.

5.2 Business Significance — Passive Components Re-rated as AI Packaging Components

MLCCs have traditionally been priced as cyclical components. When smartphone and PC demand weakens, inventory adjustments follow, and commodity products face price competition. Silicon capacitors, in contrast, are high-complexity components embedded inside AI GPU and HBM packages. They require customer qualification, integration into package design, and power-integrity validation.

For Samsung Electro-Mechanics, this contract carries strategic significance beyond the revenue figure. The company officially stated that it leveraged its accumulated precision process capabilities from the MLCC and package substrate businesses to enter the AI semiconductor core supply chain.³

5.3 Investment Significance — Possible Re-classification as “AI Server Component” Plays

The market has historically viewed Samsung Electro-Mechanics as an MLCC, camera module, and substrate company. If silicon capacitor revenue scales into mass-production levels, that classification partially changes.

Prior Market Perception	Possible Post-Silicon-Capacitor Perception
Smartphone component play	AI server component play
MLCC cycle stock	High-value power integrity component stock
Automotive MLCC growth story	AI GPU / HBM packaging supply chain entrant
Exposure to commodity passives	Exposure to high-complexity in-package components

Overclaiming risks remain, however. Silicon capacitors will not replace the entire MLCC market. Growth will come first in AI package interiors, high-performance computing, high-performance mobile chips, and select telecom and industrial applications where extreme performance is required. The vast majority of power stabilization demand in mainstream smartphones, automobiles, and home appliances will likely remain the domain of MLCCs for the foreseeable future.

6. This Week’s Tape — MLCCs Outran AI Substrates

This week (May 18–22, 2026), the market aggressively re-rated the MLCC theme. The three core MLCC names averaged +35.6%; the five core AI substrate names averaged +14.0%.

6.1 Core MLCC Three-Name Basket

Name	Weekly Return	20D	60D	2026E PER	Foreign	Institution	Foreign + Inst.	Program
Samwha Capacitor	+56.4%	+56.0%	+68.0%	40.8x	-₩5.88B	+₩16.09B	+₩10.21B	N/A
Samsung Electro-Mechanics	+32.7%	+65.0%	+200.5%	73.2x	-₩292.90B	+₩111.79B	-₩181.11B	+₩170.96B
Amotech	+17.6%	-18.4%	+55.2%	18.9x	+₩4.91B	-₩0.87B	+₩4.05B	+₩5.10B

Samsung Electro-Mechanics posted three consecutive sessions of strength: +7.5% on May 20, +13.5% on May 21, and +11.3% on May 22. Samwha Capacitor was more extreme: +23.0% on May 20, +29.9% on May 22. This is less a pure earnings-driven move than a re-rating of AI power-integrity components that originated with Samsung Electro-Mechanics and radiated outward into the MLCC ecosystem.

6.2 Core AI Substrate Five-Name Basket

Name	Weekly Return	20D	60D	2026E PER	Foreign	Institution	Foreign + Inst.	Program
심텍 (심텍)	+31.4%	+52.2%	+146.8%	38.7x	-₩24.58B	+₩61.81B	+₩37.22B	-₩5.35B
TLB	+18.0%	+44.4%	+89.2%	25.3x	-₩7.80B	+₩15.30B	+₩7.50B	-₩3.98B
Daeduck Electronics	+11.4%	+49.5%	+132.8%	38.9x	-₩2.93B	+₩29.05B	+₩26.12B	+₩5.28B
Korea Circuit	+10.7%	+5.6%	+66.7%	25.5x	-₩2.71B	-₩0.58B	-₩3.29B	+₩0.44B
ISU Petasys	-1.5%	-21.9%	+12.7%	35.6x	-₩119.17B	-₩0.74B	-₩119.90B	-₩125.28B

The AI substrate basket was not a uniformly rising tide. Simtech, TLB, and Daeduck Electronics were strong, but ISU Petasys finished the week at -1.5% amid heavy foreign and program selling. Within AI substrates, the market this week appeared to favor SoCAMM, memory module, and FC-CSP exposure over high-layer-count network boards.

6.3 Price Leadership to MLCCs; Flow Quality Favors Select AI Substrate Names

Basket	Foreign	Institution	Foreign + Inst.	Retail	Program
Core MLCC (3 names)	-₩293.86B	+₩127.01B	-₩166.85B	+₩159.40B	+₩176.05B
Core AI Substrate (5 names)	-₩157.19B	+₩104.84B	-₩52.35B	+₩52.05B	-₩128.89B
AI Substrate ex-ISU Petasys	-₩38.02B	+₩105.57B	+₩67.55B	-₩65.47B	-₩3.60B

On price, MLCCs dominated. On flow quality, the AI substrate basket excluding ISU Petasys looks cleaner. Simtech, TLB, and Daeduck show clear institutional accumulation with retail sellers or neutral retail — a more constructive configuration. In MLCC names, the ₩292.9 billion of foreign selling in Samsung Electro-Mechanics alone is too large to ignore. The stock price was strong, but this looks more like institutional, program, and retail-driven re-rating than foreign-led accumulation.

7. Investment Assessment

Name	View
MLCC	Strongest price momentum. Overheating risk and foreign selling overhang present.
AI Substrates	Lower absolute return, but institutional flow and earnings linkage are more stable.
Samsung Electro-Mechanics	Cross-category leader across MLCC, FC-BGA, and silicon capacitors. Holding is defensible; chasing is not.
Samwha Capacitor	Highest theme expansion intensity. +56% in one week makes new entry extremely difficult.
Daeduck Electronics	Core AI substrate holding. Valuation discount relative to Samsung Electro-Mechanics is meaningful.
Simtech / TLB	Led this week’s AI substrate expansion. Next week, monitor for a post-heat consolidation before re-entry.
ISU Petasys	Relative laggard within AI substrates. Lower priority until foreign flow stabilizes.

This week’s market was one in which attention expanded from AI substrates into MLCCs. The AI substrate thesis has not been invalidated. Rather, the market is now disaggregating “the next bottleneck after memory” more finely — extending the re-rating into AI server power integrity components.

In practical terms:

Hold: Samsung Electro-Mechanics, Daeduck Electronics
Do not chase: Samsung Electro-Mechanics, Samwha Capacitor
Monitor on pullback: Simtech, TLB
Await flow recovery: ISU Petasys
Key confirmation needed: Whether MLCC price increases and AI-server revenue lift 2Q–3Q earnings revisions

8. One-Line Takeaway

An MLCC is the electronic industry’s “ultra-compact ceramic water tank.” A silicon capacitor is an “ultra-proximity power stabilizer” embedded inside AI GPU and HBM packages. Silicon capacitors do not replace MLCCs. The battlefield for passive components is expanding from the PCB surface into the interior of AI semiconductor packages.

The real meaning of Samsung Electro-Mechanics’ ₩1.5 trillion contract is not “selling one more type of capacitor.” It is the signal that a passive component maker has entered the bottleneck component supply chain inside AI semiconductor packages. This shift can re-classify premium product lines at Samsung Electro-Mechanics and Murata into AI infrastructure value chain assets.

The market has, however, already responded quickly. The three core MLCC names averaged +35.6% this week; Samsung Electro-Mechanics is up +200.5% over 60 days and trades at 73.2x 2026E PER. The company is excellent and the industry shift is real. Whether today’s price is also a good new entry point is a separate question entirely. The next catalyst to monitor is not price action but 2Q–3Q earnings revisions, disaggregated AI-server revenue disclosure, follow-on silicon capacitor orders, and foreign flow recovery.

This post is for research and commentary purposes only and does not constitute investment advice. Technical descriptions of capacitors, MLCCs, and silicon capacitors are based on official materials from Samsung Electro-Mechanics, TDK, Murata, and other publicly available technical sources. The Samsung Electro-Mechanics silicon capacitor supply contract (approximately ₩1.5 trillion, January 1, 2027 through December 31, 2028) reflects the company’s official announcement. The customer has not been publicly disclosed; no specific GPU, ASIC, or cloud company has been identified as the counterparty. Price, flow, and valuation data for the May 18–22, 2026 period are sourced from the user’s Research OS local database and may differ from official exchange or brokerage data. 2026E PER, foreign/institutional/program flow figures, and basket averages for the core MLCC three-name and core AI substrate five-name groups reflect the analyst’s classification. Analysis may be wrong. Data reference date: May 22, 2026, KST.

KOSPI Surges 8%: Semiconductors Lead Korea's Risk-On Rally

Thu, 21 May 2026 23:30:00 +0900

KOSPI Surges 8% as Three Catalysts Hit at Once

South Korea’s benchmark equity index KOSPI — comprising roughly 800 listed companies across all major sectors — posted one of its sharpest single-session gains of the year on May 21, 2026, rising 8.09% to approximately 7,792 as of mid-afternoon trading. The tech-heavy KOSDAQ climbed 4.80% to 1,106. The rally was real but narrow: this was a semiconductor-driven risk-on session, not a broad market recovery.

Three distinct catalysts converged simultaneously, which is rare enough to take seriously.

What Actually Happened Today

Catalyst 1: Samsung Electronics labor risk defused. Samsung Electronics (005930.KS), South Korea’s largest semiconductor manufacturer and the single heaviest-weighted stock on the KOSPI, had been facing a threatened general strike by its union scheduled for May 21. A tentative labor-management agreement announced this morning neutralized that risk. Foreign investors responded immediately — net buying in Samsung reached approximately ₩1.07 trillion on the day, with institutional buying adding another ₩765 billion. The stock gained 8.51%, closing near ₩299,500.

Catalyst 2: Nvidia earnings beat drove memory pricing expectations higher. Nvidia (NVDA) reported FY2027 Q1 revenues of $81.6 billion, with data center revenue of $75.2 billion and networking revenue up 199% year-over-year. The Vera CPU and SoCAMM demand pipeline in Nvidia’s guidance reinforced a supply shortage thesis for LPDDR and DRAM. Mirae Asset raised its Samsung Electronics price target to ₩480,000, citing Q2 2026 LPDDR5X 16GB prices up 81% quarter-over-quarter, LPDDR4X 8GB up 72% QoQ, with NAND and UFS pricing also strengthening.

Catalyst 3: Korea’s May 1–20 semiconductor export data surprised to the upside. Early export figures for the first twenty days of May confirmed that the momentum visible in Q1 trade data has continued, reinforcing the argument that the memory upcycle is not a one-quarter story.

Why does the convergence of these three matter? Because each individually could have been dismissed as noise. Together, they shifted the market’s interpretation of Samsung from “labor-risk discount + memory uncertainty” to “labor-normalized + memory upcycle confirmed.” That repricing happened in a single session.

Sector Breakdown: Who Benefited, Who Didn’t

Strongest sectors: Electrical/electronics (semiconductors), AI PCB (printed circuit boards for AI applications), and selected power equipment names.

Weakest sectors: Real estate, food/beverage, paper/timber, and parts of biotech. This pattern is consistent with a classic semiconductor-led risk-on session where capital rotates out of defensives and into high-beta tech hardware.

The divergence was sharp enough to matter for sector positioning. Foreign and institutional investors did not spread their buying broadly — they targeted the semiconductor supply chain with notable precision.

AI PCB: The Derivative Trade Getting Attention

Beyond Samsung Electronics itself, the session highlighted the AI PCB and semiconductor materials subsector as a secondary beneficiary of the Nvidia demand signal.

Samsung Electro-Mechanics (009150.KS), South Korea’s leading manufacturer of multilayer ceramic capacitors (MLCCs) and camera modules — and increasingly a supplier of advanced PCB substrates for AI applications — posted the session’s strongest performance among widely-followed names, gaining 13.48% to ₩1,204,000. Foreign investors bought ₩43.4 billion, institutions ₩36.9 billion, with a relative strength rank of 99.1 out of 100. The RSI hit 84.3, placing the stock in technically overbought territory — meaning the question for tomorrow is whether volume and price hold, not whether the move was justified.

Daeduck Electronics (001680.KS), a specialist in high-density interconnect PCBs for memory and AI servers, gained 10.08% to ₩146,400 on foreign and institutional co-buying. Its relative strength rank recovered to 98.1, making it a name to watch if tomorrow’s trading volume sustains the move.

Simtec (222800.KS), another AI PCB and semiconductor packaging name, triggered a breakout signal in quantitative screeners, rising 21.1% on volume 1.9 times the 20-day average. Analysts tracking the AI hardware supply chain will want to see whether the move holds or fades over the next one to three sessions before treating it as a confirmed breakout.

Iotechnics (039030.KS), a laser equipment maker for semiconductor packaging, rose 18.8% on combined foreign and institutional buying, though the stock is trading near its 52-week high — a level that typically requires a pullback consolidation before sustainable continuation.

The Macro Overlay: Risk-On With Conditions

The session opened under a cautious macro regime — KRW/USD was trading near ₩1,504, and US and Japanese long-term bond yields remained elevated. By afternoon, a combination of factors shifted sentiment: progress in Iran nuclear negotiations reduced oil risk (WTI holding below $100), and rate anxiety moderated enough for the semiconductor earnings catalyst to dominate.

The takeaway is that the risk-on session was real but conditional. The dollar-won rate at ₩1,504 remains a headwind for Korean exporters’ earnings translation and for foreign investor returns. US and Japanese long-end rate volatility has not been resolved. The semiconductor rally does not override those structural pressures — it simply outweighed them for one session.

What Didn’t Work: The Relative Strength Test

Pearl Abyss (263750.KS), a South Korean video game developer known for the Black Desert Online franchise, gained only 0.11% on the day despite modest positive flows from foreign and institutional investors. On a day when KOSPI gained over 8%, a 0.11% move represents a sharp deterioration in relative strength. The stock’s RSI sits at 11.5, technically deeply oversold, and it has lost 10.18% over the prior five sessions.

Why mention a stock that barely moved? Because relative performance on strong market days is one of the most informative data points available. If a stock cannot participate in a broad semiconductor-led rally, the market is communicating something about the stock-specific thesis that price-to-book or RSI alone won’t capture. For Pearl Abyss, the key forward questions are: Can the Black Desert franchise demonstrate re-acceleration in China’s Steam rankings? Does the company’s next investor day shift market perception? Those are thesis-level questions, not technical ones.

Stocks Emerging on Quantitative Screens

Several names broke into the top-tier of relative strength rankings on today’s session that weren’t widely discussed heading into the day:

SK Square (402340.KS) — a semiconductor-focused holding company with SK Hynix (000660.KS) exposure — gained 14.6% with a relative strength rank of 98.8, though foreign flow data requires confirmation.
Fadu (440110.KS) — a data center SSD controller developer — rose 4.0% with a relative strength rank of 98.8. Volume and institutional participation need independent verification.
Jeju Semiconductor (080220.KS) — gained 24.3% with a relative strength rank of 98.6 and an RSI of 89.2. The move is too extended to chase on day one.

These names are worth tracking as potential derivative plays on the AI data center infrastructure theme, but all warrant a consolidation period before the single-day move can be treated as a sustainable signal.

Forward Checkpoints

The session established semiconductor hardware as the dominant market theme for now. The questions that will determine whether this extends or fades:

Does Samsung Electronics hold ₩299,500? And does foreign buying persist beyond one session, or was today a re-rating event rather than the start of a sustained inflow?
Can Samsung Electro-Mechanics consolidate above ₩1,204,000 before attempting another leg higher? The RSI at 84.3 suggests near-term digestion is likely.
Does Daeduck Electronics sustain volume and price at ₩146,400 on Day 2? Single-session breakouts in mid-cap names require follow-through confirmation.
What do Marvell Technology (MRVL) and other US AI networking names do tonight? If Nvidia’s guidance translates into a broader AI hardware re-rating in US markets, Korean semiconductor supply chain names will likely extend the move tomorrow.
Does the KRW/USD rate stabilize? At ₩1,504, currency drag remains a concern for foreign investors calculating returns in dollar terms.

Today was a legitimate re-rating session for Korean semiconductor hardware names. It was not, however, a signal that Korea’s bear market regime has conclusively reversed. The rally was concentrated, the macro risks are unchanged, and several domestically-oriented sectors did not participate. The next forty-eight hours of price and volume behavior will be more informative than today’s dramatic headline move.

Korea Quality Re-Rating Watch 2026-05-21: VM, Samyang NCC & Jeju Semi

Thu, 21 May 2026 16:30:00 +0900

Macro Dashboard

Indicator	Level	5-Day Δ	Signal
KOSPI	7,815.6	+322.4 pts (+4.3%)	Bull
KOSDAQ	1,106.0	−23.9 pts (−2.1%)	Weak
USD/KRW	1,507	+13.5 (+0.9%)	Stable
VIX	17.8	−0.67 (−3.6%)	Calm
Brent	$107.2	−$5.0 (−4.4%)	Easing
US 10Y	4.57%	+0.11	Watch
DXY	99.2	+0.3%	Flat

KR Regime: Bull | US Regime: Bull | Stance: Aggressively Expanding

KOSPI breadth above 50-MA sits at 30.3%; above 200-MA at 43.8%, with the internal Discovery signal on Day 19 of a fresh cycle. KOSDAQ lagged KOSPI by 6.4 percentage points over the 5-day window — a clear sign this advance is semiconductor-driven rather than broad-market. USD/KRW at 1,507 and US 10Y at 4.57% remain the two macro overhangs to watch.

Market Wrap

Today was a semiconductor-led risk-on session of uncommon force. KOSPI hit 7,792 intraday (+8.09%), closing near 7,816 — among the sharpest single-session moves of the current cycle. Three catalysts landed simultaneously: Samsung Electronics (005930.KS) reached a tentative labor agreement, eliminating the May 21 general strike threat; NVIDIA reported FY27 Q1 revenue of $81.6B (data center $75.2B, networking +199% YoY), confirming AI hardware demand is accelerating into the supply chain; and Korea’s May 1–20 semiconductor export data confirmed the pricing cycle is turning sharply. LPDDR5X 16GB pricing reportedly jumped +81% QoQ in 2Q26, with LPDDR4X and NAND/UFS also repricing higher.

The flow picture was decisive. Foreign investors pushed ₩1.07 trillion into Samsung Electronics alone; institutional desks added ₩764.9 billion and program buying reached ₩1.71 trillion in the name. Samsung Electro-Mechanics (009150.KS) was the supply-chain standout — up +13.5%, RS rank 99.1 — as analysts revised AI PCB and MLCC assumptions upward in real time. Daeduck Electronics (353200.KQ) (+10.1%) and EO Technics (039030.KQ) (+18.8%) each drew confirmed foreign and institutional co-buying, and both appear in today’s quality screener overlap (covered in Section 3 below).

Simtec (036490.KQ) broke out +21.1% on volume 1.9× average, adding it to the AI PCB/packaging watch list alongside Daeduck. Jeju Semiconductor (080220.KQ) surged +24.3% to RS 98.6 — the purest DRAM price cycle proxy of the session — though RSI reached 89.2, signalling near-term overextension that typically warrants a pause before fresh exposure.

Weak segments were unambiguous: real estate, food and beverage, paper/timber, and select biotech names sat out the move entirely. KOSDAQ’s underperformance versus KOSPI reflects this bifurcation — today the market rewarded operating leverage inside the semiconductor stack, not broad beta.

SK Square (402340.KS) (+14.6%, RS 98.8) attracted attention as a holding company proxy for SK Hynix, though institutional flow confirmation remains pending. The session’s message: Korea is in a semiconductor-led bull phase, but the advance is narrow enough that stock selection still matters more than index-level positioning.

Today’s Quality Re-Rating Candidates

The KR Meta Screener evaluated 109 tickers across five independent frameworks today. The editorial lens: good businesses where institutional money is entering and the market is beginning to reprice the story. Names hitting three or more screeners are the preferred candidates; names hitting five are the lead story.

Top Overlap Candidates

Meta Rank	Ticker	Name	Meta Score	Screener Overlap	Screeners Hit	Key Metrics
1	482630.KQ	Samyang NcChem	101.0	4	Quality · SM Quality · Cycle Rerating · SM Earnings	ROE 14.7%, OP +64.5% YoY, margin +4.3pp
2	089970.KQ	VM	98.5	5	Quality · SM Quality · Cycle Rerating · SM Earnings · PEAD	ROE 14.7%, OP +386.9% YoY, margin +29.3pp
3	080220.KQ	Jeju Semiconductor	80.2	4	Quality · SM Quality · Cycle Rerating · SM Earnings	ROE 17.4%, OP +274.4% YoY, ⚠ 4 DART risk filings
4	095610.KQ	TES	74.0	3	Quality · SM Quality · SM Earnings	ROE 14.5%, OP +50.3% YoY, F+QI 5d +₩28.1B
5	000660.KS	SK Hynix	65.4	2	Quality · Cycle Rerating	ROE 35.6%, OP +101.2% YoY, consensus z-score +1.13
6	039030.KQ	EO Technics	58.9	4	SM Quality · Cycle Rerating · SM Earnings · PEAD	Margin +11.5pp, F+QI 5d +₩57.1B
7	240810.KQ	Wonik IPS	55.8	4	SM Quality · Cycle Rerating · SM Earnings · PEAD	OP +593.6% YoY, margin +6.7pp
8	353200.KQ	Daeduck Electronics	48.5	3	SM Quality · Cycle Rerating · SM Earnings	RS 98.1, OP +335.7% YoY

Meta rank reflects a composite of quality fundamentals, institutional flow, cycle rerating, earnings improvement, and DART catalyst layers. Overlap count is the raw screener intersection.

Candidate Spotlights

#1 Samyang NcChem (482630.KQ) — Meta Score 101.0

An electronic specialty chemicals manufacturer that clears all four primary screening layers in today’s run. Quality fundamentals are confirmed: ROE 14.7%, operating profit +64.5% YoY, margin expanding +4.3pp. Institutional and foreign co-buying is active (5-day F+QI net buy +₩3.8B). The cycle rerating signal is live. A DART IR event was filed on May 18 — a positive catalyst flag. The meta screener places it at #1 despite VM having a higher raw overlap count (4 vs. 5), because the composite score across quality, flow, earnings, and catalyst layers is marginally stronger. Next check: whether the IR disclosure reveals a structural business development or is routine investor relations. The earnings improvement rank (#1 in SM Earnings) deserves attention.

#2 VM (089970.KQ) — Meta Score 98.5 | Overlap Leader at 5 Screeners

A special-purpose semiconductor equipment maker and the strongest name by raw screener intersection — clearing all five frameworks including PEAD. Operating profit growth of +386.9% YoY and margin expansion of +29.3pp point to genuine operating leverage, not just revenue scale. The PEAD composite score of +2.94 is the highest in Tier A, suggesting post-earnings drift momentum is still intact following the May 15 quarterly filing. The meta screener ranks it #2 (slightly behind Samyang NcChem on the composite) but by pure overlap logic it would lead the table. Next check: order backlog depth and whether the 29pp margin step-up is structural or reflects project-mix timing.

#3 Jeju Semiconductor (080220.KQ) — Meta Score 80.2 | ⚠ Dilution Caution

The most direct fabless DRAM/NAND beneficiary in today’s screener universe, with ROE 17.4% and operating profit +274.4% YoY — a clean read-through from today’s memory repricing session. Five-day foreign and quality institutional net buying reached +₩57.8B. However, DART logged four risk-category filings today, specifically convertible bond and bond-with-warrant exercise designations, flagging potential dilution. The meta screener discounts the score for this (-4 risk flag penalty). The price action (+24.3% intraday, RSI 89.2) also signals near-term overextension. Next check: confirm the outstanding CB/BW conversion schedule and actual dilution timeline before treating the momentum as technically clean.

All screener data sourced from 2026-05-21 runs across KR Quality Compounder, KR Smart Money Quality, KR Cycle Rerating, KR Smart Money Earnings, and KR PEAD frameworks. This post is market analysis only and does not constitute a recommendation to buy or sell any security.

After Google I/O 2026 — Search Isn't Dead. The Real Issue Is the AI Infrastructure Bottleneck Created by the Agentic OS

Thu, 21 May 2026 10:35:00 +0900

📚 Related Series Google I/O + NVIDIA Earnings Preview / Korea AI Infrastructure After NVIDIA Q1 FY27 / Samsung Electronics TSMC-Style Rerating / Samsung Electro-Mechanics Hybrid Challenger / May 17 Market Snapshot

Google I/O 2026 was not a simple model announcement. It was an event that bundled Gemini 3.5, Gemini Omni, Gemini Spark, Antigravity, AI Search, Universal Cart, and Android XR together to reengineer Google’s entire product portfolio into an agentic OS. Search moves from a window displaying links to an interface that executes tasks; shopping becomes a cart agent; developer tools become a runtime that runs multiple agents in parallel. From an investment perspective, the key is not Alphabet itself but the infrastructure bottlenecks this transformation creates: token throughput, inference traffic, data center CapEx, HBM, packaging, power integrity, high-layer PCBs, and power equipment. Post-Google I/O, Korean alpha comes not from models but from bottlenecks.

Key Takeaways

The Essence of Google I/O 2026: Google bundled Gemini 3.5 Flash, Gemini Omni Flash, Search agents, Gemini Spark, Antigravity 2.0, Managed Agents API, Universal Cart, and Android XR into a single agentic product layer.
The Search Collapse Thesis Is Not Yet Confirmed by the Numbers: Alphabet’s Q1 2026 Search & other revenue was $60.399 billion, up approximately 19.1% year-over-year from $50.702 billion in the prior year period. So far, this looks less like AI encroaching on Search and more like Google absorbing Search into an AI interface.
The Most Important Number Is Token Throughput: Google disclosed that monthly processed tokens grew from 9.7 trillion in May 2024, to approximately 480 trillion at the 2025 I/O, to over 32 quadrillion in May 2026. The agentic OS shifts the unit of usage from search queries to tokens, tasks, and agent execution counts.
Alphabet Shares Confirm Defensiveness; Korea Should Track the Bottleneck: Alphabet’s case as a full-stack AI company has strengthened, but its market cap in the $4 trillion range and an earnings multiple of around 30× already reflect significant expectations. Fresh alpha is more visible in HBM, packaging, power integrity, PCBs, and data center power.
Korean Priority: Samsung Electro-Mechanics > Samsung Electronics > SK Hynix > Test/Socket/HBM Equipment > ISU Petasys · Daeduck Electronics · TLB · Korea Circuit > Power Equipment.
Strongest Single Idea: Samsung Electro-Mechanics. The KRW 1.557 trillion silicon capacitor contract is evidence that power stabilization within AI packages has emerged as a distinct bottleneck.
Macro Conditions: If the seven macro gates from the May 17 market snapshot have not been cleared, high-P/E AI infrastructure names can be shaken even with a good thesis. Scaling in after confirmation beats chasing.

1. The Essence of Google I/O 2026 — All of Google Becomes an Agentic OS

Summarizing this I/O as “Gemini got better” understates what actually happened. What Google actually did was reengineer its entire product portfolio into an agent interface.

Google’s official I/O collection presented Gemini Omni and Gemini 3.5 as the core new models, and grouped Search agents, Gemini Spark, Universal Cart, intelligent eyewear, Workspace, and Antigravity all under agentic experiences. In other words, this was not an update to one or two features — it was a repackaging of Search, the Gemini app, YouTube, Gmail, Docs, Shopping, Android, and Cloud developer tools into a single agent layer.

Reclassifying the key announcements from an investment perspective:

Category	Announcement	Investment Significance
Models	Gemini 3.5 Flash, 3.5 Pro upcoming	AI competition axis shifts from one-shot responses to long-running task execution
Multimodal Generation	Gemini Omni Flash	Video generation/editing, YouTube Shorts, and Flow become creative infrastructure
Search	AI Mode as default model, intelligent Search box, Search agents	Search box transforms from a query box to a task interface
Personal Agent	Daily Brief, Gemini Spark	Personal OS tying Gmail, Calendar, Docs, and Gemini data together
Developer Tools	Antigravity 2.0, CLI, SDK, Managed Agents API	Beyond Cursor-style coding assistance into enterprise agent runtime
Commerce	Universal Cart	Reassembling the purchase funnel across Ads, Search, YouTube, and Gmail
XR	Android XR, smart glasses	Post-smartphone ambient computing form factor option

In this structure, Google’s true asset is not any single model. It is the distribution that encompasses Search, YouTube, Gmail, Android, Chrome, Workspace, Shopping, Maps, Cloud, and TPU. Even when OpenAI or Anthropic lead on model performance, Google can convert the surfaces where users already spend time into an agent layer. That is what “agentic OS operator” means, as opposed to merely a “Search defensive name.”

2. Search Isn’t Dead — But the Unit of Search Is Changing

The market narrative over the past two years has been simple.

AI answers replace search results
→ Users click fewer links
→ Google Search ad revenue erodes
→ Alphabet multiple contracts

This logic is intuitive, but current numbers tell the opposite story. In Alphabet’s Q1 2026 results, Google Search & other revenue was $60.399 billion, up approximately 19.1% from $50.702 billion in the prior year period.

Search & other growth rate
= 60.399 / 50.702 - 1
= approx. 19.1%

Google Cloud revenue also came in at $20.028 billion, up approximately 63.4% from $12.260 billion in the prior year period. Cloud operating income was $6.598 billion, with an OPM of 32.9%.

Cloud growth rate = 20.028 / 12.260 - 1 = approx. 63.4%
Cloud OPM = 6.598 / 20.028 = approx. 32.9%

The facts to date do not support “AI killed Search.” A more accurate framing is: “Google is defending usage by transforming Search into an AI interface.”

That said, there is no room for complacency. The economic unit of search is changing.

Legacy Search	AI Search / Agentic Search
Keyword input	Natural language, files, images, video, tab input
Link clicks	Agent collects information, summarizes, and suggests actions
CPC-centric	Intent/action auction potential
Search query count	Token throughput · agent execution count
SEO publisher traffic	Task completion within Google surfaces

Google stated that AI Mode surpassed 1 billion monthly users within a year of launch and that AI Mode queries have more than doubled quarter over quarter since launch. Google also presented a framework in which Search agents continuously monitor web, news, social, financial, shopping, and sports data in the background.

This is not the destruction of search — it is the redefinition of search. But redefinition comes with costs. Search where an agent reads the web, summarizes, reasons, and sends notifications consumes far more compute than search that displays ten links. That is the critical point for Korean semiconductors and AI infrastructure.

3. The Most Important Numbers — Token Throughput and CapEx

The most important numbers from this I/O are not model benchmarks — they are usage figures.

Google disclosed that monthly processed tokens have grown as follows.

Period	Monthly Processed Tokens	Significance
May 2024	9.7 trillion	Early-stage AI usage
2025 I/O	approx. 480 trillion	Gemini, Cloud, and developer usage expansion
May 2026	32 quadrillion+	Approx. 7× year-over-year increase

The Gemini app also grew from 400 million monthly users at the 2025 I/O to over 900 million in May 2026. Google stated that daily requests also increased more than 7× over the past year.

More important is CapEx. Sundar Pichai stated that while Google spent $31 billion in annual CapEx in 2022, this year it expects approximately $180–190 billion. At that scale, Google I/O is not a consumer product event — it is a global AI infrastructure demand event.

AI agentic OS
→ Search, Gemini, YouTube, Workspace, Android, Shopping usage increases
→ Token throughput increases
→ Inference cluster expansion
→ HBM / packaging / network / power / cooling bottlenecks widen
→ Korean AI infrastructure value chain benefits

The first-order interpretation that “Google using TPUs reduces NVIDIA’s benefit” is too shallow. Whether it’s a TPU, GPU, or ASIC, high-performance AI accelerators require HBM, advanced packaging, high-speed substrates, power stabilization components, and data center power. Korean investors should focus on shared bottlenecks rather than any specific US chip vendor.

4. Alphabet Stock — Defensiveness Confirmed, Fresh Alpha Limited

Alphabet reinforced three defensive arguments through this I/O.

First, it demonstrated that Search is not simply a platform being encroached upon by AI, but rather a platform that absorbs AI. The 19% growth in Q1 2026 Search & other revenue lends force to this argument.

Second, Google is a full-stack AI company with models, TPU, Cloud, Search, YouTube, Android, Workspace, and Ads all under one roof. Its monetization surface is broader than companies that have only a model API.

Third, Google Cloud demonstrated both high growth and high margins simultaneously. A Cloud revenue growth rate of 63% and Cloud OPM of 32.9% reduce concerns that “AI Cloud grows revenue but not profits.”

However, investment discipline demands a cold eye. Alphabet is already a company with a market cap in the $4 trillion range, and its input P/E is around 30×. Even on a simple annualized Q1 operating income basis, market cap to operating profit is approximately 30×. The Q1 2026 net income figure also includes significant other income effects such as unrealized gains on equity investments, so concluding it is cheap based on GAAP EPS alone would be a mistake.

The investment view on Alphabet is as follows.

Item	View
Business Quality	Very high
Search Defensiveness	Reinforced by I/O and Q1 2026 results
Cloud Growth	Strong
Valuation	Already largely priced in
New Entry	Watchlist / on correction
Better Alpha	AI infra second-order beneficiaries

Alphabet is a good company. But the better trade after Google I/O is not chasing Alphabet — it is finding the beneficiaries of the CapEx bottlenecks Google will create.

5. Korean Value Chain Mapping — From HBM to Power Integrity

The Korean investment implication of Google I/O is not simply “HBM demand is good.” It is that as agentic workloads increase, bottlenecks propagate in the following sequence.

AI Search / Omni / Antigravity / Spark / Universal Cart
→ Persistent inference traffic increases
→ TPU/GPU/ASIC cluster expansion
→ HBM · DRAM · eSSD demand increases
→ Packaging · substrate · power stabilization component demand increases
→ Optical/networking · data center power demand increases

Mapping the Korean value chain by node:

Node	Bottleneck	Korean Candidates	View
HBM / Memory	Bandwidth, capacity	SK Hynix, Samsung Electronics	Structural benefit continues; leaders are crowded
Advanced Packaging	HBM stack, bonding, inspection	Hanmi Semiconductor, Intest Plus, ISC, Leeno Industrial	Beneficiary thesis is valid but pricing and customers need confirmation
Power Integrity	Silicon capacitors, MLCC	Samsung Electro-Mechanics	Clearest new alpha
Substrate / PCB	FC-BGA, high-layer PCB, memory module PCB	Samsung Electro-Mechanics, ISU Petasys, Daeduck Electronics, Simmtech, TLB, Korea Circuit	Earnings confirmation needed over theme-chasing
Power Infrastructure	Transformers, distribution, power quality	HD Hyundai Electric, LS ELECTRIC, Hyosung Heavy Industries, Iljin Electric	Demand is clear; valuation discipline required
Optical / Network	Optical, SerDes, interconnect	OESOLUTION etc.	Potential beneficiary; direct revenue verification needed

The key here is not finding “direct Google suppliers.” Even if it cannot be confirmed whether Korean companies directly supply Google TPUs, the bottlenecks commonly consumed by Google, NVIDIA, AWS, Meta, and Microsoft’s AI infrastructure expansion exist. Alpha lies in identifying where those bottlenecks accrue.

6. Samsung Electro-Mechanics — The Clearest Korean AI Infrastructure Alpha After Google I/O

The most important Korean name after this I/O is Samsung Electro-Mechanics. The reason is simple: the fact that AI infrastructure bottlenecks are shifting from HBM to power stabilization within the package has been confirmed through a real order.

Samsung Electro-Mechanics disclosed on May 20, 2026, that it had entered into a silicon capacitor supply agreement worth KRW 1.5570 trillion with a major global company. The contract period runs from January 1, 2027 to December 31, 2028.

Annualized contract value = KRW 1.5570tn / 2 years = KRW 778.5bn
Total contract value as % of recent revenue per disclosure = 13.8%
Annualized revenue contribution = 13.8% / 2 = approx. 6.9%/year

More important than the revenue figure itself is the classification shift this contract represents. Samsung Electro-Mechanics was previously classified as an MLCC, camera module, and substrate company. However, silicon capacitors are components placed inside or close to AI semiconductor packages to reduce power noise and stabilize power delivery. In other words, Samsung Electro-Mechanics is entering the AI server package BOM.

Samsung Electro-Mechanics’ official technical documentation also explains that MLCC-centric power architectures are expanding due to growing power demands from AI servers and GPU modules. AI GPUs have large instantaneous power fluctuations, making low ESR/ESL, high-frequency ripple response, and space efficiency critical. This is not a simple MLCC cycle — it is a power integrity cycle.

Samsung Electro-Mechanics’ P×Q×C framework can be viewed as follows.

Axis	Variable	Significance
P	Silicon capacitor, AI MLCC, FC-BGA ASP	Higher value-add mix versus legacy IT components
Q	GPU/TPU/ASIC package count, HBM stack count, AI server board count	Directly linked to AI CapEx
C	Yield, material costs, expansion depreciation	Early ramp risk

Why this may not yet be fully priced in.

The market still strongly tends to view Samsung Electro-Mechanics as a smartphone MLCC and camera module company.
Silicon capacitors are a separate category, and it takes time for them to be incorporated into sell-side models.
Contract revenue recognition begins in 2027–2028, so it feeds into near-term EPS with a lag.
The customer is undisclosed, so it cannot be determined whether the order is from NVIDIA, Google, or a Broadcom/Marvell ASIC.

The conclusion is therefore not “chase the move” but rather “pullback buy.” Samsung Electro-Mechanics is the clearest reclassification candidate in Korean AI infrastructure following this I/O and NVIDIA’s results, but chasing a sharp spike is inefficient when the macro gates from the May 17 market snapshot have not yet opened.

7. Samsung Electronics and SK Hynix — The Other Implication of Google TPU Multipolarity

Google I/O might appear to be an event that weakens NVIDIA’s monopoly. Google has its own TPUs, and the ASIC ecosystem is broadening to include Broadcom, Marvell, and even MediaTek. But from a Korean memory perspective, the conclusion is different.

TPUs consume HBM. ASICs consume HBM. Even as inference TPUs multiply, memory bandwidth and data movement bottlenecks do not disappear.

SK Hynix

SK Hynix remains the highest-quality beta to HBM. Google I/O’s growing agentic workloads, NVIDIA Rubin ramp, and hyperscaler CapEx expansion are all positive for HBM demand.

However, the alpha has diminished. The market already views SK Hynix as the lead HBM name. Being a good company and offering a good entry price right now are different things.

View: Hold / Buy on pullback
Positives: HBM pure beta, customer trust, profitability
Watch: crowding, elevated expectations, intensifying HBM4 competition

Samsung Electronics

Samsung Electronics is more interesting. The market tends to view Samsung Electronics solely as a “late HBM follower to NVIDIA.” However, as Google/Broadcom/Marvell/ASIC multipolarity advances, the HBM customer structure also moves away from a single-NVIDIA frame. This dynamic is favorable for a variant perception on Samsung Electronics.

Samsung Electronics is an integrated IDM with HBM, DRAM, eSSD, base die, foundry, and advanced packaging all under one roof. The issue is execution. HBM4 customer qualification, base die, foundry profitability, and customer diversification must be confirmed in actual numbers before it can escape its approximately 5× P/E discount.

View: Watchlist / Conditional Buy
Positives: HBM4, base die, eSSD, foundry optionality
Watch: execution discount, foundry losses, HBM qualification delays

This is the same conclusion as the prior Samsung Electronics rerating piece. Samsung Electronics is not cheap because the market doesn’t understand it. The market is demanding proof of downcycle defensibility and customer neutrality. Google I/O showed one path to meeting that demand: not through the NVIDIA single-chain, but through TPU/ASIC multipolarity in which Samsung’s HBM and foundry optionality can come alive.

8. Broadcom vs. Marvell — Korean Investors Should Not Be Looking for a Winner

In the US, the relative performance of Broadcom and Marvell post-Google I/O was also read as significant. The key is not whether Google is abandoning Broadcom. It is a signal that Google intends to multi-source its AI ASIC supply chain.

Broadcom is Google’s core TPU partner, and there are reports of a long-term agreement with Google. Marvell drew attention from reports of discussions with Google on a new AI chip, memory processing unit, and inference TPU. This movement does not mean “Google TPU demand is weak” — it means “Google is expanding its supply chain leverage.”

What matters to Korean investors is not which US ASIC vendor wins. It is the bottlenecks that are needed in common regardless of whether Broadcom or Marvell prevails.

US ASIC Shift	Korean Common Bottleneck
TPU generation increases	HBM, base die, advanced packaging
Inference TPU expansion	HBM/LPDDR, eSSD, networking
Memory processing unit	Memory bandwidth, substrate, test socket
AI rack growth	High-layer PCB, optical, power equipment
Multi-sourcing	Value increase for companies with customer qualifications

The Korean priority is therefore not any specific US chip designer but vendor-agnostic bottlenecks. The order should be: Samsung Electro-Mechanics, Samsung Electronics, SK Hynix, test/socket, high-layer PCB, power infrastructure.

9. Practical Trading Strategy

Priority

Rank	Stock / Group	View	Rationale
1	Samsung Electro-Mechanics	Buy on pullback	Silicon capacitor contract confirms AI package power integrity bottleneck via real order
2	Samsung Electronics	Watchlist / Conditional Buy	TPU/ASIC multipolarity can narrow the late-HBM discount
3	SK Hynix	Hold / Buy on pullback	HBM direction is right but consensus is crowded
4	Hanmi Semiconductor · ISC · Leeno Industrial · Intest Plus	Watchlist	Benefit from growing HBM/ASIC SKUs and test/packaging complexity
5	ISU Petasys · Daeduck Electronics · Simmtech · TLB · Korea Circuit	Watchlist	AI server PCB beneficiary potential, but customer/product/margin confirmation needed
6	HD Hyundai Electric · LS ELECTRIC · Hyosung Heavy Industries · Iljin Electric	Wait	AI data center power demand is clear, but valuation may be stretched

Entry Conditions

Samsung Electro-Mechanics: Confirmation of follow-on silicon capacitor customers, expansion plans, and potential disaggregation of AI MLCC/FC-BGA revenue.
Samsung Electronics: HBM4 customer qualification, HBM4 base die, volume toward Google/Broadcom/ASIC customers, or signs of foundry recovery.
SK Hynix: HBM4 pricing holds, NVIDIA Rubin ramp, 10–15% correction, or easing of supply-demand overheating.
PCB / Substrates: When AI-oriented revenue mix, ASP premium, and OPM maintenance are confirmed in quarterly results.
Test / Socket: Confirmation of TPU/ASIC/HBM customer references and new socket revenue.
Power Equipment: When North American data center order backlog growth and OPM maintenance are confirmed simultaneously.

Catalyst

Gemini 3.5 Pro launch and AI Mode/Search agent adoption.
Google Cloud next-quarter revenue, backlog, and CapEx update.
Broadcom and Marvell earnings commentary on Google/ASIC pipeline.
NVIDIA Rubin / Vera Rubin NVL72 ramp.
Samsung Electro-Mechanics follow-on silicon capacitor disclosure.
Samsung Electronics HBM4 qualification, base die, and foundry customer commentary.
Korean PCB/substrate companies disaggregating AI server-oriented revenue in Q2–Q3.

Invalidation

Google Search & other growth rate decelerates sharply after AI Mode expansion.
Google Cloud growth remains high but OPM falls below the low-20% range.
AI CapEx increases translate into depreciation, power costs, and networking costs that erode Alphabet margins.
HBM ASP declines or HBM4 oversupply signals emerge.
Samsung Electro-Mechanics’ silicon capacitor contract is confirmed to be low-margin or a single-customer event.
PCB/substrate names where AI revenue increases but OPM declines.
May 17 macro gates remain uncrossed, compressing high-P/E AI infrastructure multiples.

10. Frequently Asked Questions

“Is Google I/O Negative for NVIDIA?”

In the near term, some negative interpretation is possible. If Google expands its TPU footprint and the Broadcom/Marvell ASIC ecosystem grows, NVIDIA’s monopoly premium could compress somewhat. But from a Korean supply chain perspective, multipolarity is actually positive. Whether it is a GPU, TPU, or ASIC, HBM, packaging, PCBs, and power infrastructure are all required.

“Wouldn’t It Be Better to Just Buy Alphabet?”

Alphabet is a good company. Search defensiveness, Cloud growth, and its full-stack AI structure are all strong. However, the current market cap and earnings multiple already reflect much of the winner premium. The conclusion of this piece is not that Alphabet is a bad investment, but that immediately after Google I/O, fresh alpha is larger in Korean AI infrastructure bottlenecks than in Alphabet itself.

“Why Is Samsung Electro-Mechanics the Top Pick?”

HBM is already well known to the market. By contrast, the power integrity thesis — covering power stabilization within AI packages, silicon capacitors, AI MLCC, and FC-BGA — is still not finely classified by the market. To that, a real order of KRW 1.557 trillion is attached. This is a rare moment when both the story and the numbers are present simultaneously.

“Can I Buy All the PCB/Substrate Names?”

No. This is the segment with the greatest risk of theme excess. The “AI server-oriented” label is easy to apply, but without confirmed customers, products, ASP premium, and OPM, it can end as a short-term trading theme. From here, only companies that simultaneously show “AI-oriented revenue growth + margin maintenance” should remain in the portfolio.

11. The Final Word

Google I/O 2026 was not an event that declared search dead. On the contrary, it was an event in which Google declared its intention to convert search, mail, documents, shopping, developer tools, video, and even eyewear into agent interfaces. The search box moves from a query box to a task box; developer tools move from an IDE to an agent runtime; and shopping moves from a keyword funnel to Universal Cart.

Alphabet’s Q1 2026 numbers show that this transition has not yet damaged Search. Search & other revenue grew approximately 19% year-over-year; Cloud revenue grew approximately 63%; and Cloud OPM rose to 32.9%. So far, “Google absorbs AI into its existing distribution” is a more accurate frame than “AI kills Google.”

But the real investment point is the AI infrastructure bottleneck, not Alphabet itself. Google disclosed that monthly processed tokens grew from 9.7 trillion in May 2024 to over 32 quadrillion in May 2026. It also indicated that CapEx could expand from $31 billion in 2022 to approximately $180–190 billion this year. This shift is a demand function for HBM, packaging, power integrity, PCBs, and power infrastructure.

Korean priorities are clear. Samsung Electro-Mechanics has confirmed the AI package power integrity bottleneck through a real large-scale silicon capacitor contract. Samsung Electronics is an option where TPU/ASIC multipolarity can narrow the late-HBM discount. SK Hynix remains the strongest HBM name but is already a crowded trade. PCB/substrates, test, and power equipment have significant beneficiary potential but require earnings confirmation.

The conclusion is simple. Post-Google I/O alpha comes not from “which model is smarter” but from “where does growing agent usage force spending on bottlenecks.” The answer is not Alphabet alone. It is HBM, packaging, silicon capacitors, high-layer PCBs, and data center power. Of these, the freshest Korean alpha right now is Samsung Electro-Mechanics. But until the macro gates open, scaling in after confirmation beats chasing.

This article is for research and commentary purposes only and does not constitute investment advice. Google I/O 2026 announcements (Gemini Omni, Gemini 3.5, Search agents, Gemini Spark, Universal Cart, Antigravity, Android XR) are based on the official Google I/O collection and Google’s official blog. Alphabet Q1 2026 results (revenue $109.896 billion, operating income $39.696 billion, Search & other $60.399 billion, Google Cloud $20.028 billion, Cloud operating income $6.598 billion) are based on Alphabet’s Q1 2026 SEC filings. Monthly processed tokens of 32 quadrillion+, Gemini app monthly users of 900 million+, and CapEx guidance of $180–190 billion are based on Sundar Pichai’s statements at Google I/O 2026. Samsung Electro-Mechanics’ KRW 1.5570 trillion silicon capacitor supply agreement (term: 2027/01/01–2028/12/31; 13.8% of recent revenue) is based on the company’s May 20, 2026 regulatory disclosure and major domestic news coverage. Statements related to Samsung Electro-Mechanics’ AI server power architecture are based on Samsung Electro-Mechanics’ official technical documentation. Information on Broadcom, Marvell, and Google ASIC is based on reporting by Reuters and other foreign media; confirmed contracts and negotiation reports are mixed and should be distinguished accordingly. Stock priorities, value chain mapping, entry conditions, and invalidation conditions represent the analyst’s current framework and may differ from actual outcomes. Global macro conditions (interest rates, oil prices, exchange rates, VIX, foreign investor flows) may have additional impact on stock prices. This analysis may be wrong. Data as of May 21, 2026 KST.

After NVIDIA Q1 FY27 — No AI CAPEX Peakout. Korean Alpha Is Spreading Beyond HBM

Thu, 21 May 2026 09:20:00 +0900

📚 NVIDIA Earnings Follow-Up Series Google I/O + NVIDIA Earnings Preview / AI-RAN and the Korean Supply Chain / May 17 Market Snapshot

📚 Related: Samsung Electronics & Korea AI Infrastructure Samsung Electronics TSMC-Style Re-rating Thesis / Samsung Electro-Mechanics Hybrid Challenger / AI PCB Thesis

NVIDIA Q1 FY27 delivered revenue of $81.6B, Data Center revenue of $75.2B, and Non-GAAP EPS of $1.87, beating consensus. Q2 guidance of $91.0B ±2% also topped market expectations by a wide margin. The numbers alone offer no evidence of an AI CAPEX peakout. Yet the stock slipped roughly -1% in after-hours trading immediately after the report. That was not because of weak results. The market had already priced in “NVIDIA will inevitably beat,” and now demands validation on China, ACIE durability, Vera Rubin ramp, and custom ASIC penetration. For Korean investors, the more important conclusion lies elsewhere. AI server bottlenecks are no longer confined to HBM — they are spreading into packaging, power integrity, PCB/substrates, and data center power infrastructure.

Key Summary

NVIDIA Q1 FY27 was strong. Revenue came in at $81.615B, up +20% QoQ and +85% YoY. Data Center reached $75.246B, up +21% QoQ and +92% YoY. Non-GAAP EPS was $1.87, up +140% YoY. Q2 revenue guidance is $91.0B ±2%, with the company explicitly excluding China Data Center compute revenue from its guidance base.

The after-hours weakness was not about “weak numbers” — it was about “the height of expectations.” Q1 beat, Q2 guidance beat, an additional $80B share buyback authorization, and a dividend hike were all positives. But the market no longer views NVIDIA as a simple earnings-beat story. The next question is whether this growth rate continues beyond 2027.

The most significant new disclosure was the segmentation of the Data Center customer base into Hyperscale and ACIE. Hyperscale came in at $37.869B and ACIE at $37.377B. ACIE encompasses AI Clouds, Industrial, Enterprise, and Sovereign AI. If this structure holds, NVIDIA could be re-rated not as a “big-tech CAPEX proxy” but as an “AI CAPEX tax on every industry.”

The Korean takeaway is straightforward. The HBM cycle is not over. But the fresh alpha does not lie in simply owning SK Hynix. Rather than the leading HBM name that has already rallied, the more compelling second- and third-order beneficiary nodes are Samsung Electro-Mechanics silicon capacitors, HBM equipment, AI server PCB/substrates, and data center power infrastructure.

Stock priority order: Samsung Electro-Mechanics > Samsung Electronics > SK Hynix. SK Hynix is the most direct HBM winner, but the trade is crowded. Samsung Electronics offers a catch-up option across HBM4, SOCAMM2, HBM4 base die, and Foundry. Samsung Electro-Mechanics has earned the possibility of being reclassified from “MLCC company” to “AI package power-integrity component supplier” following a major silicon capacitor supply contract.

1. The Numbers — No AI CAPEX Peakout

The most important figure in this earnings report is not Q1 itself, but the Q2 guidance. A strong Q1 was already broadly anticipated. What mattered was whether double-digit sequential growth remained achievable off a massive revenue base in the following quarter.

Item	Q1 FY27	QoQ	YoY	Interpretation
Revenue	$81.615B	+20%	+85%	Primary evidence that AI infrastructure demand has not turned
Data Center	$75.246B	+21%	+92%	92.2% of total revenue
Data Center compute	$60.4B	+18%	+77%	Core GPU and AI compute
Data Center networking	$14.8B	+35%	+199%	Key signal of AI factory systemization
GAAP operating income	$53.536B	+21%	+147%	GAAP OPM of 65.6%
Non-GAAP operating income	$53.783B	+21%	+147%	Non-GAAP OPM of 65.9%
Non-GAAP EPS	$1.87	+18%	+140%	Now includes SBC in the calculation
FCF	$48.554B	—	—	FCF margin of approximately 59.5%

Q2 guidance matters even more.

Item	Company Guidance	Implication
Revenue	$91.0B ±2%	Approximately +11.5% QoQ growth over Q1
GAAP / Non-GAAP GM	74.9% / 75.0% ±50bp	Mid-70s margins maintained even with rising Blackwell mix
GAAP / Non-GAAP Opex	$8.5B / $8.3B	Cost increase to capture growth opportunities
China Data Center compute	Excluded from guidance	China is an upside option, not the base case

These numbers are unfavorable to the “AI CAPEX peak-out” thesis. A company guiding $91.0B in the quarter following $81.6B is not a company whose cycle is sharply rolling over from its peak. Of course, the slope of high growth rates eventually flattens. But based on this quarter’s numbers alone, the argument that “it has already turned” is difficult to defend.

2. Why the Stock Was Weak

The reason the stock dipped in after-hours trading despite strong results is simple: market expectations had moved faster than the actual numbers.

This earnings report delivered on four fronts simultaneously.

Q1 revenue beat
Q1 EPS beat
Q2 revenue guidance beat
Enhanced buyback and dividend

Yet the stock did not respond immediately. The reason is that “good numbers” are no longer sufficient on their own. NVIDIA is now a company where the bar for “good enough” rises every quarter.

What the market is demanding is the following.

Market Question	Why It Matters
Can FY27 growth hold even if China revenue remains at zero?	China is an upside option but also a multiple ceiling
Is ACIE growing faster than Hyperscale?	Reduces the discount tied to big-tech CAPEX concentration
Is Vera Rubin shipping in Q3 and ramping in Q4 without delays?	Removes transition risk after Blackwell
Are custom ASICs eroding inference market share?	Hyperscaler bargaining power concern
Can revenue absorb upper-40s OpEx growth?	EPS leverage sustainability

This reaction is therefore not a negative signal so much as a natural validation phase for a high-growth stock where expectations have become elevated. It remains a great company. For it to be a great stock, it must demonstrate the quality of its beats in the quarters ahead.

3. ACIE — The Most Important New Word in This Earnings Report

Starting this quarter, NVIDIA began disclosing Data Center revenue split between Hyperscale and ACIE. This change is not a simple reporting format update. It is an event that could shift the center of gravity in the valuation argument.

Segment	Q1 FY27 Revenue	QoQ	YoY	Significance
Hyperscale	$37.869B	+12%	+115%	Large cloud and consumer internet companies
ACIE	$37.377B	+31%	+74%	AI Clouds, Industrial, Enterprise, Sovereign AI
Edge Computing	$6.369B	+10%	+29%	Agentic and physical AI at the edge

The key point is that ACIE is already nearly as large as Hyperscale. The market has long viewed NVIDIA as “a function of big-tech CAPEX.” Under that framework, the investment cycles of Microsoft, Amazon, Google, and Meta determine the ceiling on NVIDIA’s multiple.

But if ACIE continues to grow, the framework changes. NVIDIA becomes not a hyperscaler CAPEX proxy, but the common infrastructure for AI cloud, industrial AI, sovereign AI, and enterprise AI. In that case, the TAM logic expands considerably.

This matters for Korea as well. The larger ACIE becomes, the more demand grows not just for GPUs, but for server memory, HBM, eSSD, networking, power-stabilization components, PCB, and power equipment simultaneously. In other words, the benefits flowing to the Korean supply chain broaden from a single hyperscaler axis to a full industrial AI infrastructure axis.

4. The Korean Value Chain — Bottlenecks Are Spreading Beyond HBM

NVIDIA’s Q2 revenue guidance of $91.0B is not a GPU-only figure. Running those GPUs inside actual data centers simultaneously requires HBM, DRAM, eSSD, packaging, substrates, capacitors, networking, and power equipment.

The capital flow of AI capex moves as follows.

Tier	Bottleneck	Korean Listed Companies
Primary	HBM, DRAM, eSSD	SK Hynix, Samsung Electronics
Secondary	HBM packaging equipment, bonding, inspection, test	Hanmi Semiconductor, PSK Holdings, Intekplus, ISC, Leeno Industrial
Tertiary	AI server substrates, PCB, power-integrity components	Samsung Electro-Mechanics, Daeduck Electronics, Korea Circuit, Simtech, TLB
Quaternary	Data center power infrastructure	HD Hyundai Electric, Hyosung Heavy Industries, LS ELECTRIC, LS Cable & System/LS, Gaon Cable

The most important takeaway from this table is not “the primary-tier trade is over.” Primary-tier benefits continue. The point is that the freshest alpha is migrating toward the second-, third-, and fourth-order bottlenecks.

SK Hynix is the most direct beneficiary. But the market already knows this. Samsung Electronics is a catch-up option with HBM4, base die, Foundry, and eSSD all in hand. Samsung Electro-Mechanics has earned a potential reclassification into AI package power-integrity components following this cycle. Power equipment functions as an alternative AI infrastructure exposure when semiconductor beta becomes overheated.

5. Samsung Electro-Mechanics — The Freshest Korean Alpha in This Cycle

On May 20, Samsung Electro-Mechanics disclosed a Silicon Capacitor supply contract with a global large enterprise worth KRW 1.557 trillion. The contract runs from January 1, 2027 through December 31, 2028, with an average annualized revenue contribution of approximately KRW 778.5 billion.

The significance of this contract is not simply incremental revenue. The core point is that the company’s classification may change.

Prior market perception:

MLCC
Camera modules
FC-BGA
Mobile and automotive components

Emerging perception:

AI package power-integrity component supplier
Power stabilization at or near GPU/HBM package internals
Critical passive component within high-performance semiconductor BOM
AI server package supplier

AI GPU and HBM packages carry high power density, where management of voltage fluctuation, noise, and signal loss is critical. Silicon capacitors stabilize power delivery in close proximity to the package. In other words, Samsung Electro-Mechanics now has a meaningful probability of entering the AI package BOM — not merely as a generic component supplier.

In a prior piece on Samsung Electro-Mechanics at KRW 1,010,000, the key question was framed as “Is 2027E operating profit above KRW 2.7 trillion achievable?” This contract is important new evidence toward that question. The contract covers 2027–2028, meaning it contributes almost nothing to 2026 earnings. But it can have a direct impact on upward revisions to 2027 estimates and a potential multiple re-rating.

The red-team case is equally clear, however.

The counterparty is undisclosed.
It cannot be confirmed whether this is a direct NVIDIA supply, a North American big-tech custom ASIC supply chain, or another high-performance semiconductor customer.
Revenue begins in 2027.
Initial mass-production yields and margins have not yet been verified.

The conclusion is therefore not “chase it unconditionally now,” but rather “preferred candidate on any pullback.” Samsung Electro-Mechanics has become the freshest re-rating candidate within the Korean AI infrastructure value chain following these developments.

6. Samsung Electronics — The HBM Catch-Up Option Has Become More Important

From Samsung Electronics’ perspective, NVIDIA’s Q1 FY27 results are a positive. The reasons are threefold.

First, NVIDIA Data Center has returned to strong growth, raising the ceiling on demand for HBM4, DDR5, SOCAMM2, and eSSD.

Second, NVIDIA excluded China Data Center compute revenue from its Q2 guidance — meaning it guided to $91.0B revenue based on non-China AI capex alone. Korea’s memory demand quality does not depend entirely on a Chinese market reopening.

Third, Vera Rubin has become the next battleground for the second half of 2026. Samsung Electronics explicitly referenced HBM4, SOCAMM2, PCIe Gen6 SSD, and HBM4 base-die supply in its 1Q26 earnings materials. The “HBM + base die + packaging + foundry integrated option” thesis outlined in the earlier Samsung Electronics re-rating piece has become relevant again.

Samsung Electronics’ investment case is distinct from SK Hynix.

Dimension	SK Hynix	Samsung Electronics
Current position	HBM leader	HBM4 catch-up + diversified IDM
Strength	Direct NVIDIA HBM beta	HBM4, base die, eSSD, Foundry, SOCAMM2
Risk	Already a well-known winner	Execution discount
Investment character	Hold / Pullback Buy	Wait / Buy on confirmation

Samsung Electronics is not “an already-proven HBM winner.” It is instead “a catch-up option whose multiple could move more significantly upon confirmation.” What is therefore needed is one thing: concrete confirmation in the form of HBM4E samples, Vera Rubin supply chain inclusion, HBM4 base-die qualification, or external HPC customer design wins.

7. SK Hynix — Strong, But Crowded

SK Hynix is the most direct Korean beneficiary of NVIDIA’s Q1 FY27 results. The company posted 1Q26 revenue of KRW 52.5763 trillion, operating income of KRW 37.6103 trillion, and an OPM of 72%. Expanding AI infrastructure investment and growing sales of HBM, high-capacity server DRAM, and eSSD drove the results.

The issue is that the stock has already reflected much of this logic in its price. The fact that SK Hynix is the HBM leader is no longer new information. NVIDIA’s strong Q2 guidance is good for SK Hynix. But from a new-money perspective, alpha comes not from “great company,” but from “the gap between new information and price.”

The stance on SK Hynix is therefore Hold / Pullback Buy. Existing positions can be maintained. Adding new exposure is more rational after a 10–15% pullback or following HBM4/HBM4E allocation and capex updates.

8. HBM Equipment, Substrates, and Power — Lagging Orders Behind GPU Revenue

The channel through which NVIDIA revenue growth flows into Korea does not end at HBM.

HBM Equipment / Inspection / Sockets. As the industry transitions from HBM3E to HBM4 and HBM4E, the complexity of TSV, bonding, packaging, and testing continues to rise. Hanmi Semiconductor, PSK Holdings, Intekplus, ISC, and Leeno Industrial remain in the beneficiary category. However, names like Hanmi Semiconductor that have already rallied significantly are better approached after order-gap confirmation, a pullback, or customer diversification evidence rather than on new buys.

PCB / Substrates. AI servers demand high-speed signal integrity and power stability. Samsung Electro-Mechanics, Daeduck Electronics, Korea Circuit, Simtech, and TLB are candidates within the AI server PCB/substrate space. However, this segment carries high risk of thematic excess. Terms like SOCAMM, LPDDR server modules, North American customers, or NVIDIA reference design, attached to a stock name without verified orders, ASP data, and margins, remain in theme territory.

Power Infrastructure. Data center bottlenecks ultimately converge on power. HD Hyundai Electric, Hyosung Heavy Industries, LS ELECTRIC, LS Cable & System/LS, and Gaon Cable can serve as alternative AI infrastructure exposure when semiconductor beta becomes overheated. That said, many power equipment names have also rallied significantly, making valuation and margin sustainability important.

9. NVIDIA Investment View

The conclusion on NVIDIA is BUY, with a 12-month price target of $280. Based on the current price of $223.47, the upside is approximately +25.3%.

The math is straightforward.

Scenario	FY27E EPS	Applied P/E	Price Target	Assumptions
Bear	$7.30	28x	$205	Q2 guidance low end, Vera delay, ACIE deceleration, China zero throughout
Base	$8.75	32x	$280	Q2 $91.0B achieved, GM 75%, ACIE sustained high growth
Bull	$10.00	36x	$360	Vera early ramp, ACIE acceleration, partial China resumption, FY28 visibility expansion

The $280 target is not an aggressive number. It applies a 32x multiple to FY27E EPS of $8.75. That is not unreasonable for a high-growth AI infrastructure monopoly platform, but it also reflects the view that further multiple expansion is limited at market cap above $5 trillion.

Four conditions would trigger a thesis reversal.

Q2 FY27 revenue falls below $89.2B, breaching the guidance floor
Non-GAAP GM falls below 74%
The thesis that ACIE grows faster than Hyperscale is undermined
Vera Rubin Q3/Q4 ramp is delayed

10. Q2 FY27 Checklist — August Exam

The next real test is Q2 FY27 results in August. This quarter demonstrated “no AI CAPEX peakout.” The next quarter must demonstrate “is this growth rate repeatable.”

Checkpoint	Positive Signal	Negative Signal
Q2 revenue	Above $93.0B	Below $89.2B
GM	Holds at 75%	Falls below 74%
ACIE	Continues growing faster than Hyperscale	ACIE decelerates
Vera Rubin	Q3 shipments and Q4 ramp confirmed	Transition delayed
China	FY27 growth maintained at zero China; any resumption is upside	Regulatory language tightens
Networking	Grows faster than Data Center overall	Networking decelerates
OpEx	Revenue growth absorbs cost increases	EPS leverage deteriorates

For Korean equities, the checkpoints are distinct.

Korean Checkpoint	Relevant Companies
HBM4/HBM4E customer qualification	Samsung Electronics, SK Hynix
HBM equipment generational transition orders	Hanmi Semiconductor, PSK Holdings, Intekplus
Silicon capacitor additional customers, capacity expansion, margins	Samsung Electro-Mechanics
SOCAMM / AI server PCB actual revenue	TLB, Korea Circuit, Daeduck Electronics, Simtech
North American AI data center power equipment orders and OPM	HD Hyundai Electric, Hyosung Heavy Industries, LS ELECTRIC

11. Practical Conclusions

Korean semiconductor strategy following NVIDIA’s Q1 FY27 results can be summarized as follows.

Rank	Sector	View
1	AI package power-integrity components	Freshest alpha. Samsung Electro-Mechanics is the focus
2	HBM / server memory	Structural tailwind continues. Leading names are crowded
3	HBM equipment / inspection / sockets	Ongoing capex tailwind. Check valuation burden
4	AI server PCB / substrates	Order verification required on a name-by-name basis
5	Data center power infrastructure	Non-semiconductor AI infra alternative

Top pick is Samsung Electro-Mechanics. The silicon capacitor contract is an event capable of changing the company’s classification. However, chasing after a short-term spike is less appropriate than buying on a pullback following confirmation of additional customers, capacity expansion, and margin guidance.

Samsung Electronics is the HBM4 catch-up and base-die and Foundry option. It can move more meaningfully upon confirmation, but execution evidence is still needed.

SK Hynix is the HBM leader. But the best of the narrative is already heavily reflected in the price. Maintain existing positions; adding new exposure is more efficient on a pullback.

Hanmi Semiconductor and other HBM equipment names are good businesses but crowded trades. Power equipment is a valid alternative AI infrastructure exposure when semiconductor beta becomes overheated.

12. Final Line

NVIDIA Q1 FY27 refuted the AI CAPEX peakout narrative. Revenue of $81.6B. Data Center of $75.2B. Q2 guidance of $91.0B. Looking at those numbers, it is difficult to argue that the AI infrastructure cycle is over.

But it is also understandable why the stock did not surge immediately. The market had already expected strong numbers. What is now needed is validation of ACIE durability, Vera Rubin ramp execution, growth maintenance ex-China, and defense against custom ASIC encroachment. NVIDIA is still a BUY, but the risk is not the earnings — it is the expectations.

The Korean implication is more important. HBM is not over. But alpha is spreading beyond HBM. SK Hynix is the most direct winner but is crowded. Samsung Electronics is the HBM4 catch-up option. Samsung Electro-Mechanics, following its major silicon capacitor contract, has earned the possibility of being reclassified as an AI package power-integrity component supplier.

The real question is not “is NVIDIA good?” It already is. The question is “which Korean node does NVIDIA’s revenue growth translate into in new earnings?” Right now the answer is Samsung Electro-Mechanics, Samsung Electronics HBM4, HBM equipment, AI server substrates, and power infrastructure.

A great company and a great stock are different things. NVIDIA remains a great company. In Korea, the great stocks are more likely to come not from the leading names that have already rallied, but from the bottleneck nodes where GPU revenue growth has not yet been fully priced in.

This article is intended solely for research and commentary purposes and does not constitute investment advice. NVIDIA Q1 FY27 results (revenue $81.615B, Data Center revenue $75.246B, GAAP EPS $2.39, Non-GAAP EPS $1.87, Q2 revenue guidance $91.0B ±2%, Non-GAAP GM 75.0% ±50bp, China Data Center compute revenue excluded from guidance, $80B additional share repurchase authorization approved, dividend increase) are based on NVIDIA’s official Newsroom and IR materials. Hyperscale $37.869B, ACIE $37.377B, Edge Computing $6.369B, and China-related risk disclosures are based on NVIDIA’s 10-Q and earnings materials. Post-announcement stock price reaction, current price of $223.47, and after-hours price of $220.91 are based on market data sources including MarketBeat and are subject to real-time change. The price target of $280, FY27E EPS of $8.75, applied P/E of 32x, and Bear/Base/Bull scenarios represent the analyst’s subjective estimates and are not guaranteed. SK Hynix 1Q26 results (revenue KRW 52.5763 trillion, operating income KRW 37.6103 trillion, OPM 72%) are based on the company’s official earnings materials. Samsung Electronics 1Q26 results (consolidated revenue KRW 133.9 trillion, operating income KRW 57.2 trillion, DS segment revenue KRW 81.7 trillion, DS segment operating income KRW 53.7 trillion, and statements regarding HBM4, SOCAMM2, PCIe Gen6 SSD, and HBM4 base-die) are based on Samsung Electronics’ official disclosures. Samsung Electro-Mechanics’ KRW 1.5570 trillion silicon capacitor supply contract is based on the May 20, 2026 regulatory disclosure and domestic press coverage; the customer is undisclosed. Korean stock prioritization and sector views reflect the analyst’s judgment based on current price, flow, and earnings visibility, and are subject to change as market conditions evolve. Global macro conditions (U.S. interest rates, exchange rates, oil prices, VIX), China regulatory developments, hyperscaler CAPEX trends, custom ASIC penetration rates, and Vera Rubin ramp timing may further influence the conclusions herein. The analysis may prove incorrect. Data as of May 21, 2026 KST.

Samsung Electro-Mechanics' KRW 1.5T Silicon-Capacitor Win — A Reclassification, Not a Single Order. But KRW 967K Already Prices Most of It In

Thu, 21 May 2026 00:00:00 +0000

📚 Samsung Electro-Mechanics series Part 1: SEMCO — ‘The Invisible Infrastructure of AI Silicon’ (MLCC / FC-BGA / Optics deep dive, May 15) Part 2: SEMCO Hybrid Challenger — Between Murata and Ibiden, What KRW 1.01M Implies for 2027E OP (May 17) Mirae KRW 1.3M target: The valuation-frame shift (May 7) First re-rating call: Reclassification into an AI-infrastructure components name (Apr 21)

TL;DR

The headline is positive. SEMCO just proved an actual KRW 1.557T order — not a concept — inside the AI-packaging value chain.
The real point is reclassification. From “MLCC + camera + substrate” to “power-integrity component company sitting inside the AI-package."
The market priced it immediately. May 20 close KRW 1,061,000 (+7.50%); May 21 pre-market KRW 1,109,000 (+4.52%). April 20 → May 20 = +56% in a month.
But at KRW 967K the stock is already expensive. 2026E PER ~60x. New buys: Wait. Holders: Hold. Re-entry on a pullback to KRW 850–900K or a consensus 2027E EPS lift to KRW 28–30K.
Next checkpoints: named customer, application position (Top / Land / Embedded), 2027 revenue-recognition cadence, yield / CAPEX, Si-Cap margin, and the second / third design wins.

1. The contract in numbers

1.1 The basics

Item	Value	Read
Contract value (KRW)	KRW 1,557.0 billion	Very large by SEMCO new-business standards
Contract value (USD)	~USD 1,035.4 million	At ~1,503.8 KRW/USD
Term	Jan 1, 2027 – Dec 31, 2028	Recognition mostly in 2027–2028
Implied annual revenue	~KRW 778.5 billion / year	Contract ÷ 2 years
vs FY25 revenue	~13.8% total / ~6.9% annual	Per disclosure
Implied quarterly	~KRW 194.6 billion / quarter	Even-recognition assumption

Check: KRW 1,557.0B ÷ 2 = KRW 778.5B. KRW 778.5B ÷ KRW 11,314.5B (FY25 revenue) ≈ 6.9%.

1.2 Versus the global market

The global silicon-capacitor market is estimated at ~USD 2.3 billion in 2026. The contract’s annualized USD 517.7M equates to roughly 22.5% of that figure. Caveat: definitions differ across research houses and recognition is split across 2027–2028, so treat this as a sizing check, not as a literal market-share claim.

1.3 OP sensitivity — margin is not disclosed

Si-Cap margin is not public. The table below is a sensitivity, not a forecast.

Assumed OPM	Annual OP contribution	After-tax NI (24% tax)	EPS contribution (~74.7M shares)
15%	~KRW 116.8B	~KRW 88.7B	~+KRW 1,188
20%	~KRW 155.7B	~KRW 118.3B	~+KRW 1,584
25%	~KRW 194.6B	~KRW 147.9B	~+KRW 1,980
30%	~KRW 233.6B	~KRW 177.5B	~+KRW 2,376
40%	~KRW 311.4B	~KRW 236.7B	~+KRW 3,169

At 30% OPM, this single contract adds ~KRW 2,400 to EPS — about a 10% lift to Hana Securities’ standing 2027E EPS of KRW 22,874. Actual margin depends on early-stage yields, depreciation, customer-price terms, wafer-process cost, and test cost.

2. Price action — May 20 and May 21

2.1 May 20 regular session

Date	Open	Low	High	Close	Chg	Volume
May 19, 2026	1,031,000	957,000	1,040,000	987,000	-4.27%	879K
May 20, 2026	968,000	924,000	1,106,000	1,061,000	+7.50%	1.91M

The disclosure landed around 2 PM. The stock immediately reversed from the morning’s weakness, hit KRW 1,080,000 (+9.42%) intraday, and closed +7.50%. Volume = ~2.17x the prior session.

2.2 May 21 pre-market (NXT)

Time	Price	Change
May 21, 08:23 NXT	KRW 1,109,000	+4.52% vs prior close
vs May 19 close	KRW 1,109,000	+12.36%

Pre-market high was KRW 1,139,000 — a test of the 52-week high (KRW 1,133,000).

2.3 Flows — institutions absorbing foreign profit-taking

May 20 flows: foreigners -111,248 shares / institutions +82,167 shares. That is the pattern of “institutional re-rating absorbs foreign profit-taking” — not a one-day thematic pop, especially when combined with the +56.0% one-month run from April 20 to May 20.

3. Why this is a reclassification event

3.1 The bottleneck Si-Cap solves — power integrity

AI accelerators pull current very fast across hundreds of billions of parallel operations. If voltage swings, the chip mis-computes, throttles down, and heats up. The basic equation is simple.

Voltage swing ≈ L × di/dt + I × ESR L = ESL inductance, ESR = equivalent series resistance, di/dt = instantaneous current change

AI chips have very high di/dt, so the only fix is to put a storage capacitor as close to the chip as possible and minimize ESL / ESR. A silicon capacitor is thinned to ≤100 µm and placed Top-Side, Land-Side, or Embedded into the substrate. That is what relieves the bottleneck.

If a board-level MLCC is a “battery a few rooms away,” a Si-Cap is “the emergency reservoir taped to the chip’s leg.”

3.2 Aligned with the TSMC direction

TSMC describes integrating deep-trench capacitors into the silicon interposer for CoWoS-S, and integrating eDTC beneath the SoC for CoWoS-L for power management. The direction of advanced packaging is exactly “signal routing + HBM linkage + power-stabilization components, all integrated." SEMCO is now in one of those layers.

3.3 Why this matters specifically for SEMCO — only when bundled with FC-BGA

The real value is not Si-Cap as a stand-alone line. It is becoming an integrated supplier of FC-BGA + MLCC + Si-Cap.

MLCC only: faces Murata head-on.
FC-BGA only: faces Ibiden / Shinko head-on.
MLCC + FC-BGA + Si-Cap as a bundle: the only player globally with all three.

From the customer’s perspective, “design my substrate and my power-stabilization parts together” is dramatically more attractive — and once designed in, it’s hard to swap out. That lock-in is the real moat.

3.4 High qualification barrier — sticky once in

Si-Cap reliability is tied to overall AI-package reliability. A defect is not a part-level issue; it becomes a GPU / ASIC package-level reliability issue. So qualification is long, and design-in is hard to reverse. Reporting notes the field has been concentrated among a small number of qualified suppliers precisely for this reason.

4. Customer reading — separate fact, inference, and unknown

[Fact] The counterparty is not disclosed. Whether it is NVIDIA, Marvell, Broadcom, Amazon, Microsoft, or Google cannot be confirmed today.

[Fact] Earlier reporting indicated SEMCO supplies silicon capacitors into Marvell’s AI-accelerator multi-die packaging platform, with the Marvell program reportedly entering mass production in 1Q 2025.

[Inferred] Given the deal size and the 2027–2028 supply window, the end use is almost certainly AI-server GPU / ASIC or large data-center silicon, not mobile APs. Three reasons:

Deal size is too large for a mobile-only program
Si-Cap’s technical need is biggest in AI / HPC packages
SEMCO has publicly framed high-end semiconductor packaging and AI-server use as the target end markets

[Unknown] It would still be premature to call this “an NVIDIA contract” or any specific hyperscaler. Public documents do not distinguish whether the immediate counterparty is a fabless designer, a hyperscaler, or a packaging-house intermediary.

5. What the current price prices in — and what it doesn’t

5.1 Where we are

Item	Value
May 20 close	KRW 1,061,000
May 21 pre-market	KRW 1,109,000
Market cap	~KRW 72.2 trillion
Shares out	74,693,696
2026E revenue	~KRW 13.36 trillion
2026E OP	~KRW 1.57 trillion
2026E EPS	~KRW 16,315
2026E PER	~60x

This is not a cheap stock. The disclosure is not a “buy because it’s cheap” story — it is a “the high multiple now has more earnings visibility behind it” story.

5.2 Multiples are stretched; the swing variable is the EPS-revision pace

SK Securities lifted its target from KRW 1.10M → KRW 1.50M, citing FC-BGA undersupply, larger AI-package area, ASP gains, MLCC improvement, and Si-Cap / glass-substrate differentiation. 2027E OP at KRW 2.353T.
Hana Securities (April) set KRW 810K = 2027E EPS KRW 22,874 × global FC-BGA peer 35.4x.

To cleanly justify the KRW 970K–1.10M zone, 2027E EPS needs to reach the KRW 28–30K band. That requires all three of: this Si-Cap contract recognizing on schedule, FC-BGA price hikes, and the MLCC up-cycle continuing.

5.3 Bands and calls

Price band	Read	Call
850K–900K	Implies 2027E EPS 24–26K at 35x	High buying interest
950K–1,050K	Si-Cap deal starts being priced in	Conditional buy on dips
1,061K (now)	After a one-month +56%	Wait / Hold
1,100K–1,130K	52-week-high test	Chasing inefficient
Above 1,130K	Without a fresh catalyst, short-term overheat	Trim or take profit

6. Trading call — split for holders and new entrants

6.1 Existing holders

Call: Hold.

Fundamentals are intact and stronger. The KRW 1.5T contract reduces the de-rating risk one notch. But additions should follow at least one of:

2Q or 3Q earnings showing FC-BGA / MLCC ASP gains
A new Si-Cap customer disclosure
Qualitative or quantitative evidence of Si-Cap OPM at 25–30%+
Consensus 2027E EPS lifted above KRW 28,000

6.2 New buyers

Call: Wait.

Reason: a stock that is +56% in a month and is now stacking a KRW 1.5T positive on top is a poor entry point. Two cleaner entries:

Price: a pullback into the KRW 850K–900K zone. At 2027E EPS KRW 24–26K, that band offers a 35x entry.
Earnings: consensus 2027E EPS lifted to KRW 28K+ AND price held under KRW 1.0M. That combination earns the AI-package premium honestly.

6.3 Thesis-kill conditions

Re-evaluate the call on any of:

Amended disclosure reducing contract size or duration
2027 Si-Cap recognition slipping into 2028
Confirmed Si-Cap OPM under 20%
FC-BGA or MLCC pricing-cycle rollover
Single-customer concentration above 70% with no visibility on 2028+ renewal
Yield problems following large CAPEX

7. Six- to twelve-month checklist

Application position — Top Side / Land Side / Embedded? Embedded carries the largest FC-BGA bundling value.
Customer profile — GPU fabless / AI-ASIC fabless / hyperscaler / packaging house?
Revenue-recognition cadence — even in 2027, or skewed into 2028?
Margin — above the legacy MLCC / package-substrate blended margin?
CAPEX and yield — do mass-production yields hold without distorting operating leverage?
Second / third customer — does another AI design-win land?
FC-BGA bundling — is Si-Cap a stand-alone supply, or part of a substrate + power-integrity solution?
Consensus 2027E EPS path — does it move from KRW 22,874 toward KRW 28,000+?

8. How this links to other posts

SEMCO — ‘Invisible Infrastructure of AI Silicon’ (May 15) — established the three-division framing (Component / Package Solution / Optics). This article adds a fourth product line to the Component division.
SEMCO Hybrid Challenger (May 17) — the “what KRW 1.01M implies for 2027E OP KRW 2.7T” framing. The KRW 1.5T Si-Cap contract contributes roughly KRW 100–200B of incremental OP at base-case margin assumptions.
Mirae KRW 1.3M target — valuation-frame shift (May 7) — the “if SEMCO is reclassified as a global AI-component peer, KRW 1.3M is reachable” framing. This article is the first hard evidence of that reclassification.
First re-rating call — reclassification into an AI-infrastructure components name (Apr 21) — the bet placed one month ago; +56% over the month and a KRW 1.5T contract today is the validation.

9. One-line read for non-specialists

AI silicon is no longer just a “make the chip smaller” race. GPUs and ASICs draw so much electricity, so quickly, that the part that delivers stable power inches from the chip has become a bottleneck. Silicon capacitors fill that role.

A regular capacitor is a backup battery in another room. A silicon capacitor is the emergency reservoir taped to the chip’s leg. AI chips that suddenly want more current need that reservoir, or voltage shakes, performance drops, errors rise, and heat climbs.

For SEMCO, that means a category change. Historically it was an MLCC + camera + substrate company. This KRW 1.5T contract says SEMCO is now inside the AI-package itself. If it stacks with more customers and FC-BGA bundling, the multiple moves from “parts maker” toward “AI-package core component supplier.”

But the stock is already trading at KRW 967K–1.11M. It did not just become a good company — it is already priced like one. Chasing is inefficient; pullbacks are the entry.

10. The one-line bottom line

This disclosure reclassifies SEMCO from “smartphone parts maker” toward “AI-packaging core-component platform." The direction is clearly right. But the price already reflects part of that shift. From here, what matters is not the headline quality — it is the pace of the consensus EPS revisions.

This is not a stock that just became a good company. It is a stock already priced like a good company. Chasing is inefficient; pullbacks are the opportunity.

This article is research and commentary only and is not investment advice. The KRW 1,557.0B contract disclosure (May 20, 2026; term Jan 1, 2027 – Dec 31, 2028) is per the company’s official filing. The USD 1,035.4M figure at FX 1,503.8 KRW/USD is per the disclosure attachment. FY2025 revenue KRW 11,314.5B and OP KRW 913.3B are per the company’s official release. The May 20 close of KRW 1,061,000, May 19 close of KRW 987,000, and May 21 pre-market KRW 1,109,000 are per KRX, Naver Finance, and Investing.com. 2026E revenue / OP / EPS / PER reflect Naver Finance / FnGuide consensus and are subject to change. SK Securities’ KRW 1.5M target, Hana Securities’ KRW 810K and 2027E EPS KRW 22,874 are per each broker’s report. Marvell-related reporting (June 19, 2025 via Yonhap and others) is not evidence identifying the counterparty in this contract. The ~USD 2.3B silicon-capacitor market figure (2026) references Mordor Intelligence. The OPM sensitivity table (15–40%) is the author’s assumption; actual margins are not disclosed. Band-by-band reads, entry conditions, and thesis-kill conditions are the author’s views and are not guaranteed. Global macro variables (US rates, oil, FX, VIX) can independently move the stock. The analysis may be wrong. Data cut-off: May 21, 2026 08:30 KST.

Korea Quality Re-Rating Watch 2026-05-20: KOSES, Vitzrocell & Eltrontech — Smart Money Holds in a Bear-Regime Selloff

Wed, 20 May 2026 16:30:00 +0900

1. Macro Dashboard

Indicator	Level	5-Day Change	Signal
KOSPI	7,208.9	−772 pts / −9.7%	Bear
KOSDAQ	1,056.1	−135 pts / −11.3%	Bear
USD/KRW	1,505	+1.0%	KRW weak
VIX	17.9	+3.9%	Stable
US 10Y	4.67%	+0.19pp	Rising
Brent	$108.5	−0.7%	Steady
DXY	99.4	+0.1%	Flat

Regime verdict — KR: Bear / US: Neutral. KOSPI breadth sits at just 26.2% above the 50-day MA and 40.4% above the 200-day MA; the Discovery score is 65/100 at Day 20. US breadth is softer at 45.8%/42.8% but holds Neutral. The KR–US divergence — Korea deteriorating while the US holds — is the defining macro feature of the week. The spread between the two regimes argues for reduced Korea exposure and elevated cash positioning until breadth recovers.

2. Market Wrap

(Based on 2026-05-19 Korea close data, most recent available as of publication)

Tuesday’s Korea session was a broad risk-off washout, not a selective rotation. KOSPI closed near 7,304 (−2.81%) and KOSDAQ at 1,084 (−2.40%), with every macro variable pointing the wrong direction simultaneously: USD/KRW remained above 1,500, US 10-year yields held at 4.6%, and oil — though slightly off its highs on Iran-deal speculation — stayed elevated enough to pressure growth multiples.

The single most telling print was foreign outflow: −₩4.83 trillion on KOSPI in a single session. That is a figure large enough to overwhelm institutional support in even large-cap index anchors. Domestic institutions partially offset the selling in semiconductor bellwethers, but absorption was incomplete.

Sector scorecard:

Outperformed: Defense, power equipment, and a cluster of AI-infrastructure discovery names resisted the tape. These sectors carry domestic demand narratives partially decoupled from global risk sentiment and benefited from relative rotation out of high-multiple growth.
Underperformed: Large-cap semiconductors bore the brunt. The proximate catalyst was Seagate’s CEO signaling caution on new capacity expansion — commentary that rippled into memory, storage, and adjacent equipment names across the Korea complex. Robotics and high-multiple growth stocks fell sharply in the rate/dollar environment. Securities and machinery names followed lower.

Flow nuance: Institutional behavior was not uniformly negative. In large-cap semiconductors, domestic institutions absorbed meaningful blocks — suggesting the selling pressure reflects foreign portfolio rebalancing rather than a domestic conviction shift. Supporting that read: CoreWeave’s announcement of a $3.1bn GPU-backed credit facility confirms that AI infrastructure capex has not yet inflected downward. Investors should distinguish between multiple compression (a rate/FX story) and structural demand deterioration (a different story entirely).

Session character: Selective survival. Money rotated toward defense and power infrastructure, away from anything priced for growth. Heading into Wednesday, the operative mode is defensive-selective: quality-of-business screens matter more than beta.

3. Today’s Quality Re-Rating Candidates

Today’s meta screener ranked 102 candidates across five frameworks — Quality Compounder, Smart Money Quality, Cycle Rerating, Smart Money Earnings, and PEAD. Composite scores weight screener overlap, fundamental quality, official DART catalysts, and Kiwoom market surface signals. Only 4 stocks cleared the Smart Money Quality filter from a universe of 2,716 — a thin read that independently confirms Bear-regime conditions. In this environment, 4-way overlap carries meaningfully higher conviction weight than usual.

Top Candidates

Rank	Ticker	Name	Meta Score	Overlap	Screeners Hit	Key Metrics
1	089890.KQ	KOSES	84.9	4	QC · SMQ · CR · SME	ROE 17.8%, OP YoY +410%, OPM Δ +29.2pp
2	082920.KQ	Vitzrocell	73.8	4	QC · SMQ · CR · SME	ROE 17.2%, OP YoY +33%, 5d net +₩25.6bn ⚠️
3	054210.KQ	Eltrontech	65.8	4	SMQ · CR · SME · PEAD	OP YoY +168%, OPM Δ +2.8pp, IR filed 5/19
4	000660.KS	SK Hynix	65.7	2+CUR	QC · CR · CUR	ROE 35.6%, OP YoY +101%, consensus z +1.38 ⚠️
5	062040.KS	Sanil Electric	59.3	2	QC · CR	ROE 25.4%, OP YoY +30%, 2 DART catalysts
6	420770.KQ	Gigabis	58.4	2+CUR	QC · CR · CUR	OP YoY +777%, OPM Δ +29.9pp, consensus z +1.50
7	053610.KQ	Protek	58.0	2+CUR	QC · CR · CUR	ROE 13.7%, OP YoY +245%, OPM Δ +12.2pp
8	147830.KQ	Jelyong Industrial	55.3	2	QC · CR	ROE 13.9%, OP YoY +159%, OPM Δ +13.9pp

QC=Quality Compounder, SMQ=Smart Money Quality, CR=Cycle Rerating, SME=Smart Money Earnings, CUR=Consensus Up Revision. ⚠️ = notable caution flag requiring verification.

Top 3 Deep Dive

#1 — KOSES (089890.KQ) | Meta Score: 84.9 | 4-Screener Hit

KOSES manufactures precision equipment for semiconductor and display processes — a supply-chain node with direct leverage to fab utilization cycles. The quality case is strong: ROE of 17.8% and operating profit growth of +410% YoY, with a 29.2 percentage-point margin expansion that signals a genuine structural shift in earnings power, not a revenue-mix accident. On the money-flow side, combined foreign and quality-institution net buying reached +₩2.1bn over five sessions — meaningful for a name of this market cap. A DART supply-contract amendment (April 30) adds an official catalyst. What to verify next: whether the margin inflection is structural versus driven by a single concentrated order; H2 2026 order-pipeline visibility.

#2 — Vitzrocell (082920.KQ) | Meta Score: 73.8 | 4-Screener Hit

Vitzrocell manufactures primary and specialty secondary batteries, with exposure to industrial and military applications — a segment that benefits from Korea’s accelerating defense procurement cycle and rising demand for non-consumer battery chemistries. The fundamental screen clears: ROE 17.2%, OP YoY +33%, operating margin up 3.9pp. Smart-money net inflows over five days reached +₩25.6bn, the strongest absolute flow figure among this week’s candidates. Two caution flags temper the setup: a DART litigation filing (May 14) disclosing a claim above the mandatory disclosure threshold, and program-trading exposure at +20.6% with short interest at 6.1%. The litigation disclosure is the gating item. Do not treat this as a clean candidate until the nature and materiality of the claim are established.

#3 — Eltrontech (054210.KQ) | Meta Score: 65.8 | 4-Screener Hit

Eltrontech is an electronic component manufacturer posting one of the sharpest earnings inflections in today’s screener output: operating income +168% YoY, revenue +35%. The PEAD signal is active — earnings were recently announced, price is tracking above the 50-day MA, and the drift window is open. Critically, Eltrontech filed an IR event disclosure on DART on May 19, which creates a near-term catalyst that the market has not yet fully absorbed. The Kiwoom surface shows institutional and foreign activity both net positive, with short interest at a clean 2.9%. This is the most straightforward “earnings inflection with money flowing in” setup in today’s queue. What to verify: whether the operating income jump reflects a durable revenue-mix shift or a delivery-timing concentration in a single quarter.

Editorial note: The meta screener also flags SK Hynix (000660.KS) at rank #4 on consensus upgrade momentum and cycle rerating — ROE 35.6%, OP YoY +101%, consensus composite z-score +1.38. However, net foreign and quality-institution selling over five days totals −₩9.6 trillion, a caution flag significant enough to warrant watching the flow reversal before treating it as an active re-rating candidate rather than a held-quality name under distribution pressure.

All screener data as of 2026-05-20. Market close data from 2026-05-19 (most recent confirmed close). This is public market research, not investment advice.

KOSPI Drops 2.8%: What's Driving Korea's Sell-Off

Tue, 19 May 2026 23:30:00 +0900

KOSPI Sheds 2.8% as Foreign Selling Hits ₩4.83 Trillion

South Korea’s benchmark equity index, KOSPI — comprising roughly 800 listed companies across the Seoul bourse — fell 2.81% to 7,304.49 on May 19, 2026, while the tech-heavy KOSDAQ dropped 2.40% to 1,084.43. Foreign investors net-sold ₩4.83 trillion worth of KOSPI shares in a single session, one of the heavier single-day outflows of recent weeks. The Korean won weakened to the ₩1,507-per-dollar range, and the U.S. 10-year Treasury yield hovering near 4.6% added pressure on growth multiples across the board.

The day’s regime is best described as risk-off leaning neutral: the KR operational screener technically remains in bull mode, but the combination of currency pressure, rate headwinds, and commodity costs created a difficult backdrop for equities requiring high multiples to work.

What Actually Moved Today: Storage Caution Ripples Into Korean Chips

The primary catalyst for Korean semiconductor weakness was not domestic. Seagate’s CEO flagged caution on capacity expansion, triggering weakness in Micron, storage-adjacent names, and equipment stocks in the U.S. session. That sentiment carried directly into Korean semiconductor names on Tuesday.

Samsung Electronics (005930.KS), South Korea’s largest semiconductor and consumer electronics manufacturer with a global market cap exceeding ₩160 trillion, fell 1.96% to ₩275,500. Foreign investors sold ₩2.54 trillion of Samsung shares alone — a significant single-name outflow. Institutional investors partially offset that, buying roughly ₩393 billion and providing some technical support. Samsung held above its 20-day moving average, a line analysts have been watching as a near-term support level.

Samsung Electro-Mechanics (009150.KS), a leading manufacturer of multilayer ceramic capacitors (MLCCs) and semiconductor substrates used in AI server and advanced packaging applications, dropped 4.27% to ₩987,000. Foreign investors sold ₩143.6 billion while domestic institutions bought ₩25.7 billion — a pattern similar to Samsung Electronics, where institutional buyers absorbed some of the foreign exit but could not fully offset it.

Daeduck Electronics (008060.KS), a printed circuit board and substrate manufacturer that has benefited from AI-driven demand for advanced packaging materials, fell 4.34% to ₩134,300. Unlike the above two names, both foreign and institutional flows were negative — a weaker setup that analysts noted as worth monitoring if the pattern persists.

The Counternarrative: AI Capex Isn’t Dead

Against the semiconductor weakness, one piece of news cut the other way. CoreWeave, the U.S. GPU cloud provider, secured a $3.1 billion loan collateralized by GPU assets. For market participants tracking the structural AI infrastructure buildout, this confirms that hyperscaler and GPU cloud capex has not meaningfully reversed — the Seagate caution reflects storage-specific dynamics rather than a broad withdrawal from AI infrastructure spending.

This matters for assessing Korean exposure to the AI supply chain. The thesis for substrate and MLCC manufacturers supplying into AI server builds remains structurally intact; the question is near-term timing and whether foreign institutional flows stabilize.

Pearl Abyss: A Rare Bright Spot in a Down Market

Pearl Abyss (263750.KS), the Korean online game developer behind Black Desert Online, gained 2.62% to ₩47,000 — bucking the broad market decline. Foreign and institutional flows were both modestly positive, a relatively rare combination on a day when most names saw indiscriminate selling. The stock has underperformed over the trailing five days (-10.98%), and it remains below its 20-day moving average.

The near-term focus for Pearl Abyss watchers is a scheduled investor relations event on May 21. Pre-IR sessions typically attract thesis-monitoring flows, but the more meaningful signal will be the market’s reaction after the IR rather than positioning ahead of it.

SK Telecom: SEC Filings Land, No Immediate Catalyst

SK Telecom (017670.KS), South Korea’s largest wireless carrier and a significant player in AI and data center infrastructure, fell 2.79% to ₩97,700. New SEC 6-K and 20-F filings were detected through regulatory monitoring systems. However, an initial review found no new earnings data or shareholder return announcements that would materially change the investment case in the near term. The stock sits near its 20-day moving average after a 5.51% decline over the trailing five sessions.

Foreign investors were actually modest net buyers of SKT on Tuesday (₩13.9 billion), while domestic institutions sold. The mixed flow on a down day is worth noting but not yet actionable.

Screener Picks: Padu Stands Out, Jusung Pulls Back

Two names emerged from quantitative screens on Tuesday:

Padu (440110.KS), an AI storage semiconductor fabless company, rose 6.98% with foreign and program buying both positive — one of the few names to post gains on strong volume in an otherwise weak session. Its relative strength score sits at 98.6, placing it at the top of the screener rankings. The risk is a stretched RSI in the 70s and extended price action following the move; technicians tracking the name are watching for a pullback that maintains foreign buying interest before adding exposure.

Jusung Engineering (036930.KS), a domestic semiconductor equipment maker, ranked first in the KR operational screener with a relative strength score of 98.3 — but fell 2.91% on Tuesday with foreign and program selling. The ranking reflects longer-term momentum rather than today’s session, and analysts suggest waiting for selling pressure to stabilize before treating it as actionable.

Macro Headwinds: The Three Variables to Watch

Three macro factors compressed Korean equity multiples on Tuesday and remain the key variables going forward:

USD/KRW at ₩1,507: Won weakness raises import costs and pressures foreign-currency denominated earnings comparisons. A move back below ₩1,500 would be a constructive signal for multi-asset Korean exposure.
U.S. 10-year yield at ~4.6%: High-multiple growth names — game developers, semiconductor equipment, advanced materials — are most sensitive to discount rate pressure at this level.
Oil prices: Iran nuclear negotiation expectations provided some intraday relief, but energy prices remain elevated enough to sustain cost pressure on industrial and logistics-heavy names.

Key Levels to Watch on May 20

For market participants tracking Korean equities into Wednesday’s session:

Samsung Electronics: ₩275,500 as near-term support; a recovery toward ₩281,000 would signal stabilization of foreign selling pressure
Samsung Electro-Mechanics: Whether ₩1,000,000 is reclaimed; extended time below that level weakens the near-term technical picture
Daeduck Electronics: ₩134,300 support; recovery toward ₩140,400 would suggest Tuesday’s weakness was a one-day event
Padu: Whether foreign/program buying persists on a pullback following the 6.98% session gain
Macro: USD/KRW re-testing the ₩1,500 handle; U.S. 10-year yield stability; further oil price moderation

Bottom Line

Tuesday’s KOSPI session was a broad-based risk-off move driven by a combination of storage sector caution from the U.S., macro rate and currency headwinds, and concentrated foreign selling in large-cap semiconductor names. The structural case for Korean AI infrastructure exposure — semiconductors, substrates, MLCC, equipment — remains intact based on continued hyperscaler capex signals. But the near-term setup calls for patience: foreign flow stabilization and price level recovery are prerequisites before the market’s technicals support adding new positions in the affected names.

Defense, power infrastructure, and select AI-adjacent names held relative strength in Tuesday’s session. That rotation, combined with the macro variables above, will define whether Wednesday offers a stabilization session or a continuation of the current defensive posture.

Korea Quality Re-Rating Watch 2026-05-19: EOTechnics, Device, Gigabis — Quality Screeners Hold Through a Risk-Off Sell-Down

Tue, 19 May 2026 16:30:00 +0900

Section 1 — Macro Dashboard

Indicator	Level	5-Day Change	Signal
KOSPI	7,271.7	−7.3%	Bear
KOSDAQ	1,084.4	−7.9%	Bear
USD/KRW	1,508	+1.0%	Weak KRW
VIX	18.0	+0.5%	Stable
US 10Y	4.62%	+16 bps	Rising
Brent	$110.9	+4.9%	Elevated
DXY	99.2	+0.4%	Flat

Regime verdict: Korea Bear / US Neutral. The KR–US divergence is the defining tension: US breadth is holding (50MA breadth 47.2%), but Korean breadth has collapsed to 30% above the 50-day MA. The macro headwind — USD/KRW above 1,500, US 10Y at 4.62%, and Brent near $111 — is compressing multiples on growth and semiconductor names simultaneously.

Section 2 — Market Wrap

Tuesday’s Korea session was a broad risk-off sell-down with selective pockets of strength. KOSPI closed near 7,304 (−2.81% intraday) and KOSDAQ at 1,084 (−2.40%). The headline flow number: foreign investors sold a net ₩4.83 trillion on KOSPI alone — one of the heavier single-day outflow readings in recent weeks.

What drove the weakness. Three forces converged. First, macro pressure: USD/KRW held above 1,507, the US 10-year pushed 4.62%, and oil prices remained elevated despite tentative Iran negotiation headlines offering partial relief. That combination hits high-multiple growth names hardest. Second, semiconductor-specific headwinds: Seagate’s CEO signalling caution on capacity expansion triggered a negative read-across into memory and storage equipment names globally, which transmitted to Korea’s semiconductor supply chain. Third, the broader Bear regime in Korea is filtering capital away from domestic risk assets.

Sector scorecard. The session’s clear winners were defense stocks and power/energy infrastructure — both sectors where domestic policy demand and long-cycle capex are less sensitive to global rate moves. Some AI infrastructure discovery names also held up. Losers were concentrated in large-cap semiconductors, high-multiple robotics and equipment names, and brokerages.

Flow nuance. Institutional buyers provided partial defense on select large-cap names, absorbing foreign selling at the margin. However, the split — foreign selling vs. institutional buying — is not yet stabilising enough to call a floor. The operating screener environment remains formally BULL on a rules basis (Discovery score 65/100, Day 20), but the intraday picture argues for selectivity over breadth.

Context note. Brokerage research files (Hana, NH, Mirae) were not available today. The regime assessment uses intraday data and screener signals; confirmed closing data was not filed by the time of this briefing.

Section 3 — Today’s Quality Re-Rating Candidates

Today’s meta screener ranked 119 tickers across a 5-layer framework — Quality Compounder, Smart Money Quality, Cycle Rerating, Smart Money Earnings, and PEAD — against official DART filings and Kiwoom market surface data. Candidates with 3+ screener overlaps are the primary focus; 2-overlap names back the thesis.

Top Candidates Table

Meta Rank	Ticker	Name	Meta Score	Screeners Hit	Key Metrics
5	039030.KQ	EOTechnics	57.8	QC + SMQ + CR + SME + PEAD (4 of 5)	OP YoY +158.8%, OPM +11.5pp, F+QI net buy +₩8.6B (5d), RS 91
1	187870.KQ	Device	62.7	QC + CR + PEAD	OP YoY +311.9%, OPM +11.7pp, ROE 11.1%, RS 88
3	420770.KQ	Gigabis	59.6	QC + CR + Consensus Up	OP YoY +777.2%, OPM +29.9pp, consensus z +1.50, RS 95
4	356860.KQ	TLB	59.4	QC + CR + PEAD	OP YoY +664.7%, ROE 16.0%, RS 98; caution: short ratio 8.2%
11	003230.KS	Samyang Foods	46.4	SMQ + SME + PEAD	OP YoY +52.1%, ROE 31.0%, Rev YoY +36.1%, RS 74
12	079550.KQ	LIG Defense & Aerospace	46.3	SMQ + SME	OP YoY +39.0%, smart money net buy, 3 DART catalysts, RS 91
9	147830.KQ	Jeryong Industrial	49.2	QC + CR	OP YoY +158.5%, OPM +13.9pp, ROE 13.9%, RS 81
6	000660.KS	SK Hynix	57.0	QC + CR + Consensus Up	ROE 35.6%, OP YoY +101.2%, consensus z +1.38; caution: 5d foreign+QI net sell −₩10.35T

Top 3 — Brief Context

EOTechnics (039030.KQ) — Meta rank 5, overlap rank 1. EOTechnics manufactures laser processing systems used in semiconductor wafer dicing and PCB drilling — direct exposure to the advanced packaging and AI chip supply chain. It is the only name today hitting all four non-quality screeners simultaneously: Smart Money Quality (#1), Cycle Rerating (#7), Smart Money Earnings (#1), and PEAD (#14). The operating leverage story is concrete: revenue grew 18.7% YoY while operating income grew 158.8%, expanding OPM by 11.5 percentage points. Foreign and quality-institutional net buying totalled +₩8.6B over the past five days — unusual in today’s otherwise foreign-selling market. The meta screener ranked it #5 (below pure Quality Compounder names) because it lacks the Quality Compounder flag, but the money flow and earnings combination is the strongest in today’s universe. Key risk: short ratio at 9.5%, program selling pressure, and a high trailing PER of 100x that requires sustained earnings delivery to justify.

Device Co. (187870.KQ) — Meta rank 1, overlap rank 3. Device makes special-purpose machinery for semiconductor and display manufacturing processes. It tops the meta screener on the strength of operating leverage: OP YoY +311.9% on revenue growth of +78.2%, with a margin expansion of +11.7 percentage points. A merger-related DART filing (2026-05-18) introduces corporate structure change risk that warrants verification before treating the name as a clean re-rating candidate. PEAD score is solidly positive (+0.68), suggesting post-earnings drift momentum has not yet been fully priced in. The stock sits 32% below its 52-week high, providing re-rating runway.

Gigabis (420770.KQ) — Meta rank 3, overlap rank tied. Gigabis produces automated optical inspection (AOI) equipment used in PCB and panel manufacturing — a quality-control bottleneck that benefits as board complexity increases for AI server builds. The earnings acceleration is the most extreme in today’s table: OP YoY +777.2% with a margin expansion of +29.9 percentage points from a low base. Consensus estimates are being revised upward (composite z-score +1.50), making this a classic quality re-rating setup where analysts are catching up to reported numbers. The trailing PER of 98x is high, reflecting the growth optionality priced in. Institutional buying is present, but foreign and program flows are net negative — suggesting positioning is still concentrated domestically.

All financial data sourced from today’s screener runs (2026-05-19). DART filing checks cover a 21-day lookback window. This post is market analysis only and does not constitute investment advice.

iComponent — Not a 'Display Materials Stock' but a 'Barrier Film Mix Turnaround + Perovskite Option Play'. What PER 6x Really Means

Tue, 19 May 2026 11:30:00 +0900

📚 Related Series Pamicell Supply-Demand Analysis / TLB SOCAMM Module PCB Analysis / SK Hynix AI Memory Architecture Shift / May 17 Market Overview

Bucketing iComponent as a ‘display parts stock’ gets the thesis wrong from the start. In 1Q26, revenue was ₩9.74B, with COATED PET (barrier coating film) at ₩8.06B — 82.8% of total. The company’s identity has already changed. FY2025 revenue of ₩34.65B (YoY -6.6%) produced operating profit of ₩6.04B (+132.6%) and OPM of 17.4% — a clear mix-improvement signal where profits surged despite falling sales. Market cap ~₩42.1B, TTM PER ~6.0x. On raw numbers alone, this is obvious undervaluation. The core question is whether ‘COATED PET at 80% mix + OPM in the 17% range repeats in 2Q26.’

Key Summary

iComponent is a company that resists a single-keyword label. In the past it was a display parts stock centered on PC and PMMA optical films, but in 1Q26 COATED PET (barrier coating film) accounts for ₩8.06B — 82.8% of the ₩9.74B in total revenue. The implication is that the company’s identity has already been restructured into ‘barrier coating film + flexible electronic materials + next-generation solar cell option play.’

FY2025 results make this clearest. Revenue was ₩34.65B, down 6.6% YoY, yet operating profit was ₩6.04B, up 132.6%, with OPM of 17.4%. 1Q26 repeated the identical structure: revenue ₩9.74B, OP ₩1.66B, OPM 17.0%. A pattern in which profits surge even as revenue declines signals not a ‘simple cyclical recovery’ but a ‘structural shift in product mix.’

Market cap ~₩42.1B, TTM net income ~₩7.0B, TTM PER ~6.0x. PBR ~0.95x. On raw numbers alone, this is obvious undervaluation. Base scenario (2026E revenue ₩39.5B, OPM 17.5%, PER 7.5x applied) target price ₩7,500, +26.8% upside. Bull (PER 9x) ₩11,100 (+87.3%), Bear (OPM 15.5%, PER 6x) ₩5,000 (-15.7%).

However, four unverified variables exist. First, no formal sell-side target-price report within 90 days. Second, revenue breakdown by customer and product is undisclosed. Third, FX effects cannot be isolated from the 2025 profit improvement. Fourth, the direction of the treasury share disposal plan to be decided at the June 29 EGM (cancellation vs. disposal).

The real option value lies in expansion across four application layers: ESL/e-paper, perovskite/flexible solar cells, flexible OLED substrates, and battery/energy devices — the areas where business purposes will be added at the June 29 EGM. Whether those added purposes translate into actual revenue is the binary test.

Verdict: Watchlist → Conditional Buy. First entry at ₩5,400–5,700 on pullback or after closing above ₩6,150, with full conviction entry after 2Q26 confirms revenue ₩10B+, OPM 17%+, and COATED PET mix at 80%+. It belongs in the same category as the Pamicell thesis — a ‘small-cap materials reclassification gap’ — but market cap, liquidity, and coverage are all smaller, making it more volatile.

1. The Fundamental Shift in Corporate Identity

Bucketing iComponent as a ‘display parts stock’ means the first button is already misaligned. In 2020, PC and PMMA each accounted for 31.9% and 32.2% respectively, making it an optical film company where those two products covered over 64% of revenue. As of 1Q26, PC LGP Film is 8.6% and PMMA is 3.7% — combined just 12.3%. Meanwhile COATED(PET) has exceeded 80%. In under five years, the revenue composition has been completely inverted.

Product	1Q26 Revenue	Mix	Primary Use
COATED(PET)	₩8.057B	82.8%	Barrier film, ESL/e-paper, flexible solar cells
PC LGP Film	₩0.839B	8.6%	Light guide plates, optical film
COATED(PC/PMMA)	₩0.427B	4.4%	Functional coating film
PMMA	₩0.362B	3.7%	Optical / display materials

Understanding what COATED(PET) is forms the starting point of the thesis. The name is simple, but the substance is a ‘functional barrier film that blocks moisture and oxygen.’ Next-generation electronics like OLEDs, e-paper, and perovskite solar cells offer thin, lightweight, and flexible advantages, but are vulnerable to moisture and oxygen. A transparent shield is needed to protect the sensitive electronic materials inside from the external environment — that is a barrier film.

The legacy PC/PMMA optical film business is captive to the display cycle and faces heavy Chinese low-cost competition pressure. In the 1Q26 report, the company cited softening demand for extruded film and competitive pricing pressure from Chinese manufacturers. That is, the legacy business faces a structural headwind — and that is precisely why 2025 revenue fell 6.6%.

By contrast, the rising COATED PET mix has pulled margins higher. OPM jumped from 7.0% in 2024 to 17.4% in 2025, a swing of over 10 percentage points. A structure in which an OPM surge overwhelms a revenue decline is a clear ‘mix-improvement signal.’

2. 1Q26 Results — What a Repeat 17% OPM Means

1Q26 results suggest that the structural shift seen in 2025 is not temporary. Revenue ₩9.74B (YoY +42.7%), operating profit ₩1.66B (YoY +249.4%), net income ₩1.96B, OPM 17.0%. This is nearly identical to the full-year 2025 OPM of 17.4%.

Period	Revenue	Operating Profit	OPM	Notes
2023	₩27.34B	-₩2.14B	-7.8%	Cycle / mix deterioration
2024	₩37.09B	₩2.59B	7.0%	Recovery
2025	₩34.65B	₩6.04B	17.4%	Revenue decline + profit surge
1Q26	₩9.74B	₩1.66B	17.0%	Structural confirmation

Cross-check: 2025 OPM = ₩6.04B / ₩34.65B = 17.4%; 1Q26 OPM = ₩1.66B / ₩9.74B = 17.0%. Simple annualization (1Q26 × 4) gives revenue ₩38.96B and operating profit ₩6.64B, both above 2025 levels.

Why this matters: the probability that ‘17% OPM is structural rather than one-time’ is steadily rising. If FX effects or one-time project revenue were the primary cause, we would have seen meaningful swings between quarters. The fact that FY2025 and 1Q26 showed OPM at similar levels is circumstantial evidence that the product-mix change is real.

The definitive verification, however, is 2Q26 results. If 2Q26 also repeats revenue ₩10B+ and OPM 17%+, the thesis gains full confirmation. Conversely, if OPM retreats to the 13–14% range, the interpretation solidifies that much of the 2025–1Q26 strong results were driven by FX and one-time variables.

The company cited both ‘increased optical film product sales and KRW depreciation’ as factors behind the 2025 operating profit improvement. The inability to fully isolate the FX effect is the largest unverified variable in the thesis. If KRW appreciates, a portion of 2025–1Q26 profits could be given back.

3. Valuation — What PER 6x Really Means

As of 11:01 KST on May 19, share price ₩5,950, shares outstanding 7,071K, market cap ~₩42.1B. Excluding treasury shares of 611,620, the EPS-denominator share count is 6,459K.

Based on trailing four quarters (2025Q2–2026Q1) combined net income of ~₩7.00B:

TTM PER = Market cap ₩42.07B / TTM net income ₩7.00B = 6.0x
PBR = Market cap ₩42.07B / 2025 total equity ₩44.48B = 0.95x

Rounded to one decimal place. Clearly in undervaluation territory. Below half the KOSPI/KOSDAQ average PER, and a PBR below 1.0x means trading below asset value.

Target prices across three scenarios:

Scenario	2026E Revenue	OPM	Operating Profit	Net Income	EPS	PER Applied	Target Price	Upside
Bear	₩36.5B	15.5%	₩5.66B	₩5.4B	₩836	6.0x	₩5,000	-15.7%
Base	₩39.5B	17.5%	₩6.91B	₩6.5B	₩1,006	7.5x	₩7,500	+26.8%
Bull	₩43.0B	20.0%	₩8.60B	₩8.0B	₩1,239	9.0x	₩11,100	+87.3%

P/B cross-check is consistent. Estimated 2026 year-end equity ~₩49.8B (= 2025 ₩44.48B + 2026E net income ₩6.5B − dividend ₩0.97B). BPS = ₩49.8B / 7,071K shares = ~₩7,053. Applying PBR 1.0–1.1x yields ₩7,050–7,760, nearly identical to the PER-method Base of ₩7,500.

The interpretation has two branches. PER 6x is either a signal that the market views the 17% OPM as one-time, or a signal that the market has not yet recognized the company’s identity change. Either way, 2Q26 results are the binary test for the thesis.

One important criticism deserves mention here. There is no formal sell-side target-price report within 90 days. The most recent report shown on the consensus page is a 2023 IBK note. With a market cap of ₩42.1B and daily average turnover of ~₩1.13B, institutional coverage is effectively absent. This is double-edged: on one hand it is a signal that the market does not know this company, a potential source of alpha; on the other, weak price-discovery mechanisms mean results may take longer to be reflected in price.

4. Four Application Layers — Where the Real Option Value Lies

The value of barrier coating film lies in simultaneous applicability across multiple next-generation electronics, not in a single end market. Organizing the four application layers the company presents by their probability and timing of revenue realization reveals that each is different.

ESL/e-paper — the most realistic revenue base. Electronic shelf labels (ESL) and electronic paper (e-paper) are low-power, thin, lightweight displays spreading rapidly in retail, logistics, and industrial environments. Leading global ESL player VusionGroup posted 1Q26 revenue of €289M (YoY +34%) with a 2026 full-year growth guidance of +15–20%, serving 350+ large retailer customers and a 650M+ ESL installation base. Per iComponent IR materials, ESL/e-paper panel companies are cited among major customers, making it likely that a significant portion of COATED PET revenue comes from this segment. However, since customer names and per-customer revenue are undisclosed, the exact share cannot be estimated.

Perovskite / flexible solar cells — the single strongest catalyst. In 2024, the company signed an ₩8.2B barrier film supply contract with a leading European flexible solar cell company — equivalent to 23.28% of the prior year’s revenue, large for a single contract. Interview-based coverage also mentions supply to global perovskite/solar-related companies in the US, Europe, Japan, and China. The reason perovskite matters is simple: conventional silicon solar panels are heavy and rigid, but perovskite-based flexible solar cells can be affixed to building facades, car roofs, drones, satellites, and wearables. If this market genuinely opens up, high-performance film rather than glass will be needed — and iComponent is a direct beneficiary. However, the timeline for perovskite commercialization and durability validation remains an open variable.

Flexible OLED / smartphones / TVs — potential large-scale option. The company cites TVs, flexible smartphones, and flexible OLED substrates as applications for its barrier coating technology. Mass-production adoption by customers, however, is still unconfirmed. If meaningful share in the flexible OLED substrate encapsulation market is secured, revenue could step up to a new level. Competitive variables include UTG (Ultra-Thin Glass) and inorganic thin-film encapsulation as alternative technologies. This segment is best treated as a long-term option with Low–Medium conviction.

Battery / energy devices — new entry territory. The June 29, 2026 EGM agenda includes flexible energy devices, flexible display dedicated components, perovskite/solar cell dedicated components, ESL/e-paper components, battery/energy device components, and nano-scale roll-to-roll materials as new business purposes. This signals the company’s intent to extend its barrier film technology into energy and flexible electronic materials. However, adding a business purpose is merely a charter amendment; whether it translates into actual revenue and orders is the key. Historical precedents show that charter additions often end as simple housekeeping.

Synthesizing these four layers reveals the true depth of the thesis. The market views iComponent as a ‘small-cap optical film stock,’ but the actual profit structure is closer to an ‘option platform manufacturing barrier shields for multiple next-generation electronics.’ The applications differ in timing and intensity, which is why they resist a single category label.

5. Four Unverified Variables — Starting Points for Invalidation

As attractive as the thesis is, unverified variables are clearly present. Laying out four of them reveals the starting points for invalidation.

First, customer and product revenue breakdown is undisclosed. Within the ₩8.06B of COATED PET revenue, it is impossible to decompose how much is from ESL, how much from flexible solar cells, or whether OLED-related revenue actually exists. A structure where a single ₩8.2B supply contract represents 23% of annual revenue means single-customer, single-project risk is high. Sudden revenue volatility could appear in quarterly results.

Second, FX effects cannot be isolated. The company cited both increased optical film product sales and KRW depreciation as factors behind the 2025 operating profit improvement. In other words, some of the 2025 OPM of 17.4% reflects FX. Exactly how much is not disclosed. The fact that 1Q26 repeated OPM at 17.0% is circumstantial evidence against FX being the only cause, but in a scenario where KRW enters the ₩1,300s, some profit give-back should be anticipated.

Third, the June 29 EGM treasury share disposal plan. How the 611,620 treasury shares (~8.6% of total) will be handled is a significant variable. Cancellation means EPS +8.6% and a shareholder-return signal — a multiple expansion factor. Employee compensation or third-party disposal means overhang/dilution risk — a multiple contraction factor. The disclosure after June 29 is a binary signal.

Fourth, execution verification for new business purposes. Whether the four areas being added at the June 29 EGM (flexible energy devices, perovskite dedicated components, ESL/e-paper components, battery/energy device components, roll-to-roll materials) translate into actual revenue is the core of the medium-term thesis. If it ends as a simple charter addition, the market will interpret it as a ’thematic disclosure’ and momentum will fade.

Invalidation conditions can be defined clearly. 2Q26 revenue below ₩9.0B or OPM at or below 14%; COATED PET mix falling below 70%; treasury share disposal interpreted as overhang; new business purposes yielding no material revenue within one year; acceleration of entry by global materials companies (3M, Mitsubishi, etc.) into the high-end barrier market. If two or more of these are met, the thesis warrants re-evaluation.

6. Entry Guide — Chasing Is Inefficient, Scale-In Is the Answer

Verdict: Watchlist → Conditional Buy. As of 11:01 KST on May 19, trading up +5.5% at ₩5,950, with an intraday range of ₩5,540–6,150. Short-term momentum is present, making chasing inefficient. Yet an ultra-small-cap at ₩42.1B market cap has thin volume — once momentum builds it moves quickly, and when it falls it falls faster. A scale-in approach is essential.

Price Range	Action	Condition
₩5,400–5,700	First tranche buy	Partial macro gate passage
₩5,700–5,950	Hold OK, no new entry	Current range
Break above ₩6,150 + volume confirmation	Consider trend re-entry	After closing above
₩7,000–7,500 reached	Partial Base target profit-taking	Before 2Q26 confirmation
₩7,500–8,500 reached	Additional profit-taking	After Bull scenario entry confirmed
₩11,000 reached	Bull target achieved, active exit	Application expansion + FX tailwind simultaneous

Stop-loss: reduce position on break below ₩5,000; thesis weakens on break below ₩4,500 + 2Q26 OPM below 14%.

Full-conviction entry conditions are clear. 2Q26 revenue ₩10B+, OPM 17%+, COATED PET revenue ₩6.5B+ with mix maintained at 80%. When all four conditions are met simultaneously, the multiple-expansion thesis engages fully. The Bull target of ₩11,100 is only achievable when those four conditions combine with perovskite new supply contracts and a June 29 treasury share cancellation decision.

The seven macro gates from the May 17 market overview post are also preconditions. An ultra-small-cap at ₩42.1B market cap is the most vulnerable to macro shocks. In Scenario C (additional risk-off, 15–20% probability), a -20%+ correction is possible on retail flows alone without any foreign selling. Entering after confirming three to four macro gates is therefore rational.

7. Position in the Series — A New Category Called ‘Small-Cap Materials Reclassification Gap’

Clarifying where this post fits within the series sharpens the thesis comparison. It belongs in the same ‘reclassification gap’ category as Pamicell, but with a different mechanism.

Axis	Pamicell	iComponent
Market Cap	~₩1T	~₩42.1B
1Q26 Revenue	₩36.7B	₩9.74B
1Q26 OPM	35.6%	17.0%
Core Product Mix	Low-k materials 71%	COATED PET 82.8%
Market Reclassification Gap	Bio → AI CCL upstream	Display → Barrier / Perovskite
Coverage	Some (DS)	Effectively none
Volatility	Medium	High
Real Trigger	2Q26 results + supply-demand reclassification	2Q26 results + June 29 EGM

Both companies are small-cap materials names where ‘a market reclassification could expand multiples by a step.’ However, iComponent has a smaller market cap, less liquidity, and less coverage, making it more volatile. Even within the same thesis, position sizing should differ.

In comparison with TLB and Samsung Electro-Mechanics, the categories are different. TLB and Samsung Electro-Mechanics sit on the ‘AI memory architecture change’ spine of the series, while iComponent expands the series’ periphery into the ‘flexible electronic materials + perovskite option’ space. It operates on a separate track from the AI memory thesis, but both are part of the broader ‘Korean materials reclassification’ theme.

Connecting more concretely to other posts: the seven macro gates from the May 17 market overview are the common preconditions for all new entries. The SK Hynix Tech Note’s ‘memory architecture change thesis’ is an unrelated domain, so iComponent provides portfolio diversification. When AI memory names are shaken by macro shocks, iComponent responds to different variables (perovskite, FX, treasury shares), meaning correlation is likely low.

8. Red Team — The Strongest Counter-Arguments

Where is this thesis weakest? Laying out strong counter-arguments honestly:

Macro failure mode. If KRW appreciates, the FX tailwind — which the company explicitly cited as a factor in the 2025 operating profit improvement — weakens. Given that the company specifically noted KRW depreciation as a driver, if KRW enters the ₩1,300s, OPM could retreat to the 14–15% range. When the May 17 market overview’s macro gate (KRW/USD 1,500 break) is cleared, it is short-term favorable for KOSPI recovery, but for iComponent specifically, one must simultaneously consider the possibility of FX effect give-back.

Micro failure mode 1. Customer and product concentration risk is high. COATED PET at 82.8% of 1Q26 mix is a strength, but simultaneously means results could swing sharply with order timing from specific product categories. There is prior precedent of short-term result volatility from ESL value chain production disruptions or customer order delays. A structure where a single ₩8.2B supply contract accounts for 23% of annual revenue creates very high result volatility upon contract completion or delay.

Micro failure mode 2. Global materials company entry risk. If global materials players like 3M, Mitsubishi, or Toray make a full push into the high-end barrier market, iComponent’s advantage could erode quickly. This is the most common failure mode for Korean small/mid-cap materials companies. The company’s moat is ’early entry + customer qualification + mass-production know-how’ rather than permanent monopoly, meaning it could weaken over time.

Perovskite commercialization delay risk. Perovskite solar cells have been validated on efficiency but still face issues with durability, longevity, and manufacturability. If perovskite commercialization slips to 2027–2028, iComponent’s perovskite option value is delayed alongside it.

Liquidity / volatility risk. At ₩42.1B market cap and daily average turnover of ~₩1.13B, institutional buying is difficult to sustain, and once price falls, selling pressure accumulates quickly. A -30%+ correction is possible if macro Scenario C materializes. Position sizing must reflect a volatility tolerance higher than that of the average investor.

Treasury share disposal overhang risk. If the 611,620 treasury shares are distributed to employees or sold to a third party, an overhang of roughly ₩3.6B (₩5,950 × 611,620 shares) could hit the market in the short term. If the June 29 EGM resolution is unfavorable, a near-term -10–15% correction is possible.

Synthesizing these six risks makes clear that this is not a ‘stable blue-chip stock’ but a ‘small-cap materials name with option value embedded in high volatility.’ That must be reflected in position sizing.

9. Frequently Asked Questions

“Isn’t PER 6x an automatic buy?”

Making a buy decision on PER 6x alone would be a mistake. There is a reason the market assigns PER 6x. Skepticism over the sustainability of 17% OPM, customer concentration risk, inability to isolate FX effects, and absence of formal coverage within 90 days — four variables operating simultaneously. Even the Base PER of 7.5x (target ₩7,500) rests on the assumption that 2Q26 results repeat the 1Q26 level. If that assumption breaks, PER reverts to 6x or lower. PER 6x is a price that simultaneously reflects both ‘undervaluation signal’ and ‘market skepticism.’

“Compared to Pamicell, which is more attractive?”

They operate on different dimensions. Pamicell has a market cap of ~₩1T and OPM in the 35% range — higher stability, but multiple expansion upside of roughly +55%. iComponent at ₩42.1B market cap carries higher volatility but offers +87% in the Bull scenario. Risk-reward asymmetry is more pronounced, but so is the risk asymmetry itself. From a portfolio diversification standpoint, holding both may be rational. However, given that the market cap difference is 23x, position sizing should differ accordingly.

“Is the June 29 EGM really that important?”

Three things are decided simultaneously, which makes it important. First, the direction of 611,620 treasury shares (8.6% of total): cancellation means EPS +8.6% plus shareholder-return signaling — a multiple expansion factor; disposal creates overhang concerns — a multiple contraction factor. Second, the new business purpose agenda: if flexible energy devices, perovskite dedicated components, ESL/e-paper components, and roll-to-roll materials are formally added to the corporate scope, the company’s identity officially shifts from ‘optical film company’ to ‘flexible electronic materials platform.’ Third, the shareholder return / capital allocation signal. This is the inflection point for how the market classifies this company. Depending on June 29 outcomes, short-term price volatility could range from -15% to +20%.

“If perovskite fails, what happens to the thesis?”

Perovskite is one of four application layers. Even if it fails, ESL/e-paper and flexible OLED applications remain. However, since perovskite was the single largest option value, failure removes the Bull scenario of ₩11,100 and makes the Base of ₩7,500 the new ceiling. Perovskite commercialization delay itself affects the 2027–2028 option value; it does not entirely collapse the near-term thesis.

10. Final Line

Bucketing iComponent as a ‘display parts stock’ gets the thesis wrong. The accurate identity is ‘a flexible electronic materials + next-generation solar cell option platform restructured around COATED PET barrier coating film.’ The fact that COATED PET accounts for 82.8% of 1Q26 revenue of ₩9.74B is the evidence.

That revenue fell 6.6% in 2025 yet operating profit surged 132.6% and OPM jumped from 7.0% to 17.4% is not the result of a simple cyclical recovery, but of a structural product-mix change. The repeat of 17.0% OPM in 1Q26 strengthens this. Market cap ₩42.1B and TTM PER ~6.0x are obvious undervaluation on raw numbers alone, but there are clear reasons the market assigns this price: skepticism over the sustainability of 17% OPM, customer concentration risk, inability to isolate FX effects, and absence of formal coverage within 90 days.

The real option value lies in expansion across four application layers: ESL/e-paper (most realistic revenue), perovskite/flexible solar cells (2024 European ₩8.2B supply contract, single largest catalyst), flexible OLED substrates (potential large-scale option), battery/energy devices (new entry). These are areas where business purposes will be added at the June 29 EGM, but whether a charter addition translates into actual revenue is the binary test.

Valuation scenarios are clear. Base ₩7,500 (+26.8%), Bull ₩11,100 (+87.3%), Bear ₩5,000 (-15.7%). PER method and P/B cross-check are nearly identical. Risk-reward asymmetry is attractive, but six risks (FX, customer concentration, global materials company entry, perovskite delay, liquidity volatility, treasury share overhang) operate simultaneously.

Verdict: Watchlist → Conditional Buy. First entry at ₩5,400–5,700 on pullback or after closing above ₩6,150; full-conviction entry after 2Q26 simultaneously confirms revenue ₩10B+ / OPM 17%+ / COATED PET mix maintained at 80% — three conditions together. The seven macro gates from the May 17 market overview post are also preconditions. An ultra-small-cap at ₩42.1B market cap is the most vulnerable to macro shocks.

Its position in the series is the same ‘small-cap materials reclassification gap’ category as Pamicell, but with a different mechanism: low-k materials vs. barrier film, market cap ₩1T vs. ₩42.1B, OPM 35% vs. 17%. Portfolio diversification by holding both is possible, but position weighting must be differentiated in line with market cap and volatility. It is an unrelated domain to the AI memory architecture thesis in the SK Hynix Tech Note, so correlation is low and it provides a hedge effect.

The core question is simple. Does 2Q26 repeat the 1Q26 pattern? Is the June 29 EGM treasury share decision a cancellation? Does a follow-on perovskite supply contract emerge? These three binary questions determine the thesis over the next six months. It is likely a good company. The good entry price is a pullback to ₩5,400–5,700. Chasing above ₩6,000 is inefficient. The true meaning of PER 6x is that it is a price where ’the possibility that the market has not yet priced in this company’s identity change’ and ‘rational skepticism about whether that change is real’ coexist.

This post is for research and commentary purposes only and does not constitute investment advice. Performance, price, and supply-demand data are based on company disclosures, market data, and press reports and may vary by timing. Scenarios, target prices, and entry prices represent an analyst framework and are not guaranteed. 1Q26 product-level revenue is based on Investing.com reporting, not the original quarterly report; customer names, per-customer revenue, and isolated FX effect figures are not publicly disclosed. This analysis may be wrong. Data reference date: 2026-05-19 KST.

Pamicell Flow Analysis — Between Doosan and ST Pharm; Daeduck Is Still Far Away

Mon, 18 May 2026 23:40:00 +0900

📚 Pamicell Reclassification Series Part 1: Doosan Electro BG AI CCL Material Re-evaluation / Part 2: AI CCL Material Transition and Industry Cycle / Part 3: 1Q26 Earnings Confirmation / AI Substrate·PCB Hub

On fundamentals, Pamicell is already an electronic materials company. In 1Q26, low-dielectric electronic materials revenue reached KRW 26.0bn — 71% of total revenue — with an OPM in the mid-30% range. Its KRX sector classification was also changed to Electronic Components Manufacturing as of 2026-05-04. Yet looking at the same two-week flow data, Daeduck Electronics gained +22.7% with KRW 64.8bn in net foreign + institutional buying, while Pamicell fell -17.9% on a mere KRW 1.6bn. Price correlation tells the same story: Doosan at 0.597, ST Pharm at 0.507, versus Daeduck Electronics at 0.259. The business has re-rated as an electronic materials name, but the stock remains anchored to biotech inertia. That reclassification gap is the source of alpha.

Key Summary

Pamicell is already a low-dielectric electronic materials company by business structure. In FY2025, low-dielectric electronic materials revenue reached KRW 64.7 billion — 56% of total revenue, up +118% YoY. In 1Q26, that figure rose to KRW 26.0 billion — 71% of total revenue, +57% YoY, with an OPM of ~35%. KRX sector classification was also reclassified on May 4, 2026 from “Basic Pharmaceutical Manufacturing” to “Electronic Components Manufacturing.” In other words, the fundamental transition to an electronic materials company is not in the verification stage — it is effectively confirmed.

However, looking at two weeks of flow data from May 4 through May 18, the market is not yet trading Pamicell as an AI substrate leader. Daeduck Electronics rallied +22.7% with combined foreign + institutional net buying of KRW 64.8 billion, representing 5.9% of trading volume. Pamicell, by contrast, fell -17.9% with only KRW 1.6 billion in foreign + institutional net buying — just 0.6% of trading volume. Price correlation is also telling: Doosan 0.597, ST Pharm 0.507, Kolon Industries 0.421 — while Daeduck Electronics sits at 0.259. Pamicell trades closer to “the group that was sold alongside ST Pharm and Kolon Industries.”

This divergence is the source of alpha. That said, divergence does not resolve on its own. The necessary conditions are: low-dielectric electronic materials revenue of KRW 26.0 billion or more sustained in 2Q26, OPM above 30%, confirmation of co-movement during strength in the AI substrate value chain (Daeduck Electronics, Doosan Electronics, Simtech, TLB, etc.), and foreign + institutional net buying expanding to 3% or more of trading volume.

The current view is Watchlist / Conditional Buy. Chasing new positions is inefficient; existing holdings can be maintained as long as earnings remain intact. Additional buying is best approached in tranches upon confirmation of flow reclassification.

1. Fundamentals vs. Flow Gap

The core of the Pamicell thesis is not “is low-dielectric electronic materials real?” That question has already been answered. The real question is when will the market re-rate Pamicell as an AI substrate upstream materials play and assign it a higher multiple?

Four fundamental indicators all confirm the electronic materials transition. First, revenue mix: low-dielectric electronic materials reached KRW 64.7 billion in 2025, representing 56% of total revenue, rising to KRW 26.0 billion and 71% of the mix in 1Q26. Second, growth: +118% in 2025, +57% in 1Q26. Third, profitability: 1Q26 OPM stands at approximately 35%, a healthy level. Fourth, KRX classification: effective May 4, 2026, the company was reclassified into the Electronic Components Manufacturing sector — the result of a scheduled sector review triggered by the shift in revenue composition.

Yet the stock market is trading it differently. The two-week flow picture from May 4 to May 18 is as follows.

Ticker	Proxy	2W Return	Foreign	Institutional	Foreign+Inst	Foreign+Inst / Turnover
Pamicell	Subject	-17.9%	+KRW 3.0B	-KRW 1.4B	+KRW 1.6B	+0.6%
ST Pharm	RNA/CDMO	-18.8%	+KRW 14.1B	-KRW 24.0B	-KRW 9.8B	-4.8%
Kolon Industries	Chemicals/Electronic Materials	-16.3%	+KRW 44.4B	-KRW 22.9B	+KRW 21.4B	+5.1%
Doosan	Doosan Electronics proxy	-6.6%	+KRW 32.1B	-KRW 87.1B	-KRW 55.0B	-3.7%
Daeduck Electronics	Substrate proxy	+22.7%	+KRW 18.1B	+KRW 46.8B	+KRW 64.8B	+5.9%

The key is the diametrically opposite pattern between Pamicell and Daeduck Electronics. If both were being read as part of the same “AI substrate/PCB value chain,” it would be difficult to see Pamicell fall -17.9% while Daeduck rallies +22.7% simultaneously. Foreign and institutional net buying of +KRW 1.6 billion — just 0.6% of turnover — is not sector-reclassification accumulation; it is stagnant, neutral flow. Daeduck, by contrast, shows a clear accumulation pattern: retail sold KRW 58.5 billion net while foreigners and institutions absorbed KRW 64.8 billion.

Fundamentals have shifted to electronic materials; positioning remains anchored in the weak bio/chemicals cohort. That gap is the central point of this piece.

2. 1Q26 Earnings

Pamicell’s 1Q26 was a record quarter. Revenue came in at KRW 36.7 billion, up +36% YoY. Operating profit was KRW 13.0–13.1 billion, up +56% YoY. OPM was approximately 35.4–35.6%. The key story is revenue mix. Low-dielectric electronic materials revenue reached KRW 26.0 billion, representing 71% of total revenue — up +57% from KRW 16.6 billion in the same period last year.

On an annual basis, the shift is even more pronounced.

Period	Total Revenue	Low-Dielectric Electronic Materials Revenue	Mix
FY2024	KRW 64.8B	KRW 29.7B	46%
FY2025	KRW 114.0B	KRW 64.7B	56%
1Q26 annualized	~KRW 146.8B	Based on 71% share	71%

In just eighteen months, the low-dielectric electronic materials share rose from 46% to 71% — a 25 percentage-point increase. This structural shift goes well beyond the notion of “a biotech company with an electronic materials option attached.” The profit center has fully migrated to electronic materials.

The margin improvement is also difficult to attribute solely to cyclical tailwinds. According to DS Investment Securities analysis, the 35.6% OPM in 1Q26 was driven by a higher share of high-margin curing agents relative to resins. In other words, product mix improvement — not unit price increases — is the key variable. This looks less like a one-time pricing effect and more like a structural step-up in the margin floor.

The electronic materials supply agreement signed with Doosan Electronic BG on May 8 further validates the reality of 2Q26 demand. The contract is valued at approximately KRW 8.07 billion, equivalent to 7.08% of recent revenue, covering the period from May 7 to July 31, 2026. While the single-contract value represents only a portion of quarterly revenue, what matters is that AI CCL demand is actually translating into firm purchase orders.

3. Supply & Demand: May 4–18

A closer look at flow data makes the market’s perception clearer. Over the same two-week period, the divergence between foreign and institutional behavior sends a stock-by-stock message.

Pamicell’s detailed flows: securities firms −₩1.1bn, investment trusts +₩0.1bn, private equity −₩1.0bn, pension funds +₩0.6bn. Pension funds provided only modest defense against selling by securities firms and private equity. There is no trace of systematic long-term institutional accumulation. Foreign buying of +₩3.0bn is meaningful but, at roughly 1% of traded value, falls short of a full-scale entry.

Daeduck Electronics is the mirror image. Securities firms +₩7.7bn, investment trusts −₩10.6bn, private equity +₩65.8bn, pension funds −₩13.1bn. The +₩65.8bn in private equity is the key figure — a clear event-driven / momentum buying pattern — and foreign buying of +₩18.1bn joined in. The market’s conviction capital is unambiguously flowing toward Daeduck Electronics.

ST Pharm and Kolon Industries also provide useful comparisons. ST Pharm saw foreign buying of +₩14.1bn overwhelmed by institutional selling of −₩24.0bn, leaving a net foreign + institutional balance of −₩9.8bn — a sell-side bias of −4.8% relative to traded value. Kolon Industries attracted strong foreign buying of +₩44.4bn against institutional selling of −₩22.9bn, producing a net foreign + institutional balance of +₩21.4bn (+5.1% of traded value). Superficially that looks strong, yet the share price fell −16.3% — meaning foreign buying was insufficient to support the price.

Summing up Pamicell’s flow positioning: it is neither as clearly sold as ST Pharm nor as forcefully bid by foreigners as Kolon Industries — it sits in a stagnant neutral zone. The most telling signal remains the diametrically opposite pattern versus Daeduck Electronics. If both names were genuinely grouped within the same AI substrate value chain, the two stocks should have moved together. The fact that they did not is the clearest indicator of where market perception currently stands.

4. Price Co-movement

Return correlations are the most objective indicator of how the market categorizes a stock. The table below summarizes Pamicell’s daily return correlations with each peer.

Peer	Correlation
Doosan	0.597
ST Pharm	0.507
Kolon Industries	0.421
Daeduck Electronics	0.259

The highest correlation is with Doosan, but Doosan is a holding company / conglomerate proxy with significant noise. The more meaningful signal is ST Pharm 0.507 vs. Daeduck Electronics 0.259. This means Pamicell is moving nearly twice as closely with the RNA/CDMO category as with the AI substrate category.

Interpreted qualitatively, the market currently perceives Pamicell in the following order of association.

Rank	Market Perception	Interpretation
1	Doosan Group play	Doosan Electronics proxy
2	RNA/bio materials	Nucleoside · RNAi optionality
3	Chemical/electronic materials hybrid	Specialty chemicals
4	AI substrate upstream materials	Still the weakest signal

In other words, AI substrates are a fourth-tier association, and the −17.9% correction reflects the primary drag from weakness across categories 1–3.

For the market to meaningfully reclassify Pamicell as a Daeduck Electronics-type AI substrate upstream materials play, the following data points would need to shift. The correlation with Daeduck Electronics would need to rise above 0.4–0.5, while the correlation with ST Pharm would need to fall below 0.3. A recurring pattern of Pamicell posting +1–3% gains on days when Daeduck Electronics, Doosan Electronics, Simtec, and TLB rally would also need to emerge. That pattern has not yet formed.

5. Why Bio Inertia Persists

KRX sector reclassification is an administrative change, not an immediate shift in market perception. Although the reclassification to electronic components manufacturing took effect on May 4, market perception adjusts gradually through the following sequence:

KRX sector change
→ Portal/HTS sector update
→ News tag change
→ Research coverage sector change
→ Institutional basket change
→ Flow correlation shift

Pamicell is currently in the middle of this transition.

First, corporate identity inertia remains. In official press releases, Pamicell still describes itself as “a biopharmaceutical company that developed the world’s first stem cell therapy.” This reflects the coexistence of its Biomedical Division and Biochemical Division. As long as the company defines itself as both a bio and a chemical company, the market will see it that way too.

Second, the RNAi API optionality is real. In November 2025, the company obtained a patent for the manufacturing method of 2’-F-dU, a key raw material for RNAi-based gene therapies, and explicitly stated its intention to expand the nucleoside product line and strengthen its supply base in line with growth in the global RNAi therapeutics market. In other words, bio-related flows are not entirely irrational — though in terms of current earnings contribution, electronic materials remain the overwhelming core.

Third, data vendors and HTS theme tags change slowly. Some data vendors still describe Pamicell as a “company engaged in bio-pharma and chemical businesses,” and legacy bio theme tags such as Moderna and mpox remain attached. These tags influence short-term flows and retail investor search behavior, and they also affect algorithmic trading baskets. A full market perception reset following an administrative reclassification typically takes six to twelve months.

6. Position in the AI Substrate Value Chain

Viewing Pamicell as an AI substrate play is imprecise. The accurate description is AI substrate upstream materials supplier. A simplified view of the value chain is as follows.

AI server / GPU / ASIC demand
→ High-speed signal transmission required
→ Demand for low-loss, low-dielectric CCL
→ CCL manufacturers such as Doosan Electronics benefit
→ Demand for low-dielectric resin and curing agent materials
→ Pamicell benefits

This position carries two implications.

First, there is no need for Pamicell to move in lockstep with PCB and substrate manufacturers such as Daeduck Electronics. PCB manufacturers sit at the finished-product stage, while Pamicell operates two tiers upstream at the materials level. Both the timing lag and the ASP mechanism differ accordingly.

Second, however, if the market expands its lens to encompass the entire AI substrate value chain, the correlation with Daeduck Electronics, Doosan Electronics, Simtech, and TLB should rise meaningfully. The current reading of 0.259 signals that this broadening of perspective has not yet occurred.

There are five choke points.

Bottleneck	Significance
Customer qualification	Process certification by end customers is critical for high-speed, high-frequency materials
Material formulation	Resin and curing agent composition and quality consistency
Process reproducibility	Maintaining uniform physical properties at mass-production scale
Capacity	Expansion timing is the key earnings lever when demand accelerates
Dual-sourcing risk	Multiple discount applied if customers secure alternative suppliers

DS Investment & Securities indicated that Doosan’s Electronics BG has hinted at additional demand upside, and projected that Pamicell could expand production volume by approximately 20–30% versus the current level through process improvements. Plant 3 is estimated to be completed in September 2026, with operations commencing in 2027. In other words, the current bottleneck is capacity, not cost reduction. If the expansion enters a full ramp-up from 2027, revenue could step up to the next level, at which point sustaining OPM becomes the single most critical validation variable.

7. Three Market Misjudgments

The first misjudgment is “Pamicell is still a biotech stock.” This is half right, half wrong. The company’s history and its RNAi option remain intact, but the center of earnings has already shifted to low-dielectric electronic materials. In 1Q26, electronic materials likely account for 71% of revenue and an even higher share of operating profit, given that a segment with ~35% OPM dominates the mix. The perception of “still a biotech” is being pulled by the inertia of the company’s legacy identity, and that perception could gradually unravel over the next 6–12 months.

The second misjudgment is “electronic materials growth has not yet been validated.” This one is simply wrong. With low-dielectric electronic materials revenue up +118% in 2025, +57% YoY in 1Q26, and OPM in the 35% range, both growth and profitability are already substantially validated. What still needs validating is not the growth itself, but how long it can be sustained and whether margins hold after capacity expansion — a meaningfully different question.

The third misjudgment is “Pamicell has to move in lockstep with Daeduck Electronics to count as an AI substrate play.” This is also an oversimplification. Pamicell is an upstream materials company, not a PCB manufacturer; there is no reason it must track Daeduck one-for-one. That said, whether the market re-classifies it as part of the broader AI substrate value chain can be confirmed by watching its co-movement with Daeduck Electronics, Doosan Electro-Materials, Simtech, and TLB. The current correlation of 0.259 is a signal that re-classification has not yet occurred — not that it will never happen.

Because all three misjudgments operate simultaneously, Pamicell’s multiple is stuck in an awkward middle ground — neither a true biotech multiple nor a true AI substrate multiple. Which category market perception ultimately converges on will determine the direction of the multiple over the next 6–12 months.

8. Valuation

Based on the May 18 closing price of KRW 16,630 and 60,017 thousand shares outstanding, market capitalization stands at approximately KRW 998.1 billion. Using DS Investment & Securities’ 2026F estimates of KRW 160.2 billion in revenue, KRW 57.7 billion in operating profit, 36% OPM, and EPS of KRW 987, the PER calculation is as follows.

2026F PER = KRW 16,630 / KRW 987 = 16.9x
Market cap / 2026F OP = KRW 998.1B / KRW 57.7B = 17.3x
DS target price KRW 26,000 implied PER = KRW 26,000 / KRW 987 = 26.3x

The implication is straightforward. The stock currently trades at roughly 17x 2026F PER. Compared to representative multiples in the AI substrate value chain — Daeduck Electronics at 27x PER, Simtech at 20x PER, and Samsung Electro-Mechanics at 40x 2027E PER — the discount is clear. At the same time, it carries a modest premium over the 12–15x PER average for chemical and materials stocks. In other words, the market is pricing Pamicell as “chemicals plus a small AI option.”

The DS target price of KRW 26,000 requires 26x 2026F PER — a multiple in line with AI substrate peers. Moving from the current 17x to 26x requires the market to reclassify Pamicell from a specialty chemicals stock to an AI CCL upstream materials play. If that reclassification occurs, the share price could rise +55% with no change in EPS whatsoever — upside driven purely by multiple expansion.

That said, multiple expansion does not happen on its own. A catalyst is needed. The single strongest catalyst would be 2Q26 results confirming both low-dielectric electronic materials revenue of KRW 26 billion or above and OPM of 30% or above simultaneously. If confirmed, it would prove that 1Q26 was not a peak but a base, and analyst coverage framing would shift accordingly.

9. Entry Guide

The stance is Watchlist / Conditional Buy. Chasing the stock at current levels is inefficient. Existing positions can be held until fundamentals deteriorate. Adding to positions is best approached in tranches once supply-demand reclassification is confirmed.

The five conditions for a tranche-buy are as follows.

Condition	Threshold
2Q26 low-dielectric electronic materials revenue	Maintained at KRW 26.0B or above
2Q26 OPM	Maintained at 30% or above
AI substrate value-chain co-movement	Rebounds alongside Daeduck Electronics, Doosan Electro-Materials, Simtech, and TLB on at least 3 up-days
Foreign + institutional net buying	At least 3% of daily turnover, sustained for 3 or more trading days
Decoupling on biotech down-days	Relative strength maintained even on weak days for ST Pharm and other RNA/CDMO names

If 3 or more of the 5 conditions are met, increasing the position is justified. All 5 met signals an aggressive buy zone.

By price level:

Price Range	Action	Condition
Below KRW 14,000–15,000	Aggressive tranche buying	PER 14–15×; majority of macro gates cleared
KRW 15,500–16,500	Hold OK / chasing is inefficient	Current range
Break above KRW 18,000	Trend buy possible	AI substrate co-movement signals start of reclassification
Break above KRW 20,000	Strong trend buy	Foreign + institutional turnover 5%+
Reach KRW 26,000	Begin taking profits	Near DS target price

Stop-loss: reduce exposure on a break below KRW 13,000; view thesis as weakening on a break below KRW 12,000 combined with 2Q26 OP below KRW 11.0B.

Invalidation conditions are equally clear. The thesis warrants re-evaluation if: 2Q26 low-dielectric electronic materials revenue falls to KRW 22.0B or below; OPM drops under 30%; margins are impaired by a shrinking share of high-margin curing agents; orders from Doosan Electro-Materials slow or are dual-sourced; or Plant 3 commissioning and ramp-up are delayed. Additional re-evaluation triggers include Pamicell consistently underperforming while the broader AI substrate value chain (Daeduck Electronics, Doosan Electro-Materials, etc.) is strong, or buy-side interest remaining confined to the biotech narrative while sell-side research coverage of the electronic materials segment fails to expand. If 2 or more of these conditions are met, the thesis must be revisited.

10. Position Within the Series and the Macro Gate

This post is a new chapter in the Pamicell Recognition Transition Series. Where earlier entries built the fundamental thesis for why Pamicell should be reclassified as an AI CCL upstream materials play, this installment uses supply-demand data to verify that the market has not yet made that reclassification. The next entry will follow after the 2Q26 earnings release to track whether the recognition gap has begun to close.

The connections to other series are equally clear. Within the eleven-stock AI back-end processing matrix, Pamicell sits one materials layer above the PCB and substrate names — Daeduck Electronics, Simtech, and HaeSeong DS. Clearing the seven macro gates outlined in the May 17 market overview post is a prerequisite for any new entry. The stocks covered in the NVIDIA Korea 2nd- and 3rd-tier value chain post belong to a different category: those are NVIDIA beta names, whereas Pamicell is an AI CCL upstream material. TLB occupies an adjacent position in the AI memory module space, but the two serve different functions — TLB is a module PCB supplier; Pamicell is a CCL materials supplier.

From a macro gate perspective, Pamicell is currently in Scenario C, meaning it could absorb a significant blow if further risk-off materializes. Two reasons underpin this view. First, as a mid-cap with a market capitalization below KRW 1 trillion, it is more vulnerable to foreign-investor selling pressure. Second, its fundamentals are sound but positioning has stagnated, which means a deteriorating macro environment could apply an additional discount by marking it as a name lacking supply-demand conviction.

Accordingly, rather than aggressively buying at current prices, the rational approach is to build a full position after clearing three to four macro gates and confirming 2Q26 earnings results.

11. Frequently Asked Questions

“Is Pamicell really an AI substrate play?"

That’s the wrong question. The right question is: “When will the market recognize Pamicell as an AI substrate upstream materials company?” By fundamental metrics, it already qualifies. Revenue +71%, OPM 35%, and the KRX sector reclassification are all confirmed. By flow metrics, not yet. Correlation with Daeduck Electronics stands at 0.259; with ST Pharm, 0.507. That gap is the source of alpha. Buying before the market updates its perception captures multiple expansion upside; buying after means you’re simply riding EPS growth.

“Why shouldn’t I buy right now?"

Three reasons. First, the fact that AI substrate multiples are not yet priced in means the timing of reclassification is undefined — it could take six months or eighteen. Second, there is a non-zero probability that 1Q26 represents peak earnings. Conviction in the base case is low until 2Q26 results are verified. Third, buying a mid-cap before the macro gate has cleared produces unfavorable risk asymmetry. That is the rationale for scaling in gradually.

“Does Pamicell have to move in lockstep with Daeduck Electronics to qualify as an AI substrate stock?"

No. Pamicell is an upstream materials supplier, not a PCB manufacturer. The time-lag dynamics and ASP mechanisms differ, so daily co-movement is not required. However, for the market to recognize it as part of the same value chain, the monthly return correlation needs to rise to roughly 0.4–0.5. The current reading of 0.259 is a strong signal that the market does not yet place them in the same category. Quarterly changes in correlation should be tracked closely.

“What happens if 2Q26 results disappoint?"

That is precisely where the invalidation condition sits. If 2Q26 low-dielectric electronic materials revenue comes in at or below KRW 22 billion, or OPM falls below 30%, the market interpretation that 1Q26 was the peak will harden. In that scenario, a PER of 17x becomes the justifiable multiple and the KRW 26,000 price target disappears. The stock could de-rate into the KRW 12,000–14,000 range. Accordingly, 2Q26 earnings are not merely a catalyst — they are a binary test of the entire thesis.

12. One Final Line

By business fundamentals, Pamicell is already a low-dielectric electronic materials company. In 1Q26, low-dielectric electronic materials revenue reached KRW 26.0 billion, representing 71% of total revenue, with an OPM of 35.6% and YoY growth of +57%. KRX sector classification also changed to Electronic Components Manufacturing effective May 4. On a fundamental basis, the transition to an electronic materials identity is not a hypothesis under validation — it is a conclusion already confirmed.

However, the two-week flow-of-funds picture from May 4 through May 18 tells a different story. Daeduck Electronics rose +22.7% and absorbed KRW +64.8 billion in combined foreign and institutional net buying, while Pamicell fell -17.9% and attracted only KRW +1.6 billion. Price co-movement correlations over the same period stand at 0.597 for Doosan, 0.507 for ST Pharm, 0.421 for Kolon Industries, and 0.259 for Daeduck Electronics. The market is still trading Pamicell as an RNA/bio + chemicals hybrid, and has not yet reclassified it as an AI substrate upstream materials play.

This reclassification gap is the source of alpha. A multiple expansion from 17x to 26x PER alone implies +55% upside. Yet the gap does not close on its own. Corporate identity inertia, the tangible optionality of the RNAi pipeline, and lagging data-vendor and HTS tagging all work simultaneously to slow the reclassification process. After an administrative sector reassignment, a full market perception shift typically takes six to twelve months.

The conclusion is therefore clear. Chasing the stock on new positions is inefficient. Existing holders can maintain positions until earnings deteriorate. Additional accumulation is best approached in tranches once three or more of five conditions are satisfied: (1) 2Q26 low-dielectric electronic materials revenue holds at or above KRW 26.0 billion; (2) OPM stays at or above 30%; (3) at least three rallies in tandem with strength in the AI substrate value chain; (4) sustained foreign and institutional net buying exceeding 3% of daily turnover; and (5) decoupling from biotech on weak-bio days. Clearing a majority of the seven macro gates is also a prerequisite.

Pamicell is not “a beaten-down biotech.” It is “a candidate AI low-dielectric materials stock that the market has yet to classify correctly.” Until reclassification is confirmed through fund flows, a conditional accumulation approach is more appropriate than a full-scale position build-up. 2Q26 earnings are the binary test for the thesis. The market’s verdict today is cold. Pamicell sits between Doosan and ST Pharm — and Daeduck Electronics remains a long way off.

This article is intended solely for research and commentary purposes and does not constitute investment advice. Pamicell 1Q26 results (revenue KRW 36.7 billion, operating profit KRW 13.0–13.1 billion, OPM 35.4–35.6%, low-dielectric electronic materials KRW 26.0 billion representing 71% of total, YoY +57%) are sourced from the company’s press release and DS Investment & Securities materials. Full-year 2025 results (revenue KRW 114.0 billion, operating profit KRW 34.3 billion, low-dielectric electronic materials KRW 64.7 billion, YoY +118%, representing 56% of total) are sourced from company materials. The KRX sector reclassification (2026/05/04, from Basic Pharmaceutical Substance Manufacturing to Electronic Components Manufacturing, reason: periodic sector review triggered by shift in revenue-mix composition) is sourced from KIND disclosures. Flow-of-funds data for 5/4–5/18 (Pamicell -17.9%, foreign+institutional net buy KRW +1.6 billion; Daeduck Electronics +22.7%, KRW +64.8 billion; ST Pharm -18.8%, KRW -9.8 billion; Kolon Industries -16.3%, KRW +21.4 billion; Doosan -6.6%, KRW -55.0 billion) are sourced from Naver Finance daily foreign/institutional net trading records and the user-provided Research OS local database, and may vary depending on the reference time. Price co-movement figures (Doosan 0.597, ST Pharm 0.507, Kolon Industries 0.421, Daeduck Electronics 0.259) are daily return correlation coefficients calculated over the same period. The May 18 closing price (KRW 16,630), shares outstanding (60,017 thousand shares), and market cap (approximately KRW 998.1 billion) are sourced from market data. DS Investment & Securities 2026F estimates (revenue KRW 160.2 billion, operating profit KRW 57.7 billion, OPM 36%, EPS KRW 987, target price KRW 26,000) are sourced from DS Investment & Securities and may differ from other brokers’ estimates. The 2026F PER of 16.9x, market cap/OP of 17.3x, and target-price-implied PER of 26.3x are calculated based on the above estimates. The supply contract with Doosan Electronics BG for electronic materials (disclosed 2026/05/08, approximately KRW 8.07 billion, 7.08% of recent revenue, contract period 2026/05/07–07/31) is sourced from company disclosures. The November 2025 acquisition of a patent for the 2’-F-dU manufacturing process is sourced from the company’s press release. The 20–30% production capacity increase from process improvements, the third plant completion in September 2026 with production start in 2027, and the estimated capacity of KRW 200.0 billion are sourced from DS Investment & Securities materials and may differ from official company announcements. The characterization of the company as defining itself as “the world’s first developer of a stem cell therapeutic” is sourced from official company introduction materials. The five accumulation conditions, entry price ranges, and invalidation criteria represent the analyst’s framework and carry no guarantee. The estimated reclassification timeline of six to twelve months and the multiple expansion estimate of +55% are analyst scenarios and may differ materially from actual outcomes. Global macro conditions (U.S. interest rates, oil prices, FX, VIX, foreign investor flows) may exert additional influence on the share price. This analysis may prove incorrect. Data reference date: May 18, 2026 KST.

KOSPI Holds Flat as Samsung Drives Semiconductor Rally

Mon, 18 May 2026 23:30:00 +0900

Samsung Lifts KOSPI While KOSDAQ Slumps 2.2%

South Korea’s benchmark KOSPI index closed effectively flat on May 18, 2026, edging up +0.13% — but that headline number obscures a sharp divergence inside the market. The KOSDAQ, South Korea’s technology and small-cap index, tumbled −2.18%, and internal breadth told a sobering story: only 33.7% of KOSPI stocks were trading above their 50-day moving average, with 46.3% above the 200-day. Today’s session was not a broad rally. It was a large-cap semiconductor rescue mission.

The market regime sits in neutral territory with a risk-off undertone. Strong flows concentrated in semiconductors, high-bandwidth memory (HBM), CPO/photonics, and power infrastructure. Weak areas included machinery, transportation equipment, construction, consumer discretionary (including department stores), and high-beta KOSDAQ names. The pattern is familiar: when macro headwinds tighten, Korean equities bifurcate — global tech leaders hold, domestic cyclicals bleed.

Samsung Electronics: Strike Risk Absorbed, For Now

The day’s defining story was Samsung Electronics (005930.KS), South Korea’s largest semiconductor and consumer electronics conglomerate. The stock surged +3.88% to close at ₩281,000, reversing recent weakness.

The catalyst was not a positive earnings surprise — it was the market’s re-reading of a risk. Samsung’s union announced a potential 18-day strike beginning May 21. A court ruling clarified that production equipment protection obligations remain in force, limiting the practical disruption scenario. The market chose to interpret the news as a mean-reversion opportunity rather than a structural threat, with buyers stepping in at depressed levels.

The caveat is significant: foreign investors net-sold ₩1.24 trillion of Samsung shares today. That scale of outflow from a single stock in a single session is not noise — it signals continued institutional repositioning at the global level, likely tied to macro factors (USD/KRW near ₩1,498, Brent crude around $110, and U.S. 10-year Treasury yields at 4.59%). Samsung’s near-term trajectory hinges on whether today’s buying was durable accumulation or a short-covering bounce that foreign flows will overwhelm again.

The Real Winners: Daeduck Electronics and Samsung Electro-Mechanics

While Samsung grabbed headlines, two names in the semiconductor supply chain delivered cleaner price-and-flow combinations.

Daeduck Electronics (353200.KS), a manufacturer of printed circuit boards (PCBs) critical to advanced semiconductor packaging, posted the most constructive setup of the day. The stock rose +3.39% to ₩140,400, extending a +11.61% five-day gain, with simultaneous net buying from both foreign and domestic institutional investors. Its Relative Strength score reached 98.1. In a market where most stocks fell, Daeduck’s synchronized institutional sponsorship stands out as a signal worth tracking.

Samsung Electro-Mechanics (009150.KS), a global leader in multilayer ceramic capacitors (MLCCs) and camera modules, added +2.08% to close at ₩1,031,000 — a five-day advance of +14.56% with an RS score of 98.4. The stock is in a clear uptrend, though foreign investors were net sellers today while domestic institutions absorbed shares. That divergence warrants watching: the price action is strong, but a trend where only one cohort drives buying is inherently more fragile than broad institutional consensus.

AI Infrastructure Demand: The Thesis Behind the Flows

Why are semiconductor-adjacent names outperforming in a broadly weak market? The demand signal from hyperscalers is becoming more concrete. Reports circulating in Korean financial channels note that Amazon, Microsoft, and Google are increasing their general-purpose server procurement mix in Q2 2026 alongside continued AI data center buildout. Power and electrical infrastructure is tightening in step — LS Cable & System (001120.KS) reportedly secured a supply contract worth approximately ₩4 trillion, underscoring how AI-driven infrastructure investment is reaching deep into Korea’s industrial supply chain.

This connects directly to why CPO (co-packaged optics), HBM (high-bandwidth memory), and advanced PCBs are the market’s strongest themes today. Korean companies are embedded across multiple layers of the AI hardware stack, and the current phase — where hyperscaler capex is translating into actual component orders — is a tangible positive for the sector.

Emerging Candidates: Breakout Signals on the Screener

Today’s screening universe flagged several new names worth monitoring for entry timing.

Jeju Semiconductor (제주반도체) was the only stock classified as a pure breakout candidate in the quantitative screener, posting +12.52% with an RS score of 97.3 and domestic institutional buying. Foreign investors were net sellers, which creates a mixed signal — the price action is constructive, but broad institutional conviction is not yet confirmed.

Laser Cell (레이저쎌), a semiconductor equipment name, surged +23.4% with an RS of 98.2 and positive foreign and institutional flow. The move looks technically extended in the short term; first pullback behavior will be the key test.

In the venture capital and private equity space, Mirae Asset Venture Investment (+22.8%, RS 99.4) and Aju IB Investment (+18.9%, RS 99.1) saw explosive volume, apparently driven by thematic interest in SpaceX/private market exposure. These moves look event-driven rather than fundamental — momentum is real, but sustainability requires confirmation.

The Soft Spot: Pearl Abyss Continues to Underperform

Not every theme is working. Pearl Abyss (263750.KS), the Korean game developer behind Black Desert Online, fell −2.97% on the day and has lost −14.23% over five days, closing at ₩45,800. Foreign investors were modest net buyers, but the price action contradicts that flow — typically a sign that selling from other cohorts (retail, domestic funds) is overwhelming the foreign bid.

Pearl Abyss is holding an investor relations event on May 21. Until then, the market lacks a new catalyst to reverse the downtrend. The key question heading into IR: can management present a credible growth roadmap for Crimson Desert that shifts the market’s narrative from “missed launch timeline” to “next leg of IP monetization”? Without that, the stock risks further compression.

Macro Overlay: Breadth Warns Against Aggressive Positioning

The session’s internal breadth data reinforces caution. A KOSPI where only one-third of stocks trade above their 50-day moving average is not an environment that rewards broad beta exposure. The macro backdrop — USD/KRW near ₩1,498 (Korean won under pressure), elevated energy prices, and U.S. rates anchored at 4.59% — continues to weigh on risk appetite for emerging market equities.

Foreign investors have room to sell Korean equities further at current valuations if the dollar stays strong. Today’s KOSPI resilience rested almost entirely on Samsung Electronics. A market that cannot broaden beyond its largest constituent is fragile, regardless of how strong the semiconductor theme looks in isolation.

What to Watch on May 19

Samsung Electronics: Does the ₩281,000 level hold, and do foreign outflows slow? A second day of heavy selling would undercut the “oversold bounce” read.
Daeduck Electronics: Whether foreign and institutional co-buying extends to a second consecutive session — that would confirm a trend, not just a one-day spike.
Samsung Electro-Mechanics: Ability to hold above ₩1,031,000 while monitoring whether foreign selling abates.
Pearl Abyss IR (May 21): Pre-event sentiment and any preview commentary. The stock needs a catalyst, not just stabilization.
Jeju Semiconductor: Foreign flow direction on day two — a shift to net buying would validate the breakout reading.

The operative mode for Korean equities in the near term is winner confirmation, not aggressive expansion. Semiconductors are doing the heavy lifting. The question is whether that strength can broaden — or whether this remains a market of one sector carrying everyone else.

Pearl Abyss Flow Check — Not a Turn Yet. The ₩45,500 Test Between Foreign Bargain Buying and Institutional Selling

Mon, 18 May 2026 18:20:00 +0900

📚 Pearl Abyss × Crimson Desert Series 1Q Recognized Sales and the 2027 Cliff / Patch 1.07 Weekend Data / May 21 IR Checkpoint / What the DLC Comments Really Mean / Pearl Abyss Hub

Pearl Abyss materially reduced “2Q cliff” concerns with its 1Q26 results and 2Q guidance. Yet the stock kept falling after the earnings release. The reason is straightforward: the market has moved on from 2Q and is now pricing the 2027 earnings cliff, capital allocation, and flow dynamics. The tape over the past two weeks does not yet look like a reversal. Foreigners have partially returned since the May 13 forced selling, but institutions have not. This is a “flow-bottom test,” not a “flow turn.”

Key Summary

Two-week flow is not a turn yet. From May 4 to May 18, 2026, the stock fell from ₩58,500 to ₩45,800, a decline of roughly -21.7%. This reads less like a routine pullback and more like a post-1Q valuation reset.
Both foreigners and institutions were net sellers over the full two weeks. Per Naver Finance, cumulative net selling from May 4–18 was -502,638 shares for institutions and -546,424 shares for foreigners, translating to estimated net consideration of approximately -₩26.4B and -₩30.1B respectively.
That said, foreigners showed a partial reversal after the May 13 panic. Net buying came in at +56,653 shares on May 14, +206,722 on May 15, and +65,936 on May 18. Foreign re-entry at lower prices is a positive.
The institutional breakdown still looks weak. Per the Research OS local DB from May 4–15, institutional selling was led by securities firms (-₩17.3B) and private funds (-₩7.5B). Pension funds and insurers have not stepped in meaningfully; trust funds were only marginally positive at +₩0.9B.
Program trading and short selling have calmed, but loan balance risk remains. Program trading swung from -₩34.5B on May 13 to +₩3.0B on May 14, +₩9.7B on May 15, and +₩3.2B on May 18. Short-selling share fell from 19.26% to the 7% range. However, the stock loan balance continued to rise through May 15.
We are currently in stage 3 of a 4-stage process. Distribution and forced selling are behind us; foreign bargain buying and program normalization have appeared. But institutional re-entry and trend recovery have not.
The practical call is hold and observe; additional buying should wait. A genuine flow turn requires holding ₩45,500–₩46,000, reclaiming ₩47,200, and seeing trust funds, pension funds, and insurers add positions in the institutional breakdown.

Data covers May 4–18, 2026. Price, foreign, and institutional flow data are sourced from the Research OS local DB and Naver Finance Pearl Abyss foreign/institutional page.

1. The Price Action Signals a Post-Event Valuation Reset

Looking at the price alone, the correction is already significant.

Date	Close	Change	Volume
5/04	₩58,500	-0.68%	530K shares
5/13	₩51,900	-1.70%	2.23M shares
5/15	₩47,200	-6.72%	1.34M shares
5/18	₩45,800	-2.97%	740K shares

Over two weeks, the stock moved from ₩58,500 down to ₩45,800.

Decline = ₩45,800 ÷ ₩58,500 - 1
= -21.7%

A drawdown of this magnitude is hard to explain as simple post-earnings profit-taking. More precisely, even after 1Q26 results and 2Q guidance were confirmed, the market re-discounted 2027 earnings durability, DLC specificity, shareholder returns, and institutional flow.

The earnings debate around Pearl Abyss has already moved one chapter forward. The 1Q operating profit of ₩212.1B and the company’s 2Q guidance weakened the “immediate 2Q collapse” narrative. The fact that the stock still fell means the market has shifted to the next question:

2026 numbers look strong. But will earnings hold into 2027, and will those earnings return to shareholders?

This decline is therefore best understood not as a simple price correction, but as a period in which peak-earnings discounting and flow repositioning occurred simultaneously.

2. Cumulative Two-Week Flow Is Still Negative

The cumulative May 4–18 flow from Naver Finance is unambiguous.

Participant	Net Shares	Est. Net Consideration
Institutions	-502,638 shares	approx. -₩26.4B
Foreigners	-546,424 shares	approx. -₩30.1B

Foreign ownership also dropped from 6.86% on May 4 to 5.58% on May 18. Viewed across the full two weeks, both foreigners and institutions were net sellers. Overlooking this and saying only that “foreigners are coming back” would be an overly optimistic reading of the flow.

That said, the tone shifted somewhat over the last three trading sessions.

Date	Institutions	Foreigners	Interpretation
5/14	-67,686 shares	+56,653 shares	Foreign bargain buying begins
5/15	-3,902 shares	+206,722 shares	Strong foreign inflow
5/18	-18,417 shares	+65,936 shares	Foreign buying continues; institutions still selling

The key sentence is this:

Foreigners show signs of returning. Institutions do not yet.

Flow reversals rarely end with foreign buying alone. For a KOSDAQ gaming stock with significant earnings, event, and multiple debates, the reversal typically requires institutional model upgrades and position rebuilding as well. Right now we see the first traces of foreign bargain buying, but no institutional confirmation.

3. The Institutional Breakdown Points to Securities Firms and Private Funds

The Research OS local DB data from May 4–15 sharpens the picture of who is selling.

Sub-category	Net Consideration
Securities firms	-₩17.3B
Private funds	-₩7.5B
Pension funds etc.	-₩1.4B
Insurers	-₩0.3B
Trust funds	+₩0.9B
Institutions total	-₩25.7B
Retail	+₩59.4B

Securities firms and private funds drove institutional selling. Long-duration capital — pension funds and insurers — has not stepped in meaningfully. Trust funds were only marginally positive at +₩0.9B, which is insufficient to call a flow turn.

The structure looks equally poor when isolated to the sharp-decline period.

Sub-category	Net Consideration (5/13–5/15)
Securities firms	-₩4.0B
Private funds	-₩7.0B
Pension funds etc.	-₩1.0B
Institutions total	-₩12.0B
Retail	+₩33.1B

During the May 13–15 plunge, retail absorbed the selling while private funds and securities firms were the sellers. This is not a constructive accumulation structure. A healthy pattern would have retail distributing during a sharp decline while foreigners, institutions — specifically trust funds, pension funds, and insurers — step in. Pearl Abyss is not there yet.

This point matters for the investment thesis. Pearl Abyss’s 2026 earnings are already strong. But a re-rating on 2027 earnings defense and DLC/shareholder return/DokeV roadmap requires institutional models to move first. Without institutional re-entry, even strong 1Q earnings can leave the stock stuck in a “cheap but peak” range.

4. Program and Short Selling Have Normalized; Loan Balance Risk Remains

There are positive elements in the flow picture. Program trading and short-selling intensity have clearly peaked.

Item	Recent Trend	Interpretation
Program trading	5/13 -₩34.5B → 5/14 +₩3.0B → 5/15 +₩9.7B → 5/18 +₩3.2B	Normalized after forced selling
Short-selling share	5/13 19.26% → 5/15 6.90% → 5/18 7.20%	Short pressure has peaked
Stock loan balance	5/12 6.10M shares → 5/15 6.57M shares	Short position burden remains
5/18 loan balance	0 shares in local DB	Treated as data anomaly; not used

May 13 was a forced-selling episode with simultaneous program selling and elevated short activity. Since then, program trading has flipped to net buying and short-selling share has fallen from 19.26% to the 7% range. That is clearly positive.

However, the stock loan balance continued to rise through May 15, which means the short position has not been fully unwound. The more precise interpretation is:

Attack intensity has decreased. But the short position burden remains.

The distinction matters: downside pressure has not disappeared — the pace of downside attacks has simply slowed.

5. Current Stage: “Bottom Test”

The cleanest framework for Pearl Abyss flow is a four-stage model.

Stage	Description	Status
1	Distribution begins	Past
2	Forced/panic selling	Occurred 5/13–5/15
3	Foreign bargain buying; program normalization	Current
4	Institutional re-entry; trend recovery	Not yet

We are at stage 3. The May 13–15 panic is behind us, and foreign bargain buying with program normalization has appeared. A bottom-formation possibility therefore exists.

But stage 4 has not arrived. The institutional breakdown is still weak, and at ₩45,800 the stock has not reclaimed its May 15 close of ₩47,200. Calling it a flow turn requires at minimum two things.

First, price must reclaim ₩47,200. Recovering the May 15 panic close is the signal that “the forced selling failed.”

Second, trust funds, pension funds, or insurers must post net buying in the institutional breakdown. Foreign buying alone can mark a bottom test; the return of long-duration institutional capital is a trend-change signal.

6. Price Levels: ₩45,500 and ₩47,200

The levels to watch at the current price are straightforward.

Level	Significance
₩45,500–₩46,000	Near-term support
₩47,200	May 15 close; first recovery target
₩50,500–₩52,000	Bottom of the broken trading range; structural recovery
₩53,000+	Zone confirming joint foreign + institutional return
Below ₩45,500	Risk of further decline and renewed short attack

₩45,500–₩46,000 is the near-term support. Holding this range while foreign buying continues keeps the bottom test intact.

₩47,200 is the first recovery target. As the May 15 panic close, reclaiming this level signals that the forced-selling episode has been fully reabsorbed.

₩50,500–₩52,000 carries greater importance. This is the underside of the broken trading range. Recovering it marks not merely a bounce but a structural normalization of price and flow.

Above ₩53,000, joint foreign and institutional re-entry is required. If price rises while institutions continue selling, the move is likely short covering or a technical rebound rather than a genuine reversal.

Conversely, a break below ₩45,500 would reopen the risk of further decline and a renewed short attack. With stock loan balance still elevated, a downside break could re-stimulate short positioning.

7. Practical Call: Hold and Observe; Wait on Adding

Three things are currently positive for Pearl Abyss.

First, foreigners have started returning since May 14.

Second, program selling has calmed since May 13.

Third, short-selling share has peaked.

Three things are currently negative.

First, the full two-week cumulative flow shows net selling by both foreigners and institutions.

Second, securities firms and private funds drove institutional selling; long-duration capital is absent.

Third, stock loan balance risk remains.

The conclusion therefore is:

This is a “flow-bottom test,” not a “flow turn.”

Existing holders should observe while watching whether ₩45,500–₩46,000 holds. Adding to the position is better deferred until ₩47,200 is reclaimed, foreign buying continues, and the institutional breakdown improves.

The practical decision framework is as follows.

Condition	Judgment
₩45,500–₩46,000 holds + foreigners sustain net buying	Maintain position
₩47,200 reclaimed + program stays net buy	First stabilization
₩50,500–₩52,000 reclaimed + institutional breakdown improves	Flow turn probability rising
Trust funds / pension funds / insurers turn net buyers	Genuine reversal signal
Break below ₩45,500 + foreigners turn net sellers	Reduce exposure or reassess

8. How This Connects to the Existing Pearl Abyss Thesis

This is a flow note, not an earnings or product note — but it cannot be read in isolation from the broader thesis.

As covered in 1Q Recognized Sales and the 2027 Cliff, the 2Q guidance weakens the “immediate cliff” framing. The stock fell anyway because the market is looking at the 2027 earnings cliff.

Patch 1.07 Weekend Data defends the product and sales-rank side of the thesis. Patch cadence and review improvement are positives. But strong product data alone will not drive an immediate price response if flow remains poor.

May 21 IR Checkpoint is the next potential catalyst for institutional flow. If shareholder return policy, the DLC roadmap, and DokeV development stage are clarified, institutional models could start moving again.

What the DLC Comments Really Mean is the core argument for removing the 2027 cliff discount. For a flow turn to materialize, this argument needs to be translated into hard numbers or an official roadmap.

In short, what is needed now is not a single piece of good news. Price, flow, product data, and company communication all need to align.

9. Final Judgment

For Pearl Abyss, the return of foreign buyers since May 14 and the normalization of program trading are positives. The May 13 panic is not simply rolling forward. Short-selling share has come down to the 7% range, easing attack intensity.

But the institutional breakdown is still weak. Securities firms and private funds have been selling; long-duration buying from pension funds, insurers, and trust funds remains insufficient. The fact that retail primarily absorbed the sell-off during the sharp-decline period is also not a constructive accumulation structure.

Therefore, a bottom cannot be ruled out, but concluding that a flow turn has arrived would be premature.

The key levels are ₩45,500–₩46,000 and ₩47,200. Holding ₩45,500–₩46,000 while foreign buying continues, then reclaiming ₩47,200, constitutes first stabilization. Recovering ₩50,500–₩52,000 with institutional breakdown improvement is required before a full flow turn can be discussed.

The practical conclusion is simple.

Hold and observe. Additional buying after confirming ₩47,200 recovery and institutional breakdown improvement.

Now is not the time to aggressively add — it is the time to watch who is absorbing the post-panic supply. The next genuine reversal for Pearl Abyss becomes more credible when sustained foreign buying + return of long-duration institutional capital + answers at the May 21 IR that break the 2027 cliff narrative all arrive together.

This post is for research and commentary purposes only and does not constitute investment advice. Price, volume, foreign, and institutional flow data are sourced from the Research OS local DB and the Naver Finance Pearl Abyss foreign/institutional page. Institutional sub-category, program trading, short selling, and stock loan balance data are from the local DB and may differ from exchange or brokerage source data in timing. The May 18 stock loan balance reading of 0 shares is treated as a data anomaly and was excluded from the analysis. Net consideration figures are estimates based on share counts and prices during the relevant period and may differ from actual executed amounts. Price reference levels (₩45,500–₩46,000, ₩47,200, ₩50,500–₩52,000, ₩53,000+) represent the analyst’s technical and flow-based judgment and carry no guarantee. Pearl Abyss’s share price may change rapidly depending on Crimson Desert sales data, the May 21 IR, DLC/shareholder return/DokeV roadmap developments, global macro conditions, KOSDAQ flow, and short and loan positions. This analysis may be wrong. Data as of May 18, 2026 KST.

Korea Quality Re-Rating Watch 2026-05-18: Protec, VM, TES — Smart Money Concentrates in Semi Equipment

Mon, 18 May 2026 16:30:00 +0900

Macro Dashboard — 2026-05-18

Indicator	Level	5-Day Δ	Signal
KOSPI	7,516.0	−1.7%	Neutral
KOSDAQ	1,111.1	−5.8%	Weak
VIX	19.2	+6.6%	Stable
US 10Y	4.59%	+19 bps	Rising
DXY	99.2	+0.8%	Strengthening
USD/KRW	1,502	+1.8%	KRW under pressure
Brent	$111.2	+5.2%	Elevated

Regime verdict: Korea Neutral / US Bear. The KR-US divergence is the key read — breadth in Korea is narrow (50MA participation: 33.7%, 200MA: 46.3%), but the market has not broken down. US deterioration is the larger structural concern. Rate and oil pressure remain the primary macro headwinds.

Market Wrap

Monday’s session had a split personality. The KOSPI printed near flat (+0.13% intraday at 14:25) while the KOSDAQ dropped −2.18% — a 230 basis-point gap between large-cap and small-cap Korea that tells you most of the story. This was not a broad rally. It was a large-cap semiconductor session carrying a structurally weak market.

Samsung Electronics (005930.KS) surged +3.88% to ₩281,000, absorbing news of a potential 18-day union strike starting May 21. Markets interpreted the court’s production-protection ruling as containing the downside risk, and the move looked more like an oversold bounce than a fundamental re-rating. The caveat: foreign investors sold a net ₩1.24 trillion in the stock — one of the largest single-day outflows in recent weeks. Domestic institutions absorbed it.

Sector rotation was narrow and deliberate. The winning themes were: semiconductors (Memory, HBM), CPO/photonics, power infrastructure/energy equipment, and select insurance names. Everything else — machinery, transportation equipment, construction, consumer/retail, and KOSDAQ high-beta — was offered. The breadth signal from the operating screener confirmed this: 76 names passed, 22 new additions, but underlying participation was thin.

The AI infrastructure narrative continues to provide a thematic tailwind. News flow around Amazon, Microsoft, and Google increasing conventional server procurement in Q2, combined with LS Electric’s ~₩4 trillion supply contract, kept power/grid-adjacent names in focus. This is consistent with the screener results showing semiconductors and electronics components dominating the quality-overlap list.

Flow summary: Samsung Electronics dominated foreign outflows. Daedeok Electronics (353200.KS) was the cleanest combination of price strength (+3.39% on the day, +11.61% over five days) and coordinated foreign plus institutional buying. Semiconductor equipment names with strong RS percentiles (95+) drew both institutional and foreign interest in a market that otherwise saw diffuse selling.

The KOSDAQ weakness was broad-based and not isolated to any single sector. High-beta names with weak fundamentals were sold indiscriminately. This is characteristic of a risk-off session dressed up as a flat tape — the headline index was calm, the underlying market was not.

Today’s Quality Re-Rating Candidates

The KR Meta Screener flagged 117 tickers today, with 20 surfacing in the top table. The screener intersection summary identifies 8 stocks with 3+ overlap across the five core frameworks. Meta score leads the ranking; overlap count confirms signal density.

Top Candidates — Screener Overlap Table

Meta Rank	Ticker	Name	Meta Score	Overlap	Screeners Hit	Key Metrics
1	053610.KQ	Protec	95.5	4	QC · SMQ · CR · SME + Consensus Up	ROE 13.7%, OP YoY +245%, margin Δ +12.2pp
2	089970.KQ	VM	87.6	4	QC · SMQ · CR · SME	ROE 14.7%, OP YoY +387%, margin Δ +29.3pp
3	482630.KQ	Samyang NCKEM	70.0	3	QC · SMQ · SME	ROE 14.7%, OP YoY +65%, IR event today
4	095610.KQ	TES	68.3	3	QC · SMQ · SME	ROE 14.5%, OP YoY +50%, F+QI +₩51.9bn 5D
5	187870.KQ	Device	66.7	3	QC · CR · PEAD	ROE 11.1%, OP YoY +312%, margin Δ +11.7pp
6	000660.KS	SK Hynix	58.9	2	QC · CR + Consensus Up	ROE 35.6%, OP YoY +101%, margin Δ +13.1pp
7	322000.KS	HD Hyundai Energy Solutions	56.4	3	SMQ · CR · SME	F+QI +₩40.4bn 5D, margin Δ +7.5pp
8	353200.KS	Daedeok Electronics	52.5	3	SMQ · CR · SME	RS 98.1, OP YoY +336%, +3.39% today
9	147830.KQ	Jeryong Industrial	53.5	2	QC · CR	ROE 13.9%, OP YoY +159%, margin Δ +13.9pp
10	089890.KQ	Koses	52.6	2	QC · CR + DART catalyst	ROE 17.8%, OP YoY +410%, supply contract filing

QC = Quality Compounder · SMQ = Smart Money Quality · CR = Cycle Rerating · SME = Smart Money Earnings

Deep Dive: Top 3 Names

Protec (053610.KQ) — Meta Rank #1, Score 95.5

Protec makes specialized semiconductor packaging and bonding equipment — it sits in the supply chain that enables advanced chip assembly. Today it hits all five screener dimensions simultaneously: it clears the Quality Compounder filter (ROE 13.7%, operating margin expanding +12.2 percentage points year-over-year), smart money is buying (net foreign + qualified institutional inflow of +₩6.7 billion over five days), the cycle re-rating signal is live (OP YoY +244.5%, operating leverage spread 209%), and consensus estimates are being revised upward. That combination — a profitable, low-debt business where institutions are accumulating and analysts are chasing numbers higher — is exactly what the editorial framework looks for. Recent DART filings are regulatory/neutral (large shareholder report, a prior capital reduction completion). No risk-flag public disclosures.

VM (089970.KQ) — Meta Rank #2, Score 87.6

VM is another semiconductor process equipment maker, but with notably more explosive operating leverage: revenue doubled (+105.5% YoY), operating income grew +387%, and operating margin expanded +29.3 percentage points. The Smart Money Quality screener ranks it #1 in the entire universe today — meaning it leads on the combined quality-plus-flow filter. Five-day net foreign and qualified-institutional inflow of +₩9.85 billion confirms that institutional money has been entering, not exiting. The only reason VM ranks second to Protec in the meta order is Protec’s additional Consensus Up Revision signal and slightly stronger absolute quality score. Both names warrant watching together. The latest DART filing is a routine Q1 2026 quarterly report (neutral).

Samyang NCKEM (482630.KQ) — Meta Rank #3, Score 70.0

Samyang NCKEM is a specialty chemicals producer (photoresists and process chemicals for semiconductor manufacturing). It passes the Quality Compounder filter (ROE 14.7%, OP YoY +64.5%) and shows strong institutional buying signals on the Smart Money screeners, though the five-day net flow is marginally negative (−₩1.1 billion). The notable timing factor: today’s DART filings include an IR event disclosure — the company held an investor relations briefing on May 18, which is a catalyst-positive signal for near-term price discovery. The quality foundation is sound; the question is whether the IR event converts into sustained institutional positioning. Worth monitoring after the meeting follow-through.

Caution flags to note: TES (095610.KQ) carries elevated short interest (11.5% of float) alongside its strong screener signals — the net foreign and institutional buying is real, but program trading and financial-sector selling create distribution risk. SK Hynix (000660.KS), despite outstanding fundamentals (ROE 35.6%, consensus revisions moving up), saw net foreign plus qualified-institutional selling of −₩12.1 trillion over the past five days, a significant headwind that the meta screener correctly penalizes with a risk flag. The screener flagged it as a candidate for the quality-plus-cycle angle, but flow confirmation is absent at the moment.

The clearest money-in-motion story across all frameworks today is semiconductor equipment — Protec, VM, TES, and Koses (089890.KQ) all surface through the same lens. The market is early in re-pricing domestic equipment makers that benefit from HBM and advanced packaging capex, and today’s price and flow data suggest that re-pricing is active, not speculative.

Korea Alpha in the CloudMatrix Expansion Scenario — Find Where Inefficiency Accrues, Not the Direct Huawei Suppliers

Sun, 17 May 2026 23:58:00 +0900

📚 US-China Decoupling Follow-Up Series US-China Summit Decoupling Analysis / NVIDIA Korea 2nd–3rd Tier Value Chain / AI Back-End 11-Stock Data Comparison / Semiconductor Value Chain Hub

Huawei’s CloudMatrix 384 is not a simple story of China catching up to NVIDIA. It is a system that lashes together 384 Ascend chips, draws 559kW of power, and routes thousands of optical transceivers to match the GB200 NVL72 in aggregate throughput — not in efficiency. China lost on efficiency but countered with system scale. The real alpha that this inefficiency points to is not in direct Huawei suppliers. Because China cannot reach NVIDIA, it must scale up its own fab, equipment, and consumables investment — and that process generates demand for inspection, metrology, and high-difficulty consumables. Treating 티씨케이, 넥스틴, and 코미코 as the same China theme is a trap. The three companies differ in investment thesis, confirmed results, and buy timing.

Key Takeaways

CloudMatrix 384’s essence is aggregate bypass, not an efficiency victory. 384 Ascend chips, roughly 559kW of power, and large-scale optical connectivity allow it to match the GB200 NVL72 on select metrics.
The real alpha is not in direct Huawei suppliers. HBM export controls and China’s domestic substitution policy limit direct supply routes for Korean companies. What matters instead is the indirect accrual structure: as China invests in its own fabs, demand for inspection, metrology, and consumables grows.
티씨케이 is a confirmed recovery play. 1Q26 revenue of ₩95.4bn, operating income of ₩28.6bn, and an operating margin of roughly 30% reflect China customer normalization and high-layer NAND conversion benefits already in the numbers.
넥스틴 is an inflection-point bet. The China inspection equipment PO resumption and Wuxi subsidiary ramp are the key catalysts — but 1Q26 was an operating loss, and the May 15 tape showed simultaneous foreign and institutional selling. This is still a pre-confirmation zone.
코미코 is global foundry quality with a China option, not a China play. The core thesis is TSMC and global foundry cleaning/coating demand plus ESC product mix improvement.
The May 15 tape differentiated the three names. 티씨케이 saw foreign buying near the 20-day moving average despite the selloff. 넥스틴 broke below all major moving averages with retail absorbing the supply.
Current priority order is 티씨케이 > 코미코 > 넥스틴. 넥스틴 has the highest thematic purity, but purity and buy timing are different things.

1. What CloudMatrix Is Actually Pointing To

Viewing CloudMatrix 384 purely as a “Chinese GB200” is half right and half wrong. Based on reporting from Tom’s Hardware and SemiAnalysis, CloudMatrix 384 links 384 Ascend chips to deliver roughly 300 PFLOPs of BF16 performance — higher than the GB200 NVL72’s roughly 180 PFLOPs. On some metrics, its memory capacity and bandwidth are also larger.

But the cost is steep. System power for CloudMatrix 384 is reported at approximately 559kW, compared to roughly 145kW for the GB200 NVL72 — a ratio of 3.85x. China did not beat NVIDIA on efficiency; it matched aggregate throughput by using more chips, more power, and more interconnects.

Metric	Huawei CloudMatrix 384	NVIDIA GB200 NVL72	Interpretation
Accelerators	384 Ascend	72 GPU + 36 CPU	Huawei uses far more chips
BF16 Performance	~300 PFLOPs	~180 PFLOPs	Huawei leads in raw aggregate
System Power	~559kW	~145kW	NVIDIA leads in efficiency
Optical Links	6,912 × 800G-class LPO reported	Relatively fewer	China’s aggregate bypass drives sharply higher optical dependency

What investors should extract from this table is not “Huawei won.” It is where the inefficiency accrues. Weaker chips require more chips. More chips require more power, more cooling, more substrates, more optical connectivity, more inspection, and more process stability. The more inefficient China’s AI infrastructure, the higher the total input volume of components and equipment.

This is where Korean alpha emerges — not because Korean companies can freely supply HBM or key equipment directly to Huawei. That route is constrained by sanctions and domestic substitution policy. The opposite is true: precisely because direct routes are blocked, China must scale investment in SMIC, CXMT, YMTC, and the broader Huawei ecosystem through its own fabs, equipment, and consumables. The resulting demand for yield-improving inspection and metrology equipment and high-difficulty process consumables flows outward from there.

The inference chain should be reframed as follows.

Flawed Chain	Better Chain
CloudMatrix success → Huawei revenue growth → find Huawei direct suppliers	H200 withheld, sanctions sustained → China scales own fab investment → yield requirements rise, process difficulty increases → inspection, metrology, consumables demand grows

This article focuses on the second chain.

2. The Three Names Are Not the Same China Theme

Markets tend to group 티씨케이, 넥스틴, and 코미코 together as “China semiconductor self-sufficiency beneficiaries.” But the investment thesis for each is entirely different.

Item	티씨케이	넥스틴	코미코
Investment Character	Confirmed consumables recovery	2H26 China PO inflection bet	Global foundry quality + China option
CloudMatrix Directness	Low	Medium–High	Low–Medium
China AI Self-Sufficiency Purity	Medium	Highest	Medium
1Q26 Confirmed	Yes	No	Partial
Core Product	SiC Ring and other high-difficulty consumables	Wafer defect inspection equipment	Cleaning/coating, ESC
Key Catalyst	China customer normalization, high-layer NAND	China PO resumption, Wuxi subsidiary ramp	TSMC spillover effect, ESC volume ramp
Current View	Conditional Buy	Watch	Watchlist

티씨케이 is not a “direct CloudMatrix component supplier.” More precisely, it benefits from China fab utilization normalization and NAND high-layering through higher consumables usage. Inventory destocking by Chinese customers ended, urgent orders picked up, and 1Q26 results confirmed the inflection. Revenue of ₩95.4bn, operating income of ₩28.6bn, and an operating margin of roughly 30% are numbers already in the record, not mere expectations.

넥스틴 is the highest-purity bet. China building its own AI chips requires improving yield, and improving yield requires more wafer defect inspection. That is where the 넥스틴 China subsidiary and PO resumption narrative comes from. But the numbers have not been confirmed yet. 1Q26 was an operating loss, and the 2H26 China PO resumption must actually materialize.

코미코 loses its logic if it is purchased purely as a China theme. The core is global foundry cleaning and coating quality. The key drivers are TSMC and global foundry utilization, operations across US, Taiwan, and Korean facilities, and ESC product mix improvement at Mico Ceramics. The China ESC option exists, but it is difficult to reduce the entire company to a single China narrative.

This distinction matters. 티씨케이 is a current-results confirmation play, 넥스틴 is a future inflection play, and 코미코 is a global quality play. Trading all three under the same theme can cause mistimed entries and exits across all three names.

3. What the May 15 Tape Said

In the sharp selloff on May 15, the signals from the three names diverged — particularly the gap between 티씨케이 and 넥스틴.

Item	티씨케이	넥스틴
May 15 Close	₩305,000	₩64,100
May 15 Change	-12.5%	-13.4%
20-Day Return	+12.3%	-15.2%
60-Day Return	+38.0%	-21.1%
120-Day Return	+87.9%	-0.2%
vs. 52-Week High	-12.5%	-36.2%
Gap to 20-Day MA	+0.1%	-12.6%
Gap to 50-Day MA	+15.7%	-13.8%
RSI 14	53.9	36.7

Both names fell roughly 13%, but the meaning is different. For 티씨케이, it was the first support test — a pullback to near the 20-day moving average within a strong 120-day uptrend, with RSI cooling to a neutral 53.9. For 넥스틴, it is trading below the 20-day, 50-day, and 120-day moving averages simultaneously. An RSI of 36.7 may look like oversold territory, but a low RSI within a downtrend can signal trend deterioration rather than a buy opportunity.

The supply/demand picture also differed.

Item	티씨케이	넥스틴
May 15 Foreign	+₩4.91bn	-₩2.31bn
May 15 Institutional	-₩3.39bn	-₩1.95bn
May 15 Retail	-₩1.51bn	+₩4.26bn
May 15 Program	-₩7.65bn	-₩3.45bn
Short-Sale Ratio	15.16%	14.03%
Stock Lending Balance	₩212.4bn	₩56.1bn

티씨케이 attracted foreign buying even on the selloff day. Institutional and program selling was present, but the cumulative foreign accumulation trend remained intact. 넥스틴, by contrast, saw foreigners and institutions selling together while retail absorbed the supply. The tape pattern to avoid most in an event-driven setup is exactly “foreign + institutional dual selling absorbed by retail.”

Looking at 5-day and 20-day cumulative flow, the divergence is even more pronounced.

Period	티씨케이 Foreign	티씨케이 Institutional	넥스틴 Foreign	넥스틴 Institutional
5D	+₩14.06bn	+₩12.97bn	+₩5.02bn	-₩3.53bn
20D	+₩74.33bn	-₩15.68bn	+₩11.92bn	-₩15.76bn

티씨케이 pulled back with ₩74.3bn of 20-day foreign accumulation underneath it. 넥스틴 had just ₩11.9bn of 20-day foreign buying, of which ₩2.3bn left on May 15 alone. The base is entirely different. 티씨케이’s move can be read as a healthy consolidation; 넥스틴 warrants waiting for trend recovery confirmation before taking a position.

4. Four Common Misreads

First misread: connecting CloudMatrix to a search for direct Huawei suppliers. The pathway for Korean core components and equipment to reach Huawei directly is constrained by sanctions and China’s domestic substitution policy. Alpha comes from indirect accrual, not direct supply. The more inefficient China’s AI systems, the greater the demand for domestic fab investment and process stabilization — which flows into inspection, metrology, and consumables.

Second misread: treating all inspection and metrology names as one block. 넥스틴 focuses on front-end pattern inspection, Park Systems on AFM metrology, OROS on overlay metrology, and Intekplus on FC-BGA and packaging inspection. They all carry the “inspection/metrology” label, but their catalysts, customer sets, and revenue recognition timing are different.

Third misread: assuming a low RSI is always a buy signal. In an uptrend, RSI at 30–40 can mark a pullback opportunity. In a downtrend, RSI at 30–40 can signal continued weakness. 넥스틴 is trading below its 20-day, 50-day, and 120-day moving averages simultaneously. In that context, RSI 36.7 is a wait signal, not a buy signal.

Fourth misread: treating thematic purity as the primary buy criterion. 넥스틴 has the highest purity for the China AI self-sufficiency theme. But purity alone is insufficient. Buy attractiveness requires assessing purity, price positioning, supply/demand flow, and fundamental confirmation together. On current price and tape, 티씨케이 has the advantage. 넥스틴 is high-purity but unconfirmed, with weak flow.

5. Name-by-Name Entry Guide

티씨케이 can be considered a pullback buy candidate on confirmation of ₩300,000 support. A first check is support at ₩300,000–305,000 with sustained foreign net buying. Secondary targets are recovery to ₩320,000, then ₩335,000. Conversely, a break below ₩295,000 accompanied by continued foreign selling warrants trimming. If 2Q26 operating margin falls below 25%, or if signals emerge that China normalization was a one-off event, the investment thesis weakens.

넥스틴 is still better observed than bought. A recovery to ₩69,500, a resumption of foreign buying, news on China PO or Wuxi subsidiary ramp, quarterly revenue recovery, and a visible path to operating profit breakeven need to appear together. Buying near ₩64,000 on RSI alone risks catching a falling knife. If China PO is not confirmed by 3Q26 or the Wuxi ramp is delayed, the thesis itself needs to be downgraded.

코미코 is a quality company but not a standalone CloudMatrix buy. To initiate a position, confirm the ESC volume production transition, utilization rate improvement across US, Taiwan, and Anseong facilities, TSMC revenue share trends, and the degree to which Mico Ceramics profits accrue to parent shareholders. The China option is a secondary catalyst.

6. Portfolio Construction

For a balanced portfolio, 40% 티씨케이 / 30% 코미코 / 20% 넥스틴 / 10% hedge is the most straightforward allocation. 티씨케이 is the core with confirmed results and foreign accumulation. 코미코 provides global quality and ESC optionality. 넥스틴 is a satellite position for convexity when 2H26 China PO confirmation arrives.

An aggressive portfolio can raise 넥스틴’s weight — but conditions apply. Recovery to ₩69,500, foreign re-entry, and China PO news must appear simultaneously. Without those conditions, a 넥스틴-heavy aggressive allocation carries unnecessary risk.

A conservative portfolio fits closer to 60% 티씨케이 / 30% 코미코 / 10% 넥스틴. What is confirmed today is 티씨케이; 코미코 is a quality holding; 넥스틴 remains an option.

Type	티씨케이	코미코	넥스틴	Hedge
Balanced	40%	30%	20%	10%
Aggressive	35%	20%	45%	0%
Conservative	60%	30%	10%	0–10%

7. Scenario-Based Action

The highest-probability scenario is sustained US sanctions combined with accelerating Chinese self-sufficiency. If H200 shipments remain blocked and China continues building its own systems, fab and process stabilization investment across the SMIC, CXMT, YMTC, and Huawei ecosystems is likely to persist. In this scenario, 티씨케이 responds fastest. 넥스틴 can strengthen later when China PO actually resumes, and 코미코 follows more slowly through ESC and global foundry revenue.

If US sanctions ease and China returns to NVIDIA, 넥스틴’s high-purity thesis takes the hardest hit. 티씨케이 has NAND and consumables drivers that provide some cushion. 코미코’s global business is the primary driver, so the relative impact is smaller.

A scenario of further US sanction tightening and increased ally-pressure pressure is also worth considering. In this case, restrictions on Korean equipment and component exports to China could intensify. 넥스틴 faces direct exposure. 티씨케이 is relatively better positioned given the consumables nature of its products, but is not fully insulated. 코미코’s global revenue share provides defensive depth.

The possibility of a Xi Jinping visit to Seoul in September and the October truce expiration are also catalysts. A truce extension would bring near-term relief; re-escalation would create near-term shock; a major deal would generate near-term euphoria. But in any of these cases, the underlying force driving China to build its own AI systems does not disappear. The framework of inefficiency accrual created by CloudMatrix therefore remains valid across scenarios.

8. Investment Conclusion

CloudMatrix 384 is not evidence that China beat NVIDIA on efficiency. It is a system that bypasses GB200 NVL72 in aggregate by using more chips, more power, and more interconnects. That inefficiency itself is the investment point. Inefficiency creates cost, and cost accrues into equipment and consumables demand.

What Korean investors should seek is not direct Huawei suppliers. That path is narrow given HBM export controls and China’s domestic substitution policy. The focus should be on the inspection, metrology, and high-difficulty consumables needed as China scales its own fab investment and pushes yields higher.

At current prices and tape, 티씨케이 leads. 1Q26 results are confirmed, and foreigners were buying on the May 15 selloff. 코미코 is global foundry quality with a China option attached. 넥스틴 has the highest thematic purity but remains unconfirmed. A simultaneous appearance of 2H26 China PO resumption and flow recovery is needed before its buy priority rises.

The real alpha is not in the word “CloudMatrix” itself. It lies in matching where inefficiency accrues and whether the market tape is confirming that accrual structure. Right now, the cleanest combination of those two conditions sits in 티씨케이. The largest option value is in 넥스틴. The longest-duration quality is in 코미코.

Sources and Evidence Classification

Fact

CloudMatrix 384 has been reported as a structure using 384 Ascend chips, approximately 559kW of power, and large-scale optical transceivers.
티씨케이 disclosed 1Q26 revenue of approximately ₩95.39bn and operating income of approximately ₩28.6bn, implying an operating margin of roughly 30%.
May 15 supply/demand and price tape figures are based on backfill data provided by the user.

Inference

CloudMatrix’s aggregate bypass structure was judged to require expanded Chinese investment in domestic fab, equipment, and consumables.
티씨케이 was classified as a confirmed consumables recovery play, 넥스틴 as a China PO inflection bet, and 코미코 as global quality with a China option.
At current price and May 15 tape, 티씨케이 was judged to have greater buy attractiveness than 넥스틴.

Speculation

넥스틴’s 2H26 China PO resumption, Wuxi subsidiary ramp, and 코미코’s ESC revenue expansion are scenarios based on sell-side estimates and industry trend analysis.
The scenario-by-scenario impact of a September Xi visit and October truce expiration are analyst estimates.

Primary sources referenced: Tom’s Hardware CloudMatrix 384 report, SemiAnalysis CloudMatrix 384 analysis, 티씨케이 1Q26 results disclosure coverage, 티씨케이 Shinhan Investment report summary PDF, 코미코 SK Securities estimate summary.

This article is intended for research and commentary purposes only and does not constitute investment advice. CloudMatrix performance and power figures are based on published reports and analyst sources and may differ under actual operating conditions. 티씨케이, 넥스틴, and 코미코 results and outlook references draw on company disclosures, press reports, sell-side report summaries, and backfill data provided by the user. China PO resumption, ESC volume ramp, and September summit diplomacy scenarios are all subject to significant uncertainty. Name-level priority rankings and portfolio weightings reflect the analyst’s subjective judgment and carry no guarantee. This analysis may be wrong. Data as of May 17, 2026 KST.

The May 14-15 Beijing US-China Summit — Managed Truce, H200 Refusal, and What It Means for Korean Semiconductors

Sun, 17 May 2026 23:50:00 +0900

📚 US-China Summit Series Pre-Summit Guide: 10 Agendas for Korean Investors / Summit Result: No Grand Bargain / CloudMatrix Expansion Scenario / NVIDIA Earnings and Korea’s Second-Order Supply Chain / AI-RAN and the Korean Supply Chain

The May 14-15 Beijing summit was not a grand bargain. It was a managed truce. Washington and Beijing agreed to avoid immediate escalation, but the core issues — semiconductor export controls, rare earths, Taiwan, and tariff structure — were not solved. The more important signal came outside the room. The U.S. cleared H200 exports, but China has not let purchases proceed. Huawei’s CloudMatrix 384, meanwhile, shows a different Chinese response: weaker chips offset by more chips, more power, and more optical links. Decoupling is not ending. It is moving into a more expensive equilibrium.

Key Takeaways

The summit was a managed truce. It reduced near-term escalation risk but did not resolve the technology-control structure.
The H200 episode matters more than the communique. The U.S. has cleared sales to roughly 10 Chinese firms, but reports indicate no H200 chips have shipped because China has not approved purchases.
CloudMatrix 384 is a system-level workaround. Huawei is not beating NVIDIA at chip efficiency. It is using many more chips, much more power, and a larger optical fabric to reach useful system-level performance.
ITIF’s model complicates the “decoupling is bad for Korea” narrative. In a full semiconductor decoupling scenario, ITIF estimates U.S. firms could lose USD 77B in China-related sales in the first year, while South Korean firms could gain roughly USD 21B.
Korea’s exposure is time-dependent. The short term favors HBM, memory, substrates, MLCC and optical links. The medium term is mixed. The long term brings standard fragmentation, Chinese substitution, and dual-supply-chain costs.
The cleanest second-order alpha is optical interconnect. The U.S. stack needs more CPO and rack-scale optics; the Chinese stack needs even more optical links because it compensates with scale.

1. What the Summit Actually Signaled

The summit should be read as a pause, not a settlement. There were preliminary understandings, purchase intentions, and stability language. But there was no binding package that changes the core technology conflict.

That distinction matters. Tariffs can be negotiated. Semiconductor controls are strategic. If export controls remain structurally intact, then the AI infrastructure map remains split into two partially separate systems.

The political incentives explain the outcome. The U.S. wanted market stability and visible business wins. China wanted time and did not want to look as if it was accepting U.S. terms. Both sides wanted to avoid a visible breakdown; neither side had a strong reason to make a structural concession.

Xi Jinping’s reference to the Thucydides Trap should also be read in that frame. On the surface it is a call to avoid conflict. Underneath it sits a time-horizon argument. If the U.S. restricts exports, China accelerates domestic substitution. If U.S. firms lose China revenue, their R&D capacity can weaken. The conflict is not disappearing; the cost distribution is changing.

2. H200 Refusal Shows the Weakening of the Export-License Carrot

The H200 episode is the clearest signal. The U.S. created a conditional framework for H200 exports to China in January 2026, with oversight and a fee structure. By mid-May, the U.S. had reportedly cleared around 10 Chinese firms to buy H200s, including major internet platforms.

But no chips have shipped. Reporting indicates China has not allowed purchases to proceed. Trump reportedly said China wants to develop its own chips.

This is not just an administrative delay. It means export permission is no longer automatically a carrot. If Beijing refuses the product for strategic reasons, U.S. permission has no commercial value.

For NVIDIA, China H200 revenue should be treated as an option, not as the base case. That makes the next NVIDIA earnings call important. If management confirms that China data-center compute revenue is immaterial or excluded from guidance, the market will need to lower the China reopening option value.

For Korea, the meaning is more nuanced. If Chinese customers do not buy H200s, the immediate HBM linkage is weaker. But if non-China hyperscalers continue to expand AI capex, Korean HBM demand remains strong. The H200 story is less about a direct shipment boost and more about the shape of decoupling.

3. CloudMatrix 384: Losing at the Chip Level, Fighting at the System Level

Huawei’s CloudMatrix 384 shows how China is responding. According to public reports based on SemiAnalysis work, the system uses 384 Ascend chips and can reach roughly 300 PFLOPs of dense BF16 compute, compared with about 180 PFLOPs for NVIDIA’s GB200 NVL72.

But the comparison is not a clean victory. CloudMatrix reportedly consumes roughly 559kW, compared with about 145kW for GB200 NVL72. It is a brute-force design: more chips, more racks, more power, and more optical transceivers.

This tells us two things.

First, China has not achieved chip-level parity. NVIDIA’s efficiency, software ecosystem, CUDA moat, and training stack remain powerful.

Second, China may not need parity for every use case. For domestic policy markets, state cloud, sovereign AI, and “good enough” inference, a less efficient but domestically controlled system can be acceptable.

The most interesting investment implication is optical interconnect. A brute-force cluster needs a much larger communication fabric. CloudMatrix reportedly relies on thousands of high-speed optical transceivers. If weaker chips require more scale-out and scale-up connections, then optical components become a shared bottleneck.

The U.S. stack also moves in the same direction. Rubin, NVL systems, Spectrum-X, and CPO all increase rack-scale optical demand. The U.S. uses optics to improve efficiency; China uses optics to compensate for chip gaps. Different reasons, same direction.

4. ITIF’s Decoupling Model: Korea Is a Short-Term Absorber

ITIF’s November 2025 report gives a useful quantitative frame. In a hypothetical full semiconductor decoupling scenario, U.S. semiconductor firms could lose about USD 77B of China-related sales in the initial year. ITIF estimates that South Korean firms could gain about USD 21B of those lost sales, more than the EU, Taiwan, Japan, mainland China, or other regions.

The memory line is especially important. In ITIF’s table, a large share of the U.S. memory-related loss is absorbed by South Korea. This is why “decoupling is bad for Korea” is too simple.

In the short run, Korea can benefit. The country is not the direct AI accelerator rival to NVIDIA. It supplies memory, HBM, substrates, MLCC, testing, packaging, and other bottleneck components. If U.S. firms lose parts of China while non-China AI capex rises, Korean component suppliers can still benefit.

But the long-run view is less comfortable. The longer decoupling persists, the more China substitutes. Standards fragment. Korean firms may need U.S.-aligned and China-aligned product paths, compliance structures, inventory systems, and qualification cycles. That raises cost and can pressure margins.

The right answer is time-segmented:

6-12 months: favorable for Korean memory and AI components.
1-3 years: mixed, as China substitution begins to pressure pricing and standards.
3-5 years: structurally riskier, because dual supply chains and fragmented standards raise costs.

5. What It Means for Korean Stocks

The first-order beneficiaries remain SK hynix and Samsung Electronics. SK hynix has the cleanest HBM exposure. Samsung Electronics has HBM4, DDR5, eSSD, foundry optionality, and a broader AI-infrastructure platform story. But both are already heavily discovered. For new money, chasing after a large move is less attractive than buying on macro or earnings pullbacks.

The cleaner second-order alpha may sit in the less crowded supply chain.

Optical interconnect is the most interesting. OE Solutions and related optical names should not be framed only as NVIDIA suppliers. They are exposure to a world where both the U.S. and Chinese AI stacks need more high-speed connections.

Samsung Electro-Mechanics is another example. AI servers need MLCCs and FC-BGA substrates regardless of whether the rack is U.S.-standard or China-standard. The company is not the GPU winner, but it is a component bottleneck.

Hana Micron belongs in the structural-improvement bucket. If its Vina pricing structure and overseas margins continue to improve, it is not simply a memory-cycle beta. It is a post-processing margin-upgrade story.

6. Four Market Misreadings

The first misreading is that decoupling is a single big-bang event. It is not. It has been moving layer by layer since the Huawei restrictions: telecom equipment, semiconductor tools, AI chips, AI systems, and now optical links, standards, and software stacks.

The second misreading is that decoupling is uniformly bad for Korea. It is bad over a long horizon if standards fragment and China substitutes. But ITIF’s model shows that in the short run South Korean firms can absorb a meaningful share of U.S. firms’ lost sales.

The third misreading is that China’s rejection of H200 proves self-sufficiency. It does not. China still lags at the chip level. What it shows is a willingness to accept inefficiency in exchange for domestic control.

The fourth misreading is that the summit means decoupling is easing. The summit reduced escalation risk; it did not dismantle the strategic conflict.

7. The Next Catalyst: September and October

The next major window is a potential Xi visit in September and the October anniversary of the prior truce framework. The base case is truce extension: both sides want stability. A tougher scenario is a renewed escalation around export controls, rare earths, or Taiwan language. A true grand bargain remains lower probability.

For investors, the playbook is simple. If the truce is extended, existing Korean semi exposure can remain. If escalation returns, near-term drawdowns can create better entries in memory and AI component names, while optical-interconnect names may become more strategically relevant. If a grand bargain triggers a sharp relief rally, that may be a chance to trim overheated first-order winners.

8. Bottom Line

The Beijing summit did not end the U.S.-China technology conflict. It confirmed that the conflict is being managed.

The H200 episode and CloudMatrix 384 are the real signals. The U.S. can approve exports, but China can refuse them. China may not have the best chip, but it can build a system that is good enough for parts of its domestic market. That is a more expensive equilibrium, not a peaceful one.

For Korea, the answer is not simply bullish or bearish. Decoupling is positive today, mixed tomorrow, and risky later. SK hynix and Samsung Electronics remain the core short-term beneficiaries, but the more interesting medium-term angle may be optical interconnect, AI passive components, substrates, and structural OSAT improvement.

The alpha is not in saying “decoupling is good” or “decoupling is bad.” The alpha is in separating the time horizon.

Sources and Evidence Notes

Fact

ITIF estimates that in a full decoupling scenario U.S. semiconductor firms could lose about USD 77B of China-related sales in the first year, while South Korean firms could gain about USD 21B.
Reuters-cited reporting and Tom’s Hardware coverage indicate the U.S. cleared H200 sales to roughly 10 Chinese firms but no chips had shipped as of mid-May.
Tom’s Hardware and SemiAnalysis-based reporting describe Huawei CloudMatrix 384 as a 384-chip system with roughly 300 PFLOPs BF16 and about 559kW system power.

Inference

H200 non-shipment is interpreted here as a weakening of the export-license carrot.
CloudMatrix 384 is interpreted as a system-level workaround rather than a chip-level victory.
Korea’s exposure is segmented into short-term benefit, medium-term mixed impact, and long-term fragmentation risk.

Main references: ITIF Decoupling Risks report, Reuters H200 licensing report via Investing.com, Tom’s Hardware on China blocking H200 purchases, Tom’s Hardware on CloudMatrix 384, SemiAnalysis CloudMatrix note.

This post is for research and commentary only and is not investment advice. Summit details, H200 shipment status, and CloudMatrix specifications are based on public reporting as of May 17, 2026 KST. ITIF’s numbers are model outputs and may not match actual decoupling outcomes. Scenario analysis and stock implications are the author’s interpretation and may be wrong.

Pearl Abyss (263750) — 3.09M Recognized Copies in 1Q, ~3M Possible in 2Q. The Real Debate Is the 2027 Earnings Cliff

Sun, 17 May 2026 23:40:00 +0900

📚 Pearl Abyss × Crimson Desert Series 1Q26 Comprehensive / May 21 IR Watch / What the DLC Comment Really Means / Patch 1.07 Weekend Data / Pearl Abyss Hub

Pearl Abyss’s 1Q26 earnings changed the center of the debate. The question is no longer whether Crimson Desert worked. Revenue of KRW 328.5B, operating profit of KRW 212.1B, and a 64.6% operating margin answered that. The questions now are different: how many copies were actually recognized in 1Q and 2Q, and how much of 2026’s earnings power can remain in 2027.

Key Takeaways

The 3.1M-copy 1Q recognition argument is defensible. More precisely, Meritz estimates 3.089M recognized Crimson Desert copies in 1Q. Dividing KRW 266.5B in Crimson Desert revenue by that volume gives an accounting ASP of about KRW 86,300 per copy.
A roughly 3M-copy 2Q recognition case is also inside company guidance. Applying the same ASP to 2Q Crimson Desert guidance of KRW 224.2B-276.5B implies 2.6M-3.2M recognized copies, with a midpoint near 2.9M.
But recognized copies are not the same as new copies sold during the quarter. 2Q recognition likely includes both 1Q commercial sales not yet recognized in accounting and new 2Q sales.
The market debate has moved from a 2Q cliff to a 2027 earnings cliff. Meeting 2Q guidance defends downside. It does not, by itself, force a major multiple re-rating.
The 2027 bridge is the real issue. DLC, Asia/China expansion, platform extension, capital return, and DokeV visibility need to turn 2027 operating profit from the KRW 200B range toward KRW 300B+.
Investment stance: hold is defensible, additional buying should wait. A 2Q midpoint result is helpful but not enough. The better add point comes when the 2027 bridge becomes visible.

1. 1Q26 Already Proved the Core Case

Pearl Abyss’s 1Q26 results were strong at the headline level and stronger in quality.

Item	1Q26
Revenue / operating revenue	KRW 328.5B
Operating profit	KRW 212.1B
Operating margin	64.6%
Continuing net profit	~KRW 170.0B
Final net profit after discontinued operations	~KRW 158.0B
Crimson Desert revenue	KRW 266.5B
Black Desert revenue	KRW 61.6B
Overseas revenue mix	94%
North America / Europe revenue mix	81%
Crimson Desert platform mix	Console 50% / PC 50%

The key point is not just that Crimson Desert sold well. The revenue converted into operating profit at unusually high incremental margins. Much of Crimson Desert’s development cost had already been expensed in prior periods, so launch revenue flowed through with limited post-launch amortization burden.

The full-year 2026 guidance is also strong.

Item	2026 Guidance
Operating revenue	KRW 879.0B-975.4B
Black Desert revenue	KRW 234.9B-240.6B
Crimson Desert revenue	KRW 644.1B-734.8B
Operating expenses	KRW 391.4B-402.8B
Operating profit	KRW 487.6B-572.6B
Operating margin	55.5%-58.7%

So the 2026 earnings reset is no longer the main controversy. The controversy is what comes after it.

2. The 3.09M 1Q Recognized-Copy Estimate Fits the Revenue

The phrase “3.1M copies recognized in 1Q” is not an official company unit-sales disclosure. It is an accounting recognition estimate from Meritz. The precise estimate is 3.089M copies.

Item	Value
Meritz estimate of 1Q recognized Crimson Desert copies	3.089M
Rounded	~3.09M
Practical shorthand	~3.1M

This is mathematically consistent with reported 1Q Crimson Desert revenue.

1Q Crimson Desert revenue = KRW 266.5B
Meritz 1Q recognized copies = 3,089,000

Accounting recognized ASP
= KRW 266.5B / 3,089,000
= about KRW 86,300 per copy

The official commercial sales milestones and accounting recognition can differ. Company milestones refer to cumulative commercial unit sales. Income statement recognition is affected by quarter-end cutoff, PC versus console recognition, platform fees, taxes, offline-package settlement lags, regional pricing, and standard/deluxe mix.

The defensible wording is therefore:

Meritz estimates about 3.09M Crimson Desert copies were recognized in 1Q26 accounting revenue, which is consistent with KRW 266.5B in reported Crimson Desert revenue.

3. 2Q Recognition Around 3M Copies Is a Reasonable Base Case

Company 2Q guidance does not assume a sharp collapse in Crimson Desert revenue.

Item	Low	Midpoint	High
Operating revenue	KRW 271.3B	KRW 298.0B	KRW 324.7B
Black Desert revenue	KRW 47.1B	KRW 47.7B	KRW 48.2B
Crimson Desert revenue	KRW 224.2B	KRW 250.4B	KRW 276.5B
Operating expenses	KRW 141.7B	KRW 144.9B	KRW 148.0B
Operating profit	KRW 129.6B	KRW 153.2B	KRW 176.7B
Operating margin	47.8%	51.4%	54.4%

Relative to 1Q Crimson Desert revenue of KRW 266.5B, 2Q guidance ranges from -15.9% to +3.8%.

2Q low change = KRW 224.2B / KRW 266.5B - 1 = -15.9%
2Q high change = KRW 276.5B / KRW 266.5B - 1 = +3.8%

Using the same KRW 86,300 accounting ASP gives the following recognized-copy range.

Case	Crimson Desert Revenue	ASP	Implied Recognized Copies
2Q low	KRW 224.2B	KRW 86,300	~2.6M
2Q midpoint	KRW 250.4B	KRW 86,300	~2.9M
2Q high	KRW 276.5B	KRW 86,300	~3.2M

So a 3M-copy 2Q accounting recognition case is reasonable. First-half recognized copies naturally land near 6M.

1Q recognized 3.09M + 2Q recognized 2.6M-3.2M
= 1H recognized 5.69M-6.29M

Midpoint:
3.09M + 2.90M = ~5.99M

The important caveat: 3M recognized copies in 2Q does not mean 3M newly sold copies in 2Q. It likely includes both deferred recognition from commercial sales made in 1Q and new 2Q sales.

4. The Market Has Moved to the 2027 Cliff

Before the 1Q print, the market worried that 2Q would fall off immediately. Company guidance weakened that thesis. The bigger question is now:

2026 is strong. What remains in 2027?

That concern is not irrational. Meritz estimates a large decline in 2027.

Item	2026E	2027E	Change
Revenue	KRW 929.8B	KRW 502.5B	-46.0%
Operating profit	KRW 505.4B	KRW 204.1B	-59.6%
EPS	KRW 6,650	KRW 3,320	-50.1%

That is why 2026E PER of 7-8x is not enough by itself. The market is discounting 2026 earnings as launch-cycle peak earnings rather than recurring earnings.

But the market may still be too conservative. First, KRW 200B+ in 2027 operating profit would still be historically high for Pearl Abyss. Second, DLC optionality is not fully reflected. Third, Asia/China upside may remain under-modeled. Fourth, cash accumulation and capital return can lower the practical 2027 valuation burden.

5. The First 2027 Bridge Is DLC

DLC is not just incremental revenue. It is the mechanism that can weaken the 2027 cliff discount.

Assumption	Range
Cumulative base game sales	8.5M-10.0M
Recognized DLC ASP	KRW 30,000-45,000
Attach rate	25%-40%
Operating margin	60%-70%

DLC revenue sensitivity looks like this.

Cumulative Sales	Attach Rate	ASP	DLC Revenue
8.5M	25%	KRW 30,000	KRW 63.8B
8.5M	35%	KRW 35,000	KRW 104.1B
10.0M	35%	KRW 40,000	KRW 140.0B
10.0M	40%	KRW 45,000	KRW 180.0B

At a 65% operating margin, the operating profit contribution would range from KRW 41.5B to KRW 117.0B.

DLC Revenue	OP Contribution
KRW 63.8B	KRW 41.5B
KRW 104.1B	KRW 67.7B
KRW 140.0B	KRW 91.0B
KRW 180.0B	KRW 117.0B

DLC alone cannot offset the entire decline from 2026 to 2027. But it can plausibly lift 2027 operating profit from the KRW 200B range toward KRW 270B-320B. That is enough to change the market’s frame.

What matters at the May 21 IR is not another vague “we are exploring DLC” line. Investors need timing, pricing, content scale, and a clear distinction between free updates and paid expansion content.

6. The Second Bridge Is Cash and Capital Return

Pearl Abyss’s cash generation improves sharply in 2026. Some sell-side work suggests year-end cash and equivalents could approach KRW 1T after Crimson Desert and CCP-related balance-sheet simplification. This is not official company guidance, so it should be treated as inference.

At KRW 47,200 per share and 64,247,855 shares, market cap is about KRW 3.03T.

Market cap = KRW 47,200 × 64,247,855
= about KRW 3.03T

If year-end cash reaches KRW 1T, ex-cash market cap is about KRW 2.03T.

Ex-cash market cap = KRW 3.03T - KRW 1.00T
= about KRW 2.03T

Using Meritz’s 2027E net income estimate of KRW 213.3B, ex-cash PER would be about 9.5x.

Ex-cash PER = KRW 2.03T / KRW 213.3B
= about 9.5x

That is why capital allocation matters. Buybacks and cancellation defend EPS. Dividends signal confidence in cash-flow durability. Either can reduce the peak-earnings discount.

7. The Third Bridge Is Asia/China and Platform Extension

One reason 1Q quality was strong is that North America and Europe accounted for 81% of revenue. That is rare for a Korean game company and supports the global AAA re-rating thesis.

At the same time, it suggests Asia/China upside may remain. Exact China revenue mix has not been officially disclosed, so it should not be treated as fact. But if China and broader Asia are still underpenetrated, there is room for incremental sales.

Platform extension works the same way. DLC, expansion packs, bundles, seasonal discounts, additional platforms, cloud gaming, and subscription services can all reduce the 2027 revenue gap. But until management gives a roadmap, those remain options rather than modeled base-case earnings.

8. Valuation: Cheap on 2026, Debatable on 2027

At KRW 47,200 and 64,247,855 shares, the 2026 guidance-based EPS range looks cheap.

Item	Low	Mid	High
FY26 OP	KRW 487.6B	KRW 530.1B	KRW 572.6B
After-tax conversion	70.0%	72.5%	75.0%
Net income	KRW 341.3B	KRW 384.3B	KRW 429.5B
EPS	KRW 5,313	KRW 5,982	KRW 6,684
PER at KRW 47,200	8.9x	7.9x	7.1x

But using Meritz’s 2027E EPS of KRW 3,320, the stock trades at about 14.2x.

2027E PER = KRW 47,200 / KRW 3,320
= 14.2x

That is the core tension. Pearl Abyss is cheap if 2026 earnings are repeatable. It is less obviously cheap if 2027 falls back toward KRW 200B in operating profit.

Scenario framing:

Scenario	Target Price	Key Conditions
Bear	KRW 45,000	2Q below guidance low, annual sales risk below 8M, no DLC
Base	KRW 68,000	2Q midpoint, 8.5M-9.0M annual sales, some 2027 DLC contribution
Bull	KRW 95,000-100,000	10M+ annual sales, DLC formalized, Asia/China upside, capital return

The KRW 68,000 base case is not an aggressive “2026 peak earnings only” target. It blends 2026 and 2027.

Item	Value
2026E net income	KRW 370.0B
2027E net income	KRW 220.0B
Shares	64.248M
2026E EPS	KRW 5,759
2027E EPS	KRW 3,424
12M blended EPS	KRW 4,883
Target PER	14.0x

12M blended EPS
= KRW 5,759 × 62.5% + KRW 3,424 × 37.5%
= KRW 4,883

Target price
= KRW 4,883 × 14.0
= KRW 68,362 ≈ KRW 68,000

The KRW 95,000-100,000 case requires several additional proof points: 10M copies, official DLC economics, capital return, and DokeV visibility.

9. Practical Stance: Hold Is Defensible; Add Later

The practical stance is simple.

Investor Situation	Stance
Existing holders	Hold is defensible
Additional buying	Wait
New investors	Only staged entry around KRW 45,000-49,000 if 2Q data remains intact
Aggressive sizing	Wait until at least two of DLC, capital return, 6M sales, or 2Q upper-half guidance are confirmed

The add conditions are clear:

First, 2Q midpoint or better: revenue above KRW 298.0B, operating profit above KRW 153.2B, and Crimson Desert revenue above KRW 250.4B.

Second, official 6M-6.5M sales confirmation, which would strengthen the 8.5M-9.0M annual case.

Third, DLC formalization: timing, price, content scale, and paid-expansion structure.

Fourth, capital return: buybacks, cancellation, or dividend policy.

Fifth, price. Below KRW 45,000, the 2027E valuation burden becomes more forgiving, but only if 2Q sales data remains intact.

Invalidation is also clear: 2Q revenue below KRW 271.3B, operating profit below KRW 129.6B, Crimson Desert revenue below KRW 224.2B, rising risk of annual sales below 8M, no DLC formalization, or long delays in DokeV / Plan 8.

Final Takeaway

Pearl Abyss has already proven 1Q26. Crimson Desert revenue of KRW 266.5B, operating profit of KRW 212.1B, and a 64.6% operating margin are enough to end the “did it work?” debate. Meritz’s 3.089M recognized-copy estimate is mathematically consistent with reported Crimson Desert revenue and implies an accounting ASP near KRW 86,300 per copy.

2Q is also not a cliff under company guidance. Applying the same ASP to 2Q Crimson Desert guidance implies 2.6M-3.2M recognized copies, with a midpoint near 2.9M. A roughly 3M-copy accounting recognition case is reasonable. But it is not the same as 3M newly sold copies in 2Q.

The real debate is 2027. The market already knows 2026 is strong. The discount is about what remains after the launch year. Meritz’s 2027E operating profit of KRW 204.1B is down about 60% from 2026E. That is why 2026E PER of 7-8x is not enough.

Pearl Abyss needs a 2027 bridge. DLC can add KRW 40B-120B in operating profit. Capital return can lower the ex-cash valuation burden. Asia/China and platform expansion can extend the sales curve. DokeV visibility can reduce the single-IP discount.

So the conclusion is not “sell the cliff” or “buy the beat.” It is: hold is defensible, but additional buying should wait for evidence that the 2027 operating profit base can move from the KRW 200B range toward KRW 300B+. The stock is no longer about whether Crimson Desert was successful. It is about whether Pearl Abyss can cross 2027 without being treated as a one-off package-game story.

This article is for research and commentary only and is not investment advice. 1Q26 results and 2026/2Q26 guidance are based on Pearl Abyss company disclosures and earnings materials. Meritz’s recognized-copy estimate of 3.089M and 2026E/2027E estimates are based on Meritz Securities research. The 2Q recognized-copy range, KRW 86,300 accounting ASP, DLC sensitivity, ex-cash PER, and scenario target prices are analyst inferences using public data and sell-side assumptions. Official company recognized-copy volume, exact China revenue mix, DLC timing/pricing/content scale, and DokeV / Plan 8 release timing were not confirmed at the time of writing. Actual results may differ. Data as of May 17, 2026 KST.

Samsung Electro-Mechanics — Neither Murata Nor Ibiden, But a 'Hybrid Challenger'. What ₩1,010,000 Is Pricing In Is 2027E OP of ₩2.7 Trillion

Sun, 17 May 2026 23:40:00 +0900

📚 Samsung Electro-Mechanics Series Samsung Electro-Mechanics AI Infrastructure Re-rating / Mirae Asset ₩1,300,000 Target Price Analysis / MLCC · FC-BGA · Optics — Three-Division Deep Dive

Read next (May 21): Silicon-Capacitor KRW 1.5T — A Reclassification, Not a Single Order — the KRW 1.5T Si-Cap contract adds an estimated KRW 100–200B to OP under base-case margins, fitting directly into this article’s “what KRW 1.01M implies for 2027E OP KRW 2.7T” framing.

📚 Related Series AI Back-End 11-Stock Data Comparison / NVIDIA Korea 2nd–3rd Tier Value Chain / May 17 Market Overview

Samsung Electro-Mechanics is neither a stable MLCC quality compounder like Murata, nor yet a proven FCBGA scarcity leader like Ibiden. But it is a high-beta challenger that simultaneously holds both axes. The current share price of ₩1,010,000 already prices in a 2027E PER of 40.5x — meaning this is not a stock you buy because it is cheap, but one that won’t look expensive until the numbers move higher. 1Q26 OPM 8.7%, Package revenue YoY +45%, 2027E Package OPM estimated at 17.9%. The real alpha is not “many themes” — it lies in whether those themes translate into Package OPM of 20% and 2027E OP of ₩2.7 trillion or more.

Key Summary

Samsung Electro-Mechanics is a high-beta challenger simultaneously holding MLCC and FCBGA. Against Murata on the quality comparison, OPM falls short (15.3% vs. 19.4%); against Ibiden on the scarcity comparison, Package margins fall short (17.9% vs. 23.1%). However, its growth rate is faster than both peers (2027E operating profit growth +52.3% vs. Murata +34.8% · Ibiden +66.7%).

Current share price ₩1,010,000, market cap ₩75.44 trillion, 2027E PER 40.5x. Hana Securities’ target price of ₩1,000,000 is derived from FCBGA peer PER 39.5x × 2027E EPS ₩24,964, nearly in line with the current price. In other words, additional upside is limited under a simple peer multiple framework. A move to ₩1,200,000–₩1,400,000 requires 2027E OP to be revised up from the base of ₩2.4 trillion to ₩2.7 trillion; ₩1,500,000 becomes possible when OPM approaches 20%, the 2028 cycle is front-run, and Package LTA is confirmed.

The four layers of upside are clear. Autonomous driving/ADAS is an earnings upside (already reflected in 1Q26 revenue), edge computing is the largest expansion axis after AI servers, robotics is a multiple option (post-2028), and space data centers are a long-term call option (starting with high-reliability MLCC for LEO satellites). On the customer side, Hyundai Mobis/Atlas is a Physical AI reference, SpaceX is an MLCC quality certification, and Tesla is the biggest customer option capable of moving both earnings and multiples simultaneously.

Assuming macro gate clearance, the entry strategy is: accumulate in tranches on ₩950,000–₩1,000,000 support, hold off if ₩900,000 is broken, trend-buy on ₩1,100,000 breakout + 2Q OP above ₩400 billion confirmed. 2Q26 earnings (August) are the key inflection point; Package OPM above 15% is the primary trigger.

1. Identity Between Murata and Ibiden

The most common trap when classifying Samsung Electro-Mechanics is forcing it into a single category. Label it an “MLCC company” and it falls short of Murata; label it an “FCBGA company” and it falls short of Ibiden. The accurate identity is a “hybrid challenger” simultaneously holding both axes.

Comparing the three companies makes the differences clear.

	Samsung Electro-Mechanics	Murata	Ibiden
Core Identity	MLCC + FCBGA simultaneously	MLCC & passive components global quality leader	Premium FCBGA scarcity leader
2027E OPM	15.3%	19.4% (FY3/27)	23.1% (FY2027)
Operating Profit Growth	2027E +52.3%	FY3/27 +34.8%	FY2027 +66.7%
Share Price Driver	MLCC price hike + FCBGA ramp + Package margin	MLCC price, mix, utilization	AI GPU/ASIC substrate shortage · LTA
Investment Character	High-growth, high-valuation event-driven	Quality compounder	Scarcity compounder

The core insight is this: Samsung Electro-Mechanics has higher growth than Murata but lower profitability, and lower Package margins than Ibiden but a more diversified business structure. The result is that “evaluating against only one peer always creates a mismatch.” The real reason the market assigns a premium to Samsung Electro-Mechanics is the option value derived from holding both businesses simultaneously.

The 1Q26 results confirm this structure. Revenue ₩3.2091 trillion, operating profit ₩280.6 billion, OPM 8.7%. By division: Components ₩1.4085 trillion, Package ₩725.0 billion, Optics ₩1.0756 trillion. Package revenue grew the fastest at YoY +45%. The company explicitly cited increased supply of high-value package substrates for AI accelerators, server CPUs, and network applications.

Based on Hana Securities estimates, 2027E divisional OPM is: Components 20.2%, Package 17.9%, Optical Solutions 3.5%. The re-rating story therefore lives in Components + Package, not Optics. The Optical Solutions segment’s dilutive effect on mix is the primary reason Samsung Electro-Mechanics’ consolidated OPM comes in below Murata and Ibiden.

2. Valuation — What Is Already Priced In

May 15 closing price ₩1,010,000, shares outstanding 74,693,696, market cap ₩75.44 trillion. TTM PER 108.1x (cycle trough effect), 2026E PER 59.2x, 2027E PER 40.5x. This is similar territory to Murata FY3/27 guidance PER of 38.2x.

Calculating implied fair values by peer multiple makes it clear how much the market has already priced in.

Scenario	PER Applied	EPS Basis	Implied Price	vs. Current
Murata basis	38.2x	2027E ₩24,964	₩954,000	-5.5%
Hana FCBGA peer	39.5x	2027E ₩24,964	₩986,000	-2.4%
Current market multiple	40.5x	2027E ₩24,964	₩1,010,000	0%
Premium granted	43.0x	2027E ₩24,964	₩1,073,000	+6.3%
Aggressive substrate premium	47.0x	2027E ₩24,964	₩1,173,000	+16.1%

The conclusion is clear. If 2027E EPS of ₩24,964 is fixed, ₩1,200,000 or above is difficult. ₩1,200,000–₩1,500,000 requires EPS upward revision.

The EV/OP basis conveys a similar message. Samsung Electro-Mechanics market cap / 2026E OP is 47.9x, nearly identical to Ibiden’s EV / FY2026 OP of 47.6x. But on a 2027E basis, market cap / OP falls to 31.5x, approaching Ibiden’s FY2027 EV/OP of 28.5x. In other words, the current share price is front-running “2027 earnings,” not “2026 earnings.”

3. Target Price Sensitivity — 2027E OP Is the Key Variable

Samsung Electro-Mechanics’ target price is ultimately a function of 2027E operating profit and the applied multiple. Based on Hana Securities, the 2027E net income / operating profit conversion rate is 80.8% (₩1.9372 trillion ÷ ₩2.3988 trillion). Therefore, every ₩100 billion increase in 2027E operating profit translates to approximately ₩1,081 in EPS (₩100B × 80.8% ÷ 74.6937M shares).

At PER 43x, the target price impact is approximately ₩46,000; at PER 47x, approximately ₩51,000. In other words, the rule holds: 2027E OP +₩100 billion = target price +~₩46,000–₩51,000.

The conditions required for each target price level are as follows.

Target Price	Required EPS	Required Conditions	Interpretation
₩1,000,000	₩25,300	2027E OP ~₩2.4 trillion	Already priced into the current share price
₩1,200,000	₩27,900 @ PER 43x	2027E OP +₩270 billion revision	Confirmable bull case
₩1,400,000	₩29,800 @ PER 47x	2027E OP +₩450 billion revision	Strong bull case
₩1,500,000	₩31,900 @ PER 47x	2027E OP +₩640 billion revision	Aggressive, pulling in 2028

The OPM sensitivity view is even clearer. Holding Hana Securities’ 2027E revenue of ₩15.6338 trillion constant and varying consolidated OPM only: at OPM 15.3% (base), fair value is approximately ₩1,070,000; at OPM 18.0% (approaching the floor of Murata and Ibiden), approximately ₩1,380,000; at OPM 20.0% (recognized as a high-margin AI components stock), approximately ₩1,540,000. ₩1,500,000 for Samsung Electro-Mechanics is not an impossible number. But the conditions are demanding. Consolidated OPM must approach 20%, which simultaneously requires Package OPM above 20%, Components OPM in the low-to-mid 20s range, and Optical Solutions’ low-margin dilution to be contained.

SK Securities’ target price of ₩1,500,000 is better characterized as front-running the 2028 higher layer-count substrate cycle rather than explained by the 2027 base alone (FCBGA revenue forecast: 2025 ₩1.153 trillion → 2027 ₩2.625 trillion). In other words, even the ₩1,500,000 target is “not unrealistic,” but it is “not fully confirmable in 2027.”

4. Four Layers of Upside — Autonomous Driving, Edge, Robotics, Space

The real depth of the Samsung Electro-Mechanics thesis lies in “what comes after AI servers.” When the perception expands to a component platform needed as AI diffuses from data centers into the physical world, multiple expansion follows. Breaking this expansion into four layers reveals different earnings visibility and share price impact for each.

Autonomous Driving/ADAS — Most Realistic Earnings Upside. Among the four themes, autonomous driving has the highest earnings visibility. Samsung Electro-Mechanics’ 1Q26 materials already explicitly cited increased supply of automotive substrates for ADAS and autonomous driving in the Package division, and the Optical Solutions division cited increased revenue from automotive camera modules for key domestic and international customers. High-voltage MLCC in the 1,000–1,500V range for EV 800V inverters and OBCs is also in development. As vehicle electrification for autonomous driving increases MLCC content per vehicle, and ADAS/autonomous driving substrate demand and higher-spec camera modules advance simultaneously, all three of Samsung Electro-Mechanics’ divisions — Components, Package, and Optics — benefit concurrently. For Samsung Electro-Mechanics, autonomous driving is not a “story” but a “mix improvement + earnings stability” factor.

Edge Computing — The Most Important Expansion Axis After AI Servers. Edge computing is the middle layer between AI servers and the physical world. As AI inference migrates from data centers to cars, robots, drones, factory equipment, security cameras, smart glasses, and industrial gateways, Samsung Electro-Mechanics’ component demand expands. AI servers use 10–15x more MLCC than general servers; this logic applies — on a reduced scale — to edge AI equipment as well. Edge devices require power stability, miniaturization, and high-speed signal handling (though not to the extent of servers), driving demand for premium MLCC and package substrates. Since Samsung Electro-Mechanics does not manufacture edge AI chips itself, the benefit is a secondary component leverage on AI chip demand.

Robotics — Multiple Option (Post-2028). Robotics is the intersection of autonomous driving and edge computing. Humanoid and industrial robots require camera modules (visual recognition), MLCC and inductors (joint and motor control), AI SoC packages and FCBGA (local AI inference), MLCC, RF components, and substrates (communications and control boards), and high-voltage, high-reliability MLCC (battery and power management). All three Samsung Electro-Mechanics divisions participate. However, humanoid shipment volumes remain small. Earnings contribution in 2026–2027 is limited; it should be viewed as a post-2028 option. It is closer to a multiple expansion factor than a near-term EPS driver. If the market comes to perceive Samsung Electro-Mechanics as expanding from “AI servers + automotive” to “AI robotics components,” a unique premium logic emerges that differentiates it from Ibiden and Murata.

Space Data Centers — Long-Term Call Option. This segment requires nuance. Commercialization of “space data centers” remains distant, and direct revenue contribution is uncertain. However, entry into high-reliability MLCC for LEO satellites and aerospace has already begun. TheElec reported that Samsung Electro-Mechanics has begun supplying MLCC for low-orbit satellites of the world’s largest private space company (customer not disclosed). Shipment volumes have not been disclosed. The significance of the space segment lies in product grade rather than volume. The extreme temperature, vibration, radiation, and longevity requirements demand reliability certifications beyond those of standard MLCC, driving higher ASP and barriers to entry. This should be seen as an extension of the high-reliability MLCC capabilities developed for automotive and AI servers.

Quantifying the target price impact: Autonomous driving/ADAS contributes 2027E OP +₩100B–₩300B (target price +₩46,000–₩153,000); edge computing +₩100B–₩400B (+₩46,000–₩204,000); robotics +₩50B–₩150B (+₩23,000–₩77,000); space/LEO +₩10B–₩50B (+₩5,000–₩26,000). Earnings materialization order: autonomous driving > edge > robotics > space; multiple impact order is roughly the reverse.

5. Customer Perspective — Hyundai Mobis, SpaceX, Tesla

Viewed through the customer lens, the Samsung Electro-Mechanics thesis centers on three key words. The most important insight is that each has a distinctly different character.

Hyundai Mobis/Atlas — Physical AI Reference. Hyundai Mobis focuses on actuators and grippers in the humanoid segment, while Samsung Electro-Mechanics can attach on the camera, MLCC, ultra-compact motor, and substrate side. The roles are different. Hyundai Motor Group targets annual robot production of 30,000 units by 2028. Samsung Electro-Mechanics also invested in Norwegian ultra-compact high-performance electric motor company Alva Industries in December 2025 to build its humanoid business foundation, and at the 1Q26 conference call disclosed plans to supply camera modules for new robotaxi applications in Q2 and for humanoid applications in the second half (customer undisclosed). However, 30,000 units is too small a scale to materially change Samsung Electro-Mechanics’ consolidated results. The key is not the number but whether Samsung Electro-Mechanics enters the Atlas/Hyundai Motor Group robotics supply chain as a reference. Once that reference is confirmed, the market begins to perceive Samsung Electro-Mechanics not as a simple electronic components company but as a “Physical AI components platform.”

SpaceX/LEO — High-Reliability MLCC Certification Effect. The SpaceX angle is about MLCC, not FCBGA. More precisely, it is about “high-reliability MLCC for low-earth orbit satellites.” The Elec reported that Samsung Electro-Mechanics is expanding its MLCC business into aerospace targeting LEO satellites and ground terminals; other reports characterized this as supply of MLCC for SpaceX low-earth orbit satellites. However, Samsung Electro-Mechanics stated that it is difficult to specifically describe customer contracts. Therefore, while interpretation as SpaceX is possible, official customer name confirmation is limited. The reason aerospace MLCC matters is that LEO satellites and ground terminals require certifications different from standard IT MLCC due to temperature variation, vibration, radiation, and longevity demands, suppliers are limited, and ASP, margin, and customer lock-in are elevated. Smartphone MLCC is heavily affected by economic and inventory cycles, while aerospace, automotive, and AI server MLCC has stronger pricing power. SpaceX is therefore less of an EPS event and more of a demonstration that Samsung Electro-Mechanics’ MLCC quality gap versus Murata is narrowing.

Tesla — The Biggest Option to Move Both Earnings and Multiple. Tesla is the largest of the three themes, because EV, robotaxi, autonomous driving chips, and Optimus all connect to Samsung Electro-Mechanics’ product lineup. MoneyToday Broadcasting reported that Samsung Electro-Mechanics is one of Tesla’s key suppliers providing MLCC, semiconductor substrates, and camera modules, and the same report mentioned that Samsung Electro-Mechanics’ FC-BGA goes into Tesla AI5. These are industry/media reports rather than official company disclosures and should be viewed conservatively. However, the direction is correct. The company cited increased supply of substrates for AI accelerators, server CPUs, networking, and ADAS/autonomous driving in the 1Q26 Package division, and disclosed plans to begin full-scale supply of AI data center networking substrates to a new big-tech customer in Q2. Tesla matters because camera modules (eyes), MLCC (power and signal stabilization), and FCBGA (AI5/FSD/robotics computation chip packaging) attach simultaneously — three product lines at once. FCBGA in particular is high-margin, high-multiple revenue. If Samsung Electro-Mechanics’ FCBGA supply for Tesla AI5 or the robotics computation module is officially confirmed, Samsung Electro-Mechanics could be re-rated from a simple MLCC/camera company to a “Tesla Physical AI computing value chain.”

Summarizing each customer in one line: Hyundai Mobis is a reference, SpaceX is a reliability certification, and Tesla is earnings upside. Rationally ranking near-term EPS impact: Tesla > SpaceX > Hyundai Mobis; multiple impact: Tesla ≥ SpaceX > Hyundai Mobis.

6. Red Team — Exit Conditions and Risks

Technology risk. If Package division OPM of 17.9% does not quickly move toward 18–20%, the re-rating thesis weakens. If the FCBGA higher layer-count and SAP technology gap with Ibiden fails to narrow, the scarcity premium logic collapses.

Demand risk. If AI server CapEx decelerates, MLCC and FCBGA demand are simultaneously pressured. Hyperscaler guidance for 2027–2028 is the key, and NVIDIA Q1 FY27 results on May 21 are the first verification point.

Margin risk. If MLCC price hikes remain at the distribution channel level and do not spread to direct AI server customers and automotive OEMs, price increases stall. If Optical Solutions OPM of 3.5% deteriorates further, consolidated OPM could retreat from the base of 15.3%.

Macro and share price risk. If the seven macro gates covered in the May 17 market overview are not cleared, Korean foreign investor fund flows will not recover, and stocks trading above PER 40x are directly hit. If PER 40.5x de-rates one step to 35x (with EPS fixed), the share price would adjust to approximately ₩870,000, or -14%.

Fundamental exit conditions are clear. If 2Q26 OP misses KB Securities’ estimate of ₩407.3 billion by more than -10%; if Package OPM stagnates in the 12–14% range; if the 2027E OP consensus is revised down from ₩2.4 trillion to ₩2.2 trillion or below; if new big-tech customer FCBGA revenue materialization is delayed; if MLCC price increases remain at raw material cost pass-through levels — any two or more of these conditions trigger thesis review.

7. Entry Guide

The judgment is Watchlist / Conditional Buy. Additional accumulation is rational after macro gate clearance + 2Q earnings confirmation.

Action by price level is as follows.

Price Range	Action	Condition
Below ₩900,000	Aggressive staged accumulation	Majority of macro gates cleared, fundamental thesis intact
₩950,000–₩1,000,000	First tranche accumulation	Continued foreign buying, Package revenue expansion signals
₩1,000,000–₩1,100,000	Chasing is inefficient	Current range; holding is OK
₩1,100,000–₩1,150,000 breakout	Trend buy possible	2Q OP above ₩400 billion confirmed
₩1,200,000 breakout	Strong trend signal	2027E OP consensus revised up to ₩2.7 trillion+
Break below ₩900,000	Reduce position	Foreign selling + fundamental deterioration

Six additional upside triggers: 2Q26 operating profit above ₩400 billion (KB Securities estimate ₩407.3 billion); Package division margin revision (2026: above 14%, 2027: approaching 20%); 2027E OP consensus revision from ₩2.4 trillion to ₩2.7 trillion or above; MLCC price hike spreading directly to AI server customers; FCBGA LTA and customer investment support structure confirmed (Ibiden-style customer-backed capex); new big-tech customer networking, AI accelerator, and server CPU substrate revenue materialization.

Cross-referencing with related posts: this post provides the detailed overlay on Samsung Electro-Mechanics within the AI back-end 11-stock matrix. Clearance of the seven macro gates from the May 17 market overview is a prerequisite for new entry. The stocks covered in the NVIDIA Korea 2nd–3rd tier value chain post are in a different category — Samsung Electro-Mechanics is closer to “earnings-confirmed core” while that post covers “expansion candidates.” Combined with the CloudMatrix Korea alpha post, a unified view of “the relative appeal of Korean component stocks in an AI decoupling environment” emerges.

8. Plain-Language Summary

When understanding Samsung Electro-Mechanics, the Murata vs. Ibiden comparison is the most important starting point. Murata is Japan’s global MLCC leader — MLCC being the “electrical signal stabilization component” that goes into every electronic device. Ibiden is Japan’s global leader in premium semiconductor substrates (FCBGA) — FCBGA being the most expensive component that mounts AI chips like NVIDIA’s H100, B200, and GB200 onto boards.

Samsung Electro-Mechanics is the Korean company that does both simultaneously. The core point is that a single company competes with global top players in both segments. However, it is not No. 1 in either. MLCC falls short of Murata; FCBGA falls short of Ibiden.

Yet the market does not penalize Samsung Electro-Mechanics for these “shortfalls” — it assigns option value for holding both businesses simultaneously. Murata and Ibiden are each at the top of their respective domains, leaving limited room to go higher. Samsung Electro-Mechanics falls short in both — which means it can improve in both. That is the meaning of “hybrid challenger.”

Adding to this, if the market’s perception expands from “AI server component stock” to “Physical AI component stock,” multiples could rise another step. As AI diffuses from data centers to cars, robots, and space, all three Samsung Electro-Mechanics products (MLCC, FCBGA, camera modules) benefit simultaneously. Autonomous driving is already in revenue; edge computing is the biggest expansion after AI servers; robotics is a post-2028 option; and space begins with high-reliability MLCC for LEO satellites.

On the customer side, three interesting keywords: if Hyundai Mobis handles actuators in humanoids and Samsung Electro-Mechanics supplies cameras and MLCC, the picture becomes a “Korean Physical AI supply chain reference.” If SpaceX truly uses Samsung Electro-Mechanics MLCC in LEO satellites, that constitutes an “aerospace-grade certification” (customer not officially disclosed). Tesla connects all of EV, robotaxi, and Optimus to Samsung Electro-Mechanics’ product suite — if FCBGA supply for AI5/FSD/Optimus computation modules is officially confirmed, earnings and multiples can both move up a step.

From an investment perspective, a sober view is necessary. The current share price of ₩1,010,000 already front-runs 2027 earnings. This is not a stock you buy because it is cheap — it is one that won’t look expensive until the numbers move higher. To reach ₩1,200,000–₩1,400,000, 2027 operating profit must rise from the ₩2.4 trillion base to ₩2.7 trillion or above. To reach ₩1,500,000 requires either pulling in the 2028 cycle or Package OPM approaching 20%. Neither is impossible, but “buying before confirmation” carries significant risk.

The conclusion is therefore clear. Samsung Electro-Mechanics is a good company, catching up with global top players in both axes (MLCC and FCBGA). But chasing at the current price is inefficient, and meaningful staged accumulation is rational only when three conditions are met simultaneously: macro gate clearance + support confirmed at ₩950,000–₩1,000,000 + 2Q26 OP above ₩400 billion. “Many themes” is not the real trigger — “2027E OP ₩2.7 trillion + Package OPM 20%” is.

9. Final Line

Samsung Electro-Mechanics is neither a stable MLCC quality compounder like Murata, nor yet a proven FCBGA scarcity leader like Ibiden. But it is a hybrid challenger simultaneously holding both axes. The market grants a premium for the “option value of these shortfalls.”

1Q26 delivered revenue ₩3.21 trillion, operating profit ₩280.6 billion, OPM 8.7%, absorbing one-time costs, with Package revenue growing the fastest at YoY +45%. The crux is 2027E Package OPM of 17.9% and Components OPM of 20.2% — the gap versus Murata FY3/27 OPM 19.4% and Ibiden FY2027 OPM 23.1% is still the room for additional upside.

The current share price of ₩1,010,000 and market cap of ₩75.44 trillion already price in 2027E PER of 40.5x. Simple peer multiples make ₩1,200,000 or above difficult. Meaningful upside opens when 2027E operating profit is revised from the base of ₩2.4 trillion to ₩2.7 trillion (₩1,200,000–₩1,400,000), and when OPM approaches 20% and the 2028 cycle is front-run (₩1,500,000).

The four upside layers are: autonomous driving/ADAS (earnings), edge computing (next expansion after AI servers), robotics (multiple option), and space data centers (long-term call option). The three key customer keywords are: Hyundai Mobis (Physical AI reference), SpaceX (MLCC quality certification), Tesla (simultaneous earnings and multiple variable). Official confirmation of Tesla AI5/Optimus FCBGA supply is the single largest catalyst.

The entry guide is clear. Chasing is inefficient; first tranche on ₩950,000–₩1,000,000 support; hold off on ₩900,000 break; trend buy on ₩1,100,000 breakout + 2Q OP above ₩400 billion confirmed. Clearance of the seven macro gates from the May 17 market overview is a prerequisite for new entry. 2Q26 earnings (August) are the key inflection point, and Package OPM above 15% is the primary trigger. It is certainly a good company. Whether it is a good entry price is still being confirmed. “Many themes” is not the point — “2027E OP ₩2.7 trillion + Package OPM 20%” is the real test of the thesis.

This post is for research and commentary purposes only and does not constitute investment advice. Samsung Electro-Mechanics 1Q26 results (revenue ₩3.2091 trillion, operating profit ₩280.6 billion, OPM 8.7%; by division: Components ₩1.4085 trillion, Package ₩725.0 billion, Optics ₩1.0756 trillion; Package YoY +45%) are based on official company disclosures. May 15 closing price (₩1,010,000), market cap (₩75.44 trillion), shares outstanding (74,693,696), and TTM PER (108.1x) are based on market data and the company’s IR page and may vary by date. Hana Securities 2027E estimates (revenue ₩15.6338 trillion, OP ₩2.3988 trillion, net income ₩1.9372 trillion, EPS ₩24,964; divisional OPM: Components 20.2%, Package 17.9%, Optics 3.5%) and target price ₩1,000,000 are based on Hana Securities research. KB Securities 2Q26 estimates (revenue ₩3.33 trillion, OP ₩407.3 billion), SK Securities target price ₩1,500,000, and FCBGA revenue forecast (2025: ₩1.153 trillion → 2027: ₩2.625 trillion) are based on respective brokerage research. Murata FY3/27 guidance (revenue ¥1.960 trillion, OP ¥380 billion, OPM 19.4%, ROIC 12.3%) and Ibiden FY2025 revenue ¥416.2 billion / OP ¥62.0 billion, FY2026 revenue ¥500 billion / OP ¥90 billion / OPM 18.0%, FY2027 mid-term target revenue ¥650 billion / OP ¥150 billion / OPM 23.1%, and FY2030 target revenue ¥1 trillion+ / OP ¥300 billion+ / OPM 30% are based on respective official company materials. Market cap, PER, and EV/OP comparison figures are based on market data sources including StockAnalysis. Peer multiple implied fair values (Murata 38.2x → ₩954,000; Hana FCBGA peer 39.5x → ₩986,000; premium 43x → ₩1,073,000; aggressive 47x → ₩1,173,000), required EPS and OP by target price, and OPM sensitivity fair values (15.3% → ₩1,070,000; 18.0% → ₩1,380,000; 20.0% → ₩1,540,000) are based on the analyst’s own calculations and will differ under different assumptions. The four-layer impact estimates for autonomous driving/ADAS, edge computing, robotics, and space data centers (2027E OP contribution ranges and target price impact) are the analyst’s sensitivity analysis and are not company guidance. Samsung Electro-Mechanics 1,000–1,500V high-voltage MLCC development, AI server MLCC usage at 10–15x, and 1Q26 plans for full-scale networking substrate supply to a new big-tech customer are based on official company materials and news sources. Humanoid camera module supply plans for 2H 2026, Alva Industries investment (December 2025), and 1Q26 conference call disclosures (Q2 robotaxi camera modules, 2H humanoid camera modules) are based on company materials; specific customer names are not disclosed. Hyundai Mobis actuator cost share of 60%, Atlas revenue starting from 2027, and Hyundai Motor Group’s 2028 annual production target of 30,000 robots are based on company and media reports. Samsung Electro-Mechanics LEO satellite MLCC supply (reported by TheElec and domestic media, referencing the world’s largest private aerospace company) has no official customer name disclosed; the SpaceX interpretation is the analyst’s estimate. Reports of Tesla AI5 FCBGA, robotaxi, and Optimus-related component supply (MoneyToday Broadcasting, Chosun Biz) are industry/media estimates and are not official company disclosures. Global macro conditions (US interest rates, oil prices, exchange rates, VIX) may have additional impact on stock prices. The analysis may be wrong. Data reference date: May 17, 2026 KST.

2026/05/17 Macro Snapshot — The Nature of the Complex Risk-Off, 7 Recovery Triggers, and Where Korean Alpha Really Stands

Sun, 17 May 2026 21:05:00 +0900

📚 Related Series US-China Summit Tech Decoupling Analysis / CloudMatrix Korean Alpha Analysis / Long-Rate Stabilization Triggers / Macro Cycle Synthesis / Korea Foreign Investor Flow Analysis

Markets wobbled in the latter half of last week — and not from a simple rate shock. An oil price surge, rising long-term real rates, dollar strength, crowded AI positioning, and weak Chinese credit all moved in the same direction at the same time, producing a five-layer complex risk-off. WTI touched the $105 range; Brent, $109. US April PPI re-accelerated to roughly +6% YoY. The 10-year Treasury threatened 4.6%, and USD/KRW stayed anchored in the 1,500s. In this environment, buying on stock-level thesis alone raises the probability of loss. Recovery is possible, but its starting point is not “the economy is strong” — it is confirmation that things are not getting any worse.

Executive Summary

The current correction is a five-layer complex risk-off. An oil shock, rising long-term real rates, dollar strength, crowded AI positioning, and weak Chinese credit are all acting simultaneously.
The recovery triggers are well-defined. The key thresholds are WTI below $100, Brent below $105, US 10-year below 4.50–4.55%, 10-year real rate below 2.0%, DXY below 99, USD/KRW breaking below 1,500, and HYG stabilization.
This week’s policy event is the Paris G7. More important than any coordinated rate action is the messaging on energy supply, FX volatility, long-bond market functioning, and NBFI leverage management.
Do not read the Warsh Fed as simply dovish. What markets want is not an immediate cut — it is restoration of Fed credibility and market functioning.
Market rates can fall without a Fed cut. A combination of RMP, SRF, FIMA repo, Treasury buybacks, long-end issuance restraint, eSLR relief, Hormuz passage normalization, and partial tariff relief can constitute stealth easing.
Korea’s beta recovery requires tougher conditions than the US. Beyond US long rates, USD/KRW, Chinese credit, and foreign semiconductor flows all need to turn around together.
The investment action is not to expand beta. This is the phase to shorten duration toward AI infrastructure earnings names, preserve optionality in cash, and wait for the macro gates to open.

1. The Nature of the Complex Risk-Off

Reading last week’s correction as “stocks fell because US rates rose” produces the wrong prescription. The actual transmission mechanism was more layered.

Hormuz tension and an oil price spike came first. Rising oil amplified concerns about PPI and CPI re-acceleration. As inflation risk got repriced, US 10-year and 30-year yields rose together, pushing real rates above 2%. Higher real rates compressed AI growth-stock multiples, triggering deleveraging from already-crowded AI infrastructure positions. Compounding this, weak Chinese credit data and the entrenchment of USD/KRW above 1,500 deteriorated foreign investor flows into Korea.

Axis	Recent Signal	Market Implication
Oil	WTI ~$105, Brent ~$109	Inflation risk premium re-elevated
Prices	US April PPI +6.0% YoY reported	Fed credibility under pressure
Rates	US 10-year threatening 4.6%	Growth-stock multiple compression
Dollar	DXY near 100, USD/KRW ~1,500	EM / Korea flow deterioration
China	April new RMB loans –¥10bn	Weak domestic demand, real estate, credit
Positioning	High-beta AI names correcting in tandem	Crowded-position unwinding

No single factor resolving in isolation is sufficient. Even if the Fed turns slightly less hawkish, inflation expectations will drag long rates back up if oil stays at $105–110. Even if rates decline, Korean foreign flows will struggle to recover if Chinese credit and the renminbi remain weak. This correction was not the result of a single-variable shock — it was the outcome of multiple variables moving in the same direction simultaneously.

Recovery therefore also requires simultaneity. At least three to four of the following need to improve together: oil, long rates, real rates, the dollar, the Korean won, credit, and foreign equity flows.

2. The 7 Recovery Triggers

The single most important indicator right now is oil. The root of the rate shock is Hormuz and energy prices. Without oil stabilizing, long rates will not come down easily — and without long rates falling, AI multiple recovery is capped.

Trigger	Improvement Threshold	Significance	Current Reading
WTI	Below $100	Eases inflation re-acceleration concern	Not yet cleared
Brent	Below $105	Reduces Hormuz risk premium	Not yet cleared
US 10-Year	Below 4.50–4.55%	Relieves growth-stock multiple pressure	Not yet cleared
10-Year Real Rate	Below 2.0%	Creates room for AI / growth recovery	On the boundary
DXY	Below 99	Improves EM / Korea flows	Not yet cleared
USD/KRW	Break below 1,500	Restores Korea beta	Not yet cleared
HYG	Rebound / spread stability	Confirms no credit stress	Needs confirmation

Clearing one or two triggers is likely to produce only a short-term bounce. Three to four triggers clearing means volatility abates; five to six means a genuine recovery can be called; seven triggers all cleared approaches a reverse risk-on.

The one-to-three-week scenario distribution is as follows.

Scenario	Probability	Conditions	Market Response
Technical rebound / volatility relief	55–60%	Further oil/rate spikes contained; G7/China channels partially ease	KOSPI rebound of 5–8% possible; partial foreign inflow return
Sideways / sector differentiation	25–30%	Rates stay elevated but HYG/credit holds	AI infrastructure earnings names separate from zero-cash-flow AI themes
Additional risk-off	15–20%	WTI stays above $105, 10-year settles above 4.6%, USD/KRW remains fixed above 1,500	KOSPI declines further; foreign selling accelerates

The key is not “good news arrives” but “things stop getting worse.” This is not a phase to buy optimism — it is a phase to confirm whether further deterioration has stopped.

3. Paris G7: May 18–19

The first policy event this week is the G7 Finance Ministers and Central Bank Governors meeting in Paris. Based on Canadian Finance Ministry and French G7 Finance Track materials, the meeting runs May 18–19 in Paris. Japan’s finance minister noted that global bond-market volatility is likely to be discussed — meaning this is not a routine diplomatic event but also carries the character of a long-bond market stability meeting.

That said, the probability of G7 collectively agreeing to “stop raising rates” is low. Central bank independence is a constraint, and the drivers of rate increases differ across countries.

Country	Core Driver of Rate Rise	Direct Coordination Feasibility
United States	Oil, PPI, Fed credibility	Low
Japan	Weak yen, import prices, BOJ normalization	Low
United Kingdom	Fiscal credibility, gilt market fragility	Low

What is feasible is not coordination to suppress rates directly, but coordination to reduce the forces pushing rates higher: energy supply stability, maritime passage normalization, curbing excessive FX volatility, long-bond market functioning, fiscal credibility messaging, and NBFI leverage monitoring.

Key phrases to watch in the G7 communiqué:

Positive Keywords	Negative Keywords
energy supply security	Only generic “stability” language repeated
global bond market functioning	Each country’s prerogative emphasized
exchange rate volatility concern	Only “monetary independence” stressed
coordinated response	No specific energy or bond market language
Strait of Hormuz	No Hormuz mention

A boilerplate statement will likely be read by markets as “every country for itself.” Conversely, if energy, FX, and long-bond market functioning are mentioned together, markets can at minimum interpret it as a signal suppressing further spikes.

4. The Warsh Fed

Chair Powell’s term ended May 15. The Fed designated Powell as chair pro tempore while Warsh’s swearing-in is pending. Accordingly, through the June FOMC, there may be a mix of Powell’s interim posture and Warsh transition messaging.

The conventional market narrative — “Warsh is conservative, therefore market-friendly, therefore ultimately a cutter” — is tempting but dangerous. Warsh’s core preoccupations are not the level of rates per se, but Fed credibility, the balance sheet, the boundary of market intervention, and the separation of fiscal and monetary policy.

The most rational path for the Warsh Fed looks closer to the following:

Axis	Expected Message
Fed credibility	Credibility restoration before premature cuts
Inflation	Distinguishing transient oil shocks from structural inflation
AI productivity	Acknowledging medium-term dis-inflationary potential
Balance sheet	Principled normalization
Market functioning	Use of technical stabilization tools such as RMP and SRF

Reuters reported that markets have begun pricing roughly a 60% probability of a Fed rate hike by the January 2027 FOMC. Markets are not already pricing Warsh as dovish. In fact, if a premature easing signal emerges immediately after he takes office, long bonds could sell off further — it could be read as a fiscal dominance signal.

What markets want from the Warsh Fed is therefore not an immediate cut. It is a credible pragmatist: someone who leaves the policy rate unchanged but stabilizes market functioning and lowers the risk premium embedded in long rates.

5. Policy Tools to Lower Market Rates Without a Fed Cut

This is the most important insight in this piece. Market rates are not determined by the Fed funds rate alone.

Component	Policy Tools to Reduce It
Expected policy rate path	Fed communication
Liquidity premium	RMP, SRF, FIMA repo, swap lines
Term premium	Treasury long-end issuance restraint, buybacks
Fiscal premium	Fiscal credibility, issuance structure stability
Inflation premium	Oil stabilization, tariff relief, supply-chain normalization
Dealer balance sheet premium	eSLR relief, Treasury market structure improvement

In other words, there are ways to lower market rates without cutting the policy rate. I call this combination composite stealth easing.

On the Fed side, RMP, SRF, and FIMA repo are the core instruments. RMP is a technical purchase of short-term Treasuries for reserves management — closer to a technical buy to maintain ample reserves than to QE. SRF caps the upper bound of short-term funding market rates. FIMA repo and swap lines reduce the risk that foreign central banks dump Treasuries into the market to secure dollars.

On the Treasury side, long-end issuance restraint and buybacks matter most. In the May QRA, Treasury indicated it would hold coupon and FRN auction sizes for at least several quarters, with unexpected borrowing needs met via bills and CMBs. This reduces duration supply pressure at the long end. Treasury buybacks are not a policy to reduce total debt but to reduce the liquidity premium on off-the-run Treasuries — useful for easing auction tail risk and bringing MOVE lower.

On the regulatory side, eSLR relief is available. The eSLR modification effective April 1, 2026, expands large-bank and dealer capacity to intermediate Treasuries and repo. Easing dealer balance sheet constraints can compress the long-bond liquidity premium. That said, LCR relief — as Michael Barr warned — carries significant financial system resilience controversy. eSLR is feasible; going as far as LCR is a different matter.

On the executive side, Hormuz passage normalization and partial tariff relief are the most direct levers. Lower oil directly reduces the inflation premium; tariff relief directly eases PPI/CPI pressure.

Tool	Effect	Limitation
RMP	Stabilizes repo/SOFR; eases forced deleveraging	Limited direct impact on 10s/30s
SRF	Caps upper bound of short-term funding rates	Stigma management required
FIMA repo / swap lines	Prevents forced selling of Treasuries by foreign central banks	Effectiveness varies with foreign CB demand
Long-end issuance restraint	Reduces term premium	Excessive bill issuance can burden short-term markets
Treasury buybacks	Improves off-the-run liquidity	Does not reduce total debt
eSLR relief	Expands dealer capacity	Financial deregulation controversy
Hormuz normalization	Reduces oil risk premium	Diplomatic/security uncertainty
Tariff relief	Eases CPI/PPI pressure	Political cost

The dangerous tools are a separate category: long-end QE, premature rate cuts, MBS purchases, LCR relief, and excessive TGA drawdowns. What markets want is not overt monetary financing but technical stabilization that reduces market-functioning disruptions and risk premiums.

6. Hormuz: The Single Most Important Diplomatic Lever

Hormuz is the single most important variable in this episode. A common market trap is the assumption that “China is close to Iran, so it won’t cooperate on passage normalization.” In reality, China is the largest buyer. China is the primary purchaser of Iranian crude and a major energy importer — prolonged Hormuz closure hurts China too.

Per Reuters reporting, Trump discussed with Xi the possibility of lifting sanctions on Chinese companies buying Iranian oil, and USTR Greer noted that China wants the Strait of Hormuz open without restrictions. This means China now has an incentive to pressure Iran toward passage normalization.

The realistic path is not a comprehensive peace agreement — it is a narrow maritime deal that starts with partial passage and gradually brings down insurance premiums and freight rates.

Stage	Confirmation Signal
1	Repeated confirmed transits by Chinese vessels
2	Expanded passage for Qatar LNG, Iraqi crude, Pakistan-bound vessels
3	AIS normalization for neutral commercial ships
4	Insurance premium and freight rate decline
5	US–Iran narrow maritime deal
6	Oil risk premium compression

Reports citing Fars News indicated that some Chinese vessel transits have been permitted; the FT reported a Qatar LNG vessel transit case. Full normalization has not been reached, but partial passage has already emerged as a trackable market variable.

This week’s monitoring keywords are clear. Positive signals: Chinese vessels passing, LNG transit, AIS normalization, insurance premium decline, narrow maritime deal. Negative signals: vessel attacks, Iranian seizures, US–Iran combat, tanker insurance surge.

7. Korea Implications

The conditions for a Korean equity recovery are more demanding than for the US. For the US, oil, long rates, and VIX are the core variables. Korea adds USD/KRW, Chinese credit, and foreign semiconductor flows on top.

Reuters reporting that China’s April new RMB loans came in at –¥10 billion is significant — far below the +¥300 billion consensus. This signals weak Chinese domestic demand, real estate, and private credit appetite, and can translate into RMB depreciation pressure. A weaker renminbi weakens the won, which delays recovery in Korean foreign equity flows.

The recovery of Korean foreign equity flows is likely to unfold in four stages.

Stage	Condition	Market Implication
1	Oil / rates / dollar stabilization	Macro gate cleared
2	USD/KRW breaks below 1,500	Korea beta restored
3	Foreign net buying resumes in large-cap semiconductors	Downside defense for the index
4	Broadening to mid-cap AI infrastructure	Stock-level alpha recovery

Stage 1 is not yet fully satisfied. Therefore, rather than expanding index beta, the right move is to shorten duration toward AI infrastructure names with confirmed earnings.

Category	Current Action
Macro-gate-agnostic names	Cycle-agnostic B2B SaaS, lower-bracket bifurcation plays, select medical devices: holdable
Buy after macro gate clears	HBM, advanced packaging, substrates, power, cooling, networking
Immediate caution	Zero-cash-flow AI themes, high-PER duration stocks, China-credit-sensitive cyclicals
Structural improvement names	Prioritize names with numbers confirmed — e.g., 하나마이크론, 티씨케이
Inflection-point bets	Conditional approach for high-volatility pre-confirmation names such as Nextin, Jeju Semiconductor

The position of Korean alpha is not “buy because it’s cheap.” This is the phase to prepare the names that have posted earnings, can weather the macro gate, and will be the first buys when foreign flows return.

8. Asset-Level Guidance and Portfolio

In the current environment, the portfolio goal is not to expand beta. It is to preserve optionality while preparing the names to buy when the macro gates open.

Asset	Stance	Weight Guide	Rationale
Nasdaq	Below neutral	0.8x	Long-rate / real-rate burden
Korean equities	Below neutral	0.7–0.8x	USD/KRW, Chinese credit, foreign flow headwinds
AI infrastructure earnings names	Selective positive	Hold	HBM, substrates, test, power, cooling: EPS defensible
Zero-cash-flow AI themes	Negative	Reduce	Directly exposed to rising real rates
Cash	Positive	Increase	Optionality value
Dollar	Partial hold	Hold	Defensive until USD/KRW breaks below 1,500
Gold	Partial hedge	Hold	Oil / geopolitical hedge
Bitcoin	Conditional	Partial reduce	Risk if HYG weakens in tandem

Korean equities can be structured in four tiers.

Tier	Weight Example	Characteristics
Core	40%	Large-cap HBM / AI infrastructure; hold-oriented
Satellite	30%	Structural improvement / confirmed-earnings names: 오이솔루션, 티씨케이, 삼성전기, 하나마이크론
Option	15%	Conditional inflection bets: Nextin, Komico, etc.
Hedge	15%	Cash, dollar, cycle-agnostic B2B SaaS, lower-bracket bifurcation plays

Portfolio actions can also be mapped mechanically to macro gate score.

Macro Gates Cleared	Action
0–1	Trim core partially; defer new satellite/option buys; raise cash to 25–30%
2	Hold and watch; monitor stock-level tape only
3–4	Maintain core; begin partial satellite accumulation; evaluate option entries
5–6	Genuine recovery; add to AI infrastructure earnings names
7	Reverse risk-on; but limit chasing overheated names

9. Weekly Checklist

Events are concentrated this week: G7, Hormuz news, Google I/O, NVIDIA earnings, and Korea CPI/export data all land in the same week.

Date	Checkpoint
Mon 5/18	Paris G7 Day 1; Korea foreign equity flows; US long rates
Tue 5/19	Paris G7 joint communiqué; Hormuz news; US retail sales
Wed 5/20	Google I/O; Korea preliminary export data; China industrial production
Thu 5/21	NVIDIA FY2027 Q1 earnings; Korea PPI
Fri 5/22	US new home sales; Korea cumulative foreign flows; global PMI
Sat 5/23	Weekly macro gate review

Daily monitoring: WTI/Brent, US 10-year/30-year, DXY/USD-KRW, HYG, KOSPI foreign flows. Two to three times weekly: real rates, VIX/MOVE, Chinese yuan, Korea CDS, Korean semiconductor stock flows.

NVIDIA earnings is a standalone event, but in this macro context, “good earnings” alone is not sufficient. Even if Q2 guidance and gross margins impress, Korean AI infrastructure names may respond only modestly if US 10-years and oil fail to fall. Conversely, even an in-line NVIDIA print combined with simultaneous stabilization in oil, rates, and FX could create a re-entry window after an initial sell-the-news reaction.

10. Final Conclusions

Last week’s market wobble was not a simple rate shock. It was a complex risk-off in which an oil shock, rising long-term real rates, dollar strength, crowded AI positioning, and weak Chinese credit all hit simultaneously. Recovery is possible — but its starting point is not “the Fed cuts immediately.” The sequence is: Hormuz de-escalation → oil decline → long-rate stabilization → dollar weakness → restoration of foreign inflows into Korea.

The decisive event this week is the Paris G7. G7 cannot directly suppress rates. But it can deliver coordinated messaging on energy supply, maritime passage, FX volatility, long-bond market functioning, and NBFI leverage — and even that much reduces the probability of a further spike.

The Warsh Fed should not be read as simply dovish. What markets want is not an immediate cut but restoration of Fed credibility and market functioning. If RMP/SRF, FIMA repo, Treasury buybacks, long-end issuance restraint, eSLR relief, Hormuz passage normalization, and partial tariff relief all operate together, market rates can fall without cutting the policy rate. This is the core policy mechanism of the current episode.

For Korea, one additional condition applies. US long rates coming down alone is not enough. USD/KRW breaking below 1,500, Chinese credit stabilizing, and foreign semiconductor flows recovering are all needed in parallel. The strategy for now is therefore not to grow beta but to shorten duration toward AI infrastructure earnings names and preserve optionality in cash.

This correction is not a confirmed structural bear market. But it is also not a simple dip-buying window. It is worth waiting until at least three to four of the seven macro gates open. If Hormuz passage normalization and G7 bond-and-energy stability messaging arrive simultaneously, this correction can shift from structural decline to a post-crowded-position-unwind re-entry phase.

Sources and References

Reuters, Oil prices climb on fears of new US-Iran combat
Reuters, US producer prices surprise with largest increase in four years
Canada Department of Finance, G7 Finance Ministers and Central Bank Governors meeting in Paris
Reuters, Japan finance minister says G7 likely to discuss bond volatility
Federal Reserve, Powell named chair pro tempore until Warsh is sworn in
Reuters, Markets begin eyeing a Fed rate hike around the turn of the year
Federal Reserve, Implementation Note issued April 29, 2026
Federal Reserve, Standing Repurchase Agreement Operations
U.S. Treasury, Quarterly Refunding Statement
Federal Reserve, Agencies issue final rule to modify certain regulatory capital rules
Reuters, Fed’s Barr warns against lowering liquidity rules
Reuters, China April new loans unexpectedly shrink
Reuters, Trump spoke with Xi about lifting sanctions on Chinese companies that buy Iranian oil
Reuters, China wants Strait of Hormuz open free of curbs

This post is intended solely for research and commentary purposes and does not constitute investment advice. Data on WTI, Brent, US PPI, China new RMB loans, G7 schedule, Fed chair pro tempore designation, Warsh swearing-in pending status, and RMP/SRF/Treasury/eSLR details are based on public reporting and official sources. Scenario probabilities, the 7 recovery triggers, asset weight guidance, and Korean stock classifications represent the analyst’s subjective framework and may differ from actual outcomes. Hormuz passage normalization, G7 communiqué content, Warsh Fed operating style, and Chinese credit stabilization are all uncertain. The global macro environment can change daily and this analysis may prove incorrect. Data as of May 17, 2026 KST.

Google I/O and NVIDIA Earnings: Two Events Korean Semiconductor Investors Should Watch

Sun, 17 May 2026 00:00:00 +0000

📚 Related Series Two Betas of AI Back-End · AI Back-End — 11 Stocks Side by Side · Samsung Electronics vs TSMC Re-Rating Thesis · AI HBM Hub

Two very different AI events arrive next week. Google I/O starts with the main keynote at 10:00 a.m. PT on May 19, or 2:00 a.m. KST on May 20. NVIDIA then publishes its fiscal 2027 first-quarter results and CFO commentary at about 1:20 p.m. PT on May 20, followed by the conference call at 2:00 p.m. PT. In Korea, that is 5:20 a.m. and 6:00 a.m. KST on May 21.

They are both AI events, but they answer different questions. Google I/O is about AI use-case expansion: where Gemini, Android, XR, AI PCs and agents can go next. NVIDIA earnings are about AI infrastructure confirmation: whether orders are still turning into revenue and whether the gross-margin structure remains intact. For Korean semiconductors, NVIDIA is the more direct event because HBM, FC-BGA substrates, MLCC, server boards and test sockets all sit downstream from NVIDIA Data Center demand.

Key Takeaways

Schedule: Google I/O keynote is May 19 at 10:00 a.m. PT, or May 20 at 2:00 a.m. KST. NVIDIA’s FY2027 Q1 results and CFO commentary are scheduled for about May 20 at 1:20 p.m. PT, with the call at 2:00 p.m. PT. That is May 21 at 5:20 a.m. and 6:00 a.m. KST.
The events mean different things. Google I/O is about AI moving into products and devices. NVIDIA earnings are about whether AI infrastructure CAPEX remains real.
For NVIDIA, Q2 matters more than Q1. Company guidance for Q1 is $78.0 billion plus or minus 2%. Visible Alpha consensus is about $78.5 billion in total revenue and $72.8 billion in Data Center revenue. The real question is whether Q2 guidance reaches the high-$80-billion area and whether non-GAAP gross margin stays around 75%.
Korean direct beta runs through SK hynix and Samsung Electronics; second-order beta runs through Samsung Electro-Mechanics, Daeduck Electronics and test-socket names. But SK hynix is already crowded, Samsung Electronics has labor-risk overlap on May 21, and Samsung Electro-Mechanics still needs a better entry point after a large move.
The practical stance is not to chase before the events. Good numbers can still trigger sell-the-news if they fail to beat elevated expectations. Wait for the events and the macro gates to line up.

1. These Are Not the Same AI Event

Google I/O is a product and use-case event. The market will watch Gemini, Android, XR glasses, AI laptops, in-car AI and agentic app flows. For investors, it is less about immediate revenue and more about the device specification curve for 2026 and 2027.

If Gemini becomes a deeper intelligence layer inside Android, if XR glasses become more concrete, or if Google pushes an AI-first laptop format, on-device memory requirements rise. That is supportive for LPDDR, eSSD, image sensors, camera modules and low-power components. But Google’s cloud AI infrastructure has a large TPU and custom-ASIC footprint. For Korean semiconductors, the I/O read-through is medium-term device demand, not an immediate HBM revenue bridge.

NVIDIA earnings are different. This is where AI infrastructure orders show up in revenue. Higher Data Center revenue means more GPUs, more AI systems, more HBM, more high-end substrates, more server boards, more test-socket demand and more power-stability components.

Event	Investor Question	Korean Semiconductor Read-Through
Google I/O	How far are AI use cases spreading?	Medium-term device memory, camera, sensor and on-device AI demand
NVIDIA earnings	Is AI infrastructure CAPEX still converting into revenue?	Direct HBM, FC-BGA, MLCC, server PCB and test-socket demand

That is why the key next-week question for Korean semiconductors is not what Google demonstrates, but how NVIDIA guides the next quarter.

2. NVIDIA: Q2 Guidance Matters More Than Q1

NVIDIA’s fiscal 2026 fourth quarter was extremely strong. Revenue was $68.1 billion, Data Center revenue was $62.3 billion, and non-GAAP gross margin was 75.2%. The company guided fiscal 2027 first-quarter revenue to $78.0 billion plus or minus 2%, excluding any Data Center compute revenue from China.

Yet the stock fell after the print. The issue was not weak fundamentals. The issue was expectations. For NVIDIA, a consensus beat is no longer enough. The market prices the next quarter, whisper numbers and whether AI CAPEX still has upside.

The same logic applies now. S&P Global / Visible Alpha consensus sees fiscal Q1 revenue around $78.5 billion and Data Center revenue around $72.8 billion. Those are already known numbers. If Q1 lands around $78-79.5 billion, it may be treated as in-line rather than exciting.

The five checks that matter are:

Checkpoint	Positive Signal	Negative Signal
Q1 revenue	Above $80 billion	Below $78 billion
Q2 guidance	High-$80-billion area or better	Below $80 billion or cautious tone
Gross margin	Non-GAAP 75% or higher	74% or lower
Data Center revenue	Above $73 billion	Below $72 billion
China / export-control commentary	Upside option from H20/H200 channel	Extended China gap

The most important combination is Q2 guidance plus gross margin. If NVIDIA can guide strongly while keeping gross margin near 75%, the market gets confirmation of both infrastructure demand and pricing power. If revenue is strong but margin cracks, investors will ask whether the Blackwell/Rubin transition is becoming more expensive or whether pricing power is fading.

3. Scenario Map for Korean Semiconductors

Scenario	NVIDIA Outcome	Korean Semiconductor Impact
Upside surprise	Q1 above $80B, strong Q2 guide, margin around 75%	HBM, substrates and MLCC move together. SK hynix is the direct beta; Samsung Electronics and Samsung Electro-Mechanics follow.
In line	Q1 near consensus, Q2 guide within expected range	Sell-the-news risk. More likely a positioning reset than a thesis break.
Disappointment	Q1 miss or weak Q2 guide	AI CAPEX peak-out fears return. Korean memory and back-end names correct together.
Margin shock	Revenue beats but gross margin falls below expectation	Pricing-power concern. Multiple pressure on memory and components.

The base case is probably not disaster. It is “good but already expected.” That is exactly the setup where the first reaction can be colder than the underlying fundamentals. Chasing before the event is therefore less attractive than watching the first one or two trading days after the event.

4. What to Watch by Korean Stock

SK hynix — the cleanest HBM beta, but already crowded

SK hynix is the cleanest Korean read-through to NVIDIA Data Center revenue. It has the clearest HBM leadership and the most direct earnings bridge to the AI memory cycle.

The issue is not quality. The issue is price and positioning. Investors already know SK hynix is the HBM leader. If NVIDIA prints a large beat, SK hynix can react first. If NVIDIA is merely in-line, SK hynix can also become the first profit-taking vehicle.

For SK hynix, the right question is not “is this a good company?” It is “is there enough event-driven asymmetry to add before the print?” The answer looks conservative. Existing holders can verify. New capital should likely wait for post-event volatility and the next earnings guide.

Samsung Electronics — HBM4 and integration optionality, but labor risk overlaps

Samsung Electronics is more complicated. It has HBM4, base-die, foundry and packaging optionality, which is why the earlier Samsung vs TSMC re-rating note focused on whether Samsung can move from “memory cycle stock” toward “AI integration platform.”

But May 21 also overlaps with Samsung labor-risk headlines. That means two separate events can hit the stock on the same day.

The positive combination is clear: strong NVIDIA guidance plus limited strike impact or a quick resolution. That would revive the HBM4 re-rating logic. The negative combination is also clear: in-line-or-weak NVIDIA plus extended labor uncertainty. Then the market focuses less on 5x PER and more on labor, policy and cycle risk.

Samsung Electronics is therefore less of a pre-event chase and more of a volatility candidate after the event.

Samsung Electro-Mechanics — not direct HBM, but an AI-parts re-rating candidate

Samsung Electro-Mechanics is not a direct HBM name. But as AI servers and networks expand, MLCC, FC-BGA and sensing modules become structurally more important. The stock still carries some legacy “smartphone parts” perception, leaving room for reclassification as an AI infrastructure parts platform.

The caution is valuation and price action. The stock has already moved strongly, and continued earnings upgrades are required to justify the multiple. If AI component stocks spike after NVIDIA, chasing is inefficient. A better setup is to wait for overheating to cool, then verify AI-server MLCC and FC-BGA demand in the second-quarter commentary.

Back-end names — substrates and test sockets are not the same trade

Daeduck Electronics, Simmtech and Haesung DS are more sensitive to NVIDIA Data Center guidance because high-end package substrates and server boards are volume-linked. If GPUs, CPUs, ASICs and network chips ramp, high-performance substrates and boards ramp with them.

LEENO Industrial and ISC are different. Their revenue can lag by one or two quarters, but chip complexity and generation turnover drive high-margin consumable demand. As argued in the earlier back-end comparison, substrates are short-term volume beta; test sockets are chip-complexity beta. For event-day sensitivity, substrates are higher. For long-term margin quality, test sockets are cleaner.

5. Google I/O’s Second-Order Read-Through

Google I/O can still matter. The key is whether Gemini stays in the cloud, or becomes a deeper intelligence layer across Android, XR, laptops and cars.

If Google frames Gemini as an operating-system-level intelligence layer rather than a set of app features, device specs move higher. On-device inference needs more memory capacity and bandwidth. XR glasses require compact cameras, low-power parts, display components and batteries. In-car AI extends the same stack into automotive semiconductors and sensors.

For Korean equities, that makes Google I/O relevant to Samsung memory, Samsung Electro-Mechanics camera and MLCC exposure, and selected mobile / XR component names. But the impact is medium-term. It does not create the same immediate HBM revenue bridge as NVIDIA.

6. Macro Gates Still Matter

Even a strong event can be muted if macro conditions are closed. US long rates, Japanese long rates, oil and the Korean won have all been pressuring risk assets.

For next week, the framework should be two-layered:

Layer	What to Check
Event	NVIDIA Q2 guidance, gross margin, Data Center revenue, concrete device-AI signals from Google I/O
Macro	US 10-year yield, Brent crude, USD/KRW, VIX

The best setup is strong NVIDIA guidance plus stabilizing US yields and oil. If NVIDIA is strong but USD/KRW and long rates remain unstable, Korean semiconductor stocks can rally intraday and still fade by the close.

7. FAQ

Can Korean semiconductors fall even if NVIDIA reports good numbers?

Yes. If expectations were higher, if Q2 guidance misses whisper numbers, or if macro conditions remain poor, good numbers can still produce sell-the-news. The question is not whether the report is good. The question is whether it is good enough.

Is Google I/O less important than NVIDIA for Korean semiconductors?

For near-term earnings, yes. For medium-term device demand, no. If AI moves from cloud to device, LPDDR, eSSD, camera, sensor and low-power component demand all rise. But that shows up through 2026-2027 device specifications, not next week’s HBM revenue.

How should Samsung Electronics be treated on May 21?

As a two-event stock. Strong NVIDIA guidance and limited strike impact would be constructive. Weak NVIDIA guidance plus labor escalation would be negative. That argues for post-event verification rather than pre-event chasing.

What is the less crowded idea?

SK hynix is the direct beta but also the most obvious trade. Samsung Electro-Mechanics is not direct HBM exposure, but the AI infrastructure parts re-rating is still less fully understood. That does not mean chase the stock; it means verify AI-server MLCC and FC-BGA demand after the event.

Final Thought

Next week’s two events are not the same AI event. Google I/O shows where AI can spread. NVIDIA earnings show whether AI infrastructure spending is still turning into revenue. For Korean semiconductors, the second event is more direct.

For NVIDIA, the key is not a Q1 beat. It is whether Q2 guidance reaches the high-$80-billion area and whether gross margin stays near 75%. If both are confirmed, the logic for HBM, substrates, MLCC and test sockets strengthens. If guidance is soft or margin weakens, “AI CAPEX peak-out” returns quickly.

By stock, SK hynix is the most direct but crowded HBM beta. Samsung Electronics combines HBM4 re-rating optionality with May 21 labor risk. Samsung Electro-Mechanics still has AI-parts reclassification potential, but chasing after a large move is inefficient. In back-end names, substrates are the short-term volume beta, while test sockets are the high-margin chip-complexity beta.

The practical conclusion is simple: do not chase before the events. Wait for event confirmation and macro gates to open together. The right setup is not just “NVIDIA good.” It is “NVIDIA good enough, and macro no longer fighting the trade.”

Sources

This article is for research and commentary only and is not investment advice. Google I/O and NVIDIA earnings timing are based on official Google and NVIDIA sources. NVIDIA FY2026 Q4 results and FY2027 Q1 guidance are company-reported figures. Visible Alpha consensus figures are cited via S&P Global’s earnings preview and may differ from actual results. Google I/O announcements, NVIDIA Q2 guidance, gross margin, China-related commentary and Samsung Electronics labor developments remain uncertain. Scenario analysis and stock-level interpretation are analytical judgments, not guarantees. Data as of May 17, 2026 KST.

Hana Micron vs Jeju Semiconductor — The Real Difference Between Two Earnings Surprises: Structural Improvement or Cycle Peak?

Sun, 17 May 2026 00:00:00 +0000

Related series AI Back-End 11-Stock Data Comparison · Jeju Semiconductor 1Q26 Deep Dive · May 15 Flow Accumulation Screen · Korea Semiconductor Value-Chain Hub

Both stocks exploded in 1Q26. Hana Micron delivered operating profit of KRW 72.0bn; Jeju Semiconductor delivered KRW 67.1bn. On the first trading day after each announcement, the stocks gained +18.6% and +8.9% respectively. On the surface both look like “AI-era memory back-end beneficiaries.” But the two surprises are structurally different. Hana Micron is closer to a business-model improvement story — repriced cost-pass-through at the Vina entity, margin step-up at the Brazil subsidiary. Jeju Semiconductor is closer to a cyclical windfall — LPDDR4X supply scarcity amplified by tariff-driven front-loading. The same label, “earnings surprise,” masks very different durability profiles. That distinction is the whole point of this note.

Key Takeaways

Both sets of numbers were strong. Hana Micron: 1Q26 revenue KRW 507.7bn, OP KRW 72.0bn, OPM 14.2%. Jeju Semiconductor: revenue KRW 180.5bn, OP KRW 67.1bn, OPM 37.2%.
The nature of each surprise is different. Hana Micron’s upside was driven by a repriced cost-pass-through structure at the Vina entity and a margin step-up at the Brazil subsidiary. Jeju Semiconductor’s upside was driven by LPDDR4X supply scarcity and tariff-driven pre-orders.
Durability favors Hana Micron. The 1Q margin cannot all be annualized on repeat, but Hana Micron may have established a structurally higher margin floor. Jeju Semiconductor’s 37.2% OPM is difficult to call a new normal.
Multiple illusion is real for Jeju. Simply annualizing 1Q net income implies a 9x PER — but that requires four identical quarters. Adjusting for normalization pushes the expected-value multiple to 16–17x.
Both are clearly overbought near-term. Hana Micron: +58.6% over 20 days, RSI ~78. Jeju Semiconductor: +94.3% over 20 days, RSI ~81.5, and +574% from the 52-week low.
Entry priority: Hana Micron > Jeju Semiconductor. But chasing either immediately after the print is inefficient. Real alpha lies in confirming that the surprise is not one-off — i.e., 2Q26 results — and buying into any post-confirmation weakness in tranches.

1. The Numbers Alone Were Explosive for Both

Taken at face value, both companies qualify as “earnings surprises.” Hana Micron beat consensus operating profit by roughly 30%. Jeju Semiconductor grew operating profit nearly 18-fold year-over-year.

Metric	Hana Micron	Jeju Semiconductor
Market cap (May 15)	~KRW 3.52tn	~KRW 2.83tn
1Q26 Revenue	KRW 507.7bn	KRW 180.5bn
1Q26 Operating profit	KRW 72.0bn	KRW 67.1bn
1Q26 OPM	14.2%	37.2%
1Q26 Net income	KRW 72.8bn	KRW 78.1bn
Revenue YoY	+62.8%	+273%
Operating profit YoY	+513.6%	+1,714%
Post-announcement 1D return	+18.6%	+8.9%
1Q annualized Mkt cap / OP	12.2x	10.6x
1Q annualized PER	12.1x	9.1x

Looking only at this table, Jeju Semiconductor appears cheaper — higher OPM and a lower annualized PER. But that arithmetic requires 1Q26 to repeat identically in Q2, Q3, and Q4. In memory cycles, that assumption is among the most dangerous you can make. A quarter driven by supply scarcity and pre-ordering can reverse within one reporting period.

The real question, then, is not “who had the better quarter?” It is: how repeatable is that profit?

2. Hana Micron’s Surprise Is Closer to Structural Improvement

The two pillars behind Hana Micron’s KRW 72.0bn in 1Q26 operating profit are (1) a repriced cost-pass-through mechanism at the Vina entity and (2) a margin step-up at the Brazil subsidiary.

At Vina, the company appears to have established a more stable arrangement under which raw-material cost increases are reflected in the unit prices charged to SK hynix. Traditional OSAT contracts often require the packaging house to absorb input cost volatility within a fixed unit price — so when substrate or wire prices rise, revenue grows but margins compress. If raw-material escalations are now passed through more reliably, the dynamic shifts: Hana Micron begins to share price volatility with its anchor customer rather than bearing it alone. That is a business-model change, not a one-quarter fluke.

The Brazil subsidiary is equally important. 1Q26 Brazil revenue is estimated at roughly KRW 105.6bn, with OPM in the high-teens — well above the mid-to-high single digits typical for mainstream OSAT. Regional positioning, contract structure, favorable currency translation, and a higher mix of value-added packaging appear to have combined to produce this level.

Caveats are real. Non-operating FX gains of approximately KRW 27.2bn are unlikely to repeat, and incremental operating leverage diminishes as utilization approaches its ceiling. But the more important question is where the margin floor has moved. If Hana Micron’s prior cycle average OPM was 6–8%, and this cycle can sustain 11–13%, then earnings power has structurally shifted — and the stock is rerating in response to that shift, not just to a single quarter’s number.

3. Jeju Semiconductor’s Surprise Is Closer to a Cyclical Windfall

Jeju Semiconductor’s 37.2% OPM in 1Q26 is extraordinary even by fabless memory standards. The cause, however, looks more like a supply-demand shock than a structural improvement.

The first driver is LPDDR4X scarcity. Samsung, SK hynix, and Micron have been migrating production capacity toward HBM, DDR5, and LPDDR5/5X — higher-margin, AI-driven products. The natural consequence is a shrinking supply of the mature LPDDR4X node. But IoT devices, automotive applications, and a tail of industrial and mobile platforms still require LPDDR4X in volume. Supply falls; demand persists; prices rise. Jeju Semiconductor, as a focused legacy-memory fabless house, was the most direct beneficiary of that dislocation.

The second driver is tariff-driven front-loading. As uncertainty over U.S. semiconductor tariffs escalated, buyers moved to pre-build inventory. One quarter’s revenue can balloon when customers pull forward purchases — but the following quarter then works through that stockpile rather than placing new orders. Separating genuine demand growth from demand-borrowed-from-the-future is essential when reading any 1Q26 memory print.

This is not a criticism of the company. Jeju Semiconductor has real capabilities: LPDDR4X and MCP productization experience, customer qualifications, and an ability to serve niches that tier-1 suppliers have deprioritized. When those tier-1 players shift capacity away from legacy nodes, Jeju can capture pricing power quickly. The problem is that if the primary earnings driver is supply scarcity, margins fall rapidly when supply normalizes or when pre-ordered inventory is digested.

Consequently, the 37.2% OPM in 1Q26 is better read as a cycle-peak excess margin than as a new steady state. A normalized OPM in the 15–25% range over a 2–3 year horizon is a more realistic anchor for valuation.

4. Same Label, Different Durability

The sharpest summary of the contrast is this: for Hana Micron, the question is whether the pricing structure has permanently changed. For Jeju Semiconductor, the question is how long the supply shortage lasts.

Dimension	Hana Micron	Jeju Semiconductor
Type of change	Business model improvement	Cyclical excess profit
Primary driver	Vina cost-pass-through repricing	LPDDR4X supply scarcity
Secondary driver	Brazil subsidiary margin step-up	Tariff-driven front-loading
Durability	Partially sustainable	1Q level likely unsustainable
Key confirmation metric	2Q OPM ≥ 13%, Vina & Brazil margins hold	2Q OP ≥ KRW 50bn, OPM ≥ 30%
Biggest downside risk	Margin structure proves one-off	Supply normalization, front-load unwind
Investment character	Structural improvement, confirm-and-hold	Cycle-peak timing call

This distinction matters for portfolio construction. Hana Micron is a question of whether the company’s earnings power has permanently risen. Jeju Semiconductor is a question of how long a supply-demand shock sustains. Both stories are live. The risk profiles are not the same.

5. Multiples: Where Illusion and Reality Diverge

Hana Micron’s market cap as of May 15 is approximately KRW 3.52tn. If 2026 full-year operating profit is modeled at around KRW 299.5bn, the Mkt cap/OP ratio is roughly 11.7x and the PER is approximately 19x. Not cheap on an absolute basis, but not demanding if structural improvement persists into 2027.

Scenario analysis for 2027 produces an expected value above the current price:

Hana Micron Scenario	2027E OP	Est. EPS	Applied PER	Fair Value	vs. Current
Bear	KRW 310bn	KRW 2,850	16x	KRW 45,600	–13.8%
Base	KRW 375bn	KRW 3,600	17x	KRW 61,200	+15.7%
Bull	KRW 440bn	KRW 4,250	19x	KRW 80,750	+52.6%

Weighting Bear/Base/Bull at 20%/50%/30% gives an expected value of approximately KRW 63,945 — roughly 21% above the current price. Not a screaming bargain, but still investable under a structural-improvement assumption.

Jeju Semiconductor is harder to value. Annualizing 1Q net income of KRW 78.1bn yields KRW 312.4bn in annual net income. Dividing the ~KRW 2.83tn market cap produces a 9.1x PER. That looks cheap. But it requires four identical quarters — an aggressive assumption.

If 2Q and beyond decelerate, the numbers shift materially. Modeling 2026 full-year OP at around KRW 200bn and converting to after-tax income pushes the implied PER to 16–17x. That is no longer inexpensive.

Jeju Semi Scenario	2027E OP	Applied PER	Fair Value	vs. Current
Bear	KRW 135bn	14x	KRW 41,155	–50.0%
Base	KRW 195bn	16x	KRW 67,938	–17.4%
Bull	KRW 260bn	18x	KRW 101,907	+23.8%

Weighting Bear/Base/Bull at 35%/40%/25% yields an expected value of approximately KRW 67,056 — below the current price. Adding the roughly 7.7% potential dilution from outstanding CBs and BWs compresses the Bull-case fair value further. Jeju Semiconductor is not a flawed company; it is a company whose current price already embeds significant optimism.

6. Near-Term Overheating Is Clear for Both

The danger of chasing immediately after a strong print is simple: good news is already aggressively priced in.

Hana Micron is up +58.6% over 20 trading days with an RSI of approximately 78. Jeju Semiconductor is up +94.3% over 20 trading days, RSI approximately 81.5, and is +574% from its 52-week low — enough to trigger an investment-warning designation. Even with genuine fundamental improvement, a new buyer entering here begins in the middle of peak volatility.

Near-term flows are clearly constructive. On May 15, during a broader market selloff, foreigners net-bought Hana Micron shares worth approximately KRW 91.8bn while retail sold KRW 93.8bn. Jeju Semiconductor also saw coordinated foreign and institutional buying. But “flows are strong” and “now is a good entry” are different statements. Strong flows signal that the name deserves attention; the right entry price is a separate question.

7. Positioning Within the Broader Semiconductor Back-End Universe

As covered in the AI Back-End 11-Stock Data Comparison, even within “AI back-end,” substrates, test sockets, memory packaging, and legacy memory are entirely distinct businesses with different margin structures, customer dynamics, and cycle exposures.

Hana Micron is primarily an OSAT story. As demand for high-value memory packaging — HBM, DDR5, eSSD — grows, both volume and unit pricing move in Hana Micron’s favor. Jeju Semiconductor is a memory fabless company, but the current tailwind is specifically “ordinary memory made scarce by AI capex” rather than AI-driven demand for advanced products directly. Both sit on the outer ring of the AI cycle. One is a back-end structural improvement; the other is a legacy-memory supply shock.

Company	2027E PER / Metric	Core Thesis
SK hynix	~5x	HBM leader, cycle-peak concern
Samsung Electronics	~5–6x	HBM4 + foundry optionality
Hana Micron	~14–19x	Vina/Brazil structural improvement
Jeju Semiconductor	9x (1Q run-rate); ~16–17x (EV)	LPDDR4X supply cycle, no consensus
Simmtech	~20x	Substrate turnaround
HaeSeong DS	~15x	Cheapest substrate candidate
Leeno Industrial	~33x	Test socket quality premium
ISC	~43x	AI data-center test

In this peer table, Hana Micron sits at a reasonable middle-ground valuation for comparable back-end names. Jeju Semiconductor appears cheap on the 1Q run-rate but requires a normalization scenario to assess properly — making it closer to a scenario bet than a straightforward value hold.

8. Practical Checkpoints

For Hana Micron, the critical variable in 2Q26 is OPM. If 1Q’s 14.2% was genuinely one-off, the margin will fall sharply in Q2. If OPM holds at roughly 13%, that is evidence that the Vina pricing structure and Brazil subsidiary margin are real, durable changes to the company’s earnings power.

Four checkpoints to monitor:

2Q26 OPM ≥ 13%
Continued Vina revenue growth
Brazil subsidiary sustaining high-teens OPM
Rising share of DDR5/eSSD packaging within the revenue mix

For Jeju Semiconductor, the critical variable is the magnitude of 2Q deceleration. If Q2 operating profit holds above KRW 50bn with OPM above 30%, 1Q was not simply a one-time peak. If Q2 drops below KRW 40bn, the front-loading and scarcity effects are unwinding faster than hoped.

Five checkpoints to monitor:

2Q26 OP ≥ KRW 50bn
OPM ≥ 30%
Inventory growth decelerating
Accounts receivable growth decelerating
LPDDR4X spot pricing trend sustaining

On positioning, waiting for a pullback makes more sense than chasing for both names. For Hana Micron, the KRW 49,000–51,000 support zone is the first level to watch. For Jeju Semiconductor, the KRW 68,000–72,000 range or the post-2Q26 results window is a more realistic entry consideration. These are not buy-recommendation price targets — they are observation levels to gauge whether the current overheating has normalized.

9. If Forced to Choose One

For a 12-month-plus holding period with emphasis on earnings durability, Hana Micron has the edge. If the Vina pricing structure and Brazil subsidiary margins hold, 2027 earnings estimates have upside. Even in a down-cycle, the structural changes may produce a higher margin floor than in prior cycles.

For a 3-month high-volatility trade, Jeju Semiconductor offers more explosive potential movement. But that trade requires correctly calling a cycle peak. The Bull case has meaningful upside; the Bear case has -50% drawdown risk from current levels. Factor in CB/BW dilution and the risk/reward tightens further.

If holding both, a core-satellite structure is more natural: Hana Micron as the core (60–70%) and Jeju Semiconductor as the satellite (30–40%). A concentrated single-name position in Jeju Semiconductor at current prices carries substantial volatility risk.

10. One-Line Summary

Hana Micron and Jeju Semiconductor both delivered strong 1Q26 results. They are not the same kind of surprise.

Hana Micron is a structural improvement story. The Vina cost-pass-through repricing, the Brazil high-teens OPM, and growing SK hynix-related packaging volume moved together. The 1Q margin cannot be fully normalized to an annualized run-rate, but the evidence suggests a higher earnings floor than in prior cycles.

Jeju Semiconductor is closer to a cycle peak. The 37.2% OPM was produced by LPDDR4X supply scarcity and tariff-driven front-loading. The company’s product capabilities and niche-market positioning are genuine, but treating 37.2% OPM as a sustainable steady-state is the wrong frame.

On multiples, Hana Micron looks more attractive. Jeju Semiconductor’s 1Q run-rate PER of 9x is an illusion — it requires four identical quarters that are unlikely to materialize. Adjusting for normalization pushes the expected-value PER to 16–17x. Hana Micron, under a structural-improvement scenario through 2027, still offers expected-value upside from the current price.

Neither stock is a good chase right now. The most dangerous move after a strong earnings print is buying immediately because the numbers look good. Real alpha comes from distinguishing the cause of the surprise, confirming durability in 2Q26, and entering in tranches after the post-announcement heat fades.

The one-line version: Hana Micron is a structural-improvement confirmation trade; Jeju Semiconductor is a cycle-peak timing call. New money priority favors Hana Micron, but chasing either before 2Q26 results is inefficient.

This article is for research and commentary purposes only and does not constitute investment advice. Hana Micron 1Q26 results (revenue KRW 507.7bn, OP KRW 72.0bn, OPM 14.2%) are sourced from company disclosures and Korea Investment & Securities/Meritz Securities research reports. Jeju Semiconductor 1Q26 results (revenue KRW 180.5bn, OP KRW 67.1bn, net income KRW 78.1bn, OPM 37.2%) are sourced from company disclosures. The +29.7% consensus beat figure is based on Korea Investment & Securities data. Jeju Semiconductor 2026E/2027E official consensus is not available in public sources; all scenario estimates in this article are analyst projections based on 1Q results, LPDDR4X supply-demand dynamics, and AI-edge transition potential. Vina cost-pass-through structure and Brazil subsidiary high-teens OPM are interpretations based on brokerage research materials. CB/BW outstanding of approximately KRW 117.0bn and potential dilution of approximately 7.7% are based on company disclosures and media reports. Scenario fair values and probability weights are subjective analyst estimates and may differ materially from actual outcomes. The timing of LPDDR4X supply normalization, success of AI-edge memory conversion, and 2Q26 earnings durability are all uncertain. This analysis may be wrong. Data as of May 17, 2026 KST.

Hyundai Mobis Robotics / Atlas Valuation — The 30K-Atlas Case Is Largely In the Price. Next Are Grippers, External Customers, and the Boston Dynamics Stake

Sun, 17 May 2026 00:00:00 +0000

📚 Hyundai Mobis / Robotics series Previous: Hyundai Mobis — The EV-Parts Giant Powering Robotics (April 28 deep dive) Korean robotics value-chain map: Complete Korea Robotics Value Chain Map (May 11) Robot-component comparison: SPG vs Halla Cast — Robot-Component Compare (May 12) Humanoid-OEM compare: Robotis vs Rainbow Robotics (May 12)

TL;DR

Bottom line: At today’s price, Hyundai Mobis already prices in most of the Atlas 30K-units/year actuator-supply scenario. What is not fully priced is gripper / sensor / controller expansion, external (non-HMG) customer wins, and the indirect Boston Dynamics stake.
Base estimate: On 30K Atlas units/year, Mobis’s robot-parts revenue is in a KRW 2.1–3.0 trillion band, operating profit KRW 170–360 billion, and segment value KRW 3.4–7.2 trillion.
Investment call: Today’s price is not “cheap” — it is a verification window. New buys: Wait. Holders: Hold. Additions should follow confirmation of scope, units, and margin.

1. Facts and structure

1.1 Core facts

In its shareholder letter, Hyundai Motor said it would deploy Boston Dynamics’ Atlas on its manufacturing floors and build the infrastructure to scale to 30,000 robots/year by 2028, alongside Google DeepMind and NVIDIA physical-AI infrastructure tie-ins.

At CES 2026, Hyundai Mobis formalized a strategic-cooperation framework with Boston Dynamics and announced it would supply actuators for Atlas. Mobis stated that actuators account for more than 60% of a humanoid robot’s BOM cost.

At Kia’s 2026 CEO Investor Day, Kia said Atlas will be deployed at HMGMA in 2028 and expand to Kia AutoLand Georgia in 2029. This moves Atlas off the “showcase” track and onto a real plant-deployment roadmap.

Hyundai Mobis 1Q26 figures: revenue KRW 15.5605 trillion, operating profit KRW 802.6 billion. The A/S business posted revenue of KRW 3.5190 trillion and operating profit of KRW 923.9 billion, defending overall profitability while the Module / Core Components segment ran a loss.

FY2025: revenue KRW 61.1180 trillion, operating profit KRW 3.3570 trillion, net income KRW 3.6650 trillion, EPS KRW 40,861.

As of May 16, 2026: share price KRW 629,000, TTM EPS KRW 40,861, PER ~15.4x, market cap roughly KRW 56 trillion. Sell-side average target KRW 568,828, top target KRW 750,000.

2. Value-chain mapping

2.1 Hyundai Mobis’s position in the robot value chain

Physical AI / Foundation Model
 ↓
Robot OS / Task Planning / Control
 ↓
Sensors / Cameras / Perception
 ↓
Controllers / Power Electronics / Battery
 ↓
Actuators / Reducers / Motors / Grippers
 ↓
Robot Assembly / Plant Deployment / Maintenance

Hyundai Mobis’s formally disclosed position is actuator supply. KB Securities expects Boston Dynamics to source the entirety of its actuator requirement from Mobis and noted that talks on additional components — grippers in particular — are ongoing.

That makes the investment question more than “robot-parts supply.” The actual swing variables are these three:

Variable	Meaning	Investment weight
Full actuator supply	Determines per-unit Mobis content	Very high
Gripper / sensor / controller expansion	Lifts per-unit revenue	High
External (non-HMG) customer wins	Captive → platform supplier shift	Very high

3. Revenue estimate at 30K Atlas units

3.1 The math

Mobis robot-parts revenue = Atlas units × per-unit Mobis content × FX

Assumptions:

Atlas units: 30,000 / year
FX: KRW 1,500 / USD
Per-unit Mobis content: $30,000 – $80,000
Rounding: revenue to KRW 0.01T, OP to KRW 10B.

Mobis’s actual per-unit content, ASP, and gross margin on Atlas are not officially disclosed. This is genuinely uncertain. Verification will come through Mobis IR, Boston Dynamics supply-contract disclosures, broker follow-up notes, and any future stand-alone robot-parts revenue guidance. For interim decisions, $50,000–$65,000 per unit is a conservative base case.

3.2 Revenue sensitivity

Per-unit Mobis content	Annual USD revenue	KRW revenue
$30,000	$0.90 bn	KRW 1.35 T
$40,000	$1.20 bn	KRW 1.80 T
$50,000	$1.50 bn	KRW 2.25 T
$65,000	$1.95 bn	KRW 2.93 T
$80,000	$2.40 bn	KRW 3.60 T

Base range: KRW 2.1–3.0 trillion per year.

That range squares with an Atlas ASP of roughly $190,000 and Mobis capturing 25–35% of per-unit content. KB Securities models a 2035 scenario of 1.5 million premium humanoids at ~$190,000 each.

4. Operating profit and segment-value math

4.1 OP sensitivity

Robot actuators are likely higher-value than legacy auto modules, but early-ramp costs, quality validation, and CAPEX have to be allowed for. As a first cut, 8–12% OPM is the right bracket.

Per-unit content	Revenue	OP @ 8%	OP @ 12%
$30,000	KRW 1.35T	KRW 108B	KRW 162B
$40,000	KRW 1.80T	KRW 144B	KRW 216B
$50,000	KRW 2.25T	KRW 180B	KRW 270B
$65,000	KRW 2.93T	KRW 234B	KRW 351B
$80,000	KRW 3.60T	KRW 288B	KRW 432B

Base values:

Revenue: KRW 2.25–2.93 trillion
Operating profit: KRW 180–351 billion
Midpoint: revenue ~KRW 2.6T, OP ~KRW 260–300B

4.2 Appropriate multiples

Applying a robot-pure-play multiple to the entire Mobis franchise is too aggressive. The base business is auto parts + A/S cash cow; robot parts have not yet shown a stand-alone earnings trail.

Asset	Fair multiple	Read
Legacy auto parts / A/S	PER 11–13x	A/S cash cow justifies a premium to pure parts
Mobis blended	PER 14–16x	Top end if you embed the robot option
Atlas-actuator segment	EV/EBIT 18–22x	Base 20x
Robot bull case (expansion)	EV/EBIT 25–30x	Requires external customer + gripper / sensor expansion

KB Securities maintains a target price of KRW 750,000, with a DCF-implied market cap of KRW 67T, 12-month forward PER 16.0x, and P/B 1.22x. KB also flags that Hyundai Mobis owns 20.0% of Hyundai Global, and Hyundai Global in turn owns 54.7% of Boston Dynamics — giving Mobis indirect economic exposure of 10.9% to Boston Dynamics.

4.3 Segment value

Robot segment value = robot OP × EV/EBIT multiple

Per-unit content	OP @ 8%	Value @ 20x	OP @ 12%	Value @ 20x
$30,000	KRW 108B	KRW 2.16T	KRW 162B	KRW 3.24T
$40,000	KRW 144B	KRW 2.88T	KRW 216B	KRW 4.32T
$50,000	KRW 180B	KRW 3.60T	KRW 270B	KRW 5.40T
$65,000	KRW 234B	KRW 4.68T	KRW 351B	KRW 7.02T
$80,000	KRW 288B	KRW 5.76T	KRW 432B	KRW 8.64T

Base read: On 30K Atlas units, Mobis’s robot segment value sits at ~KRW 3.6–7.0 trillion.

A bull case with $80,000/unit content and 12% OPM gets to KRW 8.6 trillion, but that already embeds at least partial expansion beyond pure actuator content.

5. How much robot optionality is already in the price?

5.1 Back-solving from the current quote

At KRW 629,000 and TTM EPS of KRW 40,861, PER works out to:

629,000 ÷ 40,861 = 15.39x

5.2 SOTP on TTM EPS

Assumptions:

TTM EPS: KRW 40,861
Legacy-business fair PER: 14x
Shares outstanding: ~89.2 million
Current price: KRW 629,000

Math:

Legacy value / share = 40,861 × 14 = KRW 572,054
Residual option value / share = 629,000 − 572,054 = KRW 56,946
Aggregate residual = 56,946 × 89.2M ≈ KRW 5.1 trillion

In other words, on TTM EPS and a 14x legacy multiple, today’s price already carries ~KRW 5 trillion of robot / Boston Dynamics optionality.

5.3 SOTP on 2026E EPS

KB Securities models 2026E controlling-interest net income of KRW 4.115 trillion, implying ~KRW 46,100 per-share EPS on ~89.2M shares.

Legacy PER	Legacy value/share	Residual option value/share	Aggregate option value
11x	KRW 507,000	KRW 122,000	~KRW 10.9T
12x	KRW 553,000	KRW 76,000	~KRW 6.8T
13x	KRW 599,000	KRW 30,000	~KRW 2.7T
14x	KRW 645,000	negative	Legacy alone explains the price

This table is the key:

The robot optionality embedded in the current price runs from KRW 0 to ~11 trillion depending on the legacy multiple you assign.

A realistic legacy band is 12–14x. On that band, today’s quote embeds KRW 0–7T of robot optionality, with a midpoint around ~KRW 5T.

6. Valuation read

6.1 Is the 30K-Atlas case already in the price?

Yes — largely.

Scenario	Robot segment value
Bear	KRW 2.2–3.2 T
Base	KRW 3.6–5.4 T
Top	KRW 4.7–7.0 T
Bull	KRW 5.8–8.6 T

Embedded option value in today’s price is ~KRW 5 trillion. So the base 30K-Atlas actuator-supply case alone does not clearly make the stock “cheap” at this level.

6.2 What is not yet priced

Item	Why it matters	Trigger to look for
Gripper supply	Largest BOM share after actuators	Additional Mobis supply contracts
Sensor / controller expansion	Lifts per-unit content	Robot-parts portfolio disclosure
External (non-HMG) customers	Captive → global platform supplier shift	Customers other than Boston Dynamics
Boston Dynamics stake value	Indirect 10.9% via Hyundai Global	BD outside investment / IPO / transaction price
2035 mass-volume	Tens or hundreds of thousands, not 30K	Verified unit economics

KB Securities frames Mobis as central to the supply chain enabling Boston Dynamics’ production, with full actuator supply and an ongoing gripper-supply discussion.

7. Investment read

7.1 The call

Hyundai Mobis: Wait / Holders: Hold

This is no longer a “cheap auto-parts” quote. The A/S cash cow, the robot option, the indirect Boston Dynamics stake, and governance-improvement expectations are all at least partially priced in.

7.2 By price band

Band	Read	Call
KRW 500–550K	Legacy at 11–12x + partial robot option	High buying interest
KRW 560–600K	Partial 30K-Atlas embedded	Conditional buy
KRW 620–650K	Today’s level — substantial 30K actuator embedded	Wait / Hold
KRW 700–750K	Approaches the KB target. Needs more data	Chasing inefficient
Above KRW 750K	Long-run platform scenario pre-priced	Overheating risk

7.3 Entry conditions

A new entry makes more sense once at least two of the following land:

Atlas actuator-supply volumes are quantified.
Mobis per-unit content or annual robot-parts revenue guidance is disclosed.
Non-actuator components (grippers, sensors, controllers) get into the supply scope.
A non-HMG external customer is confirmed.
Robot-parts OPM is shown to be higher than the legacy module business.

7.4 Invalidations

The robot re-rating thesis weakens on any of:

Atlas HMGMA deployment slips beyond 2028.
Plant adoption stays at pilot scale without verified productivity ROI.
Mobis supply scope is limited to a portion of actuators.
Robot-parts yield / durability issues during production ramp.
Module / Core Components losses persist while A/S high margin normalizes.
Boston Dynamics fails to win meaningful share outside the HMG ecosystem.

8. How this links to other posts

Hyundai Mobis — The EV-Parts Giant Powering Robotics (April 28 deep dive) — the starting point. Defines the 3-pillar framing: EV parts + A/S cash cow + robotics option.
Complete Korea Robotics Value Chain Map (May 11) — locates Mobis at the actuator layer; useful for comparing layers above and below.
SPG vs Halla Cast — Robot-Component Compare (May 12) — the smaller-cap component peers; useful as the size / multiple / optionality contrast vs Mobis (captive + large).
Robotis vs Rainbow Robotics — Humanoid OEM Compare (May 12) — the Korean humanoid-OEM candidates; useful to see “parts platform” (Mobis) vs “OEM” framing.

9. Bottom line

The Hyundai Mobis robot story has substance. Hyundai Motor’s 2028 plan for 30K robots/year, Kia’s 2028–2029 plant-deployment roadmap, and Mobis’s CES 2026 actuator-supply announcement are all confirmed facts.

But today’s price already embeds a meaningful share of the base-case KRW 3.6–5.4T Atlas-actuator value. For the stock to climb further from here, the question is no longer “30K Atlas” — it is how much of Atlas’s hardware content Mobis will capture, and whether it can win external (non-HMG) customers.

The current price is not cheap. But if the long-run robot-platform-supplier transition is confirmed, it is not “done” either.

What is needed now is verification rather than buying. Hold what you own; postpone new additions until scope and margin become observable. That is what risk-adjusted return calls for here.

This article is research and commentary only and is not investment advice. Citations from Hyundai Motor, Kia, and Hyundai Mobis are based on each company’s official IR materials, the CES 2026 announcement, and the CEO Letter to Shareholders. Hyundai Mobis 1Q26 figures (revenue KRW 15.5605T, OP KRW 802.6B) and FY2025 (revenue KRW 61.1180T, OP KRW 3.3570T, EPS KRW 40,861) are per company disclosures. May 16 share price KRW 629,000, market cap, and sell-side consensus are per Investing.com. The KB Securities target price of KRW 750,000, 12-month forward PER 16.0x, the 10.9% indirect Boston Dynamics stake, and the 2035 1.5M humanoids × $190,000 scenario are per the KB Securities report and may differ from actual outcomes. Assumptions for per-unit content ($30,000–$80,000), OPM (8–12%), and EV/EBIT multiples (18–22x) are author estimates and are not officially disclosed. Actual Atlas mass-production timing, supply scope, content, and margin all remain to be verified by company disclosure and earnings. Global macro variables (US rates, oil, FX, VIX) can independently move Hyundai Mobis. The analysis may be wrong. Data cut-off: May 17, 2026 KST.

NVIDIA Earnings and AI-RAN: Why the Supply Chain Matters More Than a Single SK Telecom Bet

Sun, 17 May 2026 00:00:00 +0000

📚 Related Series SK Telecom Re-rating: From Dividend Stock to AI Infrastructure Operator / Next Week’s Google I/O and NVIDIA Earnings Preview / AI Advanced Packaging Data Comparison / Samsung Electronics TSMC-Style Re-rating

NVIDIA’s FY2027 Q1 earnings call lands in the early hours of May 21. “AI-RAN” could surface on that call, and the Korean market’s reflexive response will be to reach for SK Telecom. SKT sits inside NVIDIA’s Global 6G Coalition of twelve carriers, its AIDC revenue grew 89% YoY, and foreign investors have been buying. Even so, “AI-RAN theme = buy SKT” is the most common and least efficient inference you can make. Why? The three carriers NVIDIA called out by name at MWC 2026 were T-Mobile, SoftBank, and Indosat Ooredoo Hutchison — SKT was not on that list. More fundamentally, the economic value of AI-RAN accrues to GPUs, memory, vRAN software, RU/RF components, and optical interconnects before it ever reaches the operator. Real alpha begins not with the word “AI-RAN” but with the question: where does the money actually go?

Key Takeaways

Upcoming event: NVIDIA FY2027 Q1 earnings call at 6:00 a.m. KST on Thursday, May 21.
AI-RAN mention probability: telecom/Aerial/wireless general mention 70–80%; direct “AI-RAN” reference 50–60%; specific carrier named 15–25%; SKT named directly 5–15%.
The market’s reflex: AI-RAN mention → buy SKT. This is the most common trap.
Reality: NVIDIA’s MWC 2026 field-trial carriers = T-Mobile, SoftBank, IOH. SKT was not among them.
AI-RAN economic-value waterfall:
- Layer 1 — largest pool: GPU/HBM/memory (NVIDIA → SK hynix, Samsung Electronics)
- Layer 2: vRAN/network equipment (Samsung Networks, Nokia, Ericsson)
- Layer 3: RU/RF/power amplifiers (RFHIC, KMW)
- Layer 4: optical interconnects/CPO (Oi Solutions, Bit&Electron)
- Layer 5 — CAPEX spender: SK Telecom, T-Mobile, SoftBank
SKT’s real investment case: not AI-RAN, but AIDC + GPUaaS + sovereign AI. Q1 2026 AIDC revenue of KRW 131.4 bn is already visible on the income statement.
Actionable conclusions:
- SKT: Hold / new positions → Wait. The stock has already rallied YTD; chasing is inefficient.
- Supply-chain priority: Samsung Electronics core > RFHIC conditional > Oi Solutions as option > KMW on hold > small-cap optical as speculation.
- Most rational portfolio structure: large-cap memory as core + RFHIC/Oi Solutions as satellites.

1. The “Automatic Reflex” Trap

1.1 The “NVIDIA → AI-RAN → SKT” Inference Chain

The thinking pattern that drives this trade is straightforward.

1. NVIDIA earnings imminent
 ↓
2. Jensen Huang likely to say "telecom is next"
 ↓
3. AI-RAN theme gains attention
 ↓
4. "SKT is Korea's AI-RAN flagship"
 ↓
5. Buy SKT

The problem is that every link looks plausible. NVIDIA has genuinely talked up telecom and AI-RAN; SK Telecom has genuinely articulated an AI infrastructure and 6G vision. But whether step 4 is actually true — and whether step 5 is an efficient trade — requires separate verification.

Most thematic trades stop here. They treat “the word fits” and “the money flows here” as synonymous. They are not.

1.2 Why SKT Looks Attractive — The Facts

There are real reasons SKT appears to be Korea’s AI-RAN proxy.

✓ Member of NVIDIA's Global 6G Coalition (12 carriers)
✓ "AI Native" strategy announced at MWC 2026
✓ Supermicro Haein cluster: 1,000+ AI servers deployed
✓ Joint vRAN/AI-RAN white paper published with DOCOMO
✓ 6G standardization MOU with Ericsson through 2031
✓ Q1 2026 AIDC revenue: KRW 131.4 bn (+89% YoY)
✓ Foreign net buying past 1 week: \~KRW 87.8 bn
✓ Stock outperformed May 11–13

On the surface this is compelling: defensive telecom fundamentals, an AI infrastructure growth story, and membership in a global 6G network. Near-term technicals are positive too.

1.3 What the Market Tends to Miss

Reading the same facts differently changes the picture.

✗ 6G Coalition of 12 = "coalition member" (group mention)
✗ NVIDIA field-trial carriers named directly ≠ SKT
 (T-Mobile, SoftBank, IOH are the marquee names)
✗ SKT's MWC 2026 announcement = AI Native strategy (AI-RAN was one line)
✗ Q1 2026 AIDC revenue KRW 131.4 bn = \~3.0% of consolidated revenue
✗ Foreign buying = already reflected in the May 11–13 rally
✗ Stock is up \~+101% from its 52-week low

In other words: the linkage is thin, and much of it is already in the price. Betting on SKT alone is a thin thesis — the theme is right, but the price and the value-accrual position are the problem.

2. What Actually Happened at MWC 2026 — The Facts

2.1 The Three Carriers NVIDIA Named Directly

NVIDIA singled out three carriers as first-name field-trial partners at MWC 2026.

1. T-Mobile US

A proof-of-concept running video streaming, generative AI queries, and video captioning simultaneously using Nokia AirScale Massive MIMO and a single GH200, with over-the-air validation at the Seattle AI-RAN Innovation Center.

2. SoftBank (AITRAS)

World-first 16-layer Multi-User MIMO downlink, fully software-defined 5G vRAN, and 3× spectral efficiency versus a conventional 4-layer configuration. Of all carriers, SoftBank has the most advanced AI-RAN commercialization story.

3. Indosat Ooredoo Hutchison (IOH)

Southeast Asia’s first AI-RAN-powered Layer 3 5G call, live remote control of a robot dog over a live 5G network, and pre-commercial validation milestone.

The key point is simple: SKT was not among these three. SKT appears in group mentions, but it was not a carrier NVIDIA cited with a concrete field-trial use case.

NVIDIA’s most prominent Korea-related announcements at MWC 2026 split into two tracks — neither of which is AI-RAN per se.

The first was the Supermicro SKT Haein cluster: a deployment of 1,000+ AI servers. This is AI datacenter and sovereign AI infrastructure, not AI-RAN.

The second was Samsung Electronics vRAN demo: an AI-MIMO beamforming demonstration using Samsung’s own vRAN software, NVIDIA Grace CPU, and L4 GPUs in a multi-cell test environment.

From an AI-RAN standpoint, the Korean company NVIDIA most explicitly highlighted as a partner was Samsung Networks, not SK Telecom. SKT connects more directly through the AIDC lens. These are different theses.

2.3 Implication — Repositioning SKT

A more accurate read of SKT’s actual position within the NVIDIA AI-RAN ecosystem:

Policy / Governance layer: Strong
- AI-RAN Alliance board member
- 6G Coalition member
- Standardization activities

Field / Implementation layer: Follower
- T-Mobile, SoftBank, IOH are the marquee names
- SKT appears at the group-mention level

Commercial / Monetization layer: Unverified
- How will AI-RAN translate into revenue?
- Gap between "proof of concept" and "recognized revenue"

SKT is closer to a future-vision leader. As a short-term revenue beta, it is weak. Markets typically price the latter first.

3. Where AI-RAN’s Economic Value Actually Flows

3.1 The Five-Layer Value-Accrual Structure

When one unit of AI-RAN infrastructure is deployed, spending flows through roughly five layers.

Layer 1 — GPU/HBM/Memory: the largest pool

NVIDIA GB200/GB300, L4, RTX Pro GPUs and Grace CPUs, along with HBM and DRAM inside those chips, are the heart of the system. Revenue accrues to NVIDIA and its memory supply chain — SK hynix and Samsung Electronics. This layer could absorb 40–50% of total AI-RAN CAPEX.

Layer 2 — vRAN/Network Software

5G/6G vRAN stacks, AI channel estimation, schedulers, and network optimization software live here. Samsung Networks, Nokia, and Ericsson are the key players, accounting for roughly 15–20% of CAPEX.

Layer 3 — RU/RF/Power Amplifiers

Radio units, GaN PAs, filters, antennas, and mMIMO hardware. RFHIC, KMW, and the RU divisions of Nokia and Ericsson are involved. Share: roughly 10–15%.

Layer 4 — Optical Interconnects/CPO

800G/1.6T optical transceivers, ELSFP, CPO, Spectrum-X, and ConnectX-class networking. NVIDIA on one end; Oi Solutions, Bit&Electron, and Lycom on the Korean supply side. Share: roughly 5–10%.

Layer 5 — Operators / CAPEX Spenders

SK Telecom, T-Mobile, SoftBank, and IOH sit here. Their role is to buy layers 1–4. Revenue may eventually return to them through network efficiency gains and future edge AI service monetization — but from a pure AI-RAN CAPEX perspective, they are cost centers.

This is the most important insight. Operators are not the ones receiving the money. They are the ones spending it.

3.2 Why Operators Trend Toward Cost Centers

Every generation — 2G, 3G, 4G, 5G — has taught the same lesson.

Operators built out new networks, delivered services, fought to grow ARPU, then repeated the whole process for the next generation standard. The parties that captured value first and most reliably were equipment vendors, chip companies, memory suppliers, and device manufacturers.

Who made money each cycle:
- Equipment: Nokia, Ericsson, Samsung, Huawei
- Chips: Qualcomm, MediaTek
- Memory: Samsung, SK hynix, Micron
- Devices: Apple, Samsung

Korean telecom ROE by cycle:
- 4G cycle (2010–2015): 5–10%
- 5G cycle (2019–2024): 5–9%
- 6G cycle (2027–?): not yet established

Operator ROE failed to expand meaningfully through the 5G cycle. AI-RAN could change that pattern — but the evidence is not yet there. The thesis that “AI-RAN = operator re-rating” remains thin at this stage.

3.3 Scenarios Where Operators Capture Profit

For an operator to monetize AI-RAN meaningfully, it needs to evolve beyond being a simple network provider.

1. "Network operator" → "edge AI service operator"
 - Run AI workloads on the network and charge for them
 - AI-as-a-service model
 - Convergence of connectivity and compute

2. Sovereign AI infrastructure operator
 - Data-sovereignty guarantees for governments and enterprises
 - Long-duration contracts for in-country AI infrastructure
 - New revenue category

3. B2B AI solutions integrator
 - Autonomous vehicles, smart cities, smart factories
 - Combined 5G/6G + AI solutions
 - Telecom + platform + solutions bundled

Moving toward these models is how an operator lifts its margins. This is roughly the direction SKT is pursuing. But it takes time to recognize these revenue streams. The fastest monetization path is AIDC + GPUaaS — that is SKT’s real thesis, not AI-RAN.

4. Four Market Misconceptions — Addressed Directly

4.1 Misconception 1 — “AI-RAN theme = buy SKT”

This is an unfounded reflex.

"NVIDIA mentions telecom" → "buy Korean telecom"

Three problems compound: auto-mapping “Korean telecom” to SKT specifically; treating the operator as the beneficiary of value accrual; and placing SKT among the global marquee field-trial carriers.

The corrected inference chain:

NVIDIA mentions telecom
 ↓
Telecom CAPEX cycle accelerates
 ↓
Infrastructure component companies benefit
 ↓
Samsung Networks, RFHIC, Oi Solutions

SKT appears at layer 5. It is a lower-priority destination. Its position as a CAPEX spender must be kept in view.

4.2 Misconception 2 — “SKT is a global AI-RAN leader”

Partly right, partly wrong.

The part that is right: policy and R&D. SKT is an AI-RAN Alliance board member, is active in 6G standardization, and has clearly articulated an AI Native vision.

The part that is wrong: field implementation. T-Mobile, SoftBank, and IOH lead at the proof-of-concept level. SoftBank in particular is furthest ahead on the commercialization story.

An analogy: SKT is closer to the chair of an AI-RAN academic committee; T-Mobile and SoftBank are closer to AI-RAN’s first paying customers. The chair role has value, but markets respond faster to the first customers.

4.3 Misconception 3 — “AIDC revenue grew 89%; time to buy”

The number looks impressive in isolation: Q1 2026 AIDC revenue of KRW 131.4 bn, +89% YoY.

But context requires checking the mix.

Consolidated revenue: KRW 4,392.3 bn
AIDC revenue: KRW 131.4 bn
AIDC share = 131.4 / 4,392.3 ≈ 3.0%

AIDC is growing fast — that is real. But it is 3% of the group. Its EPS impact is smaller still.

If 89% growth sustains, the picture changes: AIDC could reach roughly 10% of revenue in two years and above 20% in three. But it takes 2–3 years to reach a meaningful group-level share. As a short-term AI beta vehicle, SKT remains weak.

4.4 Misconception 4 — “Foreign investors are buying, so I should follow”

Foreign net buying of roughly KRW 87.8 bn over the past five sessions is a clear flow signal, with heavy buying on May 11–12 sustained through subsequent consolidation.

But May 11–13 saw a strong price move. Much of the foreign buying’s price impact is already reflected; chasing it now means arriving as a late buyer.

Why foreigners are buying requires separate analysis — it could be a mix of AIDC growth, defensive telecom appeal, and advance pricing of the AI-RAN theme. Foreign buying is not a sufficient condition for chasing.

5. Korea’s AI-RAN Supply Chain — The Real Priority Order

5.1 Direct-Exposure Matrix by Name

Rank	Name	Code	AI-RAN Direct Exposure	NVIDIA Beta	Fundamentals	View
1	Samsung Electronics	005930	High (vRAN + HBM)	Very High	Strong	Watchlist → Buy on dip
2	SK hynix	000660	Medium (HBM indirect)	Very High	Strong	Hold / avoid chasing
3	RFHIC	218410	Very High (GaN RU)	Indirect	Strong (Q1 2026 OPM 18%)	Wait (valuation stretched)
4	Oi Solutions	138080	High (1.6T optical)	Indirect	Turnaround in progress	Watchlist (conditional)
5	KMW	032500	High (mMIMO)	Indirect	Weak (operating loss)	Avoid until earnings turn
6	Bit&Electron / Lycom	—	Medium (optical module)	Indirect	Weak	Speculative only
Reference	SK Telecom	017670	Low (operator)	Indirect	AIDC growth	Hold / new → Wait

5.2 Name-by-Name Thesis Summary

Samsung Electronics — #1 priority

Samsung Electronics has three layers of AI-RAN exposure: HBM and DRAM that go into NVIDIA servers; Samsung Networks vRAN software; and AI-MIMO algorithms validated in a multi-cell NVIDIA-based test environment.

The appeal is cross-layer exposure to the entire AI-RAN stack, with a memory supercycle as the base case and AI-RAN as a free option. The stock also trades at roughly 5× 2027E P/E — the cheapest name in the basket on a valuation basis.

Risks include the May 21 labor strike, a secondary position to SK hynix in HBM market share, and the integrated-device-manufacturer discount. The call is to hold on the watchlist and accumulate in tranches after macro stabilizes. If the strike resolves and NVIDIA beats simultaneously, the case for adding weight strengthens.

RFHIC — #3 priority, conditional

RFHIC supplies GaN packages for RU power amplifiers and sits inside Samsung Networks’ equipment supply chain, benefiting from 5G, AI-RAN, and defense simultaneously.

Q1 2026: revenue KRW 43.1 bn, operating profit KRW 7.7 bn, OPM 18% — the turnaround is confirmed. Absolute returns over the past twelve months have been very strong. For direct exposure to the RU/RF component layer in an AI-RAN ramp, RFHIC has the highest purity.

The issue is valuation: roughly 60× 2026E P/E and roughly 40× 2027E P/E means a great deal of expectation is already embedded. A better entry point would follow a new order announcement or a price pullback.

Oi Solutions — #4 priority, option

Oi Solutions is exposed to 1.6 Tbps OSFP optical transceivers for AI datacenters, ELSFP for CPO applications, and AI-RAN fronthaul/backhaul optical interconnects.

Positives: expected to turn profitable in 2026, growing datacom mix, simultaneous exposure to AI-RAN and AI datacenters. But ELSFP volume production visibility matters more from 2027 onward, and recent price volatility has been elevated. A more measured approach is to wait for Q3 2026 ELSFP sampling confirmation before entering.

KMW — #5 priority, on hold

KMW carries C-band 64TRx 640W mMIMO, antenna, filter, and RU beta. But Q1 2026 revenue was KRW 21.5 bn with an operating loss of KRW 5.7 bn — fundamental proof remains weak. With a lossmaking company trading at an elevated P/B, chasing before a quarterly profit turn is not advisable.

Small-cap optical names — #6 priority, speculative

Bit&Electron, Lycom, and similar companies unveiled 1.6T optical transceivers, 100G ER1 Bidi, and 6G fronthaul products at OFC 2026. But the gap between product introduction and volume production revenue is wide. A safer posture for long-oriented investors is to wait for confirmed order announcements before sizing in.

5.3 SKT’s Own Investment Case — The Real Thesis

If you own or are considering SKT, the thesis should not rest on AI-RAN. It should rest on these four pillars.

Real thesis 1: AIDC revenue ramping
- Q1 2026: KRW 131.4 bn (+89% YoY)
- 5%+ of consolidated revenue possible by 2026–2027
- Revenue is hitting the income statement now

Real thesis 2: GPUaaS business expansion
- NVIDIA GPU cluster rental to enterprise clients
- Haein cluster: 1,000+ servers deployed
- B2B revenue growth

Real thesis 3: Sovereign AI infrastructure
- Data-sovereignty infrastructure for government and enterprise
- Long-duration contracts
- A new revenue category

Real thesis 4: Core telecom stability
- 5G subscriber base stabilizing
- ARPU maintenance
- Dividend yield as a defensive floor

AIDC and GPUaaS are far more direct catalysts than AI-RAN. At current prices, however, a first-tranche buy is less rational than waiting for a pullback — or for Q2 AIDC numbers to provide further visibility.

6. Reading the NVIDIA Earnings Call

6.1 Signal-Strength Mapping

Possible language on the call	SKT impact	Supply-chain impact	Market reaction intensity
“Telecom is next” (general)	Weak	Negligible	Very weak
“Future wireless networks as computing platforms”	Weak	Weak	Weak
“Aerial RAN Computer commercial deployments”	Medium	Strong (Samsung Electronics, RFHIC)	Medium
“AI-RAN deployments with operators”	Medium	Medium	Medium
“Samsung Networks validation”	Medium	Strong (Samsung Electronics direct)	Strong
“Operator deployment with SoftBank/T-Mobile”	Weak	Strong (equipment names)	Strong
“Operator deployment with SK Telecom”	Very Strong	Strong	Very Strong
“Commercial revenue from AI-RAN”	Strong	Strong (broad)	Very Strong

The highest-probability scenario is a general Aerial/telecom/wireless reference plus coalition group mentions. In that case, SKT’s near-term move may be noise, while the supply chain — Samsung Electronics in particular — could see a more meaningful reaction.

The low-probability scenario where SK Telecom is named directly could trigger a short-term +5–10% move in SKT, but it would also carry elevated sell-the-news risk.

Another low-probability scenario: a direct Samsung Networks callout. The share-price impact on Samsung Electronics might be limited by its memory-heavy weight, but RFHIC and Oi Solutions could be more sensitive.

6.2 Practical Trading Guide

Scenario A: SKT named directly — probability 5–15%

Short-term +5–10% possible. Existing holders can consider trimming; new buyers should not chase. The better entry is after the post-announcement dip.

Scenario B: AI-RAN mentioned directly — probability 50–60%

Elevated volatility across the supply chain. SKT likely sees a noise-level response. Samsung Electronics on a pullback is the more rational buy. Avoid chasing Oi Solutions or KMW immediately after any spike.

Scenario C: General telecom mention — probability 70–80%

Most likely outcome is no meaningful market reaction. Revert to fundamentals-based decision-making. There is no reason to buy on the word alone.

Scenario D: AI-RAN/telecom not mentioned — probability 10–20%

The AI-RAN theme may cool temporarily. This could actually create an opportunity to build positions in supply-chain names such as Samsung Electronics and RFHIC at better prices.

The common principle across all scenarios: avoid chasing immediately before the call; let the price settle for 1–2 days after the result and check macro conditions simultaneously before acting.

7. Frequently Asked Questions

7.1 “If NVIDIA names SKT on the call, can it really go +10%?”

It can. But sell-the-news risk is also elevated.

On the day: +3–8% possible. The next day: further gains or profit-taking. Five sessions later: partial giveback likely. Ten to thirty days out: actual revenue and order flow will determine direction.

So “named + chase” is a short-term volatility trade. “Named + wait one week + buy the dip” is the safer approach. And the naming itself needs to connect to real revenue to generate lasting alpha; a group mention has limited information value.

7.2 “Even so, isn’t SKT still Korea’s most direct AI-RAN name?”

Depends on how you define “direct.”

From an operator/field-trial standpoint, SKT is the most engaged Korean telecom, with a clearer AI-RAN and 6G narrative than KT or LG Uplus.

From an economic-value-accrual standpoint, Samsung Electronics ranks ahead. It holds both vRAN and HBM exposure simultaneously. As the Korean partner NVIDIA has most explicitly validated in an AI-RAN context, Samsung Networks is the more direct tie.

What matters for investment is value accrual. SKT can legitimately claim the label “Korea’s AI-RAN representative” — but only as an operator, not as the primary beneficiary of AI-RAN spending.

7.3 “Are you saying don’t buy SKT at all?”

No. SKT is a solid company. The question is why you are buying it.

Good reasons: AIDC revenue growth, GPUaaS expansion, core telecom + dividend + defensive characteristics, sovereign AI infrastructure option.

Poor reasons: “NVIDIA might mention AI-RAN,” “foreigners are buying,” “it’s an AI theme stock,” “it’s the most AI-sounding telecom.”

Given that the stock is significantly above its 52-week low, new buyers are better served by waiting for a pullback — or for Q2 AIDC visibility to confirm the growth trajectory.

7.4 “How do I construct a supply-chain basket?”

It depends on your holding horizon and volatility tolerance.

Defensive: Samsung Electronics and SK hynix as the core. Lower direct AI-RAN exposure, but the safest names in the basket.

Balanced: Samsung Electronics as core with a partial allocation to Oi Solutions, RFHIC, and SK hynix. Balances direct AI-RAN exposure and stability.

Aggressive: A blend of RFHIC, Oi Solutions, Samsung Electronics, KMW, and small-cap optical names. In this case, clearing a macro gate and using tranche-based entries are essential.

The common principle: avoid over-concentrating in any single name. AI-RAN is still in early commercialization, so a basket structured around value-accrual layers is more rational than a single-name bet.

8. How This Piece Connects to the Series

The Google I/O + NVIDIA Earnings Preview covered next week’s event schedule and the significance of NVIDIA’s Q2 guidance. This piece extracts only the AI-RAN portion and expands it through a Korean supply-chain lens.

The SK Telecom Re-rating piece argued that SKT’s real identity is not a dividend stock but an AIDC, GPUaaS, and sovereign AI infrastructure operator. This piece maintains that thesis while adding a caution: chasing SKT as a pure AI-RAN play is inefficient.

The AI Advanced Packaging Data Comparison piece mapped structural differences across memory, substrate, and test-socket layers. The same principle applies to the AI-RAN supply chain — what matters is not the label but where the money sticks.

The Samsung Electronics TSMC-Style Re-rating piece examined whether Samsung Electronics can be reclassified from a memory-cycle name to an integrated AI platform. Samsung Networks vRAN is one small option within that reclassification thesis.

9. The Bottom Line

AI-RAN could come up on next week’s NVIDIA earnings call. The market’s reflex will be to buy SK Telecom. That reflex is the most common and least efficient inference available.

Why? At MWC 2026, NVIDIA named T-Mobile, SoftBank, and Indosat Ooredoo Hutchison as its marquee field-trial carriers. SKT was not on that list. More fundamentally, AI-RAN’s economic value flows to GPUs, memory, vRAN equipment, RU components, and optical interconnects — not to the operator. Operators are the CAPEX spenders. They are on the paying side of the equation. The pattern that repeated through 2G, 3G, 4G, and 5G is highly likely to repeat through 6G and AI-RAN.

The real priority order for Korea’s AI-RAN supply chain is: Samsung Electronics (memory + vRAN simultaneously) > SK hynix (HBM indirect) > RFHIC (GaN RU direct) > Oi Solutions (optical interconnect option) > KMW (mMIMO beta) > small-cap optical (speculative). SKT sits at the back of this queue, in the operator seat.

SKT does have genuine appeal. But that appeal comes from AIDC + GPUaaS + sovereign AI, not from AI-RAN. Q1 2026 AIDC revenue of KRW 131.4 bn growing 89% YoY is the evidence. At roughly 3.0% of consolidated revenue, however, it takes 2–3 years to reach a meaningful group-level contribution. At current prices, waiting for a pullback is more rational than entering now.

Three practical actions: First, if NVIDIA names SK Telecom directly on the call, existing holders trim; new buyers do not chase. Second, if Samsung Networks validation or AI-RAN commercial deployment is mentioned, revisit Samsung Electronics and RFHIC after 1–2 days of price settlement. Third, if it is a generic “telecom is next” comment, revert to fundamentals and hold off on chasing any AI-RAN theme name.

Real alpha does not start with the word “AI-RAN.” It starts with asking where the economic value accrues. Ask that question precisely and the answer becomes clear: supply chain, not operator. And within the supply chain, large-cap memory as core plus RU and optical as satellites is the most rational structure. Buying the theme label is the fastest path to losses; understanding the value-accrual architecture is the path that lasts.

This article is for research and commentary purposes only and does not constitute investment advice. NVIDIA FY2027 Q1 earnings call date (May 20, 2026 at 2:00 p.m. PT / May 21, 2026 at 6:00 a.m. KST) is per NVIDIA official IR. MWC 2026 announcements (T-Mobile, SoftBank, IOH field trials; Samsung Networks vRAN multi-cell test) are per official NVIDIA, Nokia, and Samsung announcements. SK Telecom Q1 2026 results (consolidated revenue KRW 4,392.3 bn; AIDC revenue KRW 131.4 bn, +89.3% YoY) are per the company’s official earnings materials. SKT 6G Coalition membership (12 carriers) is per NVIDIA IR. RFHIC Q1 2026 results (revenue KRW 43.1 bn, operating profit KRW 7.7 bn, OPM 18%), Oi Solutions 1.6T/ELSFP products, and KMW Q1 2026 results are per company and broker materials. AI-RAN direct-mention probabilities (SKT 5–15%, general telecom 70–80%, etc.) are the analyst’s subjective probability estimates and may differ from actual call content. AI-RAN value-accrual layer weights (GPU/HBM 40–50%, vRAN 15–20%, etc.) are the analyst’s estimates; actual allocations will vary by deployment. Stock priority rankings (Samsung Electronics #1, RFHIC #3, etc.) reflect the analyst’s current judgment and are subject to change based on market conditions and earnings results. Short-term price reaction estimates are based on historical patterns and do not guarantee future outcomes. Global macro factors (U.S. interest rates, oil prices, FX, VIX) may exert additional influence on individual names. This analysis may be wrong. Data as of May 17, 2026 KST.

Pearl Abyss After Patch 1.07 — Weekend Data Says 'Sales-Rank Recovery,' Not 'CCU Rebound'

Sun, 17 May 2026 00:00:00 +0000

📚 Pearl Abyss series Previous: What the DLC “exploring” comment really means — package → franchise re-classification First post-patch weekend baseline: Crimson Desert Post-Patch Weekend Data (1.04) Hub: Pearl Abyss and Crimson Desert research hub

TL;DR

Mid-weekend read: positive. As of May 16, 23:20 KST, CCU is 66,448, Steam global sales rank #18, US #19, Korea #7, China #16 — all recovered.
But CCU is still -12% vs the same window last weekend. The stronger signal this weekend is sales-rank and review-quality recovery, not “more users.”
The new signal — Steam China rank is structurally above global. Last-7-day average China #17.3 vs global #23.4, China leading by ~6.2 rungs. This is not 1–2 day noise — it has held for a week+. Full analysis in §3.
Read: Patch 1.07 defended the long-tail thesis. Not yet a bull-flip, but the “fast patch cadence → sales-rank recovery → trust accumulation” hypothesis remains intact. Stacking the China-optionality signal on top of it, both bulls and bears need to re-look at the data.

1. The weekend data

Source — SteamDB / SteamCharts public data
Snapshot — 2026-05-16 23:20 KST
Window — May 15 18:00 – May 16 23:20 vs prior-weekend May 8 18:00 – May 9 23:20

Metric	This weekend	Last weekend	Δ
Average CCU	45,852	52,120	-12.0%
Peak CCU	66,448	75,697	-12.2%
New reviews	+504	+457	+10.3%
New-review positive rate	94.0%	82.1%	+11.9pp
Global rank average	#24.1	#20.9	weaker
Global rank latest	#18	#17	almost recovered
US rank average	#23.0	#19.5	weaker
US rank latest	#19	#17	almost recovered
China rank latest	#16	#17	improved
Korea rank latest	#7	#5	solid

2. Reading the data

[Fact] Sales rank has clearly recovered post-Patch 1.07. The 24-hour global rank-average pre-patch was #29.0; post-patch the running average is #24.7, with the latest at #18. The US moves from a pre-patch average of #29.1 to #23.7, latest #19.

[Fact] User count is still below last weekend. Both average and peak CCU are down ~12% vs the prior weekend. This reads less as “users have rolled over” and more as “the decay rate has slowed and the weekend peak has reformed.”

[Inference] The bigger change is review quality. New-review positive rate this weekend is 94.0%, sharply up from last weekend’s 82.1%. After Patch 1.06 there was clearly community friction and downward review pressure — Patch 1.07 appears to have eased most of it.

[Fact] Patch 1.07’s content is long-tail focused. Per the official notes, 1.07 added boss rematches, new Damiane combat skills, mount additions, and an AMD-driver crash fix. It is not a pure bug-fix patch — it strengthens repeat play, combat depth, and collection loops. Source: Crimson Desert Patch Notes 1.07.00.

3. China rank divergence — a new signal

The most interesting new signal in the data is that the Steam China sales rank is consistently stronger than the global rank — and it has held for a week, not 1–2 days.

3.1 Global vs China average rank by window

Window	Global avg	China avg	Gap
Last 24h	#22.2	#18.2	China +4.0 rungs
Last 48h	#26.1	#19.3	China +6.7 rungs
Last 7d	#23.4	#17.3	China +6.2 rungs
Post Patch 1.07	#24.3	#19.0	China +5.3 rungs
Since May 11	#24.3	#17.7	China +6.6 rungs

Over the past 7 days, China was better than global on 98.6% of observations — i.e., almost always.

3.2 Why this matters — the thesis shifts by one line

The initial thesis was “global AAA hit.” If China is now stronger than global, the thesis shifts slightly.

Prior signal set:

North-America / global-led AAA sales
Held a Steam global Top Seller position
Console + PC concurrent strength

Added signal set:

Steam-China demand attaching with a lag
China rank decays less even when global decays
Post-patch Chinese-user response and purchase conversion are relatively stronger

In particular, when global averaged #26–30 between May 13–15, China held #18–21. Open-world combat, boss rematches, fast patch cadence, and high-spec PC play culture may simply map well to the Chinese Steam audience.

3.3 What it is not

[Blocked] Today’s data cannot confirm any of:

Formal mainland-China distribution (license / 판호)
WeGame release
A formal Tencent partnership
China-mainland gross revenue scale

The Steam China rank is a relative indicator of Steam-China demand, not a measure of formal mainland sales. The figure mixes mainland VPN users, Hong Kong, Taiwan, and Chinese-language SEA users.

3.4 Trading read — bullish but not yet monetized

This is a “China optionality is waking up” signal. But it is not yet a price-moving catalyst. For the market to re-price, at least one of the following needs to land:

China rank persistently above global
Steam China rank settled inside Top 15
Clear inflection in Chinese community / streaming / review traction
A formal Tencent / WeGame / license / China-publishing signal
The company explicitly addresses China traction on a print or IR call

When two or more of these stack together, Pearl Abyss gets reclassified from “package-game maker” to “global AAA IP with China optionality”. Pearl Abyss already has Tencent shareholding ties and existing China business optionality — these reads carry more weight than a generic regional-rank improvement.

4. Units estimate

Current cumulative reviews — 151,847
Reviews at the 5M anchor — 130,635
Anchor-based review/units ratio — 2.6127%

Central units estimate: ~5.81M

Sensitivity:

Review / units ratio	Implied units
3.00%	5.71M
2.80%	5.76M
2.61%	5.81M
2.47%	5.86M
2.30%	5.92M

[Inference] The gap to 6M from the central case is ~190K units. At the current review-growth pace, the implied 6M ETA is around May 27–29. A stronger-than-usual weekend effect could pull it slightly earlier.

5. Investment read

This weekend’s data rebuts the bear thesis. The picture is not “sales rank slipped beyond #30 and the structure broke.” If anything, recovering to global #18 and US #19 post-patch supports the “fast patches sustain sales” thesis.

But this is not a bull confirmation. CCU is still -12% vs last weekend and average sales rank is below last week’s. Today’s data says “decay controlled and rank recovered,” not “re-acceleration.”

Practical read:

Current positions: hold
Adding: conditional
Conditions to add: global Top 20 settled into Sunday, US Top 25 held, weekend peak near or above 70K
Warning signs: post-Monday slip back outside global #30, US stuck outside #35, new-review positive rate below 80%
Thesis-kill conditions: the formal 6M announcement slips into mid-June and DLC / roadmap / capital-return / DokeV visibility fails to land while sales rank keeps falling

The cleanest summary is that this weekend is not “we’re hanging on” — it is closer to “the patch makes it sell again.” For Pearl Abyss to be re-rated as a global AAA studio with a proprietary engine, the next step set is clear: a formal 6M, DLC or a long-tail roadmap, and a capital-return signal all have to land together. If the §3 China-optionality signal holds for another 1–2 weeks, the holding thesis is reinforced one more notch — and if it disappears in the same window, it gets reclassified as noise.

6. How this links to other posts

DLC package → franchise re-classification (May 15) — “Not KRW 60B of DLC revenue — a multiple re-classification.” This weekend’s data is an observable variable for the Scenario B path.
Patch 1.04 post-patch weekend data (Apr 27) — the first comparable patch / weekend baseline. The patch cadence between 1.04 and 1.07 is the core of the thesis.
May 21 IR watch (May 12) — capital return, DLC, DokeV. Directly tied to the “thesis-kill” conditions above.
Shinhan target-price gap analysis (Apr 29) — Shinhan’s “2027 sales-cliff” assumption. Hitting 6M plus acceptance of Patch 1.07 are the first datapoints that erode it.

📚 The full Pearl Abyss series is collected in the Pearl Abyss / Crimson Desert research hub.

This article is research and commentary only and is not investment advice. CCU, new-review, and sales-rank figures come from SteamDB / SteamCharts as of the May 16, 23:20 KST snapshot and may differ from real-time values. Cumulative reviews of 151,847 and the 130,635 figure at the 5M anchor are observed values from Steam public data. The 5.81M central units estimate applies a 2.6127% anchor-based review/units ratio and is the author’s estimate; actual units depend on company disclosure. The May 27–29 ETA for 6M is a simple extrapolation of the current review-growth pace and is not confirmed. Patch 1.07 content references Pearl Abyss’s official Crimson Desert patch notes. The investment-action calls (hold current, conditional add, etc.) are author hypotheses and are not guaranteed. Global macro variables (US rates, oil, USD/KRW, VIX) can independently move Pearl Abyss as a KOSDAQ long-duration asset. The analysis may be wrong. Data cut-off: May 17, 2026 KST.

Samsung Electronics & SK Hynix KRW 62.5 Trillion Bonus Pool — Look at Profit Redistribution in Listed Equities, Not a Broad Domestic Recovery

Sun, 17 May 2026 00:00:00 +0000

📚 Related Series Samsung Electronics Strike vs. Memory Supercycle / Samsung Electronics TSMC-Style Re-rating / AI Advanced Packaging Data Comparison

Companion reads: May 12 KOSPI Crash & Rebound — “AI Citizen Dividend,” Tax Windfall, and the Profit-Sharing Debate — the day the market first priced in the “tax-windfall redistribution” channel we model here Why Korea Part 3 — Samsung Electronics + SK Hynix’s KRW 300T+ in Annual Profits Is Upgrading Korea’s Economic Structure — the big-picture macro view one level above this article’s “K-shaped reset”

The AI semiconductor profits of Samsung Electronics and SK Hynix have moved beyond individual corporate earnings to become a macro variable reshaping Korea’s capital flows. SK Hynix has institutionalized a structure allocating 10% of operating profit to its bonus pool, and Samsung Electronics’ DS division is in ongoing discussions around a profit-linked special bonus. On 2026 consensus estimates, the combined bonus pool for the two companies could reach approximately KRW 62.5 trillion. That number is large. But inferring that this money will immediately revive Korea’s aggregate domestic consumption is incorrect. The essential insight is not “consumption recovery” but “profit redistribution.” Listed-equity investors need to track where this money flows and which sectors’ income statements it actually hits.

Key Takeaways

The bonus pool is supplementary-budget scale. Applying SK Hynix’s 10% of operating profit and Samsung Electronics DS’s 12% of operating profit to 2026 consensus estimates yields a combined bonus pool of approximately KRW 62.5 trillion — roughly 2% of nominal GDP and around 5% of private consumption.
But the actual domestic consumption impulse is far smaller. After taxes, savings, debt repayment, real-estate and equity investment, and overseas spending, the realistic domestic consumption induced under the base scenario is approximately KRW 7–8 trillion — around 0.3% of GDP.
The beneficiary pool is narrow. Core employees in Samsung Electronics DS and SK Hynix number roughly 110,000–120,000 — approximately 0.2% of Korea’s total population and around 0.5% of all households. Rather than lifting aggregate consumption, this group moves specific asset markets and specific spending categories sharply.
For listed equities, the answer is not the average consumer stock. The capital flows split across: the semiconductor principals, brokerages and wealth management, semiconductor capex, power and cleanroom infrastructure, and premium consumption. Mid-market consumer stocks sit in the most ambiguous position.
The core remains Samsung Electronics and SK Hynix. The entire discussion originates from the two companies’ excess earnings. However, the larger the bonus pool, the larger the attendant labor cost allocation, shareholder return trade-off, and policy risk.
The most direct derivative beneficiary is the brokerage sector. The semiconductor rally and bonus inflows feed into equity trading volume, margin balances, and demand for wrap accounts, pension products, and wealth management. Listed brokerages such as Kiwoom Securities, Samsung Securities, and Mirae Asset Securities capture the most measurable benefit.
The structural beneficiary is capex and power infrastructure. Bonuses are a one-year cash flow event; the Yongin, Pyeongtaek, Icheon, and Cheongju cluster investments are a decade-long commitment. Power equipment, cable, cleanroom, utilities, and gas/scrubber vendors enjoy the longer-duration order book.
Risks are well-defined. A deceleration in AI capex, HBM price decline, conflict between bonus payouts and shareholder returns, and political debate over excess tax receipts and profit repatriation — if these converge, the cycle can be discounted rapidly.

1. KRW 62.5 Trillion Is Large. But It Is Not a “Domestic Recovery.”

The bonus pool arithmetic is straightforward. Assume SK Hynix earns approximately KRW 22.78 trillion in 2026 operating profit and allocates 10% as the bonus pool — roughly KRW 2.28 trillion. Assume Samsung Electronics DS earns approximately KRW 33.13 trillion in operating profit and allocates 12% as a special bonus — roughly KRW 3.98 trillion. Combined, the figure is approximately KRW 62.5 trillion.

At that scale it becomes a macro variable. Against a 2025 nominal GDP estimate of approximately KRW 2,663 trillion, this represents 2.35%; against private consumption of approximately KRW 1,241 trillion, it is around 5.0%. On the surface it resembles a private-sector supplementary budget.

The leap directly to “domestic recovery,” however, does not hold. Bonus income is high-income earned income, not a transfer payment to lower-income households. Effective tax rates are high, marginal propensity to consume is low, and debt repayment, savings, and investment allocations are large.

Consider a base scenario. Applying an effective tax rate of approximately 42% to KRW 62.5 trillion leaves an after-tax inflow of roughly KRW 36 trillion. Assuming 25% of that flows into incremental consumption, a domestic share of 70%, and a second-round multiplier of 1.2x, the domestic consumption impulse is approximately KRW 7–8 trillion — around 0.3% of GDP.

This is not meaningless. It matters. But it does not carry enough force to revive Korea’s broader self-employed sector, general retail, or middle-class consumption in one pass. Rather than dispersing broadly, the money flows deeply into specific asset classes and specific consumption categories.

2. Four Channels for Capital Flows

The first channel is taxation. Larger bonuses mean higher personal income tax receipts; larger corporate profits mean higher corporate tax receipts. This is unambiguously positive for government finances. For investors, however, what matters is where that tax revenue is deployed. If reinvested into semiconductor infrastructure, power grids, water supply, and R&D, the industry cycle extends. If directed toward near-term transfer payments, it may support consumption but is neutral to — or occasionally a drag on — corporate multiples.

The second channel is asset markets. High-income bonus recipients allocate more to real estate and financial assets than to consumption. Housing demand may strengthen in commuter-belt areas served by company shuttles — southern Gyeonggi, Icheon, Cheongju, Bundang, Suji, and Dongtan. In equity markets, capital moves into large-cap semiconductor names, ETFs, pension accounts, wrap accounts, and cash-equivalent products.

The third channel is premium consumption. Rather than broad consumer spending, high-price categories move first: imported vehicles, premium appliances, department-store VIP, luxury goods, golf, executive health check-ups, private education, and overseas travel. This creates opportunities for premium consumer stocks such as Hyundai Department Store, Shinsegae, and Hotel Shilla — but not equally across all retail names.

The fourth channel, and arguably the most important, is capex. The decade-scale investments flowing into the Yongin, Pyeongtaek, Icheon, and Cheongju clusters dwarf the bonus pool in size and duration. Power equipment, cable, cleanroom systems, piping, specialty gases, scrubbers, inspection and metrology, and logistics absorb this capital. From a listed-equity standpoint, this channel is the most structurally durable.

3. Listed-Equity Beneficiary Map

3.1 The Core: Samsung Electronics and SK Hynix

Every strand of this analysis originates with Samsung Electronics and SK Hynix. A larger bonus pool simply reflects larger operating profit — direct evidence of the AI memory supercycle’s intensity.

Investors must, however, hold two things simultaneously. First, bonuses are positive for employees but represent a transfer of excess profit to labor from a shareholder perspective. Second, when profits become very large, the political and social pressure around excess tax receipts, national dividend proposals, and industry contribution levies tends to follow.

The real investment thesis for both companies remains HBM pricing, customer qualification, supply capacity, and downcycle resilience. The bonus pool is a consequence, not a cause. Share prices will react more decisively to 2027–2028 earnings sustainability, HBM4/HBM4E market-share dynamics, and foundry and advanced packaging competitiveness than to bonus headlines.

3.2 The Most Direct Derivative: Brokerages and Wealth Management

When bonus payments and a rising market occur simultaneously, the first place the numbers appear is in brokerage earnings. Trading volume rises, margin balances and client deposits increase, and demand for wrap accounts, pension products, and wealth management services from affluent clients expands.

Kiwoom Securities carries the most direct beta to retail trading volume growth. Samsung Securities skews toward high-net-worth wealth management and carries a dividend-stock character. Mirae Asset Securities moves on global asset allocation, pension flows, overseas equities, and investment banking simultaneously.

For listed-equity investors, this segment warrants attention. Brokerage earnings sensitivity can be more direct than that of general consumer stocks. Bonus recipients do not spend all their proceeds in department stores; some portion inevitably flows into financial accounts. This can be tracked monthly through trading volume, client deposit balances, margin balances, and overseas equity transaction data.

3.3 The Structural Beneficiary: Power, Cleanroom, and Semiconductor Infrastructure

When semiconductor profitability extends for multiple years, companies do not spend only on people. The larger allocation goes into equipment and infrastructure. AI memory in particular demands massive quantities of power, water, cleanrooms, piping, specialty gases, inspection, and back-end infrastructure.

In power infrastructure, names such as LS ELECTRIC, HD Hyundai Electric, Hyosung Heavy Industries, and Taihan Electric Wire have already absorbed significant benefit. However, given how much these stocks have already re-rated, valuation discipline is required. Chasing these names without a pullback risks earning poor returns even if earnings deliver.

The second-tier cleanroom and utility group is comparatively less crowded. Companies such as Hanyang ENG, GST, and Sungjin E&C are less prominent than the large power names, but when cluster investment translates into actual contract awards, their income statements reflect it directly. For this group, order backlog and operating margin trends matter more than thematic momentum.

3.4 Selective Exposure: Premium Consumption

Where bonus income does translate into consumption, the destination is premium categories, not average spending. Department-store VIP, imported vehicles, fine dining, hotels, duty-free, executive health check-ups, and premium education move first.

In listed equities, Hyundai Department Store, Shinsegae, and Hotel Shilla are the relevant names. Each carries distinct risks: Hyundai Department Store requires monitoring both VIP sales trends and its Zinus exposure; Shinsegae requires separating the pace of department store recovery from duty-free recovery; Hotel Shilla operates duty-free and hotel segments on different cycle dynamics.

Mid-market general consumer stocks sit in the most difficult position. High-income earners trade up to premium; ordinary consumers trade down to value. The middle market faces structural pressure from both directions. A blanket view — “large bonus pool, therefore all domestic consumer stocks outperform” — is a dangerous simplification.

4. Interpretations to Avoid

First, do not connect KRW 62.5 trillion directly to a domestic consumption recovery. The portion that actually reaches consumption is materially smaller, and even then it concentrates in high-price categories.

Second, do not extrapolate semiconductor employee benefits to all Korean households. The direct beneficiary population is approximately 0.2% of the total population. Localized and category-specific changes are real; they are not sufficient conditions for a nationwide consumption recovery.

Third, do not put excessive faith in trickle-down dynamics. A significant share of high-income earnings is absorbed into assets. The broad pass-through to general consumption is limited. Generating a genuine domestic recovery requires tax redistribution, interest rates, real-estate stabilization, and labor market quality improvements — not bonus income alone.

Fourth, do not confuse a cyclical peak with permanent structural change. AI memory demand is strong, but capex deceleration or supply additions could materialize in 2027–2028. The market may currently be in the middle of a supercycle, but the intrinsic character of the memory industry remains cyclical.

5. Three Scenarios

Scenario	Conditions	Favorable Listed Equities
Strong Extension	HBM pricing holds firm; NVIDIA and hyperscaler capex revised upward; Samsung and SK earnings further upgraded	Samsung Electronics, SK Hynix, semiconductor equipment and infrastructure, brokerages
Base Path	Strength sustained through 2026–2027; 2028 deceleration risk begins to be priced	Brokerages, cleanroom and utility second-tier, selective premium consumption
Early Deceleration	AI capex slows; HBM pricing peaks out; KRW weakness and rising rates	Defensives, strong free-cash-flow names, high-dividend financials, select long-duration infrastructure order book names

The base path is the most realistic. Through 2026–2027, semiconductor earnings, bonuses, tax receipts, and capex move together, but markets will continuously discount the probability of a 2028 slowdown. This argues for selecting listed companies with verifiable earnings delivery over pure thematic momentum plays.

6. Monitoring Checklist

The indicators listed-equity investors should track are straightforward.

First, Samsung Electronics and SK Hynix Q2 and Q3 HBM revenue, operating margin, and the quarter in which bonus costs are recognized. The timing and magnitude of bonus expense recognition alters the operating margin interpretation materially.

Second, equity market trading volume and client deposit balances. For trading-volume beta names such as Kiwoom Securities, these figures flow directly into the earnings model.

Third, semiconductor cluster contract disclosures. Monitor whether order backlogs are growing at power equipment, cleanroom, utility, and gas and scrubber companies.

Fourth, premium consumption indicators: department-store VIP sales, imported vehicle registrations, hotel average daily rates, duty-free sales, and demand for executive health check-up programs.

Fifth, policy risk. Whether excess tax receipts are recycled into industrial reinvestment, cash redistribution, or corporate levy proposals will determine where multiples settle.

7. Bottom Line

The KRW 62.5 trillion bonus pool at Samsung Electronics and SK Hynix is unambiguously large. But reading it as “the key to Korea’s domestic recovery” is too simplistic. The beneficiary pool is narrow, marginal propensity to consume is low, and capital flows more heavily into taxes, assets, capex, and premium spending than into broad consumption.

For listed-equity investors, the relevant question is not average consumption — it is the path of profit redistribution. The core is Samsung Electronics and SK Hynix. The most direct derivative is brokerages and wealth management. The most structural beneficiary is power, cleanroom, and semiconductor infrastructure. Selective exposure lies in premium consumption. Mid-market consumer stocks sit in the most ambiguous position of all.

The conclusion is simple. The KRW 62.5 trillion bonus pool is not money that revives Korea’s economy in aggregate — it is money that redistributes capital flows within the Korean economy. Investors should ask not “who receives the money” but “whose listed-company revenue and earnings lines does it actually appear on.” That distinction separates thematic buying from earnings-based investing.

This article is for research and commentary purposes only and does not constitute investment advice. SK Hynix’s profit-sharing structure — 10% of operating profit as the bonus pool, removal of the PS cap, and an 80%/20% current-year/two-year deferred payment schedule — reflects the 2025 labor-management agreement. The Samsung Electronics DS 12% special bonus proposal is based on media reports of the May 2026 National Labor Relations Commission post-mediation recommendation; actual payment terms are subject to company-union negotiation and board and management judgment. The 2026E operating profit consensus figures and the KRW 62.5 trillion bonus pool estimate are analyst calculations based on publicly available estimates and may differ materially from actual results. Domestic consumption induced, effective tax rates, marginal propensity to consume, and multiplier effects are analyst assumptions. Companies mentioned in this article are cited as illustrations of economic exposure from a listed-equity perspective and do not constitute buy or sell recommendations. AI capex, HBM pricing, policy direction, interest rates, and exchange rates are all subject to change, and this analysis may prove incorrect. Data reference date: 17 May 2026 KST.

Who Absorbed the May 15 Selloff — Flow Screening Narrows to Hana Micron, HL Mando, and심텍

Sun, 17 May 2026 00:55:00 +0900

🔗 Related reading: May 15 Closing Brief — Samsung Electronics ₩2.5T Foreign Sell, Why Now · May 15 Macro Gate — Relative Strength Survivors in the KOSPI Crash · Jeju Semiconductor 1Q26 Earnings — Benefiting from AI-Driven Legacy Memory Tightness

May 15 was a brutal session for Korean equities. Market-wide flow came in at foreigners -₩5.18T, institutions -₩1.88T, and retail +₩6.93T. Foreign selling was concentrated in large-caps, with retail absorbing the bulk of the supply.

On a day like this, looking only at the top net-buy names is not enough. The more important question is: which names had already been accumulating through May 14, and continued to see foreign and institutional buying even as the broader market sold off on May 15?

The answer from that lens is fairly clear. Hana Micron, HL Mando, Simtec, Jeju Semiconductor, and Rainbow Robotics stand out. Hana Micron and HL Mando in particular show the cleanest structure — retail selling, foreigners and institutions absorbing — during the crash.

Data covers cumulative flow from May 4 through May 14, 2026, combined with May 15 single-day flow.

Key Summary

Layering the May 4–14 accumulation signal on top of the May 15 crash-day absorption reveals better candidates than a simple net-buy ranking. The core filter is simultaneous foreign and institutional buying paired with retail selling.

The strongest structure is: foreigners and institutions buying together through May 14, retail selling throughout, and both groups continuing to absorb on May 15. Names that pass this combined filter include Hana Micron, HL Mando, Rainbow Robotics, Jeju Semiconductor, Hugel, and GS Retail.

Within semiconductors, Hana Micron stands out the most. On May 15 alone, foreigners net-bought +₩91.8B, while retail sold -₩93.8B. Simtec had strong dual-accumulation through May 14 but saw institutional wobble on the 15th. Jeju Semiconductor has strong flow but has already run sharply.

On the flip side, LG Electronics, Hyundai Motor, and Hyundai E&C showed strong cumulative flow through May 14 but saw heavy foreign selling on May 15. These are not new screening candidates — they warrant caution around distribution rather than fresh entry.

Final screening candidates: Hana Micron, HL Mando, Simtec, Jeju Semiconductor, Rainbow Robotics, and Hugel/GS Retail as secondary ideas. Rather than chasing near-term momentum, the more rational approach is to watch for pullbacks where the flow signal holds.

1. The Strongest Pattern — Consecutive Dual Accumulation

The names below form the core of this screen. Foreigners and institutions bought together through May 14, and continued absorbing retail supply on May 15 — against a backdrop of broad market-wide foreign and institutional selling. That context makes the signal more meaningful.

Name	Return thru 5/14	5/15 Change	Foreign 5/14 cum.	Inst. 5/14 cum.	Foreign 5/15	Inst. 5/15	Retail 5/15	Assessment
Rainbow Robotics	+22.8%	-3.69%	+₩252.7B	+₩64.0B	+₩6.0B	+₩15.2B	-₩20.1B	Best flow quality, stretched valuation
Hana Micron	+6.1%	+18.61%	+₩30.4B	+₩4.6B	+₩91.8B	+₩1.6B	-₩93.8B	Strongest actionable candidate
Jeju Semiconductor	+33.3%	+8.86%	+₩29.2B	+₩33.4B	+₩23.5B	+₩8.1B	-₩33.3B	Strong but overheated
HL Mando	+8.4%	-1.91%	+₩24.5B	+₩13.3B	+₩21.4B	+₩13.6B	-₩33.7B	Most comfortable entry candidate
Hugel	+9.2%	-1.05%	+₩26.4B	+₩5.5B	+₩3.4B	+₩1.9B	-₩5.1B	Defensive, steady
GS Retail	+24.9%	+2.26%	+₩12.8B	+₩10.7B	+₩5.9B	+₩2.8B	-₩8.5B	Consumer staples flow intact
Seegene	+30.6%	-2.97%	+₩24.1B	+₩4.1B	+₩2.3B	+₩2.1B	-₩4.6B	Possible event-driven move
POSCO DX	+0.8%	-5.47%	+₩11.6B	+₩6.5B	+₩2.8B	+₩1.1B	-₩3.8B	Low-price absorption type
RF Materials	+1.7%	-8.70%	+₩10.3B	+₩1.9B	+₩1.3B	+₩2.2B	-₩4.0B	Small-cap option

The two most important rows in this table are Hana Micron and HL Mando.

Hana Micron had been accumulating quietly through May 14. Then on May 15 — a crash day — foreigners net-bought +₩91.8B in a single session while retail sold -₩93.8B. The stock rose +18.61%. Flow and price exploded in the same direction simultaneously. Chasing the move carries short-term risk, but this is the strongest name from this screen.

HL Mando offers a more comfortable setup. The stock declined -1.91% on May 15, but foreigners bought +₩21.4B, institutions bought +₩13.6B, and retail sold -₩33.7B. Foreigners and institutions absorbed retail supply together during the crash. The entry burden is considerably lower than names that have already surged.

Rainbow Robotics has the best raw flow quality. However, it had already gained +22.8% through May 14 and carries significant valuation risk. “Good stock” and “good entry point” are different things. Here, waiting for a pullback is more prudent than chasing.

2. Semiconductor Names Only

Within semiconductors, the ranking is sharper. Hana Micron leads, Simtec follows, Jeju Semiconductor is strong but overheated, and Daeduck Electronics is a quality name that ranks lower on this particular flow combination.

Name	Interpretation
Hana Micron	Foreign buying of +₩91.8B on May 15 is dominant. Accumulation was quiet through May 14, then exploded on the crash day. Short-term chase risk is real, but this is the #1 name from this screen.
Simtec	Through May 14: foreigners +₩66.6B, institutions +₩33.0B, retail -₩43.7B. On May 15: foreigners +₩12.6B, institutions -₩7.6B. Foreigners held but institutions wavered.
Jeju Semiconductor	Dual accumulation through May 14 and again on May 15. But with +33.3% through May 14 and an additional +8.86% on May 15, chase risk is elevated.
Daeduck Electronics	Core signal through May 14 was institutional +₩41.1B. On May 15: foreigners +₩3.8B, institutions -₩7.5B. Flow quality below Simtec.
Philoptics	Strong through May 14, but saw simultaneous foreign and institutional selling on May 15. Downgraded from candidate list.
HPSP	Foreign and institutional flow was intact through May 14, but foreigners sold on May 15. Appropriate for secondary observation only.

Semiconductor flow conclusion: Hana Micron > Simtec > Jeju Semiconductor > Daeduck Electronics.

Jeju Semiconductor, as covered separately in the earnings note, posted 1Q26 revenue of ₩180.5B, operating profit of ₩67.1B, and an operating margin of 37.2%. Earnings and flow are aligned. However, the stock moved first and moved hard — for new screening purposes, “strong but wait for a pullback” is the more reasonable framing.

Simtec remains worth watching on the semiconductor substrate side. Cumulative flow through May 14 was solid and foreigners continued buying on May 15. The institutional wobble is the key concern. The next confirmation point is institutional re-entry.

3. Names Where Foreigners Absorbed the May 15 Dip

The following group saw foreigners buying on May 15 but with weak or mixed institutional support. They rank below the dual-accumulation group, but qualify as “crash-day absorption” candidates worth monitoring.

Name	Foreign 5/14 cum.	Foreign 5/15	Inst. 5/15	Interpretation
Fadu	+₩103.2B	+₩56.4B	-₩14.6B	Foreigners very strong. Valuation stretched
KB Financial	+₩107.9B	+₩44.5B	-₩11.9B	Defensive foreign accumulation
SK	+₩152.8B	+₩28.1B	-₩19.8B	Holding company re-rating bet
POSCO Holdings	+₩148.4B	+₩19.0B	+₩5.9B	Low-price absorption character
Samsung Fire & Marine	+₩70.9B	+₩18.2B	+₩7.6B	Top name among defensives
Hanmi Pharm	+₩44.0B	+₩10.8B	+₩19.9B	Large-cap pharma absorption
Kolon Industries	+₩21.6B	+₩24.1B	-₩14.7B	Foreigners bought the dip, institutions still negative
DL E&C	+₩71.6B	+₩14.6B	-₩3.2B	Construction dip buy
Simtec	+₩66.6B	+₩12.6B	-₩7.6B	Foreigners held, institutions wavered

This group is better characterized as “dip absorption” than “trend continuation.” Fadu has very strong foreign flow but stretched valuation. KB Financial and Samsung Fire & Marine are defensive in nature. Kolon Industries saw aggressive foreign buying but institutions remain negative.

Buy priority is lower than the dual-accumulation group. However, this list becomes useful for watching whether foreigners continue to hold through the next market wobble.

4. Patterns to Avoid — Potential May 15 Distribution

Names that looked strong through May 14 can tell a different story once May 15 is added. Names where foreigners sold heavily into a price spike while retail absorbed the supply should be excluded from the new-idea list.

Name	Foreign 5/14 cum.	Foreign 5/15	Retail 5/15	Assessment
LG Electronics	+₩337.2B	-₩261.0B	+₩256.3B	Heavy foreign selling into the rally
Hyundai Motor	+₩103.6B	-₩202.8B	+₩195.3B	Institutions held; foreigners distributed
Hyundai E&C	+₩222.5B	-₩15.2B	+₩30.1B	Flow damaged on May 15
LG CNS	+₩54.2B	-₩44.1B	+₩16.6B	Institutions absorbed but foreigners exited
Philoptics	+₩44.7B	-₩0.5B	+₩0.8B	Strong candidate downgraded
Eugene Tech	+₩30.8B	-₩2.8B	+₩3.5B	Both price and flow weak

LG Electronics is the clearest example. Foreigners had accumulated +₩337.2B through May 14. Then on May 15, foreigners sold -₩261.0B in a single session while retail bought +₩256.3B. This is a classic distribution structure — foreigners offloading into strength. A name that “was strong” is not the same as a name “to discover now.”

Hyundai Motor follows the same pattern. Institutions held their position, but foreigners sold -₩202.8B while retail bought +₩195.3B. When foreigners trim aggressively on a down day, flow recovery must be confirmed before adding these names as new candidates.

5. Final Screening Candidates

Priority	Name	Conclusion
1	Hana Micron	+₩91.8B foreign buying on the crash day vs. -₩93.8B retail selling. Strongest flow reversal on the screen
2	HL Mando	Consecutive foreign and institutional buying; limited drawdown on May 15. Relatively low chase risk
3	Simtec	Dual accumulation through May 14, foreigners held on May 15. Valid if institutional re-entry is confirmed
4	Jeju Semiconductor	Flow very strong but already overheated. Wait for a pullback
5	Rainbow Robotics	Best-in-class flow. Valuation and price burden argue against chasing
6	Hugel / GS Retail	Defensive and consumer staples secondary ideas

In practice, the candidates fall into three buckets.

First, the momentum-trend leader is Hana Micron. Foreign buying intensity on May 15 was dominant. However, after a single-day surge of +18.61%, immediate pursuit carries significant volatility risk.

Second, the less-overheated flow name is HL Mando. Foreigners and institutions have been consistent buyers, and the stock held up relatively well during the selloff. It offers the best balance between flow quality and price burden.

Third, the semiconductor substrate candidate is Simtec. Flow is softer than Hana Micron, but cumulative dual accumulation through May 14 combined with sustained foreign interest on May 15 is meaningful. Institutional re-entry would be a catalyst to move it up the list.

6. One Final Thought

On a crash day like May 15, the more important question is not “what rallied the most” — it is “who absorbed the supply.” Against a market backdrop of foreigners -₩5.18T, institutions -₩1.88T, and retail +₩6.93T, a handful of names moved in the opposite direction.

By that standard, the two cleanest names are Hana Micron and HL Mando. Hana Micron is the stronger signal; HL Mando is the more comfortable setup. Simtec is the semiconductor substrate secondary, while Jeju Semiconductor and Rainbow Robotics are compelling but carry meaningful chase risk.

Conversely, LG Electronics, Hyundai Motor, and Hyundai E&C should not be evaluated on their May 4–14 cumulative flow alone. The moment foreigners sold heavily on May 15, the conclusion changes. These names belong closer to profit-taking watch lists than new discovery candidates.

From a portfolio manager’s perspective, the genuinely strong stocks in a crash are not the top performers on the day. They are the names where retail sold and foreigners plus institutions absorbed. By that measure, Hana Micron and HL Mando passed the test most cleanly in this screen.

This article is for research and commentary purposes only and does not constitute investment advice. Flow data is based on cumulative flows from May 4–14, 2026 and single-day flows on May 15, 2026. Foreign, institutional, and retail flow figures may vary depending on exchange and market data aggregation methodology. Strong flow does not guarantee price appreciation. Names that have surged sharply in the near term carry elevated volatility and profit-taking risk. This analysis may be wrong. Data reference date: May 15, 2026 KST.

AI Back-End — 11 Stocks Side by Side. Who's Actually Cheap and Who's Expensive? Real Value on 2027E PER × Operating-Profit Growth

Sat, 16 May 2026 00:00:00 +0000

📚 AI back-end series Previous: Substrates are a “volume beta,” test sockets are a “consumables beta” — same AI tailwind, completely different structures

Yesterday’s post compared substrates and test sockets structurally. Today we go one step further and compare them numerically. Eleven AI back-end names lined up on the same yardstick — 2026 YTD return, operating-profit growth, 2026E and 2027E PER, 2026E operating margin. Even within “AI winners,” how much each has already moved and how much room is left differ wildly. One table tells you most of the story.

Key takeaways

On pure “cheapness”: SK hynix (2027E PER 5.2x), Samsung Electronics (5.6x), Haesung DS (15.4x). But the memory megas are dominated by “cycle-peak” risk.
On “growth-adjusted multiple”: Daeduck Electronics (2027E PER 27.1x vs OP +34%) and Simmtech (20.0x vs OP +59%) are the sharpest setups.
“Already done the work”: Samsung Electro-Mechanics (2026 YTD +296%, 2026E PER 61.9x), Isu Petasys (+346% in 2025 but only +7% YTD 2026), Simmtech (+352% in 2025).
“Quality, but the multiple is rich”: LEENO Industrial (OPM 48.6%, 2026E PER 41.2x), ISC (2026E PER 58.5x).
“Growth at a fair price”: TSE (2026E OP +82%, PER 29.4x).
Bottom line: within AI back-end, attractiveness ranks substrates > test sockets > memory megas. Within substrates, Daeduck and Simmtech are the cleanest.

1. How the comparison is built

1.1 The data set

Field	Source	Meaning
2025 stock return	PyKRX	How much it ran last year
2026 YTD return	PyKRX	How much it has run this year
2025 OP YoY	Company disclosures	Last year’s earnings growth
2026E OP YoY	FnGuide consensus	Market’s view on this year
2027E OP YoY	FnGuide consensus	Market’s view on next year
2026E PER	FnGuide consensus	Multiple on this year’s earnings
2027E PER	FnGuide consensus	Multiple on next year’s earnings
2026E OPM	FnGuide consensus	Business-model margin level

Reference date: May 15, 2026 close.

1.2 Why 2027E PER matters more than 2026E PER

2026E PER:
→ Multiple on this year's expected earnings
→ Already absorbing 1Q26 reality
→ Affected by short-term quarterly noise

2027E PER:
→ Multiple on next year's expected earnings
→ Reflects "will the cycle persist?"
→ Where the market's true view shows up

Examples:
Company A: 2026E PER 30x, 2027E PER 28x
→ Market expects earnings roughly flat
→ "Stable growth" archetype

Company B: 2026E PER 30x, 2027E PER 15x
→ Market expects earnings to nearly double
→ Or the market hasn't bought in yet → undervalued setup

Company C: 2026E PER 7x, 2027E PER 5x
→ Market doubts earnings durability
→ "Cycle-peak fear" — typical of memory megas.

1.3 Why OP growth has to be read with PER

PER alone is misleading.

Same PER 30x:
A: OP YoY +20% → 30 / 20 = 1.5
B: OP YoY +60% → 30 / 60 = 0.5
B is far more attractive.

This is the PEG ratio (PER / OP growth).
≤1.0 attractive, 1.0–2.0 fair, >2.0 expensive.

This post visualizes a simplified version —
a 2027E PER × 2027E OP YoY matrix.

2. Eleven stocks, one table

2.1 The basic comparison

Bucket	Name	2025 return	2026 YTD	2025 OP YoY	2026E OP YoY	2027E OP YoY	2026E PER	2027E PER	2026E OPM
Memory	Samsung Electronics	+124.5%	+125.6%	+33.2%	+688.6%	+26.7%	7.0x	5.6x	51.1%
Memory	SK hynix	+280.3%	+179.4%	+101.2%	+433.1%	+36.0%	6.8x	5.2x	75.9%
Memory	Jeju Semiconductor	+186.4%	+212.3%	+274.0%	n/a	n/a	n/a	n/a	n/a
Test socket	LEENO Industrial	+52.1%	+67.5%	+42.5%	+23.2%	+24.1%	41.2x	33.4x	48.6%
Test socket	ISC	+58.7%	+102.1%	+34.2%	+65.9%	+41.4%	58.5x	43.2x	31.8%
Test socket	TSE	+39.5%	+215.1%	+23.3%	+81.9%	+21.5%	29.4x	24.7x	17.0%
Substrate	Samsung Electro-Mechanics	+108.3%	+296.1%	+24.3%	+72.0%	+50.0%	61.9x	41.9x	11.8%
Substrate	Daeduck Electronics	+206.0%	+188.3%	+334.5%	+375.4%	+34.4%	36.1x	27.1x	15.5%
Substrate	Simmtech	+351.9%	+103.2%	loss→profit	+1,234.5%	+58.7%	31.0x	20.0x	8.6%
Substrate	Haesung DS	+140.6%	+57.3%	-18.3%	+99.6%	+36.6%	20.3x	15.4x	11.4%
Substrate	Isu Petasys	+346.5%	+7.4%	+100.9%	+57.1%	+33.0%	35.6x	26.7x	21.1%

2.2 How to read it

Down the columns:
- High 2025 return = already run a lot
- High 2026 YTD = still strong this year
- High 2026E OP YoY = earnings explosion expected
- Low 2027E PER = market doubts earnings persistence

Across the rows:
- Was the name strong last year too (+200%+)?
- Still strong this year (+100%+)?
- Yet not punitive on the multiple (2027E PER ≤ 30)?
- Decent operating margin (2026E OPM ≥ 20%)?

Names that hit all four = most attractive.

3. Reading each bucket

3.1 Memory megas — “cheap, but cycle-peak fight”

The numbers look stunning:
SK hynix: 2027E PER 5.2x, OPM 75.9%
Samsung Electronics: 2027E PER 5.6x, OPM 51.1%
→ In normal industries, PER 5x screams "bankruptcy risk"
→ Memory is different. When the cycle turns, earnings fall by half+.

Why the market still applies 5x:
"We'll grant you 2026 earnings.
 We are not convinced 2027–28 earnings will stay here."

This is the heart of memory investing:
- The stock looks cheapest right before the cycle peaks
- The moment the cycle turns, PER spikes (because EPS collapses)
- Buying "trough PER" can mean selling at the actual top.

Read:
SK hynix — best case if HBM supercycle runs into 2027.
 Large drawdown if the cycle turns.
Samsung — strike risk + HBM share doubts = discount vs. SK hynix.
Jeju Semi — no consensus available; excluded from direct compare.

3.2 Test sockets — “quality intact, multiple full”

LEENO Industrial:
- OPM 48.6% (top tier of Korean manufacturing)
- 2027E OP YoY +24.1% (steady growth)
- 2027E PER 33.4x
→ Read: highest quality, but multiple is full.

ISC:
- OPM 31.8% (very good)
- 2027E OP YoY +41.4% (strong growth)
- 2027E PER 43.2x
→ Read: cleanest AI data-center beta, but tightest multiple.

TSE:
- OPM 17.0% (lowest among the three)
- 2027E OP YoY +21.5% (decent)
- 2027E PER 24.7x (cheapest in test sockets)
- 2026 YTD +215% (already running hard)
→ Read: best price-for-growth balance, but chase risk after the YTD move.

Read:
Don't treat LEENO and ISC as the same "test-socket stock."
- LEENO = quality compounder (hold-and-own)
- ISC = direct AI data-center beta (momentum)
- TSE = the balanced "value-for-growth" option.

If you must pick one new entry:
- 1-year hold: LEENO (stability)
- 6-month momentum: TSE (value-for-growth)
- Direct AI exposure: ISC (speed).

3.3 Substrates — the most interesting bucket

Samsung Electro-Mechanics:
- 2026 YTD +296% (among the top movers)
- 2026E PER 61.9x → 2027E PER 41.9x
- OPM 11.8% (blended across MLCC + FC-BGA)
→ Great company, but chasing is inefficient.

Daeduck Electronics:
- 2025 +206%, 2026 YTD +188% (persistent strength)
- 2026E PER 36.1x → 2027E PER 27.1x
- OPM 15.5% (top of substrate group)
- 2026E OP YoY +375%, 2027E +34%
→ The numbers justify the leadership premium.

Simmtech:
- 2025 +352% (biggest mover)
- 2026 YTD +103% (still strong)
- 2026E PER 31.0x → 2027E PER 20.0x (lowest)
- 2026E OP YoY +1,234% (loss → profit)
- 2027E OP YoY +58.7%
- OPM 8.6% (margin needs to expand further)
→ Turnaround + multiple appeal — but margin durability needs proof.

Haesung DS:
- 2026 YTD +57% (lagger of the group)
- 2026E PER 20.3x → 2027E PER 15.4x (cheapest of the group)
- 2026E OP YoY +99.6% (very strong)
- OPM 11.4%
→ The cheapest candidate. Weaker leadership profile,
 but the price-for-growth is good.

Isu Petasys:
- 2025 +346% (a parabolic year)
- 2026 YTD +7% (resting this year)
- 2026E PER 35.6x → 2027E PER 26.7x
- OPM 21.1% (highest in substrates)
→ All the move came pre-emptively last year; resting.
 Wait for relative-strength recovery.

4. The growth-vs-multiple matrix — the most intuitive view

4.1 2027E PER × 2027E OP YoY

X-axis: 2027E operating-profit growth
Y-axis: 2027E PER

low ← PER → high
+5x --10x---20x---30x---40x---50x PER

+60% │ Simmtech
 │
+50% │ Samsung Electro-Mechanics
 │
+40% │ ISC
 │
+35% │ SK hynix
 │
+34% │ Daeduck Electronics
 │
+36% │ Haesung DS
 │
+33% │ Isu Petasys
 │
+27% │ Samsung Electronics
 │
+24% │ LEENO Industrial
 │
+22% │ TSE
 │
 └──────────────────────────────

How to read:
- Upper right (high PER + low growth): expensive
- Lower left (low PER + high growth): best value
- Upper left (low PER + low growth): trap
- Lower right (high PER + high growth): momentum

Most attractive zone:
Left (low PER) + top (high growth) = Simmtech, Daeduck

Most punitive zone:
Right (high PER) + mid growth = LEENO

Special zone:
Far left (very low PER) = SK hynix, Samsung Electronics
→ Looks cheap, but cycle-peak risk lives here.

4.2 Reading “real value” from the matrix

Simplified PEG (PER / growth):

Simmtech: 20.0 / 58.7 = 0.34 ★★★★★
SK hynix: 5.2 / 36.0 = 0.14 (memory exception)
Samsung Elec: 5.6 / 26.7 = 0.21 (memory exception)
Haesung DS: 15.4 / 36.6 = 0.42 ★★★★
Daeduck: 27.1 / 34.4 = 0.79 ★★★★
Isu Petasys: 26.7 / 33.0 = 0.81 ★★★
Samsung E-M: 41.9 / 50.0 = 0.84 ★★★
TSE: 24.7 / 21.5 = 1.15 ★★★
ISC: 43.2 / 41.4 = 1.04 ★★★
LEENO: 33.4 / 24.1 = 1.39 ★★

Bucketing:
- ≤ 0.5: very attractive
- 0.5–1.0: attractive
- 1.0–1.5: fair
- ≥ 1.5: expensive

Memory megas are a special class — cycle peak.
On normal terms, Simmtech > Haesung DS > Daeduck.

LEENO looks the worst on PEG, but its
earnings stability and predictability are unmatched.
It deserves a "premium multiple" — PEG alone underrates it.

5. Cross-bucket — “where should fresh capital go?”

5.1 Substrates vs test sockets vs memory

Scenario 1: "AI back-end shortage cycle continues"
→ Substrates most attractive
 shortage → ASP hike → margin expansion
 Daeduck Electronics, Simmtech are top picks

Scenario 2: "AI chip diversification has further to run"
→ Test sockets most attractive
 new chip = new socket
 LEENO Industrial, ISC are top picks

Scenario 3: "HBM supercycle extends into 2028"
→ Memory megas most attractive
 SK hynix is top (OPM 75.9%)
 But you have to accept cycle-peak risk

Safest framing:
Diversify across the three buckets — 4:3:3 or 5:3:2.

5.2 Illustrative allocation across the 11

Just an example. Not real money advice.

Aggressive (short-term momentum tilt):
Daeduck 25% / Simmtech 20% / Samsung E-M 15%
ISC 15% / TSE 10% / SK hynix 15%

Balanced (growth + stability):
Daeduck 20% / LEENO 20% / SK hynix 15%
Samsung E-M 15% / Haesung DS 10% / Simmtech 10%
TSE 10%

Defensive (quality tilt):
LEENO 30% / Samsung E-M 20% / SK hynix 20%
Daeduck 15% / Haesung DS 15%

Allocation rules:
1. Higher OPM = larger weight
2. Heavier multiple burden = smaller weight
3. Don't concentrate in a single bucket
4. Cap cycle-risk names.

6. One-line read by name

6.1 Author’s attractiveness ranking

Rank	Name	One-liner
1	Daeduck Electronics	Best balance of growth, leadership, and multiple. Direct AI MLB / FC-BGA beta.
2	Simmtech	Best growth-vs-PER setup on 2027E. Needs further 2026 margin expansion to confirm.
3	Haesung DS	Cheapest substrate candidate. Less leadership profile, but strong value-for-growth.
4	SK hynix	The memory leader. Cheap on screens; cycle-peak risk is the swing variable.
5	TSE	Best value-for-growth within the test-interface group. But 2026 YTD +215% — already in motion.
6	LEENO Industrial	Top quality, full multiple. Best as a core 1–2 year holding.
7	Samsung Electro-Mechanics	A great company, but +296% YTD and 2026E PER 61.9x make chasing inefficient.
8	ISC	Real growth, but tightest multiple. Cleanest AI data-center beta.
9	Isu Petasys	Strong long-term structure, but +7% YTD — wait for relative-strength recovery.
10	Samsung Electronics	Cheap on numbers, but discount vs. SK hynix on the HBM narrative is real.
11	Jeju Semiconductor	Strong stock and earnings, but no 2026–27 consensus → excluded from direct compare.

6.2 Pre-entry checklist

Before buying any of these:

1. Macro gate (see the prior post)
 - US 10-year below 4.45%
 - Brent below 105 USD
 - USD/KRW below 1,480
 - VIX below 18
 → 2–3 of 4 stable before new buying.

2. Stock-level catalysts
 - Quarterly print dates
 - New customer LTA announcements
 - New-fab ramp schedule (LEENO)
 - 2Q guidance

3. Scaling rules
 - No one-shot buys
 - Adjust size as macro gates flip
 - Prefer pullback entries over chasing.

4. Bucket diversification
 - Avoid single-bucket concentration
 - Sensible balance across memory / substrate / socket.

7. Frequently asked

7.1 “Why not put memory megas at #1 if PER is single-digit?”

PER 5x looks like "trough PER,"
but a multiple built on "trough EPS" is dangerous.

The memory-cycle trap:
Cycle peak: earnings spike → PER 5x
Cycle decline: earnings cut in half → same price → PER 10x
Cycle trough: earnings cut to 1/3 → same price → PER 15x+

So buying at PER 5x means:
- More cycle ahead → big gains
- Cycle turns → suddenly PER 15x at the same price → big losses

That's why memory investing is not
"how cheap does it look?"
but
"when does the cycle turn?"

To put memory at #1 here, you'd need separate proof on
HBM4 / HBM4E demand visibility, China demand recovery,
and the durability of AI capex.

This post does a clean side-by-side only —
so memory megas are placed mid-pack to reflect cycle-peak risk.

7.2 “Is Simmtech’s 20x PER real?”

Simmtech's 2027E PER 20.0x is consensus-based.

Conditional signals:
1. 2026 OP turns positive (from loss),
 producing a +1,234% optical YoY (it is base-effect, not the run rate)

2. 2027 OPM must expand beyond 8.6% for the 20x PER to be "real"

3. Need a higher share of AI memory module / SSD-controller
 substrate revenue → confirm margin mix improvement.

Risks:
- Margin recovery may lag consensus
- 8.6% OPM is among the lowest in substrates (vs Daeduck 15.5%)
- 1Q26 results could trigger consensus revisions.

Read:
20x PER is "potential" value, not yet realized value.
Margin recovery is the real test.
Wait for 1Q26 / 2Q26 confirmation before adding.

7.3 “Samsung E-M is a great company. Why is it #7?”

Samsung Electro-Mechanics fundamentals are among the best:
- Combined AI FC-BGA + MLCC + camera exposure
- 1Q26 OP +40% YoY
- Core node in the global AI parts supply chain.

But on price:
- 2026 YTD +296% (one of the biggest movers of the 11)
- 2026E PER 61.9x (heavy vs. bucket average)
- OPM 11.8% (FC-BGA alone is higher; cameras drag the blend).

The ranking is "attractiveness at current price,"
not "best long-term holding."

For a long-term core position, Samsung E-M ranks high.
For best value-for-growth at today's price, Daeduck and Simmtech lead.

8. How this links to other posts

Prior post (structure compare):
→ Substrate = "volume beta," test socket = "consumables beta"
→ Structural differences between the buckets.

This post (data compare):
→ Who is most attractive WITHIN each bucket
→ Not raw PER — "growth-adjusted multiple."

Macro-cycle synthesis post:
→ "Cycle before the stock"
→ Scale into these names after the macro gate clears.

KOSPI crash + macro-gate post:
→ After the May 15 -6.12% drop
→ Watch relative strength AND the macro gate
→ The framework used to select names here.

9. The one-line bottom line

Even among “AI back-end winners,” how much price has already done the work differs wildly. Low PER alone ≠ attractive. Low PER can mean “the market fears the cycle is topping” (memory) or “the market hasn’t bought in yet” (turnaround names).

Lined up in one table, the picture is clear. On a PER × OP-growth matrix, the sharpest setups are Simmtech, Daeduck Electronics, and Haesung DS. All three are substrate names — that is not an accident. Substrates are in the middle of a “shortage → ASP hike → margin expansion” cycle.

Within test sockets, LEENO Industrial is the quality leader but the multiple is full; ISC is the cleanest AI data-center beta but carries the tightest multiple. For best price-for-growth on a new entry, TSE is the value option.

Memory megas show stunning numbers but a cycle-peak debate at their core. SK hynix’s 2027E PER 5.2x would imply bankruptcy risk in a normal industry; in memory it means “the market doubts earnings durability.” If the cycle persists, the upside is huge; the moment it turns, the multiple expands automatically.

At today’s prices, the rational order for new capital is substrates > test sockets > memory megas. Within substrates, Daeduck Electronics (leadership premium) and Simmtech (turnaround value) are the cleanest. Whatever you choose, scale in only after the macro gate clears. Buying a great company at the right price is the actual job.

This article is research and commentary only and is not investment advice. Stock returns are per PyKRX, with 2025 returns from 2024 year-end close to 2025 year-end close, and 2026 YTD from the 2025 year-end close to the May 15 close. Operating-profit growth, PER, and operating margin are per the FnGuide CompanyGuide consensus (queried on May 15, 2026) and may differ from actual reported figures. The consensus reflects the market average and can differ from individual broker estimates, and may be revised. The PEG metric (PER ÷ growth) is a simplified valuation aid and is not sufficient by itself. The cycle-peak concern for memory megas is the author’s view; actual cycles can be longer or shorter than expected. The attractiveness ranking reflects “best at current price” and may differ from a “best long-term holding” ranking. Jeju Semiconductor is excluded from the direct comparison due to a lack of 2026–2027 consensus and warrants its own analysis. Global macro variables (US rates, oil, FX, VIX) can independently move the stocks. The analysis may be wrong. Data cut-off: May 15, 2026 close, KST.

Can Samsung Electronics Deserve 15x PER? From Memory Cycle Stock to AI Integration Platform

Sat, 16 May 2026 00:00:00 +0000

📚 Samsung Electronics Series Part 1: Citi TP KRW 460,000 — Why the Memory-Cycle Frame May Be Wrong Part 2: Strike vs Memory Supercycle — The Real Question Is Who Captures Excess Profit Related hub: AI HBM Hub · Memory Pulse Dashboard

Samsung Electronics trades at roughly 5x 2027E PER. TSMC trades around 19-22x. Both are AI beneficiaries, yet the market assigns them valuation multiples that are almost four times apart.

There are two ways to read that gap. One view says the market still does not fully price Samsung’s HBM4 recovery and foundry optionality. The other says the market knows the HBM story, but still does not believe Samsung’s 2026-2027 earnings can persist beyond the peak. This article leans toward the second interpretation. The market is not blind to HBM. It is asking Samsung to prove that it is no longer just a memory-cycle stock.

That still leaves the central question open. Can Samsung deserve 15x PER? The answer is yes, but not yet as a base case. A more realistic first-stage re-rating is 8-10x. A stronger bull case is 12x. A 15x+ multiple belongs to the upper scenario after proof of external HPC design wins, foundry profitability, and down-cycle margin resilience.

Key Takeaways

The valuation gap between Samsung Electronics and TSMC is not just a simple undervaluation story. Samsung trades near 5x 2027E PER, while TSMC trades around 19-22x. The market treats TSMC as a structural compounder and Samsung as a complex IDM still exposed to memory cycles.

The 15x PER thesis rests on HBM4. HBM is moving away from commodity DRAM. Customer-specific specs, base die, packaging, power efficiency and controller integration matter more. Samsung is the only major IDM with memory, foundry, advanced packaging and system logic capabilities under one roof.

But Samsung’s 5x PER is not because the market has missed HBM. It reflects skepticism about earnings durability after 2027. In memory investing, low PER often means peak-cycle concern, not automatic cheapness.

The real difference between TSMC’s 20x+ multiple and Samsung’s 5x multiple comes down to four factors: business purity, margin stability, ROIC durability and customer neutrality. TSMC is a 100% pure-play foundry and neutral infrastructure provider. Samsung is a memory, foundry, mobile, TV and appliance conglomerate.

The realistic re-rating path has three stages. Stage one is 8-10x PER, driven by HBM4 share gains and reduced memory-cycle discount. Stage two is 12x, requiring foundry breakeven and down-cycle margin defense. Stage three is 15x+, requiring external HPC customers, 2nm/1.4nm credibility, and structural simplification of the non-semiconductor business.

The investment conclusion is simple. Betting on Samsung moving from 5x to 8-10x is reasonable. Treating 15x as today’s fair multiple is premature. The 15x case should be earned by evidence, not assumed in advance.

1. The Current Multiple Gap

Put Samsung and TSMC next to each other and the valuation gap is stark.

Metric	Samsung Electronics	TSMC	Interpretation
2026E PER	~6-7x	~21-23x	TSMC is valued more than 3x higher
2027E PER	~5x	~19-22x	Almost 4x gap
2027E PBR	~1.8x	~5-6x	Market trusts TSMC’s return on capital more
2027E EV/EBIT	~2-3x	~14-15x	Earnings durability gap
2027E FCF yield	~20%	~3-4%	Samsung’s cash flow is priced very cheaply
1Q26 operating margin	42.8%	58.1%	Samsung is a powerful cycle; TSMC is structural high margin

Samsung’s 1Q26 revenue was KRW 133.9 trillion, with operating profit of KRW 57.2 trillion. The DS division alone posted KRW 81.7 trillion in revenue and KRW 53.7 trillion in operating profit. AI memory has completely reshaped the earnings statement. TSMC also reported a 58.1% operating margin in 1Q26 and guided for 56.5-58.5% in 2Q26.

Given those numbers, it is tempting to argue that Samsung deserves a much higher multiple. Some investors now argue that after HBM4, Samsung is no longer a commodity memory company and should trade at 15x+ PER.

The argument is not absurd. But the multiple gap should not be explained only by saying “the market does not understand Samsung.” The market understands that Samsung is back in AI memory. What it doubts is different: whether these earnings survive into 2028 and 2029, whether memory margins remain resilient when pricing cools, whether foundry stops dragging consolidated earnings, and whether external customers trust Samsung with core AI silicon.

In other words, 5x PER is not ignorance. It is skepticism.

The Low-PER Trap

In many industries, 5x PER looks like deep value. In memory semiconductors, it can mean the market thinks earnings are near a cycle peak. At the top of a memory cycle, earnings surge and PER collapses. If the cycle rolls over, EPS falls and the same share price suddenly becomes 10x or 15x.

So in memory stocks, the question is not simply “how cheap is 5x?” The question is “is this peak EPS?” That is the heart of the Samsung debate. If the market has simply underappreciated HBM, the re-rating can be fast. If the market is asking for proof of post-2027 durability, the re-rating must be earned in stages.

2. Why the 15x Thesis Is Serious

The 15x PER case sounds aggressive, but the logic behind it is meaningful. The core point is that HBM is becoming a different kind of product from old DRAM.

2.1 HBM Is No Longer Simple Commodity DRAM

Traditional DRAM was close to a standardized product. Once it met JEDEC standards, multiple customers could use it. Samsung, SK Hynix and Micron products were often broadly substitutable. Pricing was heavily influenced by spot markets and quarterly negotiations.

HBM is different. HBM4 and later generations require tighter integration with customer architecture. Base die, package design, power efficiency, thermal behavior and controller integration all matter. The memory stack sits much closer to the GPU or ASIC design.

That means HBM is moving from commodity toward customer-specific infrastructure. Certification, long-term supply agreements and co-design matter more than spot pricing. If that shift continues, HBM should be less cyclical than old DRAM.

2.2 Samsung Has a Rare Integration Option

HBM value does not stop at the memory stack. It includes memory die, base die, advanced packaging, logic, memory controller and system optimization.

This is where Samsung’s strategic option becomes interesting. Samsung can build the HBM stack. It can produce base die through its own foundry. It has 2.5D and 3D packaging capability. It owns system LSI and foundry capacity. SK Hynix is stronger in HBM, but does not own a foundry. TSMC dominates logic foundry and advanced packaging, but does not make HBM memory. Samsung is the only major player that can, at least in theory, offer a more integrated memory-compute solution.

This is the strongest part of the 15x thesis. If the market starts viewing Samsung not as a memory supplier but as an AI memory-compute integration platform, the old memory-cycle multiple is too low.

But potential is not proof. The market needs actual external customer design wins, foundry margin improvement and evidence that HBM4E/HBM5 carry customer-specific pricing power.

2.3 AI Capex Slowdown Does Not Automatically Mean Memory Collapse

Another valid argument is that AI capex should not be reduced to a one-time training GPU cycle. The installed base of AI infrastructure will keep running inference. As inference moves into agents, retrieval, long context and tool use, memory footprint should grow across the data center.

More deployed models mean more server DRAM, more HBM, more SOCAMM and more high-performance storage. If this is right, Samsung’s 2027 earnings may be the beginning of a high plateau rather than a one-year peak.

That would make 5x PER too low. But it still does not make 15x PER automatic. The balanced view is neither “AI bubble collapse” nor “straight-line AI capex forever.” It is a high plateau with volatility.

3. The Real Gap With TSMC

For Samsung to deserve 15x, it must narrow the structural gap with TSMC. That gap is not just technology. It is business mix, margin stability, capital returns and customer trust.

3.1 Business Purity

TSMC is a 100% pure-play foundry. Customers see it as neutral manufacturing infrastructure. Investors can analyze one business model: wafer demand, process leadership, utilization, capex and margins.

Samsung is different. It combines memory, foundry, system LSI, mobile, TV and appliances. In 1Q26, DS generated almost all operating profit, but the consolidated company still includes lower-margin device businesses. That creates a conglomerate discount. Even if Samsung’s HBM business deserves 15x, consolidated Samsung does not automatically deserve 15x.

To reduce this discount, the market would need clearer capital allocation, stronger shareholder returns, or structural signals around the non-semiconductor businesses.

3.2 Margin Stability

TSMC’s 1Q26 operating margin was 58.1%, and its 2Q26 guidance is 56.5-58.5%. The key point is not just the level. It is the perceived structural nature of that margin.

Samsung’s 1Q26 consolidated operating margin was also strong at 42.8%, and DS margins were exceptional. But those numbers include not only HBM strength, but also broad DRAM pricing, supply tightness and operating leverage. Whether those margins survive a cooling memory market is not yet proven.

If HBM truly reduces Samsung’s cycle volatility, the next slowdown should show a higher margin floor than past cycles. That is what the market needs to see.

3.3 ROIC and ROE Durability

Quality multiples are determined by bad-year returns, not good-year returns. TSMC has sustained high ROIC and ROE across cycles. That is why the market is willing to pay a high multiple.

Samsung’s ROE surges in recoveries but falls sharply in memory downturns. To earn a quality-compounder multiple, Samsung must show that its OPM and ROE floor has structurally risen. This cannot be proven in one quarter. It usually takes a full cycle.

3.4 Customer Neutrality

TSMC’s neutrality is one of its most valuable assets. Apple, NVIDIA, AMD and Broadcom do not compete with TSMC. They can place their most sensitive designs with TSMC because TSMC is a manufacturing platform.

Samsung has a more complicated relationship with customers. It competes in mobile devices, semiconductors, displays and other categories. Some customers may be more cautious about giving Samsung their most critical AI designs.

That is why external HPC customer wins matter so much. Internal use and group-level integration are not enough. The market needs to see major external AI ASIC or HPC customers choose Samsung’s foundry and packaging.

4. A Realistic Re-Rating Path

Samsung’s multiple is unlikely to move from 5x to 15x in one jump. The market will pay for evidence in stages.

4.1 Stage One: 5x to 8-10x

This is the most realistic stage. Samsung needs to show that it has regained meaningful HBM4 share and that the old memory-cycle discount is too harsh.

The key evidence is HBM4 revenue mix, HBM4E sample timing, customer allocation, narrowing share gap versus SK Hynix, and upward revisions to 2027 operating profit estimates.

This stage does not require Samsung to become TSMC. It only requires the market to admit that Samsung is no longer a pure old-cycle DRAM stock. The appropriate range is therefore 8-10x, not 15x.

The likely timeline is 2026-2027.

4.2 Stage Two: 8-10x to 12x

To reach 12x, Samsung needs stronger proof. HBM growth alone is not enough. Margins must hold during a memory slowdown. Foundry must approach breakeven or profitability. HBM4E/HBM5 must show higher customer-specific content. Base die and packaging integration must appear in reported earnings.

External HPC customers are also crucial. If a customer such as AMD, Broadcom or a major hyperscaler meaningfully chooses Samsung for AI ASIC, foundry or packaging, the market’s view changes.

This stage is more likely in 2027-2028. At that point, Samsung would be viewed as an IDM with reduced cycle exposure.

4.3 Stage Three: 15x+

The 15x case is the upper scenario. It requires multiple conditions.

Samsung would need several external HPC customers, credible 2nm and 1.4nm execution, meaningful foundry margin improvement, lower conglomerate discount from the DX business, and repeatable HBM + base die + packaging integration.

If those conditions are met, Samsung can be reclassified from memory company to AI integration platform. At that point, 15x+ becomes discussable. But this is a 2028+ upper case, not today’s base case.

Scenario Probabilities

Scenario	PER Range	1-3 Year Probability	Key Condition
Bear	5-6x	20%	HBM peak-earnings concern persists; foundry drag remains
Base	8-10x	50%	HBM4 share confirmed; memory-cycle discount eases
Bull	12x	25%	Down-cycle margin defense, foundry breakeven, customer-specific HBM
Extreme Bull	15x+	5%	External HPC customers, DX restructuring, repeatable integration platform

The most reasonable base case is 8-10x. A 15x outcome is possible, but it is a long shot. The better investment frame is not “15x is fair now,” but “5x is too low and 8-10x is achievable.”

5. The Traps in the 15x Bet

5.1 The “Market Does Not Know” Trap

The most common mistake is assuming the market has missed Samsung’s HBM value. Global investors and sell-side analysts track HBM4, HBM4E, base die, packaging and customer share in detail. The market is not unaware. It is skeptical.

That skepticism is rational. The market still needs proof that 2027 earnings are sustainable, margins hold in a down-cycle, foundry losses shrink, and external customers trust Samsung with critical AI silicon.

The key to re-rating is not a new narrative. It is repeated evidence.

5.2 The “Samsung Equals TSMC” Trap

Samsung’s integration option is powerful, but it does not automatically justify TSMC-like multiples. TSMC’s 20x+ PER is not just about AI exposure. It reflects purity, neutrality, margin stability, customer trust and decades of execution credibility.

The realistic peer frame for Samsung sits somewhere above traditional memory and below TSMC. Better than a pure memory-cycle multiple, but still below the best pure-play foundry multiple. That points to 8-12x. 15x is the upper edge of a special scenario.

5.3 The “AI Capex Is Forever” Trap

AI infrastructure is likely a long cycle. But it is not a guaranteed straight line. Training GPU demand can slow. HBM supply can catch up. Pricing premiums can compress as 2027-2028 capacity arrives.

The balanced view is not collapse and not infinity. It is a high plateau with volatility. In that world, 5x is too low, but 15x today is still hard to justify. The reasonable middle is 8-12x.

6. The Next Six-Month Checklist

Samsung’s re-rating should be verified with data, not just a story. The next six months matter.

HBM

Checkpoint	Why It Matters
HBM4 revenue mix	Confirms HBM recovery is flowing into earnings
HBM4E sample timing	Tests roadmap credibility
Major customer allocation	Shows quality of certification and repeat demand
Share gap versus SK Hynix	Measures whether Samsung is catching up
Custom HBM roadmap	Evidence that HBM is becoming customer-specific

Foundry

Checkpoint	Why It Matters
2nm GAA yield improvement	Basic condition for external customer trust
2nm external design win	Key condition for a 12x+ multiple
HBM4 base-die supply	Converts integration thesis into earnings
Narrowing foundry losses	Reduces consolidated drag
1.4nm roadmap	Long-term technology credibility

Memory Cycle

Checkpoint	Why It Matters
Commodity DRAM pricing	Tests how cyclical current DS margins are
NAND price recovery	Shows breadth of DS earnings
Server DRAM and SOCAMM demand	Tracks inference-driven memory footprint
DS operating margin	Shows margin power before the next slowdown

Structural Change

Checkpoint	Why It Matters
Larger shareholder returns	Reduces conglomerate discount
DX restructuring signal	Precondition for the 15x upper case
Labor agreement and bonus structure	Improves predictability of excess-profit allocation
Capex discipline	Protects ROIC after the AI boom

7. Macro Context

The 15x PER debate is not only company-specific. US and Japanese long yields are high. Global discount rates matter. Even high-quality companies struggle to expand multiples when long-duration discount rates rise.

So Samsung’s first-stage re-rating also needs a macro gate. US 10-year yields need to stabilize. Brent crude should stop pressuring inflation expectations. The won should avoid disorderly weakness. VIX should move lower. Only then can foreign investors pay a higher multiple for Korean semiconductor earnings.

If macro stabilizes and HBM4 share is confirmed, 8-10x becomes realistic. If macro stress persists and memory pricing cools at the same time, the 5x box can last longer. A stock’s multiple is the product of company quality and market discount rate. Both matter.

Macro Environment	Company Evidence	Reasonable PER Range
Stable	HBM4 share confirmed	8-10x
Stable	Foundry breakeven and down-cycle defense	11-12x
Stressed	HBM strong but discount rates high	6-8x
Stressed	Memory slowdown as well	4-5x
Stable	External HPC wins and structural simplification	14-15x

8. Links to the Existing Series

Part 1 of this series discussed Citi’s KRW 460,000 target price and the argument that the old memory-cycle frame is wrong. That connects directly to the stage-one re-rating case in this article. If Samsung is no longer just an old DRAM cycle stock, 8-10x PER becomes possible.

Part 2 discussed the strike and bonus debate. That was really about who captures excess profit. In this article, that question becomes an ROIC issue. Whether excess profit is allocated to shareholders, employees, capex or customer price cuts will shape Samsung’s long-term multiple.

The AI HBM hub makes the same point at the market level. Korea’s re-rating starts with HBM and AI memory, but individual stock multiples depend on whether each company proves structural earnings.

The macro pieces also matter. If long-yield stress continues, the 15x debate moves further out. If rates stabilize and AI memory earnings hold, the market may revisit Samsung’s low multiple.

Final View

Samsung Electronics at ~5x 2027E PER versus TSMC at 19-22x is a striking gap. The instinct to argue for Samsung at 15x is understandable. The direction is right. HBM is becoming customer-specific, and Samsung has a rare combination of HBM, base die, foundry, packaging and controller capabilities.

But 15x should not be treated as today’s fair value. Samsung’s 5x PER does not mean the market has missed HBM. It means the market doubts post-2027 earnings durability. To remove that doubt, Samsung must prove down-cycle margin resilience, foundry breakeven and external HPC customer wins.

The realistic path has three stages. Stage one is 8-10x and is the most plausible. Stage two is 12x and requires foundry and margin proof. Stage three is 15x+ and requires external HPC wins plus structural simplification.

The better starting point is not “15x is the right multiple.” It is “5x is too low.” That distinction matters. A 15x case should be added as evidence accumulates, not priced in upfront.

Sources

This article is for research and commentary only and is not investment advice. Samsung Electronics 2026E/2027E PER, TSMC 2027E PER, HBM4 share assumptions and customer-allocation assumptions are analyst estimates based on public information and market consensus. Actual earnings and share prices may differ materially depending on memory pricing, AI capex, foundry yield, customer certification, FX, interest rates and policy risk. Scenario PER ranges and probabilities are subjective estimates, not guarantees. The analysis may be wrong. Data as of May 16, 2026 KST.

US and Japan Long Yields Are Rising Together: Five Stabilization Triggers and Three Scenarios

Sat, 16 May 2026 00:00:00 +0000

📚 Macro Cycle Series

Previous: Why Everything Moves at Once — Iran, Oil, US Inflation, China, and Japan as a Single Cycle

Japan axis analysis: Japan’s PPI Shock — Not the Fear of Selling Treasuries, But the Fear of Buying No More

Korea market implications: What KOSPI’s -6.12% Crash Really Means — Why You Must Clear the Macro Gate Before Hunting Relative-Strength Stocks

The previous piece showed how Iran and the Strait of Hormuz, oil prices, US inflation, Japan’s inflation, the BOJ, US long-term yields, and the global discount rate are all bound together in a single cycle. This piece addresses the next question: when and how can this cycle stabilize?

The US 10-year yield has been approaching 4.6%, and the 30-year has crossed 5%. Japan’s 10-year yield is at its highest since 1997. A simultaneous spike in long-term yields across both countries is not a common event. One can read this as a simple bond-market seizure, or as the beginning of something more structural.

My judgment lies in between. A short-term stabilization is possible. But this is not a market that unwinds on its own. What matters is how many of the following converge simultaneously: Hormuz normalization, oil prices, US inflation, BOJ guidance, and demand at US Treasury auctions.

Key Summary

Current state: US 10-year at 4.46%, 30-year at 5.02%, Japan 10-year at 2.55% (highest since 1997), Japan 30-year near 4.0%.
Decomposing the drivers: This is not simply a product of hawkish Fed commentary. Expected short-term rates, inflation premiums, term premiums, and bond supply premiums have all moved higher simultaneously.
Stabilization potential: Possible, but conditional. The structure does not unwind automatically over time.
Five core triggers: Hormuz normalization and falling oil prices; containment of energy pass-through into US core CPI/PPI; BOJ June meeting guidance; recovery in demand at 10-year and 30-year US Treasury auctions; flight-to-safety flows triggered by risk-asset corrections.
1-to-3-month scenarios: Fast stabilization 20–25%, range-bound consolidation 45–50%, renewed seizure 25–30%.
Single most important variable: Hormuz transit normalization. This one variable touches US inflation and Japan import prices simultaneously.
Investment judgment: It is premature to place a large bet on stabilization in advance. A staged entry after two to three of the five triggers are confirmed is more defensible.

1. Where We Stand Now

Between May 14 and 15, long-term yields moved higher again. The US 10-year yield climbed from 4.46% toward 4.60%, and the 30-year surged from 5.02% to as high as 5.13%. The May 30-year auction saw weak demand, with the stop-through rate coming in higher than market expectations and the tail widening.

Japan followed the same path. The 10-year JGB yield rose into the 2.55%–2.72% range, touching the highest level since 1997. The 30-year JGB approached 4.0%, and the 40-year had already traded as high as 4.24% back in January.

This is not a US-only problem. Long-end yields in Japan, Europe, and the US are all moving together. The global discount rate for risk assets is rising in unison — which explains why Korean equities, US mega-cap tech, KOSDAQ growth names, and long-duration bond ETFs are all under simultaneous pressure.

Why This Is Not a Simple Fed Hawkishness Event

Long-term yields are best understood as the sum of four components.

Component	Current direction	Implication
Expected short-rate path	Higher	Fed rate-cut expectations have faded; BOJ hike probability has risen
Inflation premium	Higher	US CPI, US PPI, and Japan PPI have all re-accelerated simultaneously
Term premium	Higher	The compensation demanded for holding long-duration paper has increased
Bond supply premium	Higher	Weak US auction demand and diminished Japanese marginal buying of Treasuries are compounding

In the US, virtually all Fed rate-cut expectations have evaporated, and markets have even begun pricing a December rate hike. In Japan, the probability of a BOJ June hike has risen above 70%. US April CPI came in at +3.8% year-over-year; Japan April PPI at +4.9%. The NY Fed ACM model puts the US term premium at 0.68% — still below the historical median of 1.47%, but the direction is toward normalization.

Supply-demand dynamics are equally unfavorable. The US must absorb roughly $2 trillion in annual fiscal deficits alongside large Treasury maturities. As BOJ normalization pushes JGB yields higher, the hedged return on US Treasuries for Japanese insurers and pension funds has declined. China, too, is no longer a consistent large buyer of Treasuries.

For yields to stabilize, improving one factor is not sufficient. If oil prices fall but the BOJ remains hawkish, Japan-sourced pressure persists. If the Fed turns more accommodative but auctions keep coming in weak, the long end of the US curve will continue to struggle.

Why the US and Japan Are Moving Together

The first linkage is a common shock. The Hormuz bottleneck and rising oil prices push US CPI and Japan import prices higher at the same time — through gasoline and energy services in the US, and through imported raw materials and corporate goods prices in Japan. This channel is the strongest.

The second linkage is Japanese capital flows. When JGB yields rise, the hedge-adjusted yield on US Treasuries becomes less attractive to Japanese insurers and pension funds. They do not need to sell Treasuries outright. Even a reduction in new purchases weakens the marginal demand for US Treasuries and pushes long-term yields higher.

The third linkage is a global re-pricing of sovereign bond risk. Watching a country like Japan — with an extremely high government debt load — see its 10-year yield move from 2% to 3% territory raises questions about appropriate term premiums for other developed-market sovereigns: the US, Canada, Italy. Markets have started asking those questions again.

Of these three channels, the first — Hormuz and oil prices — is the most powerful. Which is why the central question in this piece ultimately returns to whether Hormuz normalizes.

2. Five Core Triggers

2.1 Hormuz Normalization and Lower Oil Prices

The most powerful trigger is Hormuz normalization. The reason is straightforward: oil prices affect US and Japanese inflation simultaneously.

A substantial portion of the US April CPI print of +3.8% came from energy. Gasoline prices and energy services pushed the headline directly higher. Japan’s exposure is even more direct. The backbone of Japan’s April PPI of +4.9% was import prices rising +17.5%, dominated by energy and raw materials.

If Hormuz transit normalizes and Brent crude stabilizes below $95 for at least one week, the probability that US and Japanese yields ease simultaneously rises meaningfully. Conversely, if Brent re-breaks above $110 and vessel attacks or seizures recur, the long-yield shock can re-intensify.

Category	Stabilization signal	Deterioration signal
Oil price	Brent stable below $95	Re-breaks $110
Hormuz	Vessel traffic recovers to 70%+ of normal	Additional attacks or seizures
Inventories	IEA/EIA draw rate decelerates	Continued draws above 5 mb/day average
War risk insurance	Rates normalize	Vessel war-risk premiums remain at 2–3%
Negotiations	Iran–US talks show progress	Reported collapse of talks

I assess the probability of near-term full normalization at 25–30%. Partial normalization is the most likely outcome at 45–50%, and prolonged disruption sits at 20–25%. In other words, the most realistic picture is not Brent falling quickly into the low $90s, but rather a range-bound environment where further sharp spikes are capped.

2.2 Containment of US CPI/PPI Pass-Through into Core Inflation

The second trigger is US inflation. More important than whether energy prices are high in isolation is whether they pass through into core inflation.

The April headline CPI was +3.8%, with a significant energy contribution. But core CPI also came in at +2.8%, or +0.4% month-over-month. If that monthly pace persists, the annualized rate approaches 5%. In that scenario, even if oil stabilizes, inflationary pressure already transmitted into services, transportation, and shelter will linger.

In the next CPI and PPI releases, what matters more than the headline is core. If core CPI steps down to +0.2%–+0.3% month-over-month and PPI transportation and services decelerate, markets can breathe. Conversely, if core CPI prints +0.4% again and PPI services rise further, Fed rate-cut expectations recede even further.

The key dates: June 11 (US May CPI), June 12 (US May PPI), and June 27 (US May PCE). One deceleration is a signal; two consecutive prints confirm a trend.

2.3 BOJ June Meeting Guidance

The third trigger is the BOJ. Markets already have a substantial probability of a 25bp hike — from 0.75% to 1.00% — at the June meeting priced in. The hike itself is therefore not the critical variable. The critical variable is the tone following the hike.

If the BOJ hikes but characterizes the path as “gradual” and “data-dependent,” markets can take comfort. Even if Governor Ueda leaves the door open to further tightening, a low-implied pace of subsequent hikes would allow both JGBs and US Treasuries to stabilize.

Conversely, if the BOJ strongly validates the path to consecutive hikes, the shock will not remain contained within Japan. JGB yields would push higher, the incentive for Japanese investors to add new positions in US Treasuries would weaken further, and the term premium embedded in the US 10-year would face additional upward pressure.

BOJ message	Market interpretation	Yield impact
Gradual, data-dependent	One hike then a pause	Stabilizing
Justifies further hikes	Consecutive hike path validated	Worsening
Emphasizes yen stability	Reduces import-price pressure expectations	Stabilizing
Emphasizes upside inflation risks	Triggers additional JGB long-end selling	Worsening

A single sentence from the BOJ June meeting can move the US 10-year yield by roughly 10bp. We are in a phase where Japan’s policy dynamics are directly linked to the US Treasury market.

2.4 Recovery in US 10-Year and 30-Year Auction Demand

The fourth trigger is US Treasury auction results. The supply-demand dimension of the long-yield shock is most transparently revealed in auction data.

The May 30-year auction was weak. Bid-to-cover was soft and the tail widened — meaning the clearing yield was higher than the market had anticipated. One weak auction can be treated as an isolated event, but a pattern of weakness reads as a structural demand shortfall.

If the next 10-year and 30-year auctions show a recovering bid-to-cover, a tighter tail, and indirect bidder participation returning above 65%, long-term yields can find a footing. If weak auctions recur, the 30-year sustaining above 5.2% becomes a real possibility.

Key dates to watch: June 10-year and 30-year auctions, and the quarterly Treasury Refunding Announcement on July 30. If the Treasury signals an increase in long-end issuance, the supply burden on long-dated yields grows heavier.

2.5 Risk-Asset Correction Triggering Flight to Safety

The fifth trigger is a paradoxical stabilization path. If long-term yields rise too rapidly, risk assets eventually break. Mortgage rates climb, corporate credit spreads widen, equity multiples compress, and consumption slows. At that point, markets begin repricing growth deceleration and the possibility of eventual Fed easing.

In this case, long-term yields can come down. But it is not a good stabilization. It is a bad equilibrium in which bonds rally only because equities and real estate have sold off.

The stabilization signals here are: an S&P 500 correction of 5–10%, VIX above 25, high-yield spreads widening by at least 50bp, softening consumption and employment data, and an increasing number of Fed officials citing growth concerns. I put the probability of this pathway at 20–30%. It is not, however, a scenario any investor should welcome.

3. One-to-Three-Month Scenarios

The most probable scenario at present is not a fast stabilization but a range-bound consolidation.

Scenario	Probability	Conditions	US 10-Year	Japan 10-Year
A. Fast stabilization	20–25%	Hormuz normalizes, Brent below $95, CPI/PPI decelerate	4.2–4.4%	2.3–2.5%
B. Range-bound consolidation	45–50%	Oil stays elevated but further spikes are capped, BOJ hikes gradually	4.4–4.7%	2.5–2.8%
C. Renewed seizure	25–30%	Brent at $110–$120+, inventory draws persist, BOJ and Fed both turn more hawkish	4.8–5.0%+	Approaching 3.0%

Scenario A requires Hormuz normalization and US disinflation to materialize simultaneously — possible but not high-probability. Scenario C requires oil prices, auction results, and the BOJ to all deteriorate together. The most likely outcome is the middle path: oil remains elevated without spiking further, the US 10-year oscillates between 4.4% and 4.7%, and the Japan 10-year moves in the 2.5%–2.8% range.

Asset-Class Implications

Asset	Fast stabilization	Range-bound	Renewed seizure
US equities	Bullish	Volatility expands	Correction
US mega-cap tech	Bullish	Sideways	Significant correction
Korean KOSPI	Recovery	Sector divergence	Foreign selling
Korean growth stocks	Recovery	Weakness	Significant correction
Korean financials	Sideways	Relative strength	Relative strength
US long-duration bonds	Bullish	Bearish	Significantly bearish
US dollar	Weakness	Sideways to strength	Strength
Gold	Partial profit-taking	Bullish	Bullish
Oil refining/LNG	Weakness	Bullish	Significantly bullish
Defense	Weakness	Bullish	Significantly bullish

4. The Traps in Betting on Stabilization

The first trap is fixating on Fed communication alone. In a conventional market environment, when the Fed turns more dovish, yields fall. But this is not a market driven solely by Fed policy expectations. Inflation premiums, term premiums, bond supply dynamics, and the BOJ are all moving simultaneously. Even if Fed communication improves, the other factors can undercut any rally.

The second trap is treating a Hormuz agreement announcement as the definitive end of the episode. If an agreement is announced, Brent crude and long-term yields could react sharply on the day. But restoring actual vessel traffic, rebuilding inventories, normalizing refined product prices, and eventually seeing core CPI decelerate — all of that takes time. The announcement-day reaction is the first response; the 30-to-60-day verification period determines the true direction.

The third trap is reading this yield surge as nothing more than a short-term convulsion. The short-term geopolitical trigger is real. But simultaneously present are structural factors: US fiscal deficits, term premium normalization, BOJ normalization, and the weakening of Japanese marginal demand for US Treasuries. A return of the 10-year below 4% the way it used to trade is therefore a low-probability scenario. The more realistic target is stabilization in a 4.0%–4.5% range.

5. Practical Monitoring Checklist

The first thing to watch is Brent crude. Brent stable below $95 for at least one week is a stabilization signal. A re-break above $110 is a deterioration signal. Second, watch US core CPI and PPI: month-over-month needs to step down to +0.2%–+0.3%. Third, watch the precise language in the BOJ June statement. “Gradual” and “data-dependent” language is reassuring; any phrasing that validates consecutive hikes is a red flag.

Next, watch the US Treasury auctions. The question is whether demand recovers in the 10-year and 30-year. Finally, watch risk assets themselves. A sharp equity selloff can paradoxically drive a flight to long-duration bonds — but that is not a good scenario.

Priority	Indicator	Stabilization threshold	Deterioration threshold
1	Brent crude	Stable below $95 for 1+ week	Re-breaks $110
2	IEA/EIA weekly inventories	Draw rate decelerates	Draws sustained above 5 mb/day
3	US core CPI/PPI	MoM +0.2–0.3%	+0.4% repeated
4	Fed hike probability	December hike probability below 20%	Above 50%
5	BOJ June guidance	Gradual, data-dependent	Consecutive hikes justified
6	Japan import prices/PPI	Deceleration	Additional surge
7	US 10-year/30-year auctions	Bid-to-cover recovers, tail narrows	Weak auctions repeated
8	USD/JPY	Stable below 155	Re-breaks 160
9	US 30-year yield	Recovers below 5.0%	Settles above 5.2%
10	Japan 30-year yield	Below 3.8%	Settles above 4.0%

Entry Conditions for a Stabilization Bet

Staged entry can be considered when at least three of the following five conditions are met:

Brent crude holds below $95 for at least one week.
The next CPI/PPI release confirms core inflation deceleration.
The BOJ June meeting delivers gradual, data-dependent tightening guidance.
Demand recovers at US 10-year and 30-year Treasury auctions.
The US 10-year fails to sustain above 4.6% and begins to turn lower.

Three or more: first tranche entry is reasonable. Two or fewer: stay in cash. Zero to one: a defensive posture or positioning for further weakness is more rational than betting on stabilization.

6. Implications for the Korean Market

In a fast-stabilization scenario, semiconductors and AI hardware, KOSDAQ growth names, and platform companies are likely to react first. When discount rates fall, long-duration assets are the fastest to rebound.

In the range-bound scenario, the picture is different. Rates do not fall sharply, but they do not worsen either. In this environment, semiconductor components, substrates, test equipment, shipbuilding/LNG, defense, and value/financial stocks hold up better on a relative basis. Sectors with earnings momentum can weather the macro overhang even if it does not fully clear.

In a renewed-seizure scenario, oil refining/LNG, defense, short-duration bonds, and cash equivalents are the defensive plays. KOSDAQ growth stocks and long-duration assets are best avoided.

Scenario	Favorable areas in Korea	Areas to avoid
Fast stabilization	Semiconductors, AI hardware, KOSDAQ growth, platforms	Oil/defense — some profit-taking
Range-bound	Semiconductor components/substrates/test, shipbuilding, defense, financials	Growth stocks without earnings
Renewed seizure	Oil refining, LNG, defense, cash equivalents	KOSDAQ long-duration assets

7. Frequently Asked Questions

Won’t everything resolve once the Fed eventually cuts?

It is not that simple. Even if the Fed cuts short-term rates, long-term yields are shaped by term premiums, bond supply dynamics, foreign buyer behavior, and fiscal deficits. In 2024, there were extended periods when the US 10-year refused to fall materially despite active Fed rate-cut expectations. Monitoring only Fed communication is insufficient this cycle as well.

If Japan dumps Treasuries, doesn’t that cause a collapse?

What matters more than outright selling is the deterioration of marginal demand. Japanese insurers and pension funds do not need to actively sell US Treasuries. Simply reducing new purchases weakens the marginal demand balance in the Treasury market. That is the quieter but more structurally significant risk.

Isn’t this a buying opportunity in long-duration bonds?

With yields elevated, the expected return on long-duration bonds has improved. But entry timing still matters. If the US 10-year climbs further from 4.6% to 4.8%, long-bond prices will experience additional drawdowns. Full allocation is therefore inferior to trigger-based staged entry: consider a first tranche when three of the five triggers are met, a second tranche at four, and complete the allocation if all five confirm.

8. One Final Thought

US 10-year at 4.46%, 30-year at 5.02%, Japan 10-year at the highest level since 1997. This is not a simple Fed hawkishness event. It is the result of expected short-term rates, inflation premiums, term premiums, and bond supply premiums all moving higher simultaneously. Improvement in a single factor will not produce stabilization.

The single most powerful variable is Hormuz and oil prices. They are the common shock that simultaneously determines US inflation and Japan’s import prices. The negotiation outcome in late May to early June is the first potential inflection point.

At this juncture, scenario probabilities stand as follows: fast stabilization 20–25%, range-bound consolidation 45–50%, renewed seizure 25–30%. The most probable outcome is the middle path. Rather than betting on a return of the 10-year below 4%, the more realistic target is stabilization in a 4.0%–4.5% range.

It is too early to bet heavily on stabilization. Staged entry after at least three of the five triggers — Hormuz, CPI/PPI, BOJ guidance, US Treasury auctions, risk-asset correction — are confirmed is the more defensible approach. The most dangerous assumption is that time alone will resolve this cycle. This episode contains not only a short-term geopolitical convulsion but also structural elements: fiscal dominance and term premium normalization.

This may not be a market returning to low rates, but rather one in which — at a higher discount rate — the gap between strong assets and weak assets widens. In that environment, the true alpha does not come from simply betting that rates fall. It comes from identifying assets with the earnings and cash flows to hold their ground even when rates stubbornly refuse to decline.

This post is for research and commentary purposes only and does not constitute investment advice. US April CPI/PPI figures are based on official BLS releases. Japan April PPI is based on the BOJ Monthly Report. US 10-year/30-year yields and JGB 10-year/30-year yields reflect market data from May 14–15. Scenario probabilities (fast stabilization 20–25%, range-bound 45–50%, renewed seizure 25–30%) represent the analyst’s subjective estimates and may differ materially from actual market outcomes. Term premium estimates (US ACM 0.68%, historical median 1.47%) are based on NY Fed publications. CME FedWatch December 25bp hike probability (36%) is as of May 15 and changes daily. BOJ June hike probability of 77% reflects market consensus per Reuters and Bloomberg. The timing of Hormuz normalization, BOJ guidance, and US CPI deceleration remain uncertain. The global macro environment can shift rapidly in response to unexpected events, including additional geopolitical conflicts or policy changes. Data reference date: May 16, 2026 KST.

KOSPI Dips as Samsung Faces Selling Wave

Fri, 15 May 2026 23:30:00 +0900

KOSPI Holds Ground, But Selling Pressure Beneath the Surface

South Korea’s benchmark equity index, KOSPI — comprising roughly 800 listed companies and widely tracked by global emerging-market funds — closed at 7,981 on May 15, 2026, up 6.5% over five sessions. The headline number looked constructive. What lay underneath did not.

A market-wide discovery screener held a neutral reading of 65/100, and institutional follow-through indicators remained intact at Day 20. By that measure, the session was not a breakdown. But across Korean semiconductor and gaming names with high analyst attention, prices dropped sharply. The KOSDAQ, South Korea’s tech-heavy secondary exchange, slipped 1.4% over the same five-day window, underperforming its larger sibling meaningfully.

The day’s defining story was a single flow number: foreign investors sold a net ₩2.5 trillion (approximately $1.8 billion) of Samsung Electronics (005930.KS), South Korea’s largest company by market capitalization and the world’s top producer of DRAM and NAND flash memory. That is not a noise-level print — it is a positioning event, and it cascaded through sentiment across the board.

Samsung Electronics: Quality Thesis Intact, Timing Is Not

Samsung Electronics fell 8.61% on the day, even as the five-day return stayed marginally positive at +0.74%. The stock was still trading above its 20-day moving average entering the session, but the scale of the single-day foreign sell-off — ₩2.5 trillion in net selling alongside institutional outflows of ₩421 billion — overwhelmed that technical cushion.

Why does this matter beyond one bad day? Samsung sits at the center of two of the biggest structural themes in Korean equities right now: AI infrastructure build-out and the memory cycle recovery. On both counts, the fundamental backdrop strengthened this week. DRAM pricing is expected to rise more than 50% quarter-over-quarter, and NAND is tracking toward a 75% quarterly gain. Cisco Systems separately raised its AI infrastructure revenue target from $5 billion to $9 billion, with networking revenue already up 24.7% year-over-year — a data point that validates continued hyperscaler capex momentum flowing into memory and logic.

The disconnect, then, is between thesis and timing. The thesis — that Samsung is a primary beneficiary of AI-driven memory demand — did not weaken on May 15. The price action and foreign flow, however, issued a clear near-term caution signal. Analysts and fund managers monitoring Korean semiconductor exposure will want to see foreign selling decelerate meaningfully before treating this as a buying opportunity rather than a warning.

Samsung Electro-Mechanics: The Standout Amid the Selloff

Not everything in the Korean tech supply chain moved lower. Samsung Electro-Mechanics (009150.KS), a leading manufacturer of multilayer ceramic capacitors (MLCCs) and camera modules critical to AI servers and smartphones, was the clearest outperformer by flow quality on the day.

The stock declined just 1.37% — shallow relative to its sector peers — while posting a five-day gain of 10.5%. More importantly, institutional investors were net buyers at ₩21.7 billion, even as foreign investors trimmed ₩33.5 billion. Relative strength metrics placed Samsung Electro-Mechanics in the top percentile of the Korean market. RSI reached 84, signaling short-term heat, but the underlying demand profile from domestic institutions remained supportive.

The implication for sector observers: within Korea’s component supply chain, MLCC and substrate-adjacent names are absorbing more demand-side conviction than large-cap memory players at this moment in the cycle. Whether that rotation persists depends partly on whether AI server build rates stay elevated through Q3 2026.

Pearl Abyss and the Gaming Sector Divergence

Pearl Abyss (263750.KS), the Korean game developer behind Black Desert Online and the highly anticipated Crimson Desert, fell 6.72% on May 15 — extending a five-day decline to 10.1%. The RSI dropped to 29, a technically oversold reading, and the stock was trading 14% below its 20-day moving average.

Foreign investors were net buyers at ₩9.7 billion, a contrarian signal worth noting. But institutional investors, who typically provide more sustained directional conviction in Korean mid-caps, were essentially flat at -₩200 million. The absence of institutional buying into oversold territory is a soft negative: it suggests professionals are not yet treating the recent drawdown as a re-entry opportunity.

For international investors tracking Korean game publishers — a sector that tends to benefit from strong IP cycles and growing overseas monetization — Pearl Abyss remains a name to watch, but the technical and flow setup argues for confirmation before new positioning. The next meaningful test is price stabilization around the ₩47,200 level.

New Entrants to Watch: Fadu and the AI Storage Theme

Among names not yet widely followed by international investors, Fadu (440110.KS) — a Korean fabless semiconductor company designing enterprise SSD controllers and CXL (Compute Express Link) memory interface chips — posted one of the cleanest flow profiles in the day’s market scan.

Relative strength ranked at 98.3 out of 100. Foreign investor flows were positive. Both quality-screened institutional and financial investment firm flows aligned on the buy side — a combination that rarely clusters. The stock pulled back 3.3% on the day, which paradoxically improves the setup: the pullback reduces momentum-chasing risk and creates a more orderly entry zone if the next session confirms a bounce above the ₩100,100 support level with volume.

Fadu’s CXL controller design work is directly relevant to next-generation AI server memory architectures, where bandwidth and latency constraints are pushing hyperscalers toward pooled memory solutions. It is a smaller-cap name with corresponding liquidity risk, but the flow quality on a weak market day stood out.

Daedeok Electronics (353200.KS), a PCB substrate manufacturer supplying into AI server and memory packaging, also retained a high relative strength rank of 98.2. Foreign investors bought a net ₩3.8 billion, but domestic institutions sold ₩7.5 billion — a flow divergence that warrants monitoring before drawing directional conclusions.

The Broader Setup: Defensive Rotation or Temporary Noise?

The May 15 session illustrated a dynamic that Korean equity investors encounter periodically: macro and theme catalysts point one direction while near-term price and flow point another.

AI infrastructure spending momentum is accelerating — Cisco’s guidance revision, Cerebras Systems’ IPO pipeline, and continued NVIDIA outperformance all reinforce that. Korean names in the AI supply chain — memory, substrates, MLCCs, CXL controllers — retain credible thesis support. June biotech conference season also adds a near-term catalyst layer for select CDMO and cell therapy names.

But foreign investor flows in large-cap Korean tech, particularly the ₩2.5 trillion Samsung print, create a near-term headwind that is difficult to dismiss. When foreign selling at that scale occurs without an obvious company-specific negative catalyst, it typically reflects global fund rebalancing or risk reduction — and it can take several sessions to fully clear.

The rotation showing up within Korean equities — from large-cap memory toward components, domestic consumption, and robotics names — is worth tracking. LG Electronics (066570.KS) and Doosan Robotics (454910.KS) appeared in high-volume screener lists alongside consumer internet names, suggesting sector rotation is already underway at the institutional level.

For international investors with Korean equity exposure, May 16 is a session to watch for whether foreign selling in Samsung moderates, whether oversold names like Pearl Abyss attract institutional support, and whether high-relative-strength names like Samsung Electro-Mechanics and Daedeok Electronics consolidate constructively or give back gains. The answers to those questions will clarify whether May 15 was a distribution day or simply a turbulent session within an intact uptrend.

Samyang Foods 1Q26: ₩714.4bn Revenue / 24.8% OPM — Buldak Pulled the Entire Korean Consumer Complex. Why F&B, Department Stores, and Cosmetics All Moved Together

Fri, 15 May 2026 20:30:00 +0900

🔗 Related: Why Korea Part 2 — How Korean Cosmetics Became a Top-Three Export Industry · APR / Medicube Deep Dive · Olive Young / PharmaResearch / K-Beauty Hub · ROE-25% Screen — Samyang Foods Passes 9 Filters · US-China Summit Result

Samyang Foods posted an all-time-high quarterly print — revenue ₩714.4bn, OP ₩177.1bn, OPM 24.8%. Buldak Ramyun grew +215% in Europe, +37% in US, +36% in China. Exports = 82% of revenue. This is no longer “a Korean ramen company” — it’s a global Buldak-brand export business. The print didn’t just lift Samyang. It pulled 19 Korean consumer names: F&B (+7.6%), department stores / duty-free (+9.7%), cosmetics (+5.4%). First time this year that the entire consumer complex moved together in a single day. Why — and will it continue?

TL;DR

Samyang 1Q print. Revenue ₩714.4bn (+35% YoY), OP ₩177.1bn (+32%), OPM 24.8%. Beat consensus by 8% — a clear surprise.
82% of revenue is exports. Noodle-snack exports ₩565.7bn. Europe +215%, US +37%, China +36%. Not a single geography — simultaneous growth across regions.
The entire consumer complex moved. Samyang +11.1%, CJ CheilJedang +10.6%, Shinsegae +10.9%, LG H&H +10.3%, Hotel Shilla +8.0%. 19 names across F&B / department stores / cosmetics averaged +7–10%.
Why simultaneous. (1) Samyang’s print anchored F&B re-rating; (2) semiconductor-concentration fatigue rotated capital out; (3) residual post-summit hope on China-consumer optionality fired the cosmetics / duty-free leg.
Caveat. This is not “Korean consumption is recovering.” It’s “capital widening into earnings-validated consumer names on day one." Persistence requires 2Q earnings and follow-through flow.

1. Samyang Foods — numbers refuted the skepticism

1.1 All-time-high quarter

Item	1Q26	YoY	QoQ
Revenue	₩714.4bn	+35.1%	+12.0%
OP	₩177.1bn	+32.2%	+27.2%
NI	₩144.5bn	+46.0%	+51.2%
OPM	24.8%	-0.5pp	+3.0pp

1.2 Vs. consensus

Item	Actual	Consensus	Surprise
Revenue	₩714.4bn	~₩674.6bn	+5.9%
OP	₩177.1bn	~₩163.8bn	+8.1%
NI	₩144.5bn	~₩127.8bn	+13.1%

OP beat consensus by 8%, NI by 13%. A clean surprise.

1.3 The core — Buldak is now a global brand

The core of the print is overseas noodle-snack exports:

Noodle-snack exports: ₩565.7bn (79% of total revenue)
Total exports: ₩585.0bn (82% of total revenue)

Geography growth:
Europe: +215% (3× over)
US: +37%
China: +36%
→ Not a single region — simultaneous growth across the world

Don’t think of Samyang as a “Korean ramen company.” Think of it as a global Buldak-brand export business where 82% of revenue is overseas — Korean F&B’s BTS equivalent: international revenue is 4× domestic.

The ROE-25% screen identified Samyang Foods as one of only 4 Korean names passing 9 simultaneous filters (ROE 35.3%, OPM 23.4%). This 1Q print numerically validates that screening result.

1.4 What the market was skeptical about before the print

Pre-print, the stock had corrected from ₩1,377,000 (April 20) to ₩1,295,000 (May 13) — about -6%. Concerns were three-fold:

China inventory accumulation
Export growth deceleration
Top-of-range valuation discomfort

1Q refuted all three. China still +36%, exports actually accelerated, OPM held near 25%.

1.5 Price and flow reaction

May 14 close: +11.1% to ₩1,439,000. Turnover 2.75× the normal level.

Participant	May 14 net
Foreign	+₩34.8bn
Institutional	+₩44.4bn
Program	+₩38.7bn
Retail	-₩76.5bn

Foreign + institutional + program net buying; retail net selling. A high-quality flow combination. Within institutional: PE funds +₩29.1bn, investment trusts +₩7.0bn, pension funds +₩6.0bn, insurers +₩2.5bn — multiple participants simultaneously.

Sell-side targets: KB Securities ₩1,950,000, DS Investment ₩1,900,000, Kyobo ₩1,860,000. Upside from ₩1,439,000 remains, but immediate chasing after a +11% day is inefficient.

2. The entire consumer complex moved — anatomy of the rotation

2.1 What happened in a single day

It wasn’t just Samyang. The move propagated sequentially: F&B → department stores / duty-free → cosmetics.

Sector	# of names	Avg return	Foreign net	Institutional net
F&B	9	+7.6%	+₩62.6bn	+₩54.8bn
Dept stores / duty-free / retail	4	+9.7%	+₩0.8bn	+₩34.2bn
Cosmetics / beauty	6	+5.4%	+₩40.7bn	+₩27.3bn

Selected names:

Name	Return	Foreign	Institutional	Why
Samyang Foods	+11.1%	+₩34.8bn	+₩44.4bn	Earnings surprise
CJ CheilJedang	+10.6%	+₩12.4bn	+₩1.6bn	F&B-basket co-buying
Nongshim	+7.2%	+₩10.2bn	+₩2.0bn	Ramen peer spread
Shinsegae	+10.9%	+₩4.1bn	+₩1.6bn	Under-valued retail bounce
Hyundae Dept Store	+9.9%	-₩4.8bn	+₩8.5bn	Institutional rotation
Lotte Shopping	+9.7%	-₩4.8bn	+₩13.9bn	Institutional retail re-rating
Hotel Shilla	+8.0%	+₩6.3bn	+₩10.2bn	China / duty-free optionality
LG H&H	+10.3%	+₩8.3bn	+₩6.4bn	China-consumer rebound
AmorePacific	+8.1%	+₩1.8bn	+₩10.4bn	Cosmetics large-cap rotation
APR	+5.7%	+₩26.9bn	+₩8.6bn	K-beauty growth-name re-buy

2.2 Why this happened — four reasons

Reason 1: Samyang anchored the F&B sector
→ With earnings confirmed, the market re-priced F&B from
 "defensive" to "global K-food export growth"
→ Samyang up → CJ CheilJedang / Nongshim co-rally

Reason 2: Semiconductor concentration eased temporarily
→ Of KOSPI's May +19%, semis +39% and autos +29%
→ Only 2 of 26 sectors beat KOSPI — first since 2005
→ Chase-fatigue → capital rotated to non-semi earnings names
→ This is "rotation"

Reason 3: Post-summit residual hope fired China-consumer optionality
→ Shinsegae, Hotel Shilla, LG H&H, AmorePacific carry "China tourism / 
 duty-free / cosmetics recovery" optionality
→ US-China summit hope + China-exposed asset preference
→ Long-neglected China-consumer names saw short-cover + rotation entry

Reason 4: Flow was institutional + foreign, not retail
→ 19 consumer names aggregate: foreign +₩104.1bn, institutional +₩116.3bn,
 retail -₩218.2bn
→ Retail selling absorbed by institutions / foreigners = high-quality flow

The US-China summit result piece noted “no grand bargain” — but the Hormuz agreement and the residual China-relations-normalization hope kept partial optionality alive. That residual hope is what fired the China-consumer leg (Shinsegae, Hotel Shilla, LG H&H).

2.3 “Rotation” — primer for newcomers

What rotation means:
Capital in the market must always be somewhere.
When semis rallied hard, eventually
"semis have moved too much / others have good earnings"
→ part of the capital exits to other sectors
Money "rotates" sector to sector.

The May 14 structure:
Semis (+39% from May start) → fatigue
Samyang earnings surprise → "consumer also has earnings"
→ F&B → department stores → cosmetics spread sequentially
→ This is rotation

3. Don’t read this as “Korean consumption recovery”

3.1 Sector personalities differ

Sector	Nature	Persistence
F&B	Samyang-earnings-led earnings-based strength	Depends on 2Q print
Dept stores / retail	Under-valuation bounce + institutional rotation	Value bounce, not structural growth
Cosmetics	China-consumer hope + K-beauty re-buy	High name-level dispersion

F&B’s core is the Samyang print — driven by “Buldak selling globally,” not “consumer demand recovery.” Retail’s core is “too cheap, institutions briefly bought” — not a signal that Korean domestic consumption rose. Cosmetics rest on “China might re-open” — still unconfirmed.

Why Korea Part 2 — Korean Cosmetics framed K-beauty as a fast-experimentation ecosystem built around ODM + Olive Young + digital distribution — structurally independent of any single China-market recovery. So within cosmetics, distinguish “China-recovery-option names” (LG H&H, AmorePacific) from “ecosystem-growth names” (APR, Silicon2, PharmaResearch).

3.2 Tests for whether this is one-day or trend

To distinguish “one-day rotation” from “trend”:

Does Samyang hold the ₩1,439,000 high zone?
Do F&B peers see sustained foreign + institutional buying?
Do cosmetics / duty-free see follow-through with turnover-backed second-leg rallies, not just short-covers?

Failure on any one = one-day rotation only.

4. Name level — where the quality is, where the risk is

4.1 Highest quality — earnings-backed

Samyang Foods: revenue ₩714bn, OPM 25%, exports 82%, Europe +215%. Earnings-led growth. But at ₩1,439,000 after a +11% day, prefer a pullback or a turnover-backed breakout to immediate chasing.

APR: K-beauty growth name. Foreign +₩26.9bn, institutional +₩8.6bn co-buying. Self-brand (Medicube) global expansion story. But valuation is elevated and quarterly earnings verification is needed. Full analysis: APR / Medicube deep dive.

4.2 Tactical rotation — value, not earnings

Shinsegae, Hyundae Dept Store, Lotte Shopping: retail has been depressed and looks cheap on earnings. Institutional buying was on “value bounce” framing. Read as discount-closing bounces, not structural growth.

Hotel Shilla: duty-free + China-tourism optionality. Rises when US-China relations improve, fades when that hope fades. Event-driven character.

4.3 To handle carefully — narrative-only with no earnings

Cosmetics laggards riding only on “China recovery” without earnings verification are risky. Samyang had earnings; LG H&H / AmorePacific are large-cap and get institutional rotation flow — but small-cap cosmetics names will reverse the moment “China hope” cools.

Name-level comparisons (APR, Silicon2, PharmaResearch, Classys) are documented at the K-Beauty Hub.

5. Bottom line

Samyang Foods’s 1Q26 print is strong. Revenue ₩714.4bn, OPM 24.8%, exports 82%. Buldak grew 3× in Europe and 35%+ in both US and China. That print became the trigger that lifted the entire consumer complex by +7–10% in a single day.

But this is not a signal of “Korean consumption recovering.” Samyang is earnings-led, retail is value-bounce, cosmetics is China-hope. Different natures.

What matters most is what Samyang demonstrated: “Korean consumer goods can also become global export-growth names." Buldak proved not “noodle market growth” but that Korean F&B brands can earn premium pricing globally. If 2Q confirms this in numbers again, Samyang re-classifies from “F&B name” into “global consumer growth name."

Next checkpoint: does 2Q OP hold near ₩170bn? If yes, the capital migration into “earnings-validated consumer names” is a trend, not a single day.

FAQ

Q: Is an 8% consensus beat really that meaningful for Samyang? A: The headline beat matters less than the fact that the print refuted all three pre-print concerns simultaneously. The stock had corrected -6% on (1) China inventory worry, (2) export-deceleration worry, (3) valuation discomfort. 1Q showed China still +36%, exports accelerating, OPM holding at 25% — all three concerns resolved at once.

Q: 82% exports — is this effectively a foreign company? A: Structurally, yes. Of ₩714.4bn revenue, ₩585bn (82%) is overseas — Europe +215%, US +37%, China +36% growing simultaneously across regions. “Korean ramen company” understates it; “global Buldak-brand export business” is more accurate.

Q: Is the 19-name same-day rally a trend or one-day? A: Three tests: (1) Samyang holds ₩1,439,000, (2) F&B peers sustain foreign + institutional buying, (3) cosmetics / duty-free see follow-through with turnover-backed second-leg rallies. Failure on any one = one-day rotation only. May 15 flow is the first checkpoint.

Q: Department stores +9.7% — has Korean consumption recovered? A: No. Retail had been depressed; institutions briefly bought on “value bounce” framing. Not a structural recovery — a discount-closing bounce. Reading it as a macro signal would be wrong.

Q: Does LG H&H / AmorePacific rallying signal China recovery? A: A signal, not a confirmation. The summit didn’t deliver direct China-consumer-market opening agreements, but residual China-relations-normalization hope (post-Hormuz) lifted China-consumer optionality. Actual China cosmetics-demand recovery data hasn’t been confirmed yet.

Q: APR also rallied +5.7% — is K-beauty broadly going again? A: There’s dispersion within K-beauty. Names with own growth drivers (APR’s Medicube global expansion, Silicon2’s overseas distribution platform, PharmaResearch’s PN skin-booster, etc.) differ from “China-recovery option” names. Detailed comparisons at the K-Beauty Hub.

Q: Should one chase Samyang now? A: Immediate chasing is inefficient. After a +11% day, the average sell-side target (₩1,900,000) implies +32% upside but short-term profit-taking pressure exists. 2Q OP holding near ₩170bn (August) would confirm trend-growth re-rating; before that, prefer a pullback or a turnover-backed breakout.

This article is for research and informational purposes only and does not constitute investment advice. Samyang Foods 1Q figures from DART quarterly report (submitted 2026-05-13). Consensus (revenue ₩674.6bn, OP ₩163.8bn) is average of sell-side estimates. Geography growth (Europe +215%, US +37%, China +36%) from The Fact reporting. Sector returns / flow data from Kiwoom Securities DB (May 14 close). Sell-side targets (KB ₩1,950,000, DS ₩1,900,000, Kyobo ₩1,860,000) per each broker’s report. Rotation persistence is conditional, not confirmed. Individual-name valuations require separate verification. Analysis can be wrong. Data cut: May 15, 2026 KST.

Korea Quality Re-Rating Watch 2026-05-15: Device, Gigavis & Vitzrocell — Smart Money Holds Ground in Bear Korea

Fri, 15 May 2026 16:30:00 +0900

Macro Dashboard

Indicator	Level	5-Day Δ	Signal
KOSPI	7,493.2	−4.2%	Bearish
KOSDAQ	1,129.8	−6.4%	Bearish
USD/KRW	1,497	+2.5%	KRW weak
VIX	18.6	+1.0%	Stable
US 10Y	4.46%	+0.10pp	Drifting higher
Brent	$108.0	+3.6%	Rising
DXY	99.1	+1.2%	Firming

Regime verdict — KR: Bear / US: Bull. The divergence is the story. Korean breadth is thin (only 3.4% of names trading above their 50-day MA), while US breadth holds at 52.5%. The implication: the macro environment supports US risk-on exposure while Korea warrants selective, defensive positioning. Won weakness and a higher oil price add pressure on Korea’s import-heavy industrial base.

Market Wrap

Thursday’s Korea session was a tale of two markets inside one index. The discovery screener registered NEUTRAL at 65/100 with Follow-Through Day count still intact at Day 20 — technically, the broader market structure has not broken down. But the session felt like distribution. Large-cap semiconductor names absorbed heavy selling as foreign investors rotated out of bellwether positions, with the KOSPI finishing near five-day lows.

The thematic divide was stark. AI/semiconductor components, power electronics, and robotics held relative strength as markets continued to price in the accelerating AI infrastructure buildout. Cisco’s announcement raising its AI infrastructure order target from $5 billion to $9 billion (networking revenue +24.7% YoY) reinforced the narrative, and DRAM/NAND pricing optimism — DRAM +53% QoQ and NAND +75% QoQ expectations circulating in market commentary — kept the memory cycle thesis alive even as leading-edge names sold off.

On the weak side: large-cap memory foundry names faced significant distribution, with foreign net selling running into the multi-trillion-won range. Gaming, biotech names ahead of the June BIO USA conference window, and brokerage stocks underperformed. The divergence between quality of fundamentals and short-term price action was unusually wide — a pattern that tends to surface cleaner re-rating setups in the screeners precisely because money is moving selectively rather than broadly.

Sector-level flow takeaways:

Samsung Electro-Mechanics (009150.KS) — RS 98.5, +10.5% over five sessions, institutional buying; standout among KR tech names resisting distribution
Daeduck Electronics (대덕전자, 008060.KS) — RS 98.2, foreign buying present, but institutional selling in the session; mixed conviction
Volume leaders on the session: LG Electronics, Doosan Robotics, SK Networks, and semiconductor component names saw volume-backed moves consistent with the K-consumer and industrial robot rotation flagged in broker morning notes
Padu (파두, 440110.KQ) — RS 98.3, appeared in screener intersection with the cleanest institutional/foreign flow alignment; high price caution after a recent run

The broader takeaway: today was not a quality breakdown — the macro and earnings cycles still point constructively for selected names. It was a timing warning. Smart money is not leaving the Korea market uniformly; it is concentrating into fewer, cleaner names.

Today’s Quality Re-Rating Candidates

The meta screener ranked 121 names today, with 20 surfacing as article-grade candidates. The editorial priority goes to the 5-screener overlap at the top — a rare configuration where quality fundamentals, institutional/foreign flow, cycle re-rating, earnings improvement, and post-earnings drift all fire simultaneously.

Top Candidate Table

Ticker	Name	Meta Score	Screeners Hit	RS%	Key Metrics
187870.KQ	Device (디바이스)	102.0	QC, SMQ, CR, SME, PEAD	89.9	ROE 11.1%, OP YoY +311.9%, OPM Δ +11.7pp, F+QI 5d +₩8.8B
420770.KQ	Gigavis (기가비스)	87.0	QC, SMQ, CR, SME	95.5	ROE 7.3%, OP YoY +777.2%, OPM Δ +29.9pp, F+QI 5d +₩18.7B
080220.KQ	Jeju Semiconductor (제주반도체)	78.5	QC, SMQ, CR, SME	96.9	ROE 17.4%, OP YoY +274.4%, OPM Δ +6.0pp, F+QI 5d +₩64.6B
082920.KQ	Vitzrocell (비츠로셀)	74.9	QC, SMQ, CR, SME	96.3	ROE 17.2%, OP YoY +33.4%, OPM Δ +3.9pp, F+QI 5d +₩30.7B
356860.KQ	TLB (티엘비)	70.5	SMQ, CR, SME, PEAD	94.4	OP YoY +664.7%, OPM Δ +8.2pp, F+QI 5d +₩25.9B, IR held today
089970.KQ	VM (브이엠)	64.3	QC, CR, PEAD	93.8	ROE 14.7%, OP YoY +386.9%, OPM Δ +29.3pp, consensus up-revision
095610.KQ	TES (테스)	62.9	QC, SMQ, SME	95.8	ROE 14.5%, OP YoY +50.3%, F+QI 5d +₩65.9B; short interest 6.7%
000660.KS	SK Hynix (SK하이닉스)	56.7	QC, CR	98.3	ROE 35.6%, OP YoY +101.2%, OPM Δ +13.1pp; 5d F+QI −₩124.9B ⚠️

QC = Quality Compounder; SMQ = Smart Money Quality; CR = Cycle Rerating; SME = Smart Money Earnings. F+QI = combined foreign + quality-institutional 5-day net flow.

Top 3 In Depth

#1 — Device (187870.KQ, score 102.0) is the only name in today’s universe to clear all five screener layers. The company manufactures special-purpose industrial machinery — semiconductor and display process equipment in its core segment. The 3-layer thesis: quality (OP YoY +311.9%, margin expansion of 11.7pp over the trailing year, ROE above quality threshold), money flow (F+QI net +₩8.8B over five days; ranked #1 in Smart Money Earnings and #1 in Smart Money Quality), re-rating catalyst (DART filing shows a 무상증자 bonus issue rights-off date on April 27, which typically introduces a mechanical price reset followed by fresh positioning). The stock’s 20-day return is elevated (+30% ret20d per PEAD data), so near-term momentum chasing carries risk — the setup is better watched for a pullback-and-confirm than chased at current levels. DART shows a Q1 quarterly report filed today (20260515); verify revenue composition and order backlog in the filing before treating this as a buy signal.

#2 — Gigavis (420770.KQ, score 87.0) is flagged in four screeners with the strongest operating leverage in the universe: OP YoY +777.2% on revenue growth of +100.6%, translating into a 29.9pp margin expansion. The company operates in special-purpose machinery, likely PCB or semiconductor inspection systems given its cycle co-movement with memory capex. F+QI five-day net inflow was +₩18.7B. One caution the meta screener explicitly flags: Gigavis also appeared in the Consensus Down Revision screener, meaning analysts have been trimming forward estimates even as recent actuals beat. This creates an unusual combination — operating leverage is clearly materializing, but the street is not yet resetting numbers upward. Check the Q1 quarterly report filed today (20260515) and the two insider ownership reports from May 14 before forming a view on consensus trajectory.

#3 — Vitzrocell (082920.KQ, score 74.9) manufactures primary and secondary batteries, with strong positioning in specialty lithium cells for defense, IoT, and industrial applications. The quality layer is clean: ROE 17.2%, OPM 28.5%, debt ratio just 12.2% — one of the lowest leverage profiles in the screener universe. F+QI net inflow over five days was +₩30.7B, and the stock sits in Quality Compounder, Smart Money Quality, Cycle Rerating, and Smart Money Earnings simultaneously. The single flag: a litigation filing appeared in DART on May 14 (claim value above disclosure threshold). This does not disqualify the name but warrants reading the filing to assess materiality before the position is sized. The stock’s RS percentile of 96.3 and margin expansion of +3.9pp make it one of the more defensively grounded candidates given its low leverage and non-semiconductor exposure in a session where memory names were under heavy distribution.

This post is market analysis and does not constitute investment advice. All data sourced from today’s screener runs (2026-05-15). DART filings referenced are publicly available on dart.fss.or.kr.

US-China Summit Result — No 'Grand Bargain.' Hormuz Agreement, H200 Licenses, $30B Tariff-Easing Talks Did Land — but KOSPI 7,900 Already Priced It

Fri, 15 May 2026 16:00:00 +0900

📚 US-China summit series. Previous: Pre-Analysis — 10 Agendas and Scenario Strategy Follow-up: The May 14-15 Beijing US-China Summit — Managed Truce, H200 Refusal, and What It Means for Korean Semiconductors

🔗 Related: KOSPI May 13 V-Reversal — All-Time High Through ₩3.76tn Foreign Selling · Samsung Electronics Citi TP ₩460,000 — Memory-Cycle Frame Reset · Why Korea Part 4 — $6.7B Inflows, Korea-Discount Dissolving or Value Trap?

The summit concluded. The pre-analysis said “the most-likely outcome is a ‘soybeans + Boeing + trade-board’-level symbolic agreement” and “this is already in the price.” Both came true. Hormuz reopening, H200 licenses, and $30B tariff discussions did land, but there was no joint communiqué, no rare-earth pause extension, no semiconductor-control easing. And KOSPI is already +19% in May — semiconductors +39%. The question now is not “should I buy more?” but “which sectors haven’t moved yet?”

TL;DR

Outcome. 135-minute summit. Hormuz reopening + Iran-no-nuclear agreement, NVIDIA H200 China-sales licenses (10 firms, up to 75K units each), $30B non-sensitive tariff-easing discussions. But no joint communiqué, no rare-earth-pause extension, no semiconductor-equipment-control easing.
Vs. pre-analysis. A blend of “Scenario A (expectations met, 50%)” + “Scenario D (Hormuz easing)” + “Scenario E (FX-stability language).” “Scenario B (tech-for-rare-earths grand bargain)” did not materialize.
Market reaction. KOSPI 7,912 (+0.87%); China onshore down 1.5–2.5% (expectations exhausted). Good news arrived, but was already in the price.
Core read. Chasing is inefficient. KOSPI May +19%, semis +39% is extreme concentration. Only 2 of 26 sectors beat KOSPI (semis + autos) — first time since 2005. The asymmetric structure favors finding the least-priced-in beneficiary.

1. How did the pre-analysis perform — scenario validation

1.1 Four pre-analysis scenarios

Comparing the pre-analysis scenarios against actual outcome:

Scenario	Pre-prob	Actual	Match
A. Expectations met (soybeans / Boeing / committee + Iran signal)	50%	✅ closest	Match
B. Tech-for-rare-earths grand bargain	20%	❌ did not happen	Miss
C. Downside (Taiwan tension + control tightening)	25%	△ Taiwan warning but no escalation	Partial
D. Hormuz easing	15–20%	✅ Hormuz reopening agreement	Match
E. FX-stability comment	25–35%	△ direct language unconfirmed	Partial

The pre-analysis core message — “betting before the event is gambling; positioning after is investing” — was validated. Chinese equities actually fell on summit day.

1.2 What was confirmed — exactly what landed

Outcome	Detail	Nature
Hormuz reopening agreement	Both sides agreed on keeping the strait open and Iran no-nuclear	Most substantive outcome
H200 sales licenses	10 Chinese firms (Alibaba, Tencent, ByteDance, etc.) cleared to purchase NVIDIA H200	License granted; zero actual shipments
$30B tariff-easing discussions	Non-sensitive items focus. Items not yet defined	Discussion stage, not agreement
‘Strategic stability’ framing	Both sides agree on constructive relations	Phrase only, non-binding
Xi’s Taiwan warning	“Conflict possible” warning; US says “no policy change”	Tension preserved, not escalated

1.3 What did NOT land — this matters more

Unresolved	Why it matters
Joint communiqué	No binding agreement means no enforcement mechanism
Rare-earth pause extension	November time bomb intact
Semiconductor equipment export-control easing	Korean memory’s “geopolitical scarcity premium” preserved
Specific tariff-reduction items	$30B remains directionally undefined
Boeing / agriculture purchase scale	Risk of Trump election-cycle inflation

2. Why a “small deal” — structural reasons

2.1 Neither side can concede the core

What Trump wants: visible election-cycle wins (soybeans, beef, Boeing)
What Xi wants: semiconductor-control easing, Taiwan security

Neither can give the other their core ask:
→ US can't ease semi/AI controls (national security)
→ China can't concede Taiwan (regime legitimacy)

Result: partial face-saving trades + big picture deferred to follow-up sessions

2.2 What the absence of a joint communiqué means

No communiqué means no binding agreement was created. Hormuz, H200, tariffs — all administrative-action level. Reversible based on political conditions.

The pre-analysis warned that “2017 Trump China visit also produced large MOUs but many were non-binding.” Same pattern repeats here.

2.3 CSIS assessment

Scott Kennedy at CSIS: “a largely superficial truce that tilts in China’s favor." US preserved security cards, but on economic gestures, the direction moved toward what China wanted.

3. Sector-by-sector impact on Korea

3.1 Semiconductors — good news, but already priced in

The H200 licensing logic:
→ NVIDIA H200 can now be sold to Chinese firms
→ H200 contains HBM
→ HBM is SK hynix's market — 64% share, #1
→ Therefore additional demand for Korean memory

But:
→ Semis already up +39% in May
→ Only 2 of 26 KOSPI sectors beat KOSPI (first since 2005)
→ Foreign net sell on May 14: ₩705.2bn

Key issues:
→ "License ≠ actual shipment." Zero H200 actually delivered
→ Beijing restraining NVIDIA inflow to protect domestic AI chips
→ Real HBM-incremental-demand realization lags by quarters

Read: hold-the-semiconductor-weight makes sense; chasing is inefficient. The Samsung Citi piece framed that breaking the “memory cycle” framing requires both sustained price uplift and HBM4E customer qualifications. H200 licensing is one piece of that. Track actual H200 delivery, NVIDIA guidance, and HBM order-backlog change.

3.2 Rare earths — what didn’t happen matters more

The pre-analysis flagged the “November time bomb” of the rare-earth pause expiration. The expected extension was not announced. Two readings:

Read 1 (short-term negative):
→ Rare-earth theme-stocks' near-term catalyst is exhausted
→ The "summit will resolve it" expectation broke
→ Short-term profit-taking pressure

Read 2 (mid-term positive):
→ China continues holding rare earths as a negotiating card
→ Uncertainty persists through November expiration
→ Non-China supply-chain build-out necessity is reinforced
→ Firms with actual supply-chain build capability earn a premium

Read: short-term theme stocks (UnionMatel etc.) in profit-taking. Mid-term, non-China permanent-magnet / rare-earth supply chains remain structurally relevant.

3.3 Hormuz / energy — the most substantive outcome, the least-priced-in beneficiary

The Hormuz reopening is the summit’s most substantive deliverable. Signal that China will engage diplomatically on Iran pressure. The end-of-war scenario probability rose.

What Hormuz opening does for Korea:

1. Oil-price downward pressure
 → Energy-cost relief for Korea (high Middle-East crude dependency)

2. Naphtha price decline
 → NCC-based petrochemical cost burden relief
 → Margin recovery potential for Lotte Chemical, LG Chem

3. KRW strength pressure
 → Lower energy imports → current-account improvement → KRW stability
 → Better foreign flow → constructive for KOSPI as a whole

4. Aviation / chemical cost easing
 → Korean Air, Asiana fuel-cost reduction

Why “least-priced-in”: the market interpreted the summit through the “semiconductor beneficiary” lens first. But semis already moved +39%; petrochemicals are still near the cycle bottom. The Hormuz second-order effect (naphtha price decline → chemical-margin normalization) hasn’t been absorbed by the market yet.

3.4 Sector summary

Sector	Summit impact	Already priced?	Read
Semiconductors	H200 licensing → HBM demand expectation	Already +39%	Chasing inefficient; hold weight
Rare-earth themes	Pause extension absent → catalyst exhausted	Partial	Short-term profit-taking
Petrochemicals	Hormuz → naphtha cost relief	Not yet priced	Watch list
Autos	Neutral. Supply-chain stability +; China EV tariffs preserved	Already +29%	Chasing inefficient
Aviation / chemicals	Oil decline → cost relief	Partial	Watch
Defense	Hormuz end-of-war signal → short-term momentum weakens	Not priced	Short-term pullback → potential opportunity
Cosmetics / tourism	China opening expectation → indirect positive	Not priced	Watch

4. “KOSPI 7,900 already priced the good news” — asymmetric setup

4.1 Extreme concentration

Since start of May:
KOSPI overall: +18.9%
Semiconductors: +38.6%
Autos: +29.1%
Other 24 sectors: under-performed KOSPI

→ Only 2 of 26 sectors beat KOSPI
→ Extreme concentration not seen since 2005

What this means: KOSPI +19% is not “Korean economy strength” but “Samsung Electronics + SK hynix strength.” Most else didn’t move. Such concentration can persist, but historically it’s often a rotation precursor.

4.2 Foreign-flow warning

May 14 foreign net selling: ₩705.2bn. Foreign selling into a rising tape — is it “expectation-exhaustion selling” or “overall position-reduction”? Worth tracking. The May 13 V-reversal piece noted ₩3.76tn of foreign net selling absorbed by domestic flows — same pattern. The durability of “foreigners out, domestic capital catching” needs verification.

May 15 foreign flow is the key checkpoint. Continued selling → “catalyst-exhaustion correction” possible. Net-buying flip → trend preservation.

4.3 What China onshore showed

China onshore equities fell 1.5–2.5% on summit day. The classic “expectation rally → fade on outcome” pattern. Korea could see similar.

5. The least-priced-in spot — Hormuz second-order effect

5.1 Why petrochemicals are on the watch list

Logic chain:

Hormuz reopening agreement (confirmed)
→ End-of-war scenario probability rises
→ Oil-price downward pressure
→ Naphtha price decline
→ NCC-based petrochemical cost burden relief
→ Ethylene-naphtha spread normalization
→ Earnings leverage for trough-level chemicals

Conditions for this logic to work:
→ Dubai oil settles below $70
→ Naphtha-ethylene spread recovers above $250/tonne
→ Verifiable within 4–8 weeks

Why the market hasn’t priced it: the market absorbed the summit through the “semiconductor news” lens first. Hormuz’s chemical-margin effect was covered mainly in English-language press; Korean sell-side estimate revisions haven’t followed yet.

5.2 Cautions

The Hormuz agreement is declarative. Actual military blockade lifting is separate, and if Iran end-of-war fails, it can reverse immediately. Also, Chinese petrochemical oversupply is structural, capping spread recovery.

So the read is not “buy chemicals now” but “monitor Dubai oil and spreads over 4–8 weeks, conditional approach”.

6. Validation of the pre-analysis core messages

6.1 What was right

Pre-analysis message	Outcome
“Most-likely outcome is a ‘soybeans + Boeing + committee’ level”	✅ Was a small deal
“Betting before the event is gambling”	✅ China onshore fell on summit day
“KOSPI 7,500 already priced the good outcome”	✅ Additional upside limited
“Korea is not a counterparty but the maximum stakeholder”	✅ Semis / energy / rare earths all directly affect Korea
“November time bomb (rare earths + truce simultaneous expiry)”	✅ Rare-earth pause extension absent — risk intact

6.2 What needs revision

Pre-analysis message	Revision
“H200 China sales = summit outcome”	→ H200 licensing started Dec 2025. This summit was a “reaffirmation” rather than new licensing
Scenario D (Hormuz) probability 15–20%	→ Hormuz agreement actually landed. Probability was too low

7. Calendar — what to watch next

Window	Event	Why
5/15	Foreign flow check	Continued selling = catalyst-exhaustion correction; flip = trend preservation
5/15	Day-2 joint statement text	Additional agreement detail?
5/21–22	Pearl Abyss Meritz NDR	Institutional re-rating window (separate issue)
June	Dubai oil + naphtha spread trajectory	Hormuz agreement’s real-world effect
July	Samsung Electronics + SK hynix 2Q earnings	H200 incremental demand showing up in results?
August	Half-year reporting season	Hormuz effect appearing in chemical / energy earnings
November	Busan truce expiry + rare-earth pause expiry	The real time bomb. Extension or not = H2’s largest macro variable

8. Bottom line

The US-China summit produced a “managed truce,” not a “grand bargain.” Hormuz agreement, H200 licenses, $30B tariff discussions landed — but no joint communiqué, no rare-earth pause extension, no semiconductor-control easing. Closest to the pre-analysis Scenario A (expectations met).

The problem is KOSPI is already +19% with semis +39% and autos +29%. Good news arrived but is already in the price. Only 2 of 26 sectors beating KOSPI is the most extreme concentration since 2005.

This is not a chase setup. It’s a “find the least-priced-in beneficiary” setup. Semis have moved; rare-earth theme stocks have exhausted the near-term catalyst. The least-priced-in spot is the Hormuz second-order effect — naphtha-cost decline driving petrochemical margin normalization. And the real time bomb hasn’t gone off yet — the November simultaneous expiry of rare-earth pause + trade truce.

FAQ

Q: Why did Chinese onshore equities fall after a “successful” summit? A: The classic “rally on expectation, sell on outcome.” Pre-summit buying anticipated a “grand bargain”; the actual “small deal” outcome triggered profit-taking. Korea faces the same risk.

Q: Is H200 licensing materially positive for Korean memory? A: Limited near-term. License ≠ actual shipment. Beijing has reasons to restrain NVIDIA inflows in favor of domestic AI chips, and real HBM incremental demand lags by quarters. Plus semis are already +39% in May.

Q: Why isn’t the absence of rare-earth pause extension a bigger problem? A: November expiration is 6 months out — additional negotiation opportunities exist, and unilateral Chinese tightening would invite retaliation. But short-term theme stocks (UnionMatel etc.) have exhausted the summit-hope catalyst — profit-taking pressure.

Q: Why is petrochemicals “the least-priced-in spot”? A: The market absorbed the summit through the “semiconductor lens” first. The naphtha-cost-decline → chemical-margin normalization story is lagged-effect and Korean sell-side estimates haven’t been revised yet. But it’s conditional — requires Dubai oil settling below $70 and naphtha-ethylene spread recovering above $250/tonne over 4–8 weeks.

Q: What is the “November time bomb”? A: Busan trade truce (47% tariff regime) and rare-earth export-control pause both expire in November. Without resolution, tariffs reset to 57% and rare earths can be re-weaponized. The best-case May summit deliverable would have been a “roadmap to November” framework — that didn’t land.

Q: KOSPI +19%, semis +39% — buy more? A: Only 2 of 26 sectors beating KOSPI is extreme (first since 2005). Historically, this kind of concentration tends to be a rotation precursor. The disciplined read is to watch under-priced sectors (petrochemicals, aviation, parts of cosmetics / tourism) rather than chase.

Q: Foreigners sold ₩705bn on a rising KOSPI — why? A: Domestic flows (retail + institutional) absorbed it. Same pattern appeared May 13 (₩3.76tn foreign net selling absorbed by domestic flows). The durability of this “foreigners out, domestic capital catching” structure is the key variable — May 15 foreign-flow direction is the next checkpoint.

This article is for research and informational purposes only and does not constitute investment advice. Summit outcome details are based on Yonhap, Reuters, China MOFA, White House releases, CNBC, CNN reporting. H200 licensing (10 firms, up to 75K units each) is per Reuters reporting; zero actual shipments to date. $30B tariff easing is in discussion stage, not agreement. The Hormuz agreement is declarative; actual military-blockade lifting is separate. Day-2 may produce additional disclosures. KOSPI / sector returns reference Kiwoom Securities citations. Petrochemical margin recovery is conditional on oil and naphtha-spread evolution, not confirmed. Analysis can be wrong. Data cut: May 15, 2026 KST.

Japan's PPI Shock — The Real Fear Is Not Selling US Treasuries, but No Longer Buying Them. How BOJ Tightening Resets Global Rates

Fri, 15 May 2026 00:00:00 +0000

The market is talking about “Japan dumping Treasuries.” Japan’s PPI came in hot, BOJ rate-hike odds went up, the story goes that Japanese capital sells Treasuries and goes home. That story is only half right. The real risk is not “Japan dumping Treasuries” but “Japan no longer buying them." When the world’s largest foreign Treasury holder stops adding, prices fall even without selling. And when the US 10-year yield rises, the global discount rate rises with it — which reprices every risk asset, in the US and in Korea.

Key takeaways

The mechanism in one line: Japan PPI shock → BOJ hike probability ↑ → JGB yield ↑ + JPY-strength expectations ↑ → post-hedge Treasury return ↓ → Japanese marginal bid for Treasuries ↓ → UST term premium ↑ → global discount rate ↑.
The core idea: this is not a “selling event,” it is marginal-demand erosion. Japan does not need to dump anything. Simply not adding is enough to push prices lower (yields higher).
Why it matters: Japan is the world’s largest foreign holder of US Treasuries, with ~USD 1.13 trillion. When that marginal bid weakens, the supply-demand balance breaks.
Second-order effects: US 10-year up → global discount rate repriced → US big-tech multiples compressed → Korean equity risk premium affected.
Checkpoints: BOJ policy meetings, the JGB 10-year, USD/JPY, GPIF / Japanese life-insurer asset-allocation language, NY Fed 10-year term-premium estimates.

1. What people get wrong — this is not “selling,” it is “no new buying”

1.1 The wrong story going around

The wrong narrative:
"Japan inflation surprised → BOJ hikes →
 Japanese money sells Treasuries and goes home →
 US yields spike → stocks crash."

Why it's inaccurate:
1. Japanese lifers and pension funds hold USTs as quasi-collateral,
 not as a tactical risk position — they can't easily sell.
2. Outright large-scale selling would shock both FX and rates markets,
 so any move would be gradual.
3. The realistic transmission is not "dumping" — it is "buying less."

1.2 The actual mechanism

The correct chain:

Japan PPI shock
 ↓
BOJ tightening probability ↑
 ↓
JGB yields ↑ + JPY-strength expectations ↑
 ↓
Post-hedge UST yield falls (from the Japanese investor's POV)
 ↓
Japanese lifers / pension funds reduce NEW UST allocations
 ↓
Marginal UST demand weakens
 ↓
UST term premium ↑ (= long-end yields ↑)
 ↓
Global discount rate repriced
 ↓
Valuation pressure across risk assets

The point: not “selling,” but “weaker new buying." Even if Japan never sells a single Treasury, simply slowing additions pushes UST prices down (yields up).

2. Four concepts for the first-time reader

2.1 Currency hedging

What is FX hedging?

When a Japanese life insurer buys a US Treasury:
1. Convert yen to dollars
2. Buy the Treasury in dollars
3. At maturity, convert dollars back to yen

Problem: if USD/JPY moves in the interim, the JPY-equivalent value
can fall.

Solution: "currency hedge" — lock in the future FX rate upfront.

Analogy: buying foreign currency before a trip.
→ Eliminates FX risk
→ But costs a fee (the hedge cost)

The hedge cost is essentially the rate differential:
US rate − Japan rate ≈ hedge cost

Example: US 5%, Japan 0.5%
→ Hedge cost ≈ 4.5%
→ 10-yr UST yield 4.5% − hedge cost 4.5% = 0%
→ The Japanese investor's "post-hedge return" is effectively zero.

2.2 What happens to Japanese investors when BOJ hikes

When BOJ raises rates:

1. JGB yields rise
 → Decent yield is available at home
 → Less reason to reach overseas

2. Hedge cost narrows
 → US rate − Japan rate = hedge cost
 → As Japanese rates rise, the hedge cost falls
 → BUT a higher JGB yield is itself more attractive

3. JPY appreciation expectations
 → If the yen is expected to strengthen,
 the unhedged USD asset will translate back into fewer yen
 → "JPY-strength risk" gets priced on top of hedge cost

Net result:
The post-hedge return on USTs can turn negative for Japanese buyers.
→ JGB looks much better
→ Japanese investors slow NEW UST purchases.

2.3 Term premium

What is term premium?

Long-bond yields decompose into two parts:
1. The average of expected future short rates (the policy-rate path)
2. The "term premium" — compensation for the risk of locking up money
 for years

In plain English:
"Who knows what 10 years from now looks like? The premium is what you
get paid for that uncertainty."

When does term premium rise?
- Inflation uncertainty ↑
- Fiscal-deficit concerns ↑
- Supply/demand worsens (fewer buyers)
- Volatility ↑

If term premium rises 100bp:
→ 10-year yield rises \~100bp
→ Short rates can be unchanged while the long end climbs
→ "Curve steepening."

2.4 Japan = world’s largest foreign holder of US Treasuries

Japan's US Treasury holdings:

End-2024: \~USD 1.13 trillion
#1 globally (#2 China \~USD 0.79T, #3 UK \~USD 0.75T)

Composition:
- Japanese life insurers
- Japanese pension funds (GPIF, etc.)
- Japanese megabanks
- Japanese asset managers

Characteristics:
- Long-duration holders
- High FX-hedge ratio (\~50–70%)
- "Marginal buyer" — allocates fresh inflows to USTs every year

→ When that marginal bid weakens, the UST supply-demand balance breaks.

3. The mechanism with numbers

3.1 Post-hedge UST yield for a Japanese investor

Post-hedge 10-year UST yield (Japanese investor view):

UST 10-yr yield − hedge cost = effective return

Hedge cost ≈ short-rate differential (US − Japan)

Today (illustrative):
US short rate: 4.50%
Japan short rate: 0.50%
US 10-yr yield: 4.30%

Hedge cost = 4.50% − 0.50% = 4.00%
Post-hedge UST yield = 4.30% − 4.00% = 0.30%

Compare:
JGB 10-yr yield: 1.50%
→ For a Japanese investor: JGB 1.50% > hedged UST 0.30%
→ Domestic bonds beat US bonds.

3.2 What changes if BOJ hikes 50bp

BOJ policy rate 0.50% → 1.00% (+50bp)
JGB 10-yr 1.50% → 2.00% (assumed)
US rates unchanged (assumed)

Hedge cost = 4.50% − 1.00% = 3.50% (50bp lower)
Post-hedge UST yield = 4.30% − 3.50% = 0.80%

Compare:
JGB 2.00% vs hedged UST 0.80%
→ Gap widens to 1.20pp
→ The relative appeal of USTs for Japanese investors gets worse

Layering in JPY-strength expectations:
→ The unhedged portion of the UST exposure faces JPY translation losses
→ Net incentive to buy more USTs falls further.

3.3 How weaker marginal demand pushes UST yields up

UST market supply-demand:

Annual net issuance: \~USD 1.5–2.0 trillion
Main buyers:
- Federal Reserve (currently in QT — actually shrinking holdings)
- Domestic US (banks, money-market funds)
- Foreign (Japan, China, UK, euro area)
- Hedge funds, asset managers

Japan's typical annual net UST buying: \~USD 50–100 billion

If that drops by half:
→ \~USD 25–50 billion of marginal bid disappears
→ Who replaces it?
→ Domestic US investors or other foreign holders must absorb it
→ For them to do so, yields must rise
→ Term premium widens.

Estimated impact:
Weaker Japanese marginal demand alone could lift the US 10-year
by \~20–40bp.
(Midpoint of academic estimates — the exact number is hard to verify.)

4. Why this matters for global asset prices

4.1 The US 10-year = global discount-rate anchor

Every asset is, ultimately, the present value of future cash flows.

Price = future cash flows / (1 + discount rate)^t

The discount-rate anchor is the US 10-year Treasury yield.

If US 10-yr rises from 4% → 5%:
- US equity PERs compress (especially long-duration growth)
- EM equity risk premia get repriced
- Real-estate valuations face pressure
- Corporate-credit spreads widen

Big-tech is most affected:
→ Big-tech is heavily weighted toward distant future cash flows.
→ A higher discount rate haircuts those distant flows more.
→ AI-linked high-multiple names are the most sensitive.

4.2 How it lands on Korean equities

Direct channels:
1. Foreign-investor flows — higher US rates compress EM weights
2. Korean 10-yr yields move with the US — Korean PER compresses
3. USD/KRW upward pressure (when USD strengthens)

Indirect channels:
1. Some Japanese capital sits in Korean bonds too
 → If it repatriates, Korean bond demand softens
2. Korean growth names (Naver, Kakao, biotech) are sensitive to
 the US discount rate
3. Semiconductors are AI-cycle driven, so relatively defensive,
 but the multiple still gets squeezed

Relative winners / losers:
Friendly: financials, value, short-duration cash-flow names
Hostile: growth, biotech, high-PER thematic names

5. Is this the same story as “yen carry-trade unwind”?

Partly. Both are mechanisms where Japanese monetary / rate policy changes shift global capital flows, but the emphasis differs.

Yen carry-trade unwind:
→ "Sell global assets bought with cheap borrowed yen and repay the yen"
→ Emphasis: selling (unwind)
→ Hit in August 2024 (a brief global equity drawdown)

This Japan PPI shock:
→ "Japanese lifers / pension funds slow NEW UST buying"
→ Emphasis: weaker marginal demand (no new buys, not selling)
→ More gradual but more structural

What they share:
→ BOJ tightening is the trigger in both
→ Both come with JPY strengthening

How they differ:
→ Carry unwind is a short shock (days to weeks)
→ Marginal-demand erosion accumulates over quarters and years
→ Carry unwind hits volatile assets
→ Marginal-demand erosion hits UST yields and the global discount rate

6. What to watch from here

6.1 Japan-side indicators

Indicator	Why it matters
BOJ policy decision	The next meeting is pivotal. Is the path priced in?
Japan PPI / CPI	Transient or structural inflation
JGB 10-year yield	Higher JGB = lower post-hedge UST appeal
USD/JPY	JPY strength raises the hedge-burden for Japanese investors
GPIF / lifer asset-allocation language	The actual behavioral signal

6.2 US-side indicators

Indicator	Why it matters
US 10-year yield	Most direct readout
NY Fed term-premium estimates	ACM / KW models — close to 100bp is dangerous
MOVE (UST option volatility)	Treasury-market stress gauge
TIC foreign-holdings data	Hard data on Japanese buying / selling (2-month lag)

6.3 Scenarios

Scenario	US 10-year	Korean equities
A. Gradual BOJ hikes + US demand absorbs the slack	4.0–4.5% range	Limited impact
B. BOJ accelerates + term premium climbs	4.5–5.0%	Pressure on growth / high-PER
C. Actual Japanese selling materializes	5.0%+	Broad risk-asset correction

Base case sits between A and B. C is low probability, high impact.

7. How this links to other posts

Samsung Electronics piece: "The memory supercycle is the real story"
→ AI semis are structurally demanded, so relatively resilient to
 short-term rate shocks
→ But the multiple (PER) can still get squeezed.

US-China summit piece: "May 15 foreign-flow data is the next check"
→ Foreign flows respond to US rates, USD, and USD/JPY together
→ Japan PPI shock is yet another foreign-flow variable.

Consumer-rotation piece: "Capital is rotating out of semi-concentration"
→ Higher US rates accelerate the unwind of growth crowding
→ Relative tailwind for value and financials.

8. The one-line bottom line

Japan’s PPI shock is not the “fear of Japan dumping US Treasuries.” It is “the fear that the world’s largest foreign holder might stop adding” — marginal-demand erosion.

This is not a selling event; it is a structural shift in supply-demand. More gradual, but more durable. Even without selling, when the marginal buyer disappears, prices fall (yields rise). The term premium widens, that lifts the US 10-year, that resets the global discount rate, that flows through to every risk asset.

The precise structural equation:

Japan PPI shock
→ BOJ hike probability ↑
→ JGB yield ↑ / JPY expectation ↑
→ hedged UST return ↓
→ Japanese marginal bid for UST ↓
→ UST term premium ↑
→ global discount rate ↑

What we actually need to watch is not “will Japan sell Treasuries?” but “where will Japanese lifers and pension funds allocate fresh money?" The BOJ meeting, the JGB yield, and GPIF / lifer language — together those three are the new control panel for the global risk-asset discount rate.

This article is research and commentary only and is not investment advice. Japan’s US Treasury holdings of ~USD 1.13 trillion are per US Treasury TIC data and vary by reporting period. The hedge-cost calculation is a simplified model; in practice currency basis, CDS spreads, and other factors apply. Term-premium estimates reference the NY Fed’s ACM (Adrian-Crump-Moench) and KW (Kim-Wright) models. The assumption that a 50bp BOJ policy-rate hike maps to a 50bp move in the 10-year JGB is illustrative — the actual bond-market reaction can differ. The +20–40bp estimate for the US 10-year impact from Japanese marginal-demand erosion is a midpoint of academic ranges and is hard to verify precisely. The comparison with the yen-carry unwind is a mechanism contrast — the two phenomena can also occur simultaneously. The analysis may be wrong. Data cut-off: May 15, 2026 KST.

Memory Pulse Dashboard Now Live — Real-Time Signal Board for Samsung Electronics and SK hynix. Connecting the Blog's Analysis to Live Data

Fri, 15 May 2026 09:00:00 +0900

🔗 Related: HBM / KOSPI Investment Hub · Korea Daily Market Hub · Samsung Electronics Citi TP ₩460,000 — Memory-Cycle Frame Reset · Why Korea Part 4 — $6.7B Inflows, Korea-Discount Dissolving or Value Trap?

🌐 Visit: memory.koreainvestinsights.com — Memory Pulse Live Signal Board

The memory.koreainvestinsights.com custom domain is officially live. ‘Memory Pulse’ is a real-time signal dashboard built around Samsung Electronics and SK hynix. If the blog’s analysis pieces explain “why” on quarterly / weekly horizons, this dashboard shows “right now” — where flow is moving intraday, how US semiconductors traded overnight, how FX and macro-risk are tracking — all on one screen.

TL;DR

What it is. A real-time signal board specifically scoped to Samsung Electronics (005930) and SK hynix (000660). Korea-time anchored.
Why we built it. To close the gap between blog analysis (quarterly / weekly) and live market data. “How is yesterday’s thesis playing out today?” — visible on one screen.
How to use it. Flow: executive summary → market snapshot → company cards → signal drivers → instrument-moves table.
Data-trust transparency. Every row carries a Live / Delayed / Fallback / Missing label. You can tell live quotes apart from delayed or fallback values at a glance.
Blog integration. Direct links to the HBM Hub, Daily Market Hub, Telegram channel, and Substack. The real-time extension of the same analytical frame.

1. Why a separate dashboard

1.1 Blog and dashboard are different tools

Blog posts explain “why." When the Samsung Citi TP ₩460,000 piece analyzed whether the memory-cycle framing can break, that was a quarterly / weekly-horizon deep analysis — the thesis stays intact for months once written.

The dashboard shows “right now." Whether the thesis you read about is being validated by today’s tape, which signals are firing, which are fading — minute by minute.

Blog analysis: "Why the memory-cycle frame might break" (quarterly/weekly, thesis-centric)
Memory Pulse: "How that thesis is showing up in live data" (minute-by-minute, data-centric)

→ Complementary tools
→ Read the analysis, open the dashboard — the thesis's current state appears immediately

1.2 Why focus on Samsung + SK hynix

These two names anchor a large share of KOSPI market cap. A substantial portion of the index’s move is explained by them. Korea-side AI memory-cycle exposure also runs through these two.

As Why Korea Part 4 noted, the $6.7bn YTD foreign ETF inflow into Korea was absorbed in significant part by these two names. So tracking their minute-by-minute flow gives a strong read on the broader Korean tape.

2. Dashboard structure — what fits on one screen

2.1 Seven sections

Section	What it shows
1. Executive Summary	One-line intraday read — current market tone at a glance
2. Market Strip	KOSPI / KOSDAQ / FX / SOX snapshot
3. Source Health Strip	Whether each data source is functioning normally
4. Company Signals	Samsung Electronics + SK hynix cards (detailed per-name signals)
5. Driver Categories	Categories of forces driving the signal — US semis / FX / macro / memory cycle
6. Instrument Moves	Sortable name-move table (Korea / US / Memory / Macro·Risk filters)
7. Footer	Links back to blog / Telegram / Substack

2.2 Data-trust labels — the most important feature

Every row carries a freshness label.

Live : real-time data (≤1 min)
Delayed : delayed (e.g., 15 min, 1 hour — explicitly noted)
Fallback : fallback value (e.g., prior close)
Missing : data outage at source

Why this matters: market data delay varies by source. Some quotes are live; others are prior-day close; others are temporarily unavailable. Without labels, users assume uniform freshness. Memory Pulse makes freshness transparent so that users can verify the data itself before acting on signals.

2.3 Four filters — different lenses

The instrument-moves table sorts by 4 filters:

Filter	Lens
Korea	KOSPI / KOSDAQ names — domestic tape flow
US	NVIDIA / Micron / SOX, etc. — US semi read-through
Memory	HBM / DDR / NAND price proxies — cycle direct read
Macro / Risk	FX / oil / rates / VIX — macro environment

How to use it: e.g., pre-open, scan the ‘Korea’ tab for yesterday’s close + after-hours, the ‘US’ tab for the US memory / semi read overnight, then the ‘Macro / Risk’ tab for FX / VIX. With three views in one screen, a first-pass hypothesis on “how the open is likely to set up” comes together quickly.

3. Usage — by time of day

3.1 Pre-open (08:30–09:00 KST)

Flow:
1. Executive Summary → today's market tone in one line
2. Market Strip → headline indicators at a glance
3. US filter → US semi close (SOX, NVIDIA, Micron, etc.)
4. Macro / Risk filter → FX, VIX, oil
5. Company Signals → Samsung / hynix after-hours move

Judgment: 'how is today's Korean memory open likely to set up?'

3.2 During the session (09:00–15:30 KST)

Checkpoints:
- Executive Summary: real-time tone change
- Driver Categories: which category is firing
- Instrument Moves: track live-label rows for current pricing

If memory moves intraday:
→ Check the US filter for Micron / Western Digital corroboration
→ Check the Memory filter for HBM price proxy
→ Check the Korea filter for spillover to substrate (Samsung Electro-Mechanics, Daeduck Electronics)

3.3 Post-close (after 15:30 KST)

Flow:
1. Close-based instrument moves
2. Foreign / institutional flow from separate data sources
3. US filter → wait for US open signals

Overnight:
- When the US session opens, the US filter is central
- NVIDIA / Micron guidance or SOX flow has direct read-through to next-day Korean memory

4. Connection to the blog — real-time extension of the same frame

4.1 Direct blog links

4.2 Analysis → dashboard → analysis loop

Step 1: New analysis post on the blog
 e.g., "Samsung Electronics Citi TP ₩460,000 — Memory-Cycle Frame Reset"
 → Thesis established

Step 2: Real-time verification in the dashboard
 → Track memory price action, HBM4E-related news, FX, foreign flow
 → Current state of the thesis

Step 3: New analysis post on verified result
 e.g., "1Q earnings confirm thesis" / "Tracking-variable change — thesis adjusted"
 → Analysis updated

→ Analysis and data loop and refine the thesis over time

4.3 Two honest limitations

Limitation 1: the dashboard does not issue buy / sell signals. As stated in the footer: “educational market intelligence, not investment advice; does not issue buy, sell, or position-size instructions.” How to interpret signals is the user’s responsibility.

Limitation 2: dependent on data sources. When Fallback or Missing labels are frequent (e.g., specific US data-vendor outage windows), dashboard accuracy is impacted. This is exactly why the labels are explicit.

5. FAQ

Q: Is Memory Pulse free? A: Yes. Open to anyone at memory.koreainvestinsights.com.

Q: Does it work on mobile? A: Responsive design — mobile, tablet, desktop all supported. The instrument-moves table shows more at once on wider screens.

Q: Korean and English? A: Top-right ‘A / 가’ toggle switches instantly. Same for dark / light theme.

Q: How should I interpret a ‘Fallback’ label? A: It means the real-time data source has an issue and the value shown is a fallback (e.g., prior close). Always check the freshness label before acting. Only ‘Live’ values are real-time-reliable.

Q: Why only Samsung and hynix? Will other names be added? A: These two are the twin pillars of Korean memory and a large fraction of KOSPI. Concentrating on two names initially maximized signal quality. Expansion will follow user feedback and data-source stability.

Q: Can I use the dashboard before reading the blog analysis? A: Yes. The dashboard works standalone. But knowing the “context” of the signals helps interpretation, so reading the HBM Hub or a recent analysis first makes dashboard use more efficient.

Q: Can I cite or redistribute dashboard data? A: License terms differ by source — safer to cite with source attribution and the freshness label. Screen captures of the dashboard with “memory.koreainvestinsights.com” attribution are acceptable.

Q: What’s next? A: Data sources, signal categories, and instrument scope will expand based on user feedback. Priority is deepening the link between blog analysis and dashboard signals. New features will be announced via separate blog posts.

6. Bottom line

The blog explains “why.” The dashboard shows “right now.” When both tools live under one domain and link to each other, you get analytical depth and data speed at the same time.

Keep memory.koreainvestinsights.com open and use it alongside the HBM Hub or recent analysis posts — you can track Korean memory-cycle flow at a minute-level cadence. Data trust is labeled transparently, so signal interpretation isn’t muddied by questions about what you’re actually looking at.

The dashboard doesn’t give answers. It gathers signals on one screen so that users can quickly assemble their own read.

This article announces the official launch of the Memory Pulse dashboard. The dashboard itself is educational market intelligence and not investment advice. It does not include any buy, sell, or position-size instruction. All data carries Live / Delayed / Fallback / Missing freshness labels — check the label before acting on a signal. During source outages, some signals may be inaccurate. Data cut: May 16, 2026 KST.

Samsung Electro-Mechanics — The Invisible Infrastructure of AI Silicon. MLCC (Power Stability), FC-BGA (Chip Substrate), and Camera Modules Dissected

Fri, 15 May 2026 00:00:00 +0000

Samsung Electro-Mechanics (SEMCO, KRX 009150) crossed KRW 3 trillion in quarterly revenue for the first time. 1Q26 revenue KRW 3.21T, operating profit KRW 280.6B. Package substrate revenue grew +45% year over year, becoming the lead actor of the rerating. The stock rose from KRW 830,000 in late April to KRW 1,024,000 in mid-May — a +23% move. Sell-side target prices jumped from KRW 1.0M to KRW 1.5M within weeks. Yet few investors can clearly state what SEMCO actually makes, why it is an “AI beneficiary,” and what each of its three divisions does. This article dissects SEMCO at the division level.

Key takeaways

SEMCO is not a chip maker. It makes the power-stability components (MLCC) and the substrates that connect AI chips to motherboards (FC-BGA) that AI silicon needs to function inside a server.
Three divisions: Component (MLCC, 46% of revenue, 67% of profit), Package Solution (FC-BGA, 20% of revenue, 15% of profit), Optics (camera modules, 34% of revenue, 18% of profit).
Core shift: previously a “smartphone parts maker.” Today, the rerating thesis is AI-server / xEV high-end MLCC + AI-accelerator FC-BGA in structural supply shortage.
1Q26 results: revenue KRW 3.21T (+17% YoY), operating profit KRW 280.6B (+40% YoY). Stripping a one-off severance charge of KRW 71.4B, underlying operating profit was KRW 352.0B.
The division that matters most: Component (MLCC) defends earnings. Package Solution (FC-BGA) drives the share price. Optics is the optionality bucket.
At KRW 1.02M today: expensive on 2026 numbers, defensible only if the AI-substrate cycle keeps running through 2027–2028. A patient approach — waiting for next prints — is more reasonable than chasing the move.

1. Start here — why is SEMCO an “AI beneficiary”?

1.1 The invisible infrastructure of AI silicon

In an earlier piece on Samsung Electronics, I called HBM “the workbench of AI silicon.” But even the best workbench (memory) and the best computer (GPU) won’t work if the power supply is unstable (the system crashes) or if the chip isn’t properly wired to the motherboard (data flow breaks).

What SEMCO makes is exactly this kind of “invisible infrastructure.”

What SEMCO parts do inside an AI server:

1. MLCC — the tiny power dam
 → Stores electricity for an instant and releases it smoothly
 → AI chips spike in power draw under compute load;
 MLCCs absorb those swings
 → A single AI server uses tens of thousands of MLCCs

2. FC-BGA — the chip's foothold
 → A high-density circuit substrate that connects AI chips
 (GPUs, CPUs) to the motherboard
 → Carries both power delivery and signal between chip and board
 → As AI chips get larger and more complex,
 substrates get bigger and stack more layers

3. Camera modules — the eyes
 → Cameras for phones, cars, robots
 → Not directly AI-driven, but a real option as the franchise
 expands into automotive and robotics

1.2 Why is supply tight right now?

In the Samsung Electronics piece, I noted that HBM is “absorbing fab capacity.” A similar phenomenon is unfolding in MLCC and FC-BGA.

MLCC:
→ AI servers draw a lot of power and swing hard
→ MLCC content per server has jumped
→ At the same time, xEV demand is also exploding
→ Supply can't be added quickly — high-voltage, high-temperature,
 high-reliability parts take long qualification cycles
→ Prices are starting to rise

FC-BGA:
→ Bigger AI chips need bigger substrates
→ Bigger substrates → more layers → lower yields
 (the percentage of substrates produced without defects falls)
→ Larger panels also fit fewer units per processing line
→ Greenfield substrate fabs take 2–3 years to build
→ Demand is rising now; meaningful new supply is a 2028+ event
→ From now through 2027 is the "supply-short" window

One-line summary: SEMCO isn’t an AI chipmaker; it makes the power and connectivity parts AI chips cannot run without — and those parts are in shortage today.

2. Dissecting the three divisions

2.1 SEMCO’s structure

Division	Key products	2025 revenue	Mix	2025 profit	Profit mix	Margin
Component	MLCC, inductors, chip resistors	KRW 5.20T	46%	KRW 609.4B	67%	11.7%
Package Solution	FC-BGA package substrates	KRW 2.30T	20%	KRW 135.2B	15%	5.9%
Optics	Camera modules	KRW 3.81T	34%	KRW 168.7B	18%	4.4%
Total		KRW 11.31T	100%	KRW 913.3B	100%	8.1%

Role-split inside SEMCO:
Component (MLCC) = defends profit (67% of operating profit)
Package Solution (FCBGA) = moves the share price (+45% YoY growth)
Optics (cameras) = builds top-line scale + future optionality

3. Component (MLCC) — the profit moat

3.1 What is an MLCC?

MLCC = Multi-Layer Ceramic Capacitor. The name is technical, but the job is simple — it stores electricity for an instant and supplies it smoothly to nearby chips.

Analogy: a water tank in a plumbing system

If tap water surges and dips, the flow is unstable.
A tank stores some water and releases it evenly.

MLCC does the same with electricity:
If current surges and dips, the chip misbehaves.
The MLCC briefly stores charge and releases it smoothly.

Size: smaller than a grain of rice (0.1mm–3mm)
Count: \~1,000 in a smartphone, tens of thousands in an AI server

3.2 Why does SEMCO MLCC matter?

MLCC market structure:
#1 Murata (Japan) \~33% share
#2 Samsung Electro-Mechanics \~23%
#3 Taiyo Yuden (Japan)
#4 Yageo (Taiwan)

→ The top four control most of the market
→ High entry barrier: hundreds of micrometer-thin ceramic layers
 stacked uniformly, fired at high temperature without defects
→ SEMCO holds up to 600-layer stacking technology

Why AI-server MLCC is especially hard:
→ Higher voltage, capacitance, and thermal tolerance than phone parts
→ Long qualification cycles; field failure can crash a whole server
→ The list of qualified suppliers is narrow
→ That narrowness translates into pricing power and margin

3.3 1Q26 print and outlook

1Q26 Component revenue was KRW 1.41T (+16% YoY), with utilization at 93% (vs. 70% Package and 66% Optics — the highest in the company).

Management said high-end MLCC demand for AI servers and data centers remains strong into 2Q. Both MLCC shipments and average selling price (ASP) are expected to rise quarter on quarter.

The most important change: selected MLCC products are starting to see price increases. Murata and Taiyo Yuden reported book-to-bill ratios of 1.36 and 1.31 — the highest in five years (post-2021). “BB > 1” simply means orders exceed shipments; the gap is now wide enough to talk about real undersupply.

4. Package Solution (FC-BGA) — the body of the rerating

4.1 What is FC-BGA?

FC-BGA = Flip Chip Ball Grid Array — a high-density circuit substrate that connects AI chips (GPUs, CPUs) to the motherboard.

Analogy: the chip's foothold

AI chips are small and dense, with tens of billions of transistors.
You can't solder them straight onto the server motherboard.
You need an intermediate "foothold" that delivers power and signal
between chip and board. That foothold is the FC-BGA substrate.

Why it's hard:
→ Bigger AI chips → bigger substrates
 (server substrates are >4× the area of consumer PC substrates,
 with 20+ layers)
→ Bigger substrates warp → chips lose contact
→ More layers must align precisely → yield drops sharply
→ Fine traces must be etched at 5–10 micrometers
 (about 1/10th the width of a human hair)

SEMCO became the first Korean company to mass-produce server FC-BGA,
in October 2022. In 2025, it publicly announced a data-center
substrate supply partnership with AMD.

4.2 Why this is “the body of the rerating”

The reason SEMCO went from KRW 830,000 to KRW 1,024,000 is Package Solution.

1Q26 Package Solution:
Revenue: KRW 725B (+45% YoY, +12% QoQ)
Growth drivers: high-end FC-BGA for AI accelerators, server CPUs,
 networking

Why the market got excited:
1. Customers want more than current capacity
 → demand > supply → pricing power sits with SEMCO

2. New big-tech customer awarded an AI-data-center networking
 substrate program
 → more customers = bigger future revenue

3. Long-term agreement (LTA) discussions are underway
 → not a one-quarter spike — structural demand → multiple expansion

4. Management signaled 2026 capex could double versus 2025
 → ramping investment = a confidence signal that demand is real

4.3 What only SEMCO can do

Typical substrate maker: makes FC-BGA only
SEMCO: makes FC-BGA AND MLCC

Why this matters:

AI servers reduce power loss by placing power-delivery components
directly under the chip. That means stacking — or embedding — MLCCs
on or inside the substrate.

Because SEMCO makes both, it can develop "MLCC-in-substrate"
integration internally.

Murata (MLCC-only) and Ibiden (substrate-only) can't replicate this
in-house.

→ This dual capability is potentially SEMCO's most durable long-run moat.

4.4 What is not yet proven

Package Solution’s 2025 operating margin was just 5.9%. Revenue grew +13% YoY, but profit fell -14%. Why? Raw-material costs (CCL, prepreg) rose +19% while average selling prices rose only +4%.

The key question: in 2026, will revenue growth come with margin recovery? If only the top line grows and margins stagnate, the current price is hard to justify. Whether 2Q Package Solution margin lifts above 12% is the single biggest gate for the rerating to continue.

5. Optics (camera modules) — big revenue, not the protagonist

Optics is SEMCO’s smartphone- and automotive-camera-module business. 2025 revenue KRW 3.81T (34% of company), margin 4.4%.

State of play:
→ Revenue is large (a third of the company)
→ Margin is the lowest in the portfolio (4.4%)
→ Sensitive to the smartphone flagship cycle
→ Not the protagonist of the rerating

Future optionality:
→ ADAS cameras for EVs (number per vehicle is rising)
→ In-cabin driver-monitoring cameras
→ Vision cameras for robots and humanoids
→ Management flagged a robotaxi product ramp starting in 2Q

Current market share: \~9% (down from \~11% in 2024)
→ Real technology, but not the rerating thesis

Investor view: Optics adds revenue stability but does not move the stock. MLCC and FC-BGA do. Automotive and robotics cameras are long-dated options — not something to pay for at today’s multiple.

6. Valuation — already expensive, but for a reason

6.1 Where the stock sits

As of May 15, KRW 1,024,000 / share, market cap ~KRW 76.5T. Up +23% from late April.

Reference	PER	Reading
Trailing 12 months	~110x	Optically very expensive
2026E earnings	~58x	Once profit recovery is reflected, lower
2027E earnings	~38x	Only makes sense if the AI-substrate cycle runs into 2027

6.2 Sell-side target prices have run

Feb : Samsung Sec KRW 450,000 (on 2027 BPS)
Mar : Hana Sec KRW 550,000 (2026 EPS × 37.5x)
Early-Apr: SK Sec KRW 600,000 (global peer PER 40x)
Mid-Apr: Hana Sec KRW 810,000 (2027 EPS × 35x)
Early-May: Hana Sec KRW 1,000,000 (2027 EPS × 40x)
Mid-May: KB Sec KRW 1,400,000
 NH IB / SK Sec KRW 1,500,000

→ Target prices more than tripled in roughly three months.

Why so fast? The valuation frame itself shifted. SEMCO used to trade as a “Korean electronic-parts maker.” Today, the sell side benchmarks it against global AI-substrate and passive-component peers — Japan’s Ibiden, Taiwan’s Unimicron, Japan’s Murata. Apply those multiples and targets jump.

6.3 Scenario-based fair value

Scenario	2027E EPS	Applied PER	Fair value	vs. spot
Bear (MLCC price hike contained, FC-BGA capex weighs)	KRW 22,000	33x	KRW 730,000	-29%
Base (AI cycle persists + Package margin recovers)	KRW 26,800	43x	KRW 1,150,000	+12%
Bull (MLCC + FC-BGA price hikes together + LTA)	KRW 30,000	50x	KRW 1,500,000	+47%

Today’s KRW 1.02M offers +12% upside in the base case — and -29% downside if the bear case prints, +47% upside if the bull case prints. At this price, waiting for the next earnings confirmation is more reasonable than chasing the rally.

6.4 The key framing: not “cheap,” but “expensive without continuous earnings upgrades”

Where SEMCO sits today:
"Is it a good company?" → Yes
"Is the stock cheap?" → No
"So I shouldn't buy?" → Not necessarily — conditionally yes

Conditions:
→ 2Q operating profit above KRW 400B (vs. consensus KRW 378.8B)
→ MLCC price hike spreading from distributors to direct customers
→ Package Solution margin moving above 12%
→ LTA or customer pre-investment formalizing

Two of these = the buy thesis works at today's price.
Zero of these = waiting for a pullback is the right call.

7. What to watch from here

7.1 Earnings triggers

Trigger	Bull line	Bear line	When
2Q operating profit	above KRW 400B	below KRW 360B	Late July
MLCC price hike	Spreads to direct AI customers	Stays at cost pass-through	2Q
FC-BGA utilization	Full utilization in 2H	New-customer ramp delays	3Q
Package Solution margin	12%+ in 2Q	Below 10%	Late July
MLCC channel inventory	~4 weeks (tight)	6+ weeks (loosening)	Quarterly

7.2 Multiple triggers

Trigger	Why it matters
Sustained AI infrastructure capex	The end demand for both FC-BGA and MLCC
Customer long-term agreements (LTAs)	Proves “not one-off” — structural demand
Murata / Taiyo Yuden BB ratio	Leading indicator of MLCC cycle (5-year high today)
Global substrate capacity plans	How long the supply shortage lasts
Glass Core substrate JV	A next-generation packaging option (2027+)

7.3 Risks to keep in view

Risk	Description
Already expensive	2027 PER ~38x. If earnings estimates roll over, the stock falls hard
FC-BGA capex burden	Doubling capex doubles depreciation; revenue must follow or margin compresses
Pass-through failure	If CCL, gold, copper inflation can’t be passed on, margins suffer
Smartphone weakness	Camera modules and some MLCC SKUs remain phone-cycle exposed
AI capex peak	If AI server investment slows, FC-BGA + MLCC demand bases weaken

8. How this links to other posts

Samsung Electronics piece: "HBM is the workbench of the AI server"
→ SEMCO: "MLCC is the power stabilizer, FC-BGA is the chip's foothold"

Jeju Semiconductor piece: "Commodity DRAM is short because of AI"
→ SEMCO: "Power components and chip substrates are short because of AI"

Robotics value-chain piece: "Part 1 flagged SEMCO ramping a
humanoid camera in 2H"
→ SEMCO: that ramp lives inside the Optics optionality bucket

All three sit on the same logic:
AI capex → shortage of components, materials, substrates →
pricing power for the bottleneck supplier.

9. The one-line bottom line

SEMCO is the invisible infrastructure layer of AI silicon. MLCC stabilizes power; FC-BGA connects the chip to the board. Both are short of supply. MLCC prices have started to rise; FC-BGA demand has overtaken capacity. If this structure runs through 2027, SEMCO gets rerated from “smartphone parts maker” to “AI infrastructure parts platform.”

But at KRW 1.02M, much of that thesis is already in the price. On 2027 earnings, PER is ~38x — “great company” and “great entry price” are different questions. To chase the move from here, the numbers have to confirm: 2Q operating profit above KRW 400B, MLCC price hikes spreading, FC-BGA margins recovering.

One-line: this is not a cheap stock — it is a stock that is only “not expensive” if earnings upgrades keep coming. The pace of revision matters more than the headline news flow.

This article is research and commentary only and is not investment advice. Divisional revenue and profit figures are based on Samsung Electro-Mechanics’ 2025 annual report and 1Q26 earnings release. The 1Q26 operating profit of KRW 280.6B includes a one-off severance charge of KRW 71.4B. Utilization figures (Component 93%, Package 70%, Optics 66%) are 2025 full-year averages. The ~23% MLCC market share figure is the company’s own estimate. The AMD substrate partnership is per SEMCO’s official disclosure. The new big-tech customer name is not publicly disclosed. The valuation scenarios (Base KRW 1.15M, Bull KRW 1.50M, Bear KRW 730K) are the author’s estimates and may be wrong. Sell-side targets (NH / SK KRW 1.5M, KB KRW 1.4M, Hana KRW 1.0M, etc.) reflect each broker’s report and are subject to change. Glass-core JV mass production is a 2027+ plan and not confirmed. MLCC price hikes are at an early stage in select distributor channels; broader pass-through is not yet confirmed. The analysis may be wrong. Data cut-off: May 15, 2026 KST.

Samsung Electronics Strike vs Memory Supercycle — Can Higher Prices Really Offset the Walkout? The Real Question Is 'Who Captures the Excess Profit'

Fri, 15 May 2026 00:00:00 +0000

📚 Samsung Electronics series Previous: Samsung Electronics and Citi’s KRW 460,000 target — starting from what HBM actually is

Samsung Electronics’ union has announced an 18-day general strike from May 21 — up to 50,000 participants, the largest in the company’s history. At the same time, memory prices are exploding: 2Q DRAM contract +58–63%, NAND +70–75%. JP Morgan calls it “price gains offset the strike.” That logic is only half right. The real fight is not “will the fab stop?” but “in this memory supercycle, who captures the excess profit — labor or shareholders?”

Key takeaways

Where we stand: Samsung Electronics union has notified a general strike for May 21 – June 7. Core demands: a fixed payout of 15% of DS (semiconductor) division operating profit as performance bonus, plus removal of the 50%-of-base-salary cap. Defenses still in play: May 16 NLRC mediation, court injunctions, and the government’s emergency arbitration power.
Loss estimates: JP Morgan KRW 21–35 trillion; industry estimates cited by Yonhap range up to KRW 100 trillion. The wide spread reflects unknowns in participation, line disruption, and restart time.
Price-offset logic: with DRAM +58–63% and NAND +70–75% in 2Q, a 1–2% production hit is mathematically offsettable. JP Morgan keeps its KRW 350,000 target on this basis.
But the offset is not complete: (1) structural labor cost from a bonus formula can’t be papered over by price; (2) if customers shift orders to SK hynix or Micron, Samsung captures less of the price rally.
The real question: not “strike or no strike,” but “how the profit-sharing formula changes — that matters more for medium- and long-term valuation."

1. What is happening right now

1.1 Scale of the strike

The union notified an 18-day general strike from May 21 to June 7. Expected participation up to 50,000 employees — the largest strike risk Samsung Electronics has ever faced.

NLRC mediation sessions on May 11–12 collapsed in the early hours of May 13. The union’s position: “without concrete institutionalization, there is no reason to talk.”

1.2 What the union actually wants — not a simple pay hike

Core union demands:

1. A FIXED bonus = 15% of DS (semiconductor) operating profit
 → Current: bonus set at management discretion (EVA-based OPI)
 → Demand: 15% of operating profit codified as a formula

2. Remove the 50%-of-base-salary cap
 → Current: total bonus cannot exceed 50% of annual base salary
 → Demand: no cap. If profit is large, bonus scales accordingly.

3. Expand OPI stock-based compensation

Why this matters:
→ Samsung 1Q DS operating profit was KRW 53.7 trillion
→ 15% = \~KRW 8.1 trillion of bonus pool
→ Even 10% = \~KRW 5.4 trillion
→ The gap vs. the existing OPI scheme is measured in trillions

1.3 Why this issue blew up

First, the comp gap with SK hynix. Following its HBM success, SK hynix has materially expanded employee bonuses. Inside the same memory industry, Samsung employees feel relatively underpaid.

Second, the profit pool got too large to ignore. 1Q DS operating profit alone was KRW 53.7T at a 65.7% margin. When excess returns are this large, the question “who gets it” becomes politically unavoidable.

Third, fab production is uniquely fragile. A semiconductor fab runs 24/7 continuous flow. Stops mean scrapped wafers, lower yields, and long restart times. Even the threat of a strike already affects operations. Press reports indicate Samsung has begun a “warm-down” — cutting new wafer starts in anticipation.

2. How big is the loss — and why the spread is so wide

2.1 The estimates

Source	Estimate	What it captures
JP Morgan	KRW 21–35T	Labor cost + production disruption
JP Morgan revenue opportunity loss	~KRW 4.5T	Direct revenue loss only
Industry estimates (via Yonhap)	KRW 40T direct, up to KRW 100T including indirect	Worst case

2.2 Why the range is so wide

Unknown variables:
1. Real participation rate — of the 50k, how many actually walk out
2. Critical line staff — do DS process engineers join
3. Essential personnel — does the court injunction carve out safety
 and critical-line workers
4. Duration — full 18 days, or settled in a few days
5. Restart time — once a line stops, how long to bring it back

Quick sanity check:
Yonhap's reference: "about KRW 2.2T loss per day"
= KRW 40T direct ÷ 18 days = KRW 2.2T / day

2018 Pyeongtaek power outage: \~KRW 50B for 28 minutes
→ Day-rate equivalent \~KRW 2.6T
The two numbers are the same order of magnitude — a useful cross-check,
though the 2018 outage and a 2026 strike are structurally different,
so don't apply it mechanically.

3. Can price gains really offset the strike — JP Morgan’s logic

3.1 What JP Morgan said

In its May 13 note, JP Morgan (Jay Kwon’s team) maintained Overweight, target KRW 350,000. Core argument: “memory prices are stronger than expected and can absorb much of the strike loss.”

Item	JP Morgan
2Q DRAM contract price	+58–63%
2Q NAND contract price	+70–75%
Prior expectation	+40–50%
Production disruption (% of DS revenue)	1–2%

3.2 Does the math actually work

Memory revenue = shipments × average selling price (ASP)

If the strike reduces shipments,
revenue is defended whenever ASP gains exceed the volume loss.

Offset condition:
(1 + ASP growth) × (1 - volume loss) ≥ 1

If production loss is 1–2%:
→ ASP growth needed to defend revenue ≈ 1–2%
→ Actual ASP change: DRAM +58–63%, NAND +70–75%
→ Price gains dwarf the needed amount by tens of times

∴ For DIRECT production loss alone, the math offsets cleanly.

The more important calculation — the “excess” price gain vs prior expectations:

Prior expectation: DRAM +45%, NAND +45% (midpoint)
Updated view: DRAM +60.5%, NAND +72.5% (midpoint)

Incremental ASP:
DRAM: 1.605 / 1.45 - 1 = +10.7%
NAND: 1.725 / 1.45 - 1 = +19.0%

→ DRAM ASP is \~11% stronger than the prior model;
 NAND ASP is \~19% stronger.
→ This delta is what offsets the strike loss.

3.3 What’s right and what’s missing

Right: direct production loss (~1–2% of DS revenue) is comfortably covered by price gains. 2Q estimates could actually move up. For short-term earnings, JP Morgan’s logic holds.

Three things it misses:

1. Structural labor cost is not a "price" problem.
 If the demand for "15% of DS operating profit as a fixed bonus"
 is codified, then every future quarter has 15% of operating profit
 bleed into compensation. That's not offset by higher prices —
 prices and labor cost rise together.

 At 1Q levels: 15% = KRW 8.1T
 → This is not "offset by price." It's a SHIFT in how profit
 is divided between labor and shareholders.

2. If customers shift orders, the price rally bypasses Samsung.
 HBM, high-capacity server DRAM, enterprise SSD — once a customer
 moves a qualified supplier slot to SK hynix or Micron, it is
 slow and expensive to move it back.

 Samsung strike → delivery uncertainty → customers re-allocate
 to SK hynix / Micron
 → Memory prices keep rising, but Samsung captures less of it

 TrendForce flagged exactly this: order shift, not direct loss,
 is the bigger risk.

3. The memory cycle eventually turns.
 A fixed-ratio bonus is fine in an upcycle — and rigid in a downcycle.
 That's a long-run cost-flexibility loss → structurally lower ROIC.

4. Will the strike actually happen — there are firebreaks

4.1 Key events over the next week

Date	Event	Significance
May 15, 10:00	Deadline for the CEO’s response to the union	Tone can pivot here
May 16	NLRC follow-up mediation reopens	Last formal negotiation window
Around May 20	Court ruling on injunction	Whether essential workers are excluded
May 21	Strike start date	Actual participation rate is the swing factor

4.2 A full shutdown is unlikely

Firebreak 1: NLRC mediation (May 16)
→ Last formal chance to settle — both sides have face-saving incentives

Firebreak 2: Court injunction (around May 20)
→ Carving out safety and essential-process staff keeps critical lines running

Firebreak 3: Government emergency arbitration
→ If invoked, forces a 30-day suspension of strike action
→ Industry minister has publicly said "if the strike happens, it's unavoidable"

Firebreak 4: Both sides' own incentives
→ Union: 18 full days off-payroll hurts members too
→ Company: a shutdown damages customer trust, the share price, and exports
→ Government: semis were 37% of April exports — national economic stakes

Base case: a full shutdown is likely to be averted. But “no shutdown” and “no risk” are different statements.

5.1 Three scenarios and price impact

Scenario	Description	Price impact
A. Settlement / suspension	Strike averted via NLRC mediation or government intervention. Bonus stays under existing OPI + special payout	Short-term bounce — risk premium removed, memory cycle back in focus
B. Partial strike	Some staff walk out; critical lines maintained; participation modest	Neutral to slightly negative — noise persists, earnings impact limited
C. Major strike	DS critical staff participate; warm-down deepens; customers re-allocate	Material drawdown — estimate cuts plus multiple de-rating

5.2 More important than the scenario — the SHAPE of the bonus deal

Even after the strike ends, the question stays:
"How was the bonus institutionalized?"

Best case (shareholder-friendly):
→ Keep existing OPI framework + cycle-driven discretionary top-up
→ Workers get a share of excess profit, but not a fixed ratio
→ No cost rigidity in a downcycle

Worst case (shareholder-unfriendly):
→ 15% of DS operating profit codified as a fixed bonus
→ Every quarter, 15% of operating profit flows to comp before shareholders
→ Fine in an upcycle; in a downcycle, harder to defend earnings
→ Structural decline in long-run ROIC

How this formula gets decided matters more for Samsung’s medium- and long-term valuation than whether the strike happens. A strike that ends in a day still locks in lower shareholder economics if the bonus is fixed at 15%.

6. Where does the stock actually sit

6.1 Earnings are very strong

As covered in the prior post:

1Q revenue KRW 133.9T, operating profit KRW 57.2T
DS operating profit KRW 53.7T, margin 65.7%
Annualized operating profit ≈ KRW 228.8T → trailing-style PER ~9.6x at the current price

6.2 Sell-side outlook

Broker	2Q operating profit	Target price
Kiwoom Securities	KRW 100T	KRW 330,000
Yuanta Securities	KRW 86.8T	KRW 340,000
JP Morgan	—	KRW 350,000

6.3 Where the price sits

May 14 close: KRW 296,000. Implied upside to Kiwoom’s target = +11.5%.

What the current price reflects:
→ The memory supercycle (1Q margin 65.7% confirms it)
→ Strike risk — partially (the "price offsets" hope is baked in)
→ Structural change to bonus formula — NOT priced

Upside +11.5% against a real event risk:
→ The asymmetry for a new buy is weak
→ Target hit = +11%; strike escalation = -10–15% plausible
→ This is a "great company," not a "great entry price"

7. What to watch — checklist

#	Event	What to check
1	May 15, 10:00 CEO response	Union reaction — chance to reopen talks
2	May 16 NLRC mediation	Does a middle-ground bonus proposal emerge
3	Around May 20 Court ruling	Scope of essential-worker carve-out
4	May 21 Strike start	Actual participation, DS critical-line impact
5	After the strike	Wafer starts, shipment delays, customer allocation changes
6	Bonus settlement	Fixed-ratio vs discretionary — drives long-run ROIC

8. How this links to other posts

Samsung Electronics part 1 (Citi target KRW 460,000):
→ "The memory-cycle frame is wrong — AI is creating structural demand"
→ If that thesis still holds, the strike is noise.

Samsung Electro-Mechanics piece:
→ "MLCC / FC-BGA are also AI-infrastructure components in shortage"
→ A Samsung strike that cuts chip supply → DRAM / NAND price up
→ Indirectly tightens demand for SEMCO AI-server MLCCs as well

Jeju Semiconductor piece:
→ "Commodity DRAM is short because of AI"
→ A Samsung strike → further legacy DRAM supply tightness → prices up
→ Could actually help Jeju Semiconductor

US-China summit piece:
→ "We're in the regime where you have to find the least-priced exposure"
→ Samsung strike risk could translate into relative SK hynix upside

9. The one-line bottom line

The Samsung Electronics strike is not really about “does the fab stop.” It is about “who captures the excess profit of the memory supercycle — workers or shareholders."

In the short term, JP Morgan’s logic holds: with DRAM +60% and NAND +73%, a 1–2% production loss is mathematically offset. But if 15% of DS operating profit gets locked in as a fixed bonus, that is not a one-quarter cost to “offset” — that is a change in how profit is divided. And if customers move HBM and server-DRAM allocations to SK hynix or Micron, Samsung captures less of the rally, even though prices keep rising.

A full shutdown is unlikely — NLRC mediation, court injunctions, and government emergency power all sit as firebreaks. But “no shutdown” and “no risk” are not the same thing. Holders need not overreact; new buyers are better off waiting for the May 16 mediation outcome and the May 21 strike decision.

The need today is event-risk management, not courage.

This article is research and commentary only and is not investment advice. Samsung Electronics 1Q figures (revenue KRW 133.9T, operating profit KRW 57.2T, DS operating profit KRW 53.7T) are from the company’s official newsroom disclosure. The strike notice (May 21 – June 7, 18 days) and union demands (15% of DS operating profit as a fixed bonus) are based on Yonhap and Yonhap Infomax reporting. JP Morgan’s analysis (target KRW 350,000, KRW 21–35T impact) is per Money Today and Reuters coverage; full text of the note has limited public access. DRAM +58–63% and NAND +70–75% are TrendForce estimates. Industry estimates of KRW 40T direct / up to KRW 100T including indirect are per Yonhap and reflect a worst-case scenario. Emergency arbitration authority rests with the Minister of Employment and Labor under Korea’s Trade Union Act. Strike participation rate, line disruption, and customer allocation changes are not knowable at this point. The analysis may be wrong. Data cut-off: May 15, 2026 KST.

Two Betas of AI Back-End — Substrates Are a 'Volume Beta,' Test Sockets Are a 'Consumables Beta.' Same AI Tailwind, Completely Different Structures

Fri, 15 May 2026 00:00:00 +0000

📚 AI back-end series Previously: Samsung Electro-Mechanics MLCC and FC-BGA deep dive, Jeju Semiconductor legacy-memory thesis Read next: AI back-end — 11 stocks side by side on 2027E PER × OP growth

Who actually makes money when AI silicon sells? Nvidia (GPU) and SK hynix (HBM) are the names that come to mind first. But the “back end” of the supply chain has major winners too. We call it AI back-end. Two regions matter most: substrates and test sockets. Both are parts that an AI chip passes through before completion — but the investment structures are completely different. Substrates are a bet on “AI server units.” Test sockets are a bet on “AI chip complexity.” Margins are roughly 3x higher in test sockets, and momentum is faster in substrates. Which is the better investment depends on how long you plan to hold.

Key takeaways

Two regions in AI back-end: substrates and test sockets.
Substrate essence: a “direct CAPEX beta” on AI-server build-out. More units → more revenue; larger packages → higher ASP.
Test-socket essence: a “high-margin consumables beta” on chip complexity. More complex chips → harder testing → more / harder sockets.
Profitability gap: on 1Q26, Daeduck Electronics 14.8% vs ISC 35% vs LEENO Industrial 47.4%. Test sockets dominate.
Momentum gap: substrate earnings revisions are faster — ASP hikes, new big-tech LTAs, capacity tightness are all near-term catalysts.
Cycle risk: substrates are a CAPEX cycle (shortage → CAPEX → glut). Test sockets, being consumables, run with lower cyclicality.
Bottom line: short-term momentum trades → substrates (Daeduck Electronics, Samsung Electro-Mechanics). 1–2 year compounding → test sockets (LEENO Industrial, ISC). These are not the same “AI theme.”

1. Setup — what substrates and test sockets actually are

1.1 How AI silicon gets made

Production flow of AI silicon:

1. Design (Nvidia, AMD, Google, Meta, etc.)
2. Wafer fab (TSMC, Samsung Foundry)
3. Dicing
4. Packaging (HBM, interposer, package substrate join)
 ← "substrates" used here
5. Test (performance / reliability / burn-in)
 ← "test sockets" used here
6. Shipment

Substrate = "the pedestal the chip sits on"
Test socket = "the slot a chip plugs into briefly for testing"

Both are parts AI silicon passes through before completion,
but their roles and usage patterns are completely different.

1.2 Substrate — what is it

A substrate connects the microscopic circuitry inside a chip
(nanometer scale) to the outside world (millimeter scale).

What it does:
1. Electrical connection (thousands of pins → board)
2. Mechanical support (stable mounting of large packages)
3. Thermal management (heat-flow path away from the chip)
4. Signal integrity (high-speed signals without loss)

Why AI matters:
- AI chips are much larger than typical silicon
 (Nvidia H100 = 814 mm², 2–3x a typical CPU)
- Thousands to tens of thousands of pins
- High-speed data movement
- Heavy heat budget (700W+)

→ Specialized substrates (FC-BGA) are required
→ Harder to make and more expensive than ordinary substrates
→ Korean players: Samsung Electro-Mechanics, Daeduck Electronics,
 Isu Petasys.

1.3 Test socket — what is it

A test socket is a slot used briefly to test a finished chip.

What it does:
1. Electrically connects every chip pin to the tester
2. Verifies the chip works under signal / power / temperature stress
3. Filters out failing units
4. Reliability validation (long-duration testing)

Analogy:
Chip = car
Test socket = the diagnostic port at the inspection bay
→ Connected briefly during inspection, then disconnected
→ The same port is used for many different cars
→ Ports wear out over time and have to be replaced

Why AI matters:
- AI chips are expensive (\~USD 30k for an H100)
- A single bad unit costs a lot → test intensity rises
- High-current / high-speed / high-temperature environments
 → tougher socket spec
- Each chip generation needs new sockets → recurring revenue

Korean players: LEENO Industrial, ISC, TSE.

1.4 Why they are not the same “AI theme”

Substrate:
- Made once, shipped with the chip (capital-good character)
- Roughly one substrate per AI server
- Revenue = volume × ASP
- Adding capacity = CAPEX burden

Test socket:
- Reused many times, then replaced (consumable character)
- Roughly 50–200 sockets to test 10,000 chips
- Revenue = chip output × test intensity × socket replacement
- Capacity additions are smaller

Margin structure:
Substrate: OPM \~10–15% (CAPEX, depreciation drag)
Test socket: OPM \~30–50% (custom design, consumable economics)

→ Both benefit from AI,
→ but revenue mix, margin profile, and cycle sensitivity differ.

2. The substrate side — “AI server volume beta”

2.1 What 1Q26 showed

Samsung Electro-Mechanics (009150) — Package Solution segment
(FC-BGA-led):
1Q26 revenue: KRW 725B
YoY: +45%
QoQ: +12%

Daeduck Electronics (353200):
1Q26 revenue: KRW 346.3B (YoY +61%)
1Q26 OP: KRW 51.3B (turn-around to black)
OPM: 14.8%

Product mix:
- FCCSP (mobile-grade package substrate): 39%
- FCBGA (PC / server-grade): 23%
- CSP (memory): 22%
- MLB (multilayer board, AI server): 16%

YoY growth:
- Package substrates: +65%
- MLB: +43%

→ AI-server package substrates + network boards growing together
→ A direct beta on AI infrastructure.

2.2 Why substrates are short

Characteristics of AI-server FC-BGA:
- Surface area 2–3x a PC FC-BGA
- Multilayer (20–30 layers)
- High-speed signaling (PCIe 5.0 / 6.0)
- Strict thermal management

Companies that can make these:
- Korea: Samsung Electro-Mechanics, Daeduck, LG Innotek, Isu Petasys
- Taiwan: Unimicron, Nan Ya PCB
- Japan: Ibiden, Shinko Denki

The problem: capacity (CAPA) can't keep up with demand
- Big-tech AI-server orders have exploded
- Greenfield substrate fabs take 2–3 years to build
- Short supply → ASP hikes

Press cited:
"Samsung Electro-Mechanics FC-BGA demand exceeds capacity by \~50%;
 price-hike negotiations under way."

→ This is why substrate stocks have been strong.

2.3 Substrate cycle risk

Substrate is a classic CAPEX-cycle industry:

Up-cycle (today):
shortage → ASP hike → margin expansion → stock strength
→ companies announce sizable CAPEX

Build phase (2026–2027):
fabs under construction → revenue rising, CAPEX burden rising
→ stocks still trending strong

Build complete (2027–2028):
new fabs come online → supply surges
→ if demand doesn't keep up, prices fall
→ margins squeeze → stocks weaken

Historical patterns:
2017–2018 memory cycle: same shape
2021–2022 MLCC cycle: same shape

→ For substrate investing, "where in the CAPEX cycle" is the key
→ Today: early- to mid-stage of an up-cycle
→ Until build-out completes, upside is intact.

3. The test-socket side — “chip-complexity beta”

3.1 What 1Q26 showed

ISC (095340):
1Q26 revenue: KRW 68.3B
1Q26 OP: KRW 23.6B
OPM: 35%
Math: 23.6 / 68.3 = 34.55%

Revenue mix:
- AI revenue: KRW 55.3B (81% of total)
- Data-center revenue: KRW 54.2B (79% of total)
→ Already an "AI data-center company."

LEENO Industrial (058470):
1Q26 revenue: KRW 99.77B
1Q26 OP: KRW 47.30B
OPM: 47.4%
Math: 47.30 / 99.77 = 47.41%

Revenue mix:
- IC test sockets: 64.10%
- LEENO pins (pogo): 24.65%
- Medical components: 10.46%

→ Custom-designed sockets + vertically integrated pin manufacturing
→ 47% OPM is at the top of Korean manufacturing.

3.2 Why margins are this high

Structural features of the test-socket industry:

1. Customer-specific designs
 → pin layout, size, signal conditions vary by chip
 → hard to standardize → pricing power

2. High qualification cost
 → each new chip requires socket design / test / qualification
 → once qualified, locked in until end-of-life of that chip
 → reliability competition, not price competition

3. Consumable nature
 → sockets wear out / degrade and need periodic replacement
 → recurring revenue

4. Refresh with every chip generation
 → new chip = new socket
 → faster chip cadence = faster revenue cadence

5. Small TAM (a few tens of USD billion)
 → too small for mega-cap entrants
 → oligopoly of specialist players

All five conditions together enable 30–50% OPM.

3.3 Why AI made test sockets more important

AI chip characteristics:

1. Expensive (USD 10k–30k per unit)
 → cost of a single bad unit is high
 → test intensity ↑

2. High pin count (thousands–tens of thousands)
 → more complex sockets
 → ASP ↑

3. High current / temperature / speed
 → tougher socket durability requirements
 → shorter replacement cycle

4. SLT (System-Level Test) share rising
 → from simple functional test → full system test
 → longer test times
 → more socket usage

5. New modules: HBM, SOCAMM2, etc.
 → new socket categories
 → fresh revenue lines

→ In the AI era, test-socket demand grows
 faster than chip output.

3.4 ISC vs LEENO Industrial

Both are "test-socket companies," but their structures differ.

ISC (rubber-socket strength):
- Core tech: silicone rubber sockets
- Strength: AI data-center mass-production testing, SLT
- Customers: global GPU / ASIC majors
- Exposure: AI data center 81%
- Profile: direct AI beta (volatility higher)
- 1Q26 OPM: 35%

LEENO Industrial (pogo-pin strength):
- Core tech: pogo pins, in-house pin manufacturing
- Strength: R&D, mobile AP, RF, ASIC development
- Customers: diversified (hundreds of accounts)
- Exposure: many chip categories (lower AI share than ISC)
- Profile: quality compounder (more stable)
- 1Q26 OPM: 47.4%

→ Don't lump them as "the same test-socket stock."
→ ISC = AI data-center beta
→ LEENO = diversified high-margin platform

They are complements, not substitutes.

4. Head-to-head

4.1 1Q26 operating margins

Same "AI back-end beneficiaries," different margin profiles:

Daeduck Electronics (substrate): 14.8% ████
Samsung E-M Package Solutions: \~12% ███
ISC (test socket): 35.0% █████████
LEENO Industrial (socket): 47.4% ████████████

KRW 100 of revenue → OP:
Daeduck: KRW 14.8
ISC: KRW 35.0
LEENO: KRW 47.4

→ Roughly 3x spread on the same revenue
→ The result of structural industry differences.

4.2 Growth pace vs margin stability

Metric	Substrate (Daeduck)	Test socket (ISC)	Test socket (LEENO)
1Q26 revenue YoY	+61%	Strong YoY	+18%
1Q26 OP YoY	turnaround	Strong YoY	+35%
OPM	14.8%	35.0%	47.4%
AI exposure	Mid–high	Very high (81%)	Mid
Volatility	High	Mid	Low
Cycle sensitivity	High	Mid	Low

Summary:
- Revenue growth: substrate (Daeduck) > test socket
- Margin level: test socket > substrate
- Margin stability: LEENO > ISC > substrate
- Direct AI beta: ISC > Daeduck > LEENO.

4.3 Cycle risk

Substrate cycle risk (high):
- Shortage → CAPEX → glut, in repeat
- 2–3 year build lead time
- When the cycle rolls, utilization and ASP fall together
- Depreciation pressures margin

Test-socket cycle risk (low):
- Consumable economics dampens revenue volatility
- Customer-specific design keeps pricing stable
- New chips keep being released
- Quarterly volatility still exists

Long-hold perspective:
→ Test sockets are far more stable
→ avoids the substrate CAPEX cycle

Short-momentum perspective:
→ Substrate momentum is stronger
→ ASP-hike / capacity-tight headlines are more frequent.

5. Investment priority — different answers for different horizons

5.1 Short-term momentum (3–6 months) — substrates win

Why:
- Earnings-revision pace is faster
- Steady stream of ASP-hike headlines
- Probability of new big-tech LTA
- AI-server shipments accelerating

Order of preference:
1. Daeduck Electronics: turnaround + AI MLB / FC-BGA beta
2. Samsung Electro-Mechanics: AI FC-BGA + MLCC combo
3. Isu Petasys (complement): ultra-high-layer MLB strength

Caveats:
- Substrates have already moved a lot
- Check the CAPEX-cycle position
- Wait for the macro gate to clear (see earlier post).

5.2 Hold 1–2 years (quality growth) — test sockets win

Why:
- Margin structure is structurally superior
- Smaller cycle risk
- AI chip diversification = more socket categories = revenue diversification
- Active capacity expansion (new fabs)

Order of preference:
1. LEENO Industrial: highest quality, but the multiple is rich
2. ISC: direct AI data-center beta
3. TSE (complement): growth at a more reasonable price

Caveats:
- LEENO faces margin-impact risk during the new-fab move
- ISC has quarterly volatility (1Q26 QoQ -6%)
- Both trade at 30–45x PER.

5.3 A combined AI back-end portfolio

Aggressive (momentum tilt):
- Daeduck Electronics 40%
- ISC 30%
- Samsung Electro-Mechanics 20%
- LEENO Industrial 10%

Balanced (growth + stability):
- LEENO Industrial 30%
- Samsung Electro-Mechanics 25%
- ISC 25%
- Daeduck Electronics 20%

Defensive (quality tilt):
- LEENO Industrial 40%
- Samsung Electro-Mechanics 30%
- ISC 20%
- Daeduck Electronics 10%

→ If you must pick one, by horizon and volatility tolerance:
→ 1y+ hold: LEENO Industrial
→ 3–6 months: Daeduck Electronics
→ Direct AI data-center beta: ISC.

6. Four common misconceptions

6.1 #1: “Both are AI plays, so they’re the same”

As shown:
- OPM differs by 3x
- Cycle sensitivity differs
- Revenue mix differs

Grouping them as one bucket weakens diversification.
→ Volatility moves together
→ Pairing with another sector is more effective.

6.2 #2: “Pogo sockets are being replaced by rubber sockets”

Only partially true:

Where the shift happens:
- AI GPU / ASIC large-die chips
- High-current / high-speed signal testing
- SLT (System-Level Test)
→ Rubber-socket penetration rises (ISC strength)

Where it doesn't:
- High-precision R&D testing
- Mobile APs, RF chips
- Small-batch / many-SKU production
→ Pogo retains share (LEENO strength)

→ Not "the whole market converts" — it is "market segmentation"
→ Therefore ISC and LEENO are complements
→ It is rare for only one of them to do well.

6.3 #3: “Substrate stocks have run too much”

The right question is not the absolute level,
it is "where in the CAPEX cycle are we."

Now:
- Shortage still in progress
- ASP-hike negotiations under way
- LTAs with big-tech under discussion
- CAPEX announcements have started, but ramps are 2–3 years away

→ Early- to mid-stage of an up-cycle
→ Price moved, but the cycle has not topped.

That said, from these levels:
- Scaled entries recommended
- Wait for the macro gate to clear
- Chasing is inefficient.

6.4 #4: “The test-socket TAM is too small”

True on the headline number, but the read is wrong:

Test-socket TAM:
- \~USD 3–4 billion
- Less than 2% of the memory market (\~USD 200B)
- Small in absolute terms

Why that's actually a feature:
- Too small for mega-cap entrants
- Sustains the specialist oligopoly
- Price competition stays mild
- 30–50% OPM achievable

Compare:
A USD 200B business at 5% OPM vs
a USD 4B business at 45% OPM
→ Comparable absolute OPDollars
→ The second is more stable and more predictable.

This is the structure LEENO and ISC benefit from.

7. The next six months — checklist

7.1 Substrates (cycle-intact confirmation)

Positive signals:
□ FC-BGA ASP hikes continue
□ Samsung E-M / Daeduck add new big-tech customers
□ Large-die / high-layer mix lifts in revenue
□ MLB demand persists at 800G / 1.6T networking
□ Utilization stays above 90% even with CAPEX rising

Negative signals:
□ ASP hike negotiations delayed / fail
□ Big-tech AI server orders decelerate
□ Pace of new CAPEX announcements accelerates (glut concern)
□ Utilization drops below 80%

Frequency: quarterly earnings + mid-quarter IR commentary.

7.2 Test sockets (growth-intact confirmation)

ISC:
□ 3Q26 revenue re-acceleration (2Q may be a ramp-prep quarter)
□ New data-center infrastructure customer first-volume ramp
□ SOCAMM2 mass-production testing revenue begins
□ SLT mix stays above 70%
□ AI revenue mix stays above 80%

LEENO Industrial:
□ Progress on the new-fab move
□ No margin damage during the move (OPM 45%+ holds)
□ Customer diversification (Apple / TI / HPC / ASIC expansions)
□ Strong R&D socket demand
□ No overhang / governance issues

Frequency: quarterly earnings.

7.3 The macro backdrop

The macro gate from the earlier post:
- US 10-year below 4.45%
- Brent below 105
- USD/KRW below 1,480
- VIX below 18

These must clear for:
- Broad risk-asset recovery
- Back-end stocks to ride the trend
- Good earnings to translate into good price.

If the gate doesn't clear:
- Multiples stay pressured independent of fundamentals
- Wait for confirming close / turnover before scaling in.

8. How this connects to other posts

Samsung Electro-Mechanics piece:
→ MLCC + FC-BGA 1Q26 dual beneficiary
→ The most detailed stock covered in this article's substrate side
→ Scenario PERs for how far the multiple can stretch

Jeju Semiconductor piece:
→ "Commodity memory in shortage because of AI"
→ Another AI back-end shape (memory shortage beta)

Samsung Electronics strike piece:
→ Key variable for the memory supercycle
→ AI silicon → AI server → memory / substrate / test socket
→ Disruption in one ripples into the back end

KOSPI crash + macro-gate piece:
→ "Cycle before the stock"
→ Names in this post also rational only after the macro gate clears.

9. The one-line bottom line

When AI silicon sells, the first beneficiaries are the GPU and HBM makers. But the back end also has two big winners — substrates and test sockets.

Both are grouped as “AI back-end,” but the structures are completely different. Substrates are an “AI server volume beta." Volumes rise, packages get bigger, ASPs lift — direct upside. Momentum is fast and operating margins expand from a ~12–15% base in 1Q26. But this is a CAPEX-cycle industry, and the build-out-complete date is the risk.

Test sockets are a “chip-complexity beta." As chips get more complex, testing gets harder, and new sockets are needed each time. OPM sits at 35% for ISC and 47% for LEENO — roughly 3x substrate margins. Consumable economics dampen cyclicality. The downsides: the TAM is small and multiples are already rich.

Don’t lump them as one “AI theme." For short-term momentum, substrates (Daeduck Electronics, Samsung Electro-Mechanics) lead. For 1–2 year holding, test sockets (LEENO Industrial, ISC) make more sense. The best play is to hold both as complements — when substrates shake at a cycle peak, test sockets defend; when test sockets feel multiple pressure, substrates’ momentum carries.

Same AI tailwind, completely different structures — understanding just that lifts the quality of back-end investment decisions by a full step.

This article is research and commentary only and is not investment advice. Samsung Electro-Mechanics 1Q26 figures are per the company’s official IR release. Daeduck Electronics 1Q26 (revenue KRW 346.3B, OP KRW 51.3B, OPM 14.8%) are per its IR materials. ISC 1Q26 (revenue KRW 68.3B, OP KRW 23.6B, OPM 35%, AI revenue mix 81%, data-center mix 79%) are per its IR materials. LEENO Industrial 1Q26 (revenue KRW 99.77B, OP KRW 47.30B, OPM 47.4%) is per preliminary disclosure reporting; product-mix figures (IC test sockets 64.10%, LEENO pins 24.65%, medical components 10.46%) are per the quarterly report. OPM uses operating profit divided by revenue, rounded to one decimal place. AI / data-center revenue shares are derived from company disclosures. The operating-margin comparison reflects a single quarter (1Q26); annual averages may differ. CAPEX-cycle risk, new-fab transition margin risk, and AI demand volatility are author judgments and not certainties. Global macro variables (US rates, oil, FX, VIX) can independently move the stocks. The analysis may be wrong. Data cut-off: May 15, 2026 KST.

What Pearl Abyss's DLC Comment Really Means — Not the KRW 60B in Sales, but the End of the '2027 Cliff' Discount

Fri, 15 May 2026 00:00:00 +0000

📚 Pearl Abyss series Previously covered: post-launch reception, 1Q26 earnings summary, Crimson Desert revenue-recognition structure Read next: Patch 1.07 weekend data check — global #18 recovered, 6M ETA May 27–29

Pearl Abyss printed a record 1Q26: revenue KRW 328.5B, operating profit KRW 212.1B. Yet the stock has been weak. On the same earnings call the company said it is “exploring next-stage expansion including DLC” for Crimson Desert. Foreign outlets read that as “DLC confirmed,” the market twitched, and then drifted back lower. The meaning of that DLC line is widely misread. The headline is not “KRW 60B in DLC revenue.” The headline is that this comment removes the “2027 cliff discount” the market has been applying to Pearl Abyss. That moves the multiple from single-digit PER toward the global game-publisher average (15–25x) — and that reclassification matters far more than the dollars do.

Key takeaways

Where we stand: Pearl Abyss printed record 1Q26 (revenue KRW 328.5B, operating profit KRW 212.1B), yet the stock has been weak. The PER sits in the single digits.
The DLC comment: management formally said it is “exploring expansion including DLC.” But this is not a confirmation — no pricing, no release date, no content scope yet.
Market misread #1: “DLC sales are huge.” Realistically incremental revenue is KRW 25–40B (bear), KRW 55–75B (base), KRW 100–140B (bull). Big numbers, but not decisive against a ~KRW 3T market cap.
Market misread #2: “They already sold 5M, the story is over.” 1Q26 revenue was 81% Crimson Desert; FY26 company guidance for the title is KRW 644.1–734.8B.
The real point: DLC matters not for the dollars but for the reclassification — package-game maker → franchise IP. That moves the PER from the single digits toward the global game-publisher average (15–25x).
What’s priced in today: the assumption that “2026 base-game revenue is fine but 2027 is a cliff.”
What DLC formalization breaks: precisely that cliff assumption. The direct EPS lift (≈ +KRW 625/share) is small versus the multiple-expansion lift (PER 14–16x).

1. How the market sees Pearl Abyss today

1.1 Why a record quarter still sold off

Pearl Abyss 1Q26:
- Revenue: KRW 328.5B (record)
- Operating profit: KRW 212.1B (record)
- Net income: KRW 170B
- Crimson Desert revenue: KRW 266.5B (81.2% of total)
- PC / console revenue mix: roughly 50:50

Price action (directional):
- Briefly strong post-launch, then drifting weaker
- 5M-units milestone mid-April produced no fresh momentum
- After the May 12 print, distribution accelerated
- May 15 KOSPI -6.12% crash pulled it down further

→ Record print, weak tape — why?
→ Because the market is not pricing 1Q26. It is pricing 2027+.

1.2 What the market fears — the “package-game cliff”

Classic package-game revenue pattern:

Quarter of launch: blow-out (50–70% of lifetime)
Q+1 to Q+3: rapid deceleration
Q+4 onwards: trends toward zero
→ "One-shot hit" shape.

Mental model:
- Launch quarter: 100
- Q+1: 50
- Q+2: 20
- Q+3 onward: <10

When the market sees Pearl Abyss's 1Q26 of KRW 328.5B,
the first thought is: "What's 1Q27 going to look like?"

This is why the PER sits in single digits:
→ "2026 earnings are real"
→ "But 2027 is a cliff"
→ "So we discount future earnings hard."

1.3 What a single-digit PER actually implies

Current Pearl Abyss applied PER:
→ Single digits (\~8x area)
→ On 2026E EPS ≈ KRW 5,700

Reference:
General game-stock average PER: 15–25x
Global AAA publishers (Take-Two, EA, etc.): 20–30x
Korean quality game peers: 12–18x

Pearl Abyss at a single-digit PER = less than half the average
→ A deep discount baked into the price.

Why the discount:
1. Doubt about post-2027 earnings durability ← biggest
2. Single-IP dependence (Crimson Desert 81%)
3. Time gap to the next title (DokeV)
4. Fear of "one-and-done" package economics

→ DLC formalization breaks reason #1.

2. What the DLC comment actually says — not a confirmation

2.1 What management actually said

From the 1Q26 earnings letter:
"Exploring various ways — including DLC, ongoing updates,
 and platform expansion — for the next stage of Crimson Desert"

Original English line:
"exploring various ways including DLC"
"once concrete plans are set, will be shared"

→ "Exploring," not "confirmed"
→ Pricing, timing, scope all undisclosed
→ The GamesRadar "DLC confirmed" framing is overstated.

2.2 Even “exploring” carries weight

What it means for a company to put "exploring DLC" in writing
in an IR document:

1. Internally they take DLC seriously
2. They have decided to use DLC as an IR "card" with the market
3. Not denying = probability is meaningfully high
4. Likely to be concretized next quarter or shortly after

Game-publisher IR language is conservative:
→ "Exploring" usually means it is 50–70% under way
→ "Confirmed" announcements come after pricing and timing lock
→ So even "exploring" is a strong signal in this context.

2.3 Why the market didn’t react more

Reason 1: Global macro pressure
→ May 15 KOSPI -6.12% crash
→ KOSDAQ -5.14%
→ Broad game-stock weakness on top of which Pearl Abyss sold harder

Reason 2: "Exploring" ≠ "confirmed"
→ The market wants specific pricing and timing
→ At the "exploring" stage, fresh buying interest is muted

Reason 3: Pre-existing selling pressure
→ Short-interest 15–18% on May 12–14 (elevated)
→ May 15 turnover spike = stop-outs and disappointment selling

Reason 4: Weak follow-through in foreign flows
→ Foreigners turned net buyers May 15 but weakly
→ The marginal bid was not strong enough to defend price.

3. What DLC actually creates — direct revenue vs. multiple change

3.1 The direct revenue effect (smaller)

Formula:
DLC revenue = installed base × attach rate × DLC price × recognition rate
DLC profit = DLC revenue × incremental margin (60–70%)

Assumptions:
- Year-end installed base: 7.5–8.5M units
 (back-solved from FY26 Crimson Desert guidance KRW 644.1–734.8B)
- DLC attach rate: 15–30%
- DLC price: KRW 20,000–35,000
- Recognition rate: 75% (after 25% platform fee).

Scenario table for incremental revenue:

Scenario	Installed base	Attach	DLC price	Incr. revenue	Incr. op profit
Bear	7.5M	15%	KRW 20k	KRW 25–40B	KRW 15–25B
Base	8.5M	25%	KRW 28k	KRW 55–75B	KRW 35–50B
Bull	10.0M	35%	KRW 35k	KRW 100–140B	KRW 65–95B

Base scenario (DLC revenue KRW 55–75B, op profit KRW 35–50B):

EPS lift:
KRW 50B × 80% (after tax) / 64.25M shares ≈ +KRW 625

Price impact at 14x PER:
+KRW 625 × 14 ≈ +KRW 9,000

→ A single DLC adds roughly +KRW 9,000 / share
→ Significant — but not decisive.

3.3 The multiple-change effect (larger)

This is the real story:

Current market classification:
"Pearl Abyss = one-shot package-game maker"
→ Applied PER: single digits (\~8x)

Post-DLC formalization classification:
"Pearl Abyss = franchise-IP holder with recurring revenue"
→ Applied PER: 14–16x (global AAA publisher average)

Same EPS, different multiple — how the fair value moves:

Take PER 8x (today) as a base of 100:
PER 12x: ≈ 150 (+50%)
PER 14x: ≈ 175 (+75%)
PER 16x: ≈ 200 (+100%)

→ With EPS unchanged,
→ a re-rating alone moves fair value +50% to +100%.

The direct EPS lift (\~+KRW 9,000 / share)
is a subset of the multiple-expansion lift (+50–100% vs current).
→ The multiple effect is several times larger than the revenue effect.

3.4 Why understanding this difference matters

Wrong frame:
"DLC revenue KRW 60B is only 2% of a KRW 3T market cap"
→ "So it doesn't matter."
→ Looking at direct revenue only.

Right frame:
"DLC formalization changes the multiple classification."
→ "PER single-digits → 14x = fair value +75%."
→ Reading it as a "reclassification event."

Historical analogues in game stocks:
- CD Projekt RED re-rated after Cyberpunk 2077 DLC
- Larian's value surged after Baldur's Gate 3
- The long-run CDPR premium was built by The Witcher DLC

→ DLC isn't a one-quarter revenue bump.
 It changes how the IP is *classified*.

4. Scenario fair values

4.1 Scenarios on applied PER

Scenario A (bear — stays a package game):
- DLC remains a free QoL drip
- Year-end base-game units fall short of 7M
- 2027 revenue gap becomes visible
- Applied PER: 6–8x (today or worse)
- Fair value range: around or below current price

Scenario B (base — partial franchise-ization):
- One paid DLC formally launched
- Year-end base-game units reach 8.5M
- Partial 2027 revenue visibility
- Applied PER: 12–14x
- Fair value range: +30% to +50% above current

Scenario C (bull — full franchise IP):
- Major expansion + DLC roadmap
- Year-end base-game units cross 10M
- Crimson Desert 2 or DokeV visibility confirmed
- Applied PER: 15–18x
- Fair value range: +80% to +120% above current.

4.2 Where price sits — asymmetry

The current quote sits close to scenario A.
→ Downside is limited (price has already corrected)
→ Move to scenario B alone is +30–50%
→ Move to scenario C is +80–120%.

Asymmetry:
Downside width: relatively small
Upside width: 1.5–5x the downside

→ Upside dominates,
→ but the stock is vulnerable to short-term macro (e.g., today's
 KOSPI crash)
→ Macro stability check has to come first.

5. The next 6 months — what to watch

5.1 What the company has to show

Necessary signals for A → B transition:

1. A formal 6M or 7M units announcement
 → Currently estimated around 5.8M
 → A 6M call-out around late May / early June is appropriate
 → A delay is a negative signal.

2. DLC concretization
 → From "exploring" to "scheduled for release"
 → Pricing, timing, scope disclosed
 → Likely candidate: 2Q earnings (August).

3. H2 revenue guidance held
 → Company FY26 Crimson Desert guide: KRW 644.1–734.8B
 → 1Q26 KRW 266.5B, 2Q26 guide KRW 224.2–276.5B
 → Implied H2 residual: \~KRW 153.4–191.8B
 → Holding inside that range proves "long-tail is alive."

4. Capital return messaging
 → Buyback or higher dividend
 → A signal that cash is being put to use.

5. DokeV development visibility
 → Public progress checkpoint
 → Signal of a 2027–2028 launch window.

5.2 What the game itself has to show

User metrics:
1. Steam CCU recovery
 - Current: \~47,000 (7-day avg)
 - Target: 60,000+ recovery

2. Steam global ranking
 - Current: \~#26
 - Target: inside top 20

3. Recent-review positive rate
 - Current: \~79.4% (last 7 days)
 - Target: 85%+ recovery

4. Patch cadence
 - May 15 patch 1.07.00 (boss rematches, Damian unarmed kit, etc.)
 - June needs another sizable update.

5.3 The macro backdrop

The macro gate from the prior post:
- US 10-year below 4.45%
- Brent below 105 USD
- USD/KRW below 1,480
- VIX below 18

These have to clear for:
→ Broad risk-asset buying to resume
→ KOSDAQ long-duration names like Pearl Abyss to recover
→ Good news from the company to actually move the price.

If the macro doesn't clear:
→ Price can stay weak even on improved fundamentals
→ "Good company, but not yet a good price."

6. Trading decision — holders vs. new buyers

6.1 If you already hold

Where you are:
- Possible mark-to-market loss after the post-launch drift
- Position size may already be sizable

Call: Hold

Why:
- Fundamental thesis is intact
- Macro pressure is depressing the price
- DLC, 6M unit milestones, DokeV — multiple cards remain
- Current price sits near scenario A = downside is bounded

Conditions to ADD:
- 2Q revenue at or above the lower-bound guide (KRW 224.2B)
- 6M or 7M unit count announced
- DLC pricing / timing / scope concretized
- 2+ macro gates clear

Conditions to TRIM:
- 2Q revenue below the lower-bound guide
- Steam global rank stuck outside #40
- Recent-review positive rate stays under 75%
- Company cuts H2 guidance.

6.2 If you are a new buyer

Call: Wait

Why:
- Price is attractive on paper, but chasing is inefficient
- Macro gate hasn't cleared
- Wait for the company's next card to drop

Entry conditions:
- 2+ macro gates green
- 2Q revenue guide confirmed (early August)
- OR DLC concretization announcement

Expected returns once in:
- Move to scenario B: +30–50%
- Move to scenario C: +80–120%
- Reversion to scenario A: around current or worse

→ Upside is 1.5–5x downside
→ Scaled entries after macro stabilizes is the rational play.

6.3 The core read — “this is a ‘company-must-play-a-card’ window”

The essence of Pearl Abyss right now:
"The fundamentals are alive, but the market hasn't been convinced yet."

Crimson Desert has sold (5M+).
1Q26 was a record quarter (operating profit KRW 212.1B).
FY26 guidance is strong (KRW 487.6–572.6B).

Yet the applied PER stays in single digits.

→ What the market wants in addition:
 1. Show me how 2027 revenue stays (= DLC)
 2. Show me when DokeV ships
 3. Reignite base-game momentum (6M, 7M)
 4. Tell me what you'll do with the cash (capital return)

If any one of these materializes:
→ Today's price is oversold
→ Move to scenario B can happen fast.

If none of them do:
→ The low multiple stays justified
→ The stock stays in scenario A.

7. How this connects to other posts

Macro-cycle synthesis piece:
→ Today's KOSPI -6.12% is the cycle's settlement day
→ Pearl Abyss is not immune
→ KOSDAQ long-duration assets feel the most pressure.

KOSPI crash + relative-strength piece:
→ "Macro gate before stock work"
→ Pearl Abyss also rational to enter only after the gate clears
→ \~30% of stock-research time on macro is enough.

US-China summit piece:
→ "Find the least-priced pocket"
→ Pearl Abyss already corrected meaningfully
→ But needs its own catalyst (DLC formalization) to re-rate.

8. The one-line bottom line

The real meaning of Pearl Abyss’s DLC line is not “KRW 60B of DLC revenue.” It is the reclassification of Crimson Desert from a one-shot package game to a recurring-revenue franchise IP.

The current PER sits in the single digits for a simple reason: “2026 earnings are real, but 2027 is a cliff.” That single assumption is pinning the multiple below half the global game-publisher average (15–25x).

When DLC is formalized, the assumption breaks. The direct EPS lift (≈ +KRW 625/share, ≈ +KRW 9,000/share at 14x) is real — but secondary. The real upside is the multiple itself re-rating from package-game maker to franchise publisher. With the same EPS, applying a 14x PER instead of 8x lifts fair value by roughly +75%. Multiples, not revenue, drive the gap.

Even so, chasing here is inefficient. The macro gate has not cleared, and DLC is “exploring,” not “confirmed.” For holders, Hold is the rational call. For new buyers, wait for at least two of: ① macro-gate stabilization, ② 2Q revenue clearing the lower-bound guide, ③ DLC concretization — then add in tranches.

Investing is buying a good company at a good price. Pearl Abyss is a good company, and the price is at attractive levels — but the good entry point comes when the next card is played. When it is, the classification flips from package-game to franchise-publisher. The reclassification — not the revenue — is the alpha.

This article is research and commentary only and is not investment advice. Pearl Abyss’s 1Q26 figures and FY2026 guidance are per the company’s official earnings letter. The Crimson Desert revenue-recognition treatment (PC net of taxes; console net of platform fees) is per the company’s disclosure. The “exploring various ways including DLC” line is verbatim from the 1Q earnings letter; “confirmed” is not used. The ~5.8M cumulative-units estimate is inferred from Steam review data; the official milestone is 5M as of April 15. The DLC revenue scenarios (installed base, attach rate, price) are author estimates and are not confirmed. The single-digit applied PER is an estimate based on the May 15 price and consensus 2026E EPS. The 6–18x scenario-PER ranges are author estimates and are not guaranteed. Global macro variables (US rates, oil, FX, VIX) can independently move the stock. The analysis may be wrong. Data cut-off: May 15, 2026 KST.

What the -6.12% KOSPI Crash Really Means — Why You Have to Check the Macro Gate Before Hunting Relative-Strength Names

Fri, 15 May 2026 00:00:00 +0000

On May 15 the KOSPI fell -6.12% — the day right after the all-time high of 8,000. KOSDAQ -5.14%. The sell-side sidecar circuit breaker fired. On a day like this, investors face two temptations. One is to hunt for names that didn’t fall, looking for the next leadership stock. The other is to chase, saying “everything fell, so it’s cheap.” Both are dangerous. The right read is not stocks — it is the macro gate: US long rates, oil, USD/KRW, VIX. Until those four stabilize, chasing even a great stock is inefficient.

Key takeaways

May 15 Korea: KOSPI 7,493.18 (-6.12%), KOSDAQ 1,129.82 (-5.14%). Sell-side sidecar fired right after a fresh all-time high.
Cause: not simple profit-taking. US PPI +6.0% (largest since 2022), Japan PPI +4.9%, JGB 10-yr at the highest since 1997, US 10-yr 4.46%, Brent 108, USD/KRW near 1,500 — a global macro cycle pressuring risk assets simultaneously.
There are RS names: Hana Micron +18.6%, Jeju Semiconductor +8.9%, SFA Semiconductor +4.0%, Samsung Electro-Mechanics -1.4% (RS vs index +4.7pp). Doosan Robotics +19.3%, Robostar +9.9%.
But chasing is wrong: “didn’t fall today = next leader” is a fallacy. If macro doesn’t normalize, RS names follow the index down the next session.
The real read — macro gate: US 10-yr below 4.45%, Brent below 105, USD/KRW below 1,480, VIX below 18. At least 2–3 of those must clear before new buying is rational.
What today’s RS screen is actually for: not a “buy signal,” but a “watchlist-generation signal."

1. What happened

1.1 Korea

May 15 close:
KOSPI: 7,493.18 (-6.12%)
KOSDAQ: 1,129.82 (-5.14%)

Notes:
- Briefly hit a fresh ATH intraday, then reversed sharply
- Sell-side sidecar (5-min suspension of program selling) fired
- Massive foreign and institutional net selling; retail absorbed
- A 6%+ single-day decline — the largest since August 2024.

This is too large to read as ordinary profit-taking.

1.2 Korea wasn’t alone

Same day, global markets:
- US Nasdaq futures down sharply
- Nikkei lower
- Mainland China sold on summit disappointment

Drivers:
US April PPI +6.0% (largest since March 2022)
US April CPI +3.8% (highest since May 2023)
Japan April PPI +4.9% (3-year high)
JGB 10-yr 2.55% (highest since 1997)
US 10-yr 4.46%, 30-yr 5.02%
Brent \~108 USD
USD/KRW near 1,500.

1.3 Is this just volatility, or is it the cycle?

The “global macro cycle” we mapped in an earlier post actually executed.

The cycle in motion:

Iran / Hormuz → oil 108 USD
 ↓
US PPI +6.0%, CPI +3.8%
 ↓
Fed-cut hopes priced out (Dec hike prob 36%)
 ↓
US 10-yr 4.46%, 30-yr 5.02%
 ↓
Japan PPI +4.9% (import prices +17.5%)
 ↓
BOJ June-hike prob 77%
 ↓
JGB at the highest since 1997
 ↓
Global discount rate up
 ↓
Risk assets squeezed simultaneously
 ↓
KOSPI -6.12%

→ Not a one-off shock —
 the "settlement day" of a multi-day cycle.

2. The trap of “RS hunting” on a day like this

2.1 What relative strength is

Relative Strength (RS):
stock return − index return

If KOSPI fell -6.12% and a stock rose +5%:
→ RS = +5% − (-6.12%) = +11.12pp
→ "11.12pp stronger than the index"

Why look at it:
→ Strong demand survives the index decline
→ Tends to lead on the recovery
→ A screening tool for next-leadership candidates.

2.2 Today’s top RS names

Rank	Name	Return	RS vs index	Turnover
1	Samti	+30.00%	+35.1pp	KRW 84B
2	Samji Electronics	+29.97%	+35.1pp	KRW 77.5B
3	Doosan Robotics	+19.29%	+25.4pp	KRW 1.08T
4	Hanson Engineering	+19.18%	+24.3pp	KRW 60.6B
5	Hana Micron	+18.61%	+23.8pp	KRW 688.8B
6	Jahwa Electronics	+16.53%	+22.7pp	KRW 84.8B
10	LG Electronics	+10.83%	+17.0pp	KRW 1.44T
12	Robostar	+9.90%	+15.0pp	KRW 188.6B
13	Jeju Semiconductor	+8.86%	+14.0pp	KRW 884.6B
17	SFA Semiconductor	+3.99%	+9.1pp	KRW 280.8B
18	Samsung SDS	+3.34%	+9.5pp	KRW 366.7B
20	Samsung Electro-Mechanics	-1.37%	+4.7pp	KRW 1.41T

2.3 Trap 1 — confusing “limit-up momentum” with “earnings strength”

The RS top list mixes two very different things:

A. Earnings-backed strength (good RS):
 - Hana Micron: 1Q26 OP KRW 72B (+513.6%)
 - Jeju Semiconductor: 1Q26 OP KRW 67.1B (+1,713%)
 - Samsung Electro-Mechanics: MLCC price hike + AI parts thesis
 → Real fundamentals.

B. Theme / news pop (risky RS):
 - Samti limit-up +30%
 - Samji Electronics limit-up +30%
 - Hanson Engineering +19%
 → One-day catalyst or flow event.

Treating limit-up names as "RS #1–2" is dangerous:
→ Likely to gap down and form a red candle the next day
→ Not real leadership candidates.

2.4 Trap 2 — without “close location,” every RS looks alike

Close location = (close − low) / (high − low)

Same +5% gain, very different read:
- Close location 90% → buyers held through the close (good)
- Close location 20% → distributed from intraday highs (weak)

In today's data:

Name Return Close loc Read
Hana Micron +18.6% 71.7% ★ buying held late session
Jeju Semi +8.9% 62.2% Good, but watch-listed risk
SFA Semi +4.0% 19.5% ✗ pushed intraday (low-quality RS)
Samsung E-M -1.4% 19.6% ✗ outperformed index but
 distribution on the candle

→ By return alone, SFA looks "strong"
→ By close location, sellers actually dominated
→ Same RS rank, very different quality.

2.5 Trap 3 — if macro doesn’t clear, RS names get dragged down the next session

Historical pattern after a major down day:

A. Scenario (macro recovers):
 → Index rebounds
 → RS names lead higher (+10–15% possible)
 → They graduate into actual leadership

B. Scenario (macro worsens):
 → Index falls more
 → RS names get chase-sold
 → Yesterday's strong names mean-revert -10–15%
 → Short-term traders stop out

The differentiator between A and B is macro, not the stock.
→ You have to read macro BEFORE you trade the stock.

3. The macro gate — what to actually watch on Monday

3.1 Gate 1: US long rates

Now:
US 10-yr: 4.46%
US 30-yr: 5.02%

Recovery signals:
US 10-yr below 4.45% (first)
US 30-yr below 5.0% (stronger)

Why it matters:
→ Anchor of the global discount rate
→ Key driver of foreign flows into Korea
→ Sets the multiple for growth / semis / robotics.

Current: 🔴 Red

3.2 Gate 2: Oil / geopolitics

Now:
Brent: 108–109 USD
Hormuz: still functionally closed

Recovery signals:
Brent below 105 USD (first)
Brent below 100 USD (stronger)
Hormuz reopening headlines

Why it matters:
→ Origin of the US / Japan inflation pulse
→ Trigger variable for the whole cycle
→ Above 100 USD, Fed-cut hopes don't return.

Current: 🔴 Red

3.3 Gate 3: USD/KRW

Now:
USD/KRW \~1,495–1,500

Recovery signals:
USD/KRW below 1,480 (first)
USD/KRW below 1,450 (stronger)

Why it matters:
→ Direct gauge of foreign-selling pressure
→ KRW weakness = lower Korean asset appeal
→ Above 1,500 invites more selling.

Current: 🟡 Yellow → Red

3.4 Gate 4: VIX

Now:
VIX \~18.6

Recovery signals:
VIX below 18 (first)
VIX below 16 (stronger)

Why it matters:
→ Gauge of global risk aversion
→ Above 20 = panic
→ Below 16 = risk-on.

Current: 🟡 Yellow

3.5 Gate summary

Status of the 4 gates:
🔴 Rates: Red
🔴 Oil: Red
🟡 FX: Yellow → Red
🟡 VIX: Yellow

→ 0 of 4 are Green
→ Inefficient to add risk
→ Even RS names should NOT be chased.

4. A macro-gated trading frame

4.1 Gate-pass conditions

To size up new buying, at least 2–3 of the 4 gates must clear.

Specifically:
1. US 10-yr falls below 4.45%
2. Brent falls below 105 USD
3. USD/KRW stabilizes below 1,480
4. VIX falls below 18
5. KOSPI / KOSDAQ holds the prior session's low and closes green

At least 2–3 met → partial buying allowed.
1 or fewer met → no buying.

4.2 Action by scenario

Scenario	Conditions	Action
A. Macro recovers	3+ of 4 gates clear	Scaled buying in RS names; max 30–40% of full size
B. Macro neutral	1–2 gates clear	Just observe the close; minimize new buying
C. Macro worsens	Gates deteriorate further	No new buying; only re-evaluate existing positions for thesis damage

4.3 Relationship between “RS buys” and the macro gate

Wrong order:
RS scan → buy → check macro

Right order:
Check macro gate → if gates clear, screen RS → then add risk

Why:
- If macro doesn't clear, RS names follow the index down
- Buying great stocks at the wrong price = bad outcome
- Real leadership isn't built in a day — it shows up over days / weeks
 of sustained relative strength.

5. If the macro gate clears — name priority

When the gates pass, work the list in this order.

5.1 #1 — Hana Micron

Fundamentals:
1Q26 revenue KRW 507.7B (+62.8% YoY)
1Q26 OP KRW 72B (+513.6% YoY)
Margin 14.2%
Beats consensus operating profit by +29.7%

Close location: 71.7% (strong end-of-day demand)
Turnover: KRW 688.8B (ample)

Read:
- Cleanest earnings-backed RS candidate
- KIS target KRW 55,000 is close (current 52,900, +4% upside)
- Prefer a pullback or sustained turnover over chasing

Entry conditions:
- Hold 52,900 on the close
- Turnover stays above KRW 340B (50% of today)
- RS vs KOSDAQ +3pp or better.

5.2 #2 — Samsung Electro-Mechanics

Fundamentals:
1Q26 revenue KRW 3.21T
1Q26 OP KRW 280.6B (+40% YoY)
MLCC price-hike cycle starting
NH target lifted to KRW 1.5M

Close location: 19.6% (intraday distribution)
Turnover: KRW 1.41T (large-cap volume)

Read:
- Structurally the best large-cap candidate
- After +23% in May, today is a down candle
- Wait for KRW 1,024,000 (prior close) to be reclaimed

Entry conditions:
- Reclaim KRW 1,024,000 (yesterday's close)
- Stronger signal: hold above KRW 1,061,000 (today's open).

5.3 #3 — Jeju Semiconductor

Fundamentals:
1Q26 revenue KRW 180.4B (+273%)
1Q26 OP KRW 67.1B (+1,713%)
Margin 37.2%

Risk:
"Investment caution" designation
Trading-halt risk on further surges
Customer concentration (A 72%, China 73%)

Read:
- Numbers are strong but trading is difficult
- Volatility makes chasing inefficient
- Want a settle in the KRW 80k zone + 2–3 sessions of digestion candles.

5.4 #4 — SFA Semiconductor

Fundamentals:
4Q25 turned profitable
1Q26 expected to stay profitable (estimate only — not yet confirmed)

Close location: 19.5% (intraday push-down)
Turnover: KRW 280.8B

Read:
- Back-end thesis is right, but earnings proof is still light
- Low close quality hurts the RS quality
- Wait for 1Q26 disclosure.

5.5 Names to exclude from chasing

High RS, but no chase:

1. Limit-up pops (Samti, Samji Electronics):
 - One-day catalyst / flow event
 - Likely to gap down

2. Needs verification (Hanson Engineering):
 - Turnover only KRW 60B
 - Could be a one-off rather than a trend

3. Low close quality (SFA, today's Samsung E-M candle):
 - RS-positive on the day
 - But low close location = sellers dominated.

6. Monday checklist

Before the open (KST 08:00–09:00):

1. Did US 10-yr drop below 4.45%?
2. Did Brent drop below 105 USD?
3. Did USD/KRW move into the 1,480s?
4. Did VIX drop below 18?

→ How many of these are Green?

Intraday (09:00–15:30):

5. Did KOSPI / KOSDAQ gap down and recover, or gap up and fade?
6. Are foreigners net-buying or at least slowing selling?
7. Are Samsung Electronics / SK hynix recovering steadily?

→ How many of these are positive?

Close (15:30–16:00):

8. Did Hana Micron hold KRW 52,900?
9. Did Samsung E-M reclaim KRW 1,024,000?
10. Did turnover stay at 50%+ of today?

→ All 3 positive AND 2–3 macro gates green = scaled buying possible
→ 1 or fewer macro gates green = no new buying.

7. The core lessons

7.1 “Didn’t fall today” ≠ “next leader”

Wrong inference:
"The market fell, but this stock didn't"
→ "It's strong"
→ "It'll be strong tomorrow too"
→ "Buy"

Right inference:
"The market fell, but this stock didn't"
→ "There was real demand today"
→ "For tomorrow to be strong too, macro has to stabilize"
→ "If macro confirms AND turnover holds, then it's a buy candidate".

7.2 Leadership is built over many sessions, not one

Strong for 1 day ≠ leadership
Strong for 1 week = candidate
Strong for 1 month = leadership

Today's RS #1 is not automatically tomorrow's leader.
Only after sustained RS over the next 5–10 sessions
does a name actually earn the leadership label.

→ Today's buying is "betting on a candidate," not "buying the leader."
→ If macro doesn't clear, candidates all give back.

7.3 Stock work < macro gate

The core message of this post:
"Cycle before the stock."

A great company bought at a high price at the wrong time
ends poorly.
A mediocre name bought as macro recovers
ends fine.

→ Spend even 30% of your stock-research energy
→ on the 4 macro gates every day
→ and the quality of your entry timing improves disproportionately.

8. How this links to other posts

The macro cycle synthesis piece:
→ "Iran / Hormuz → oil → US inflation → long rates → global discount rate"
→ Today's KOSPI -6% is the "settlement day" of that cycle.

The Japan PPI piece:
→ Mechanism for the Japan axis — weaker marginal bid for USTs
→ Connected to today's foreign selling.

The US-China summit piece:
→ "Find the least-priced pocket"
→ Today's RS names are exactly those candidates.

The Samsung Electro-Mechanics piece:
→ "At KRW 1.02M, chasing is inefficient"
→ Today -1.4%; the 1,024,000 level broke
→ Wait for a reclaim before buying.

The consumer-rotation piece:
→ "Capital broadened to earnings-backed consumer staples"
→ Today Samyang Foods +0.35% (defensive)
→ Still a valid rotation candidate.

9. The one-line bottom line

May 15’s KOSPI -6.12% is not simple profit-taking. It is the day the global macro cycle settled in the Korean market. US PPI +6%, Japan PPI +4.9%, US 10-yr 4.46%, Brent 108, USD/KRW near 1,500 — every one of those variables pressed Korean risk assets in the same direction.

On days like this, there are two temptations. One is to hunt the names that didn’t fall in search of the next leader. The other is to chase, saying “everything fell, so it’s cheap.” Both are inefficient if the macro gate doesn’t clear.

The first thing to check Monday is not Hana Micron, Samsung E-M, or Jeju Semiconductor. It is the US 10-year, Brent, USD/KRW, and the VIX. Until at least 2–3 of those stabilize, chasing even a great stock is inefficient.

Today’s RS screen is not a “buy signal” — it is a “watchlist-generation signal." The next leadership stocks are not discovered AFTER the macro recovers. They are the names that did NOT fall while the macro was bad, recorded, and bought once the macro recovers.

Cycle before the stock. Macro gate before stock work.

This article is research and commentary only and is not investment advice. May 15 KOSPI / KOSDAQ moves are per the exchange’s official close. Stock-level returns and turnover figures are per Naver Stock and Korea Economic Daily market data. Close-location uses (close − low) / (high − low) and does not guarantee a perfect read on demand strength. US 10-year, Brent, USD/KRW, and VIX figures reflect May 14–15 market data. Gate thresholds (4.45%, 105 USD, 1,480 KRW, 18) are the author’s judgment and are not absolute. Company financials and target prices are per broker reports and DART filings and may change. Scenarios and entry conditions are the author’s hypotheses. The analysis may be wrong. Data cut-off: May 15, 2026 KST.

Why Everything Is Moving Together — Reading Iran, Oil, US CPI, China, and Japan as a Single Cycle

Fri, 15 May 2026 00:00:00 +0000

What actually happened between May 14 and 15? US April PPI printed +6.0% YoY, the largest since 2022. Japan’s PPI hit +4.9%; the JGB 10-year reached its highest since 1997. The US 10-year is at 4.46%; the 30-year at 5.02%. Brent is 108 USD. The Strait of Hormuz has been effectively closed for two months. The Xi–Trump summit ended in a “small deal.” The market is reading these as separate news items. That is the wrong frame. These five variables are not independent events — they are one cycle. Iran is the origin; a higher global discount rate is the terminus. Draw the picture once, and the next headline will tell you where it will land in other asset classes.

Key takeaways

What the market is actually pricing: the variables that shook May 14–15 markets — Iran/Hormuz, oil, US inflation, the US-China summit, Japan PPI — are one cycle, not five independent events.
Origin: Iran/Hormuz bottleneck → global crude supply shock.
Terminus: higher US long rates → repriced global discount rate → multiple compression across risk assets.
Five transmission links: ① Iran/Hormuz → ② oil → ③ US CPI/PPI → ④ Fed-cut-pricing fades + US long rates rise → ⑤ Japan PPI → BOJ tightening → weaker Japanese marginal bid for USTs → further long-end yield rise (self-reinforcing).
The role of the US-China summit: not a cause of the cycle but a side variable that speeds it up or slows it down.
For Korean equities: the semi / auto concentration unwinds; refining, LNG, and defense get relative tailwinds; growth and long-duration names face pressure.
The single most important variable: when Hormuz normalizes. It decides where the cycle ends.

1. The picture first — how five variables resolve into one cycle

 ┌─── Iran / Hormuz ───┐
 │ (cycle origin) │
 └──────────┬──────────┘
 │
 ▼
 ┌──── Global Oil ────┐
 │ (1st transmission) │
 └──────────┬─────────┘
 │
 ┌──────────────┴──────────────┐
 │ │
 ▼ ▼
 ┌──── US CPI / PPI ────┐ ┌──── Japan PPI ────┐
 │ (2nd transmission) │ │ (2nd transmission) │
 └─────────┬────────────┘ └─────────┬─────────┘
 │ │
 ▼ ▼
 ┌─ Fed-cut hopes ↓ ─┐ ┌─── BOJ tightening ↑ ──┐
 │ (3rd transmission)│ │ (3rd transmission) │
 └─────────┬─────────┘ └─────────┬─────────────┘
 │ │
 ▼ ▼
 ┌─ US short rates hold ┐ ┌──── JGB yield ↑ ─────┐
 │ │ │ JPY appreciation ↑ │
 └─────────┬─────────────┘ └─────────┬───────────┘
 │ │
 │ ▼
 │ ┌─ Post-hedge UST appeal ↓ ──┐
 │ │ Japanese marginal bid ↓ │
 │ └─────────┬─────────────────┘
 │ │
 └────────┬───────────────┘
 ▼
 ┌─── US long rates ↑ ────┐
 │ (term premium up) │
 │ (cycle terminus) │
 └────────────┬───────────┘
 │
 ▼
 ┌──── Global discount rate ↑ ────┐
 │ Risk-asset multiples compress │
 │ Growth / long-duration ↓ │
 └─────────────────────────────────┘

US-China summit = side variable that accelerates or delays the cycle.

Once you see this picture, the next headline will tell you which asset class it will land in.

2. Variable 1 — Iran / Hormuz: the cycle’s origin

2.1 Why Hormuz is the origin

Why the Strait of Hormuz matters:

Geography: a narrow strait between Iran and Oman (\~33 km at narrowest)
Function: the ONLY maritime exit for Persian Gulf crude and LNG to
 reach the rest of the world

Numbers:
- \~25–35% of global seaborne crude passes through
- \~20 million barrels per day
- Saudi Arabia, UAE, Kuwait, Iraq, Iran all depend on this strait

When Hormuz closes:
→ \~10.5 mb/d of Gulf-6 crude shut in
→ US SPR releases offer partial offset
→ Global inventories draw fast (\~250 mn bbl in Mar–Apr)
→ Brent stays above USD 100

Why this is not "just geopolitics":
→ Oil itself is the anchor of global inflation
→ Direct pass-through to CPI / PPI everywhere
→ Direct input into central-bank policy
→ Knock-on effects across rates and risk

2.2 Where we are on May 15

Hormuz has been functionally closed since February 28
Brent ~108 USD
IEA: 2026 global supply -3.9 mb/d, demand -0.42 mb/d, net deficit 1.78 mb/d
EIA: 2Q global inventory draw of ~8.5 mb/d on average
Iran foreign minister: “we don’t trust the US — won’t negotiate if they’re not serious”

2.3 Key judgment

Even when Hormuz normalizes, prices normalize later.

Reasons:
1. Bringing 10.5 mb/d of shut-in capacity back takes time
2. Infrastructure damage, mine-clearing, insurance / shipping restart
3. IEA: "refining margins likely stay historically elevated even
 after restart"
4. Inventory rebuild demand (the \~250 mn bbl draw needs replacing)

EIA STEO: full recovery of the shut-in 10.5 mb/d = "late 2026 / early 2027"

→ The market is pricing "Hormuz deal = oil down"
→ Reality: even with a deal, price normalization runs 6–9 months late.

3. Variable 2 — how oil transmits to US inflation

3.1 First-order: headline CPI / PPI

US April:

Indicator	MoM	YoY
Headline CPI	+0.6%	+3.8% (highest since May 2023)
Core CPI	+0.4%	+2.8%
PPI	+1.4%	+6.0% (largest since Mar 2022)
Gasoline (PPI)	+15.6%	—
Energy (CPI)	+3.8%	+17.9%

A 0.6% monthly print annualizes to ~7.4%; PPI’s +1.4% annualizes to ~18%. Single-month annualization overstates the run rate, but what the bond market reacts to is direction.

3.2 Second-order: services and shelter

Oil doesn’t just push headline.

Oil up → transport costs up → every good gets repriced
Oil up → airfares, travel up → services inflation
Oil up → heating, electricity up → shelter cost
Oil up → fertilizer up → food prices

What we saw in the April print:
- Airfares up
- Shelter +0.6%
- Food-price pressure
- Transportation services up

→ Not "just energy spiked"
→ A signal of broad-based inflation
→ The kind of mix the Fed can't ignore.

3.3 Third-order: Fed policy expectations reset

Three months ago: market expected 2–3 cuts of 25bp in 2026.

May 15 today:
- Cut expectations all but priced out
- December "hike" probability 36% (was 16% a week ago)

→ The policy cycle is being reset in the opposite direction
→ As the "cuts in the price" come out
→ The whole curve, short to long, gets reset.

4. Variable 3 — higher US long rates: the cycle terminus

4.1 The short end and the long end are different

Short rates (2-year):
- Move almost mechanically with the Fed policy rate
- Don't move much if the Fed stays on hold

Long rates (10-year, 30-year):
- Average of expected future short rates + term premium
- Term premium = "compensation for locking your money up for 10 years"

What lifts term premium:
1. Inflation uncertainty ↑
2. Fiscal-deficit concerns ↑
3. Supply / demand deterioration (fewer buyers)
4. Volatility ↑

→ Right now 1, 3, and 4 are all firing.

4.2 Where we are on May 15

US 10-year: 4.46%
US 30-year: 5.02%

This is NOT just "the Fed sounded hawkish."

Decomposition:
- Inflation concerns (CPI / PPI re-acceleration)
- Fiscal deficit (\~USD 2T per year)
- Issuance pressure (USD 9T of maturities + USD 2T net new)
- Weaker marginal bid from foreign buyers (Japan, China) ← key
- A weak 30-year auction this month

→ All of it compresses into one word: "term premium up."

5. Variable 4 — Japan PPI: the second axis that reinforces the cycle

5.1 The global price shock hits Japan too

Japan is a net importer of energy and raw materials:
- Crude: 99% imported
- LNG: mostly imported
- Fertilizers, grains: heavily imported

Hormuz bottleneck → Japan import-price surge
- Japan April import prices +17.5% YoY (in JPY)
- Japan April PPI +4.9% (highest in three years)
- JGB 10-year 2.55% (highest since 1997)

→ The BOJ can no longer hold the accommodative line
→ Market prices a 77% probability of a June BOJ hike.

5.2 How BOJ tightening transmits into US Treasuries

The chain:

BOJ hike probability ↑
 ↓
JGB 10-year yield ↑ + JPY-strength expectations ↑
 ↓
Post-hedge UST return ↓ (for Japanese investors)
 ↓
NEW UST buying falls (selling not required — marginal bid weakens)
 ↓
UST supply-demand balance breaks
 ↓
US 10-year yield rises further

* This is the "no new buy" scenario, not a "dumping" scenario
* More gradual, but more structural.

The detailed mechanism is in a separate post — the key point is that Japan is the UST market’s marginal buyer. Even without selling, when the marginal bid weakens, prices fall (yields rise).

5.3 The self-reinforcing loop

Oil up (Iran / Hormuz)
 ↓
US CPI / PPI up ←──────┐
 ↓ │
US long rates up │
 ↓ │
Japan import prices up │ (feedback)
 ↓ │
Japan PPI up │
 ↓ │
BOJ tightening expected │
 ↓ │
Japanese UST demand ↓ │
 ↓ │
US long rates up further─┘

→ Once started, the loop reinforces itself.
→ Hard to break without Hormuz normalizing.

6. Variable 5 — US-China summit: a side variable

6.1 Recap of the summit

From an earlier post:

Outcome	Detail
Agreed	Hormuz reopening understanding, H200 export licenses to China (10 firms), discussion of USD 30B tariff relief on non-sensitive items
Not agreed	Joint statement, rare-earth waiver extension, semiconductor-equipment control easing, Taiwan arms-sale changes
Xi quote	“Taiwan is the most important issue in US-China relations”
CSIS view	“A largely China-friendly, surface-level truce”

6.2 How the summit affects the cycle

Scenario A (Hormuz normalization accelerates):
→ Xi joins diplomatic pressure on Iran
→ Partial Hormuz reopening
→ Oil settles below 95 USD
→ The cycle "slows"
→ US inflation pressure eases
→ Cap on long-rate upside

Scenario B (small deal ends here, Hormuz unresolved):
→ Summit impact limited
→ Cycle keeps running
→ Oil stays in the 100s
→ US long-rate pressure persists
→ Global discount rate rises further

As of May 15:
We are closer to scenario B.
→ Hormuz remains functionally closed
→ Iran says "we don't trust the US"
→ No clear trigger to stop the cycle.

6.3 Why the summit is a “side variable,” not the engine

The summit didn’t create the cycle. The cycle was already running, with Iran/Hormuz as its origin. The summit is a variable that accelerates or slows the cycle.

Accelerators:
- US proceeds with Taiwan arms sales
- China tightens rare-earth controls
- Summit collapse or unilateral outcomes

Decelerators:
- China joins diplomatic pressure on Iran
- A Hormuz reopening deal
- USD 30B tariff easing actually implemented

→ The market focuses on the summit itself,
 but the right question is "what does the summit do to Hormuz?"

7. Korean equities — how the cycle works through here

7.1 Direct transmission

US long rates up
 ↓
Global risk-asset discount rate up
 ↓
KOSPI / KOSDAQ PER compression

Specifically:
- Foreign flows trim EM weights
- Korean 10-year yields rise in sympathy
- USD/KRW pressure up (with USD strength)
- Korean long-duration assets (growth, biotech) under pressure.

7.2 Relative strength by sector

Sector	Cycle effect	Read
Refining	Crack spreads at all-time highs; direct oil-up beneficiary	Relative strength
LNG / shipbuilding	Hormuz reroute demand, energy security	Relative strength
Defense	Iran + Taiwan risk overlap	Relative strength (already run a lot)
Semiconductors	AI demand is the body; multiple still squeezed	Neutral
Banks	Higher rates expand NIM	Relative strength
Consumer staples	Earnings-driven names (Samyang Foods etc.)	Selective strength
Autos	Already +29% in May	Neutral to weak
Petrochemicals	Naphtha cost up, margin pressure	Relative weakness
Airlines	Oil + rates double-whammy	Relative weakness
High-PER growth	Most sensitive to discount-rate moves	Relative weakness
Biotech	Long-duration squeeze	Weakness

7.3 The “least-priced” pocket of the cycle

As the earlier US-China summit piece argued — the market digested the summit as a “semiconductor news,” while the second-order effect of a Hormuz deal (lower naphtha → recovering petrochemical margins) is still unpriced.

As of today:
→ Semiconductors +39% (MTD): priced
→ Autos +29%: priced
→ Rare-earth theme: catalyst spent
→ Refining / LNG / defense: partly priced
→ Petrochemicals: latent recovery if Hormuz normalizes (NOT priced)
→ Banks: BOK hike potential (NOT priced)

Most asymmetric setups:
- If Hormuz normalizes quickly: petrochemicals, airlines
- If Hormuz drags on: refining, LNG, defense
- If the cycle stops: banks (on a BOK hike).

8. Scenarios and checkpoints

8.1 Three scenarios

Scenario	Premise	US 10-year	Korean equities
A. Cycle ends	Hormuz normalizes, oil sub-90, BOJ delays	4.0–4.3%	Relief in risk. Growth rebounds
B. Cycle persists (base)	Hormuz drags, oil 100–110, BOJ hikes in June	4.4–4.7%	Volatility widens; sector dispersion
C. Cycle accelerates	Hormuz worsens, oil 120+, actual Japanese selling	5.0%+	Broad risk-off, foreign outflows

Base case is B. A requires a Hormuz normalization. C requires confirmed Japanese UST selling.

8.2 Checkpoints (priority order)

1. Hormuz traffic normalization ★★★★★
 - Single most important variable
 - Decides where the cycle ends
 - Watch: daily tanker counts, Iran statements, OPEC+ comments

2. Does the US 10-year settle around 4.6%? ★★★★
 - The visible gauge of term-premium expansion
 - Watch: daily bond markets, auction results

3. Brent in the 95–110 range? ★★★★
 - Decides the strength of first-order transmission
 - Watch: daily oil prices, EIA / IEA weekly inventories

4. BOJ June meeting outcome ★★★
 - The Japan-axis decision point
 - Watch: June BOJ statement, JGB 10-year reaction

5. Japanese UST holdings data (TIC) ★★★
 - Actual confirmation of the marginal-demand erosion
 - Watch: US Treasury monthly data (2-month lag)

6. US May / June CPI / PPI ★★★
 - Whether the cycle keeps running
 - Watch: June 11 May-CPI, June 12 PPI

7. US-China follow-up negotiations ★★
 - Direction of the side variable
 - Watch: trade-committee product lists, rare-earth waiver decision.

9. How this connects to the whole series — every post sits on this cycle

The cycle in this post is the common background for the recent series:

Samsung Electronics piece (Citi target KRW 460,000):
→ "AI is structural demand," so relatively resilient to the cycle
→ But the multiple (rates) still squeezes.

Samsung Electronics piece (strike vs memory supercycle):
→ Price gains partly offset strike losses
→ The price spike itself is a CHILD of this cycle (supply tight + prices up).

Samsung Electro-Mechanics piece (MLCC, FC-BGA):
→ AI thesis intact
→ But at KRW 1.02M, multiple pressure bites directly.

Jeju Semiconductor piece (commodity memory):
→ Memory shortage from AI + the cycle pulls oil up too
→ Double tailwind, but multiple still squeezed.

Pearl Abyss piece:
→ Strong fundamentals, but a KOSDAQ long-duration asset
→ Cycle acceleration = multiple compression risk.

Samyang Foods + consumer-rotation piece:
→ Semi crowding unwinds → capital rotates
→ Stronger cycle = relative tailwind for value / earnings names.

Robotics (Robotis / Rainbow):
→ High-multiple growth
→ Cycle acceleration = the worst duration squeeze.

US-China summit piece:
→ Side-variable analysis
→ The logic of finding the "least-priced" pocket.

Japan PPI piece:
→ Detailed mechanism for the Japan axis of the cycle.

→ Every post is a piece or an application of this cycle.

10. The one-line bottom line

The events of May 14–15 — US April PPI +6.0%, Japan PPI +4.9%, JGB 10-year at a 28-year high, US 10-year 4.46%, 30-year 5.02%, Brent 108, a small-deal US-China summit — are one cycle, not five independent events.

The origin is Iran / Hormuz. The first transmission is oil, the second is US and Japanese inflation, the third is central-bank policy shifts, the fourth is higher US long rates, and the terminus is a repricing of the global discount rate → multiple compression in risk assets. The US-China summit didn’t create the cycle; it can only accelerate or delay it.

Once you draw the picture, every future headline tells you which asset class it will land in. The single most important checkpoint is when Hormuz normalizes. That’s where the cycle ends. If Hormuz normalizes within June, the cycle ends and a risk-on relief move is possible. If it drags, US long rates have more room, and a discount-rate repricing intensifies.

Chasing risk now without checking the cycle is inefficient. First locate the cycle — where it is, where it’s going — then evaluate asymmetric bets at the stock level. Great companies get derated when the cycle accelerates; mediocre companies can rebound when the cycle stops. The cycle comes before the stock.

This article is research and commentary only and is not investment advice. US April CPI / PPI are per the BLS official release. Japan April PPI / import prices are per the BOJ Monthly Report. US 10-year and 30-year yields, JGB 10-year, and Brent prices reflect May 14–15 market data. The Hormuz shut-in and inventory draws are per the IEA Oil Market Report (May 2026) and the EIA STEO (May 2026). The US-China summit outcomes are per Chinese MFA, the White House, Reuters, and Yonhap reporting. Scenarios, checkpoints, and sector views are the author’s judgments and not confirmed outcomes. The cycle description is a general mechanism — actual asset prices respond to many other factors. The timing of Hormuz normalization, the BOJ hike, and any actual Japanese UST selling are all uncertain. The analysis may be wrong. Data cut-off: May 15, 2026 KST.

Jeju Semiconductor — Revenue ₩180.5bn, OP ₩67.1bn, OPM 37.2%, +28% in One Day. Not an 'AI Chip Stock' — a Beneficiary of the 'Legacy Memory Squeezed by AI'

Fri, 15 May 2026 08:30:00 +0900

🔗 Related: Samsung Electronics Citi TP ₩460,000 — Memory-Cycle Frame Reset · KOSPI May 13 V-Reversal — All-Time High Through ₩3.76tn Foreign Selling · Why Korea Part 4 — $6.7B Inflows, Korea-Discount Dissolving or Value Trap?

Follow-up: Is FADU Korea’s Sandisk Beta? Foreign Flows and the AI Storage Bottleneck

Jeju Semiconductor closed +28% on May 14. The trigger is a 1Q26 print that came in extraordinarily strong — revenue ₩180.5bn, OP ₩67.1bn, OPM 37.2%; YoY revenue +273% and OP +1,713%. Most market participants are unfamiliar with what the company actually does. It isn’t Samsung or SK hynix. It doesn’t make HBM. It’s a fabless designer of low-power memory (MCP, LPDDR) for IoT devices, mid-tier smartphones, and automotive electronics. So why is it earning so much money?

TL;DR

1Q26 print. Revenue ₩180.5bn (+273% YoY), OP ₩67.1bn (+1,713%), NI ₩78.1bn. OPM 37.2%.
Stock reaction. +28.4% on May 14, close ₩75,600 (near limit-up ₩76,500). Turnover ~₩760.9bn.
What this company actually does. Not an AI-server HBM business. A fabless designer of low-power memory (MCP, LPDDR) for IoT, mid-tier smartphones, and automotive electronics. No fabs — design only.
Why earnings exploded. Samsung, SK hynix, and Micron concentrated fab capacity on HBM, shrinking supply of ordinary memory (LPDDR4X, legacy DRAM). Demand from IoT, smartphones, and autos kept going — supply down + demand steady = prices up. Jeju captured a large slice of that.
The single test. Is this earnings level repeatable? If 1Q was a one-off peak, the current price is rich. If 2Q sustains it, the price still has room.
Caveats. Customer A = 72.1% of revenue; China = 72.9% of revenue. Customer and geography concentration are extreme. Convertible-bond (CB) dilution overhang exists. The stock is under an “investment caution” designation by the exchange.

1. First — what does Jeju Semiconductor actually do

1.1 A “memory company” but completely different from Samsung / SK hynix

The Samsung Electronics analysis framed memory as the “work desk (DRAM)” and “filing cabinet (NAND).” Samsung and SK hynix operate the giant fabs that produce both, plus the AI-server-grade HBM at the high end.

Jeju Semiconductor plays in a completely different market.

Samsung Electronics / SK hynix:
→ Own their fabs
→ AI servers, PCs, smartphones — large-capacity memory
→ HBM, DDR5, high-performance NAND are the core
→ Global #1-3 megacaps

Jeju Semiconductor:
→ No fab (fabless = designs only, manufacturing outsourced)
→ IoT, mid-tier smartphones, automotive electronics — small-capacity memory
→ MCP, LPDDR, eMCP are the core
→ Market cap \~₩2.6tn (small-to-mid cap)

1.2 MCP, LPDDR — what these are

Two concepts unlock Jeju’s core products.

LPDDR (Low Power DDR): “low-power DRAM.” Uses less power than standard DRAM. Essential for battery-powered devices — smartphones, tablets, IoT.

MCP (Multi-Chip Package): multiple memory chip types packaged into one.

Analogy: a bento box
Standard memory = buying rice, side dishes, soup separately
MCP = rice + sides + soup all in one bento

NAND MCP = NAND (storage) + LPDDR (memory) bundled
eMCP = eMMC (embedded storage) + LPDDR bundled

Why it matters:
→ Fewer components → smaller, cheaper devices
→ Ideal for IoT sensors, smartwatches, mid-tier smartphones
→ Especially useful where space and power are constrained

1.3 Why an “ordinary memory” company suddenly earned this much

In the Samsung analysis, the explanation was that “HBM absorbs fab capacity.” That phenomenon directly impacted Jeju.

The AI-era memory supply structure:

Samsung / SK hynix / Micron fabs:
→ Concentrate output on HBM, DDR5, high-performance server memory
→ These carry much higher prices and margins
→ LPDDR4X and legacy DRAM output declines

Meanwhile, on the demand side:
→ IoT devices keep growing (5G IoT, smart-home, industrial sensors)
→ Mid-tier smartphone demand continues (especially China / SE Asia / LATAM)
→ Auto electronics — memory content per vehicle keeps rising

Supply down + demand steady → prices up

Jeju Semiconductor = specialist designer in that "ordinary memory" segment
→ Price uplift = revenue and earnings explosion simultaneously

One-line summary: Jeju is not an “AI chip stock.” It’s a beneficiary of the “legacy memory squeezed by AI."

2. 1Q26 print — how strong are the numbers

2.1 Headline numbers

Item	1Q26	YoY	QoQ
Revenue	₩180.5bn	+273%	+97%
OP	₩67.1bn	+1,713%	+383%
NI	₩78.1bn	+1,716%	+417%
OPM	37.2%	+29.5pp	+22.0pp
NPM	43.3%	+34.4pp	+26.8pp

Cross-check: OPM = 67.1 / 180.5 = 37.2% ✓

2.2 Why the numbers are remarkable

OP rose 18× YoY. From ₩3.7bn a year ago to ₩67.1bn. Revenue rose 3.7×, but OP rose 18× — extreme operating leverage at work.

What operating leverage means:

Fabless (no-fab) cost structure:
→ Fixed: salaries, office, R&D — doesn't move much with revenue
→ Variable: contract manufacturing, raw materials — scales with revenue

If revenue doubles:
→ Variable cost doubles
→ Fixed cost stays the same
→ Profit grows much more than double

For Jeju:
Revenue +273% → OP +1,713%
→ Revenue 3.7×, OP 18×
→ Operating leverage was extreme

Adding to that, memory prices rose. Same unit volume × higher unit price = revenue rises. Costs don’t rise as fast — margin spikes.

2.3 Why NI exceeds OP

NI (₩78.1bn) is higher than OP (₩67.1bn). Normally NI is lower due to taxes and interest. NI > OP means non-operating income (FX gains, financial income, etc.) was material. The breakdown will appear in quarterly-report footnotes. As a company with >90% export ratio, KRW weakness produces FX gains.

Caveat: if the non-operating gain is one-off, annualizing NI directly is risky.

3. Revenue quality — where and to whom

3.1 Geography — China is 73%

Region	1Q revenue	Share
China	~₩131.6bn	72.9%
Other	~₩48.9bn	27.1%

3.2 Customer — Customer A is 72%

Customer	1Q revenue	Share
Customer A	~₩130.1bn	72.1%
Other	~₩50.4bn	27.9%

3.3 Application (2025 annual basis)

Application	Share	YoY growth
IoT	51.9%	+46%
Consumer	26.3%	+400%
Mobile	8.2%	+16%
Auto	8.0%	+126%
Network	5.5%	+6%

3.4 What the data says

The positive: IoT, consumer, and auto are growing simultaneously. Consumer at +400% YoY is striking. Auto at +126%.

The concerning: Customer A alone = 72%. China alone = 73%. Extreme concentration in one customer + one geography. If that customer reduces orders or China-bound exports face friction, earnings collapse fast.

The “200 global customers” narrative exists, but the data shows the core is one customer. Whether that customer’s order pattern repeats is the central 2Q variable.

4. Technical moat — real but not absolute

4.1 What Jeju is good at

Strength	Description
MCP productization	Designs combining NAND + LPDDR in a single package. Optimal for space/power/cost-constrained IoT and mobile
Platform certification	Qualcomm / MediaTek 5G IoT chipset certification. Without this, you can’t supply memory compatible with those chipsets
Multi-SKU coverage	200+ customers, covering IoT, mobile, auto, networking
Outsourced production network	Powerchip, Winbond partners. Mass production without owning fabs

4.2 What Jeju is not good at

Weakness	Description
No own fab	Dependent on partner capacity. When supply is tight, you can’t make more even if you want to
Not cutting-edge	LPDDR4X / legacy on mature processes, not HBM, DDR5, or latest LPDDR5X
Pricing power is cycle-dependent	Earns big when supply is tight; loses pricing when supply normalizes

4.3 Core read

Jeju’s moat exists — platform certification, MCP productization, customer-handling experience are not easily replicated. But it’s not the absolute moat that Samsung / SK hynix HBM or TSMC leading-edge processes offer. If large memory players re-engage with legacy or Chinese players catch up, competition intensifies.

So Jeju is more accurately framed as “a niche memory specialist whose earnings explode when cycle and product mix align," not as a “tech-monopoly” name.

5. Fair value — looks cheap but contains a trap

5.1 Simple math

1Q NI ₩78.1bn × 4 = ₩312.4bn (naive annualization)
Market cap \~₩2.6tn ÷ ₩312.4bn = PER \~8.3×
→ PER 8× looks very cheap

But this math is dangerous. If 1Q is the cycle peak, annualization is meaningless. 2Q earnings could collapse.

5.2 Scenario fair value

Scenario	2026 revenue	OPM	OP	NI estimate	EPS	Applied PER	Implied price
Bear (1Q peak)	₩540.0bn	24%	₩129.6bn	₩95.0bn	₩2,758	15×	~₩41,000
Base	₩680.0bn	31%	₩210.8bn	₩165.0bn	₩4,790	16.5×	~₩80,000
Bull (cycle sustained)	₩820.0bn	35%	₩287.0bn	₩225.0bn	₩6,532	18×	~₩118,000

Cross-checks: Base EPS = ₩165bn / 34.44M shares ≈ ₩4,790 ✓; implied price = 4,790 × 16.5 = 79,035 ≈ ₩80,000 ✓

5.3 Where the current price sits

₩75,600 is very close to the base-case implied price (₩80,000). Upside ~+5.8%.

Upside / downside:
Bear case: -46%
Base case: +6%
Bull case: +56%

→ Even if the base case is right, additional upside is small
→ If bear case is right, the drawdown is large
→ Chasing the day after +28% is unfavorable risk-reward

6. Caveats — three risks

6.1 Customer / geography concentration

Customer A = 72.1%, China = 72.9%. One customer’s order pattern drives the whole P&L. Whether Customer A is a long-term contract or a one-off large order isn’t disclosed.

6.2 Convertible-bond (CB) dilution

Jeju issued CB and BW (warrants) totaling ₩117.0bn. Conversion / exercise price ₩44,300, ~2.64M latent shares (~7.7% of existing share count). With the stock well above the strike, conversion incentive is strong — diluting EPS.

CB dilution effect:

Current share count: \~34.44M
Latent conversion: \~2.64M (7.7%)

If converted:
→ Shares 34.44M → 37.08M
→ EPS \~7% lower

6.3 Investment-caution designation

KRX has designated Jeju as an investment-caution stock. Further sharp rises can trigger a 1-day trading halt (investment-warning). This affects short-term flow independent of earnings.

7. Flow — foreigners selling, retail buying

May 14 flow:

Participant	Net
Foreign	-₩13.4bn (net sell)
Institutional	+₩9.5bn
Retail	+₩5.0bn
Program	-₩14.7bn (net sell)

Foreigners and program sold; retail bought. This is closer to “retail-driven momentum buying” than “clean institutional accumulation.” Sustained move requires foreign / program flow to turn net-positive.

The May 13 KOSPI V-reversal piece noted the same pattern at the index level — ₩3.76tn foreign selling absorbed by domestic flows. Jeju mirrors that structure: domestic capital catching what foreigners drop. How long it lasts is the key short-term variable.

8. What to watch — 2Q decides

8.1 1Q vs 2Q

Item	1Q (confirmed)	2Q (to verify)
Revenue	₩180.5bn	≥₩170bn = positive; <₩130bn = 1Q-as-peak confirmed
OPM	37.2%	≥33% = structural; <25% = cycle peak
Customer A orders	72.1%	Repeatability is core
Memory pricing	Up	Sustained / rising = good; reversal = bad

8.2 Key triggers

Trigger	Window	Why
LPDDR4X price trajectory	Monthly	Prices holding → 2Q stays strong; rolling over → margin collapses
2Q print	August	Verdict on 1Q (one-off vs structural)
Proprietary LPDDR4X mass production	2H onward	Transition from sourcing margin to design-margin
Investment-caution removal	Days	Direct effect on short-term flow
CB / BW conversion activity	Ongoing	EPS-dilution monitoring

8.3 Linking back to the Samsung analysis

The Samsung piece argued “HBM absorbs fab capacity → legacy memory prices rise.” Jeju is the direct beneficiary of that legacy price uplift.

Samsung: HBM / DDR5 / SOCAMM2 → makes and sells directly
Jeju: LPDDR4X / MCP → benefits where Samsung isn't producing

Samsung AI memory capex → legacy memory undersupply → Jeju pricing leverage

This linkage breaks if:
→ Samsung / SK hynix resume legacy production at scale, OR
→ LPDDR4X demand contracts, OR
→ Chinese memory players ramp supply

9. Bottom line

Jeju Semiconductor’s 1Q26 print is strong. Revenue ₩180.5bn, OP ₩67.1bn, OPM 37.2%. OP up 18× YoY. The numbers themselves are undeniable.

Whether this is a “repeatable structure” or a “one-off cycle peak” is settled in 2Q. Customer A = 72%, China = 73% are extreme concentrations. CB dilution and the investment-caution designation are real risks. After a +28% day, the chase upside is only ~+6% even in the base case.

Jeju is not an “AI chip stock.” It’s a beneficiary of the “legacy memory squeezed by AI." How long that benefit lasts depends on LPDDR4X pricing, Customer A’s order repeatability, and the pace of transition to proprietary-design products. The most disciplined approach is to wait for the 2Q print (August) and confirm OPM staying ≥33%.

Whether 1Q is the “start” or the “peak” — that answer hasn’t arrived yet.

FAQ

Q: Is Jeju Semiconductor an AI chip company? A: Not directly. It doesn’t make AI-server HBM or cutting-edge GPU memory. It’s a fabless designer of low-power memory (MCP, LPDDR) for IoT, mid-tier smartphones, and automotive electronics. But Samsung/SK hynix concentrating capacity on HBM tightened supply of ordinary memory, which lifted prices and benefited Jeju.

Q: Is OP rising 18× from ₩3.7bn to ₩67.1bn normal? A: Two effects coincided. (1) Revenue rose +273%. (2) Memory prices rose. Fabless has small fixed costs, so revenue growth produces disproportionate earnings growth (operating leverage). But the structure runs in reverse too — revenue declines compress earnings disproportionately.

Q: After a +28% day, is it safe to buy now? A: Base-case implied price is ~₩80,000. Vs current ₩75,600, upside is only ~+6%. Bear-case downside (1Q-as-peak) is -46%. The asymmetry is unfavorable for chasers. Waiting for the 2Q print is more disciplined.

Q: Who is Customer A? A: Not disclosed. Only the 72.1% revenue share appears in the quarterly report. Market speculation points to global 5G IoT chipset-related OEMs (Qualcomm / MediaTek ecosystem) or Chinese IoT/mobile OEMs. Whether Customer A’s order pattern repeats or proves one-off is the central 2Q variable.

Q: What does the investment-caution designation mean? A: KRX designates stocks with sharp short-term gains. Further sharp rises can escalate to “investment-warning” with a 1-day trading halt. Affects short-term flow regardless of earnings.

Q: How significant is the CB dilution? A: Latent conversion is ~2.64M shares (7.7% of existing count). With the strike at ₩44,300 and the stock well above, conversion incentive is strong. Conversion lowers EPS ~7%. Fair-value calculations should reflect dilution for conservatism.

Q: What’s the fastest signal in 2Q? A: Two. (1) Monthly LPDDR4X market pricing — sustained or rising. (2) 2Q print (August) — revenue ≥₩170bn and OPM ≥33%. Both confirm “structural”; either failing supports “1Q-as-peak.”

This article is for research and informational purposes only and does not constitute investment advice. 1Q figures based on the DART quarterly report (submitted 2026-05-14). Customer and geographic concentration percentages are from the quarterly report; Customer A’s identity is undisclosed. Application-mix figures are 2025 annual basis (USD), not 1Q. Fair-value scenarios are analyst estimates; the company has not published guidance. Public consensus is thin, limiting quantitative versus-consensus comparison. CB/BW dilution scale (~2.64M shares, 7.7%) is from press reports; exact conversion conditions require separate verification. The stock is under an investment-caution designation, with potential trading halt on further sharp rises. May 14 flow showed foreigners and program as net sellers. Analysis can be wrong. Data cut: May 15, 2026 KST.

KOSPI Nears 8,000 as Retail Absorbs Foreign Selling

Thu, 14 May 2026 23:30:00 +0900

KOSPI Closes at 7,981 — But the Quality of This Rally Deserves Scrutiny

South Korea’s KOSPI benchmark index climbed 1.75% to 7,981.41 on May 14, 2026, while the tech-heavy KOSDAQ gained 1.20% to 1,191.09. On the surface, it looks like a clean risk-on session. Beneath it, the story is more complicated — and understanding that distinction matters for investors tracking Korean equities.

Foreign investors sold a net 2.2 trillion KRW on the KOSPI and 170 billion KRW on the KOSDAQ during the session. Retail buyers absorbed almost all of it. That dynamic — strong price action driven by domestic retail rather than institutional or foreign conviction — is a meaningful quality flag. Korean markets have historically been more durable when foreign and institutional flows align with price direction. Today, they did not.

The rally was real. The leadership was not broad.

Samsung Electronics Breaks to a Near 52-Week High

The session’s headline mover among large caps was Samsung Electronics (005930.KS), South Korea’s largest company by market cap and the world’s leading memory chip and smartphone manufacturer. Shares rose 4.23% to 296,000 KRW, putting the stock within 1.2% of its 52-week high. Unusually for a retail-dominated session, Samsung Electronics drew simultaneous net buying from foreign investors (+223.3 billion KRW), institutional investors (+74.6 billion KRW), and program trading (+1.75 trillion KRW).

That triple-flow convergence is rare and tends to be a reliable signal. The key technical level now is 300,000 KRW — a psychological and chart milestone. Whether Samsung can close above and hold that level in the coming sessions will define whether this move extends or stalls.

The catalyst backdrop supports the move. Expectations around a potential US-China summit have reduced near-term geopolitical risk premium for Korean tech exporters. Micron Technology’s continued strength in US markets is providing positive read-through for Korean DRAM and NAND suppliers. The memory supercycle narrative — built on HBM demand for AI training infrastructure and recovering NAND pricing — remains the dominant institutional thesis for Korean semiconductors in 2026.

The one outstanding risk for Samsung is labor. Union negotiations remain unresolved, and any escalation could disrupt production cadence at a moment when the market has priced in execution.

Jeju Semiconductor: Q1 Earnings Validate the MCP Thesis

The sharpest single-day move among stocks in the analysis universe was Jeju Semiconductor (080220.KQ), a South Korean fabless chip designer specializing in multi-chip package (MCP) memory products. Shares surged 28.35% to 75,600 KRW on volume of 815.7 billion KRW, following Q1 2026 results that beat expectations: revenue of 180.5 billion KRW and operating profit of 67.1 billion KRW.

MCP products — which stack DRAM and NAND dies in a single package for mobile and IoT applications — are a direct beneficiary of rising DRAM spot prices. Jeju Semiconductor’s Q1 print confirms that the pricing tailwind is flowing through to smaller, more agile fabless players, not just the memory giants.

The caveat: foreign investors sold 13.4 billion KRW net on the session, and program trading was also negative. A 28% single-day gain following an earnings catalyst is historically prone to give back 5–10% in the days immediately following. For investors interested in the MCP/DRAM pricing theme but wary of chasing, the setup to watch is a controlled pullback with sustained trading volume and decelerating foreign outflow.

Daedeok Electronics Continues Its Breakout Run

Daedeok Electronics (000240.KS), a printed circuit board and substrate manufacturer that supplies Samsung Electronics and other memory chip producers, extended its recent breakout with a further gain of 1.19% to 144,600 KRW. The five-day return stands at +19.8%. Relative strength ranking is 98.2 — placing it among the top 2% of KOSPI/KOSDAQ stocks by momentum.

The catalyst that triggered this move — a new capital expenditure disclosure filed with South Korea’s DART (Data Analysis, Retrieval and Transfer System) — is still being digested by the market. When a substrate supplier announces facility investment alongside rising memory prices and strong Samsung Electronics flow, the read-through is straightforward: production capacity is being built to meet expected demand. The stock is confirming that thesis with price.

The level to monitor is 140,000 KRW. A close below that on sustained volume would suggest the breakout is exhausting itself.

Game Sector Rotation: Earnings Confirmation Matters

South Korea’s gaming sector showed a telling divergence on May 14. Pearl Abyss (263750.KQ), the developer behind the Black Desert franchise, fell 2.50% to 50,600 KRW and has declined 3.62% over the past five sessions. Short interest sits at 15.3%, and institutional investors were net sellers during the session.

Meanwhile, Mirae Asset published a research note upgrading NCSoft to a buy recommendation based on a Q1 2026 earnings beat and new title momentum. That kind of intra-sector rotation — capital moving from pre-revenue catalyst names toward earnings-confirmed ones — is a classic pattern in Korean small and mid-cap gaming stocks.

For Pearl Abyss, the thesis still rests on new title optionality. But with no confirmed release timeline and institutional flows turning negative, the risk-reward asymmetry is deteriorating relative to gaming peers with near-term catalysts already in the numbers. The 50,000 KRW level is now the critical support to watch.

Samsung Electro-Mechanics and NAVER: Divergent Signals Worth Tracking

Two large-cap names showed conflicting flow patterns that merit attention.

Samsung Electro-Mechanics (009150.KS), a leading manufacturer of multilayer ceramic capacitors (MLCCs) and camera modules, declined 0.49% to 1,024,000 KRW despite a five-day return of +11.67%. Today’s session saw institutional buying of 46.4 billion KRW offset by foreign selling of 40.2 billion KRW and program outflows of 49.3 billion KRW. That institutional-versus-foreign divergence at a recent high is a pattern consistent with domestic fund rotation rather than fresh conviction. Until foreign flow turns positive, the stock appears to be in consolidation mode after a sharp rally.

NAVER (035420.KS), South Korea’s dominant internet and search platform, jumped 5.71% to 213,000 KRW — but foreign investors sold 35.7 billion KRW net and institutions sold another 15.6 billion KRW. Price up, flows down. That combination typically resolves in one of two ways: either the selling pressure absorbs and the stock reaccelerates, or the rally fades as the buyers who drove it take profits. The next two to three sessions will clarify which.

Macro Overlay: Risk-On, But With Friction

The macro backdrop complicates a straightforward risk-on read. USD/KRW held in the 1,490 range during the session — a level that historically pressures Korean exporters’ earnings guidance and creates hesitation for foreign equity allocators. Brent crude at $105 per barrel adds to cost-side concerns for energy-intensive manufacturers.

Both of these factors help explain why foreign investors were net sellers even as Korean stocks rallied. The price action was domestic-driven. That is not inherently bad — retail demand is real demand — but it does suggest the current move is more fragile than the index level implies.

Korean equity markets tend to sustain rallies most durably when foreign flows are either neutral or positive. On May 14, that condition was not met.

What to Watch Going Into May 15

The near-term watchlist for Korean equity investors is clean and specific:

Samsung Electronics at 300,000 KRW: A confirmed close above this level on institutional or foreign buying would be a meaningful breakout signal for the broader semiconductor sector.
Jeju Semiconductor pullback: The Q1 earnings story is real. The entry timing after a +28% session is not. A 5–10% decline with volume contraction would represent a more constructive risk setup.
Pearl Abyss at 50,000 KRW: If this support breaks on volume, the sector rotation away from pre-catalyst gaming names will likely accelerate.
Macro: USD/KRW and Brent: If both remain elevated, the quality of any further KOSPI rally will stay constrained by foreign selling pressure.
Kiwoom Securities (039490.KS): The brokerage disclosed a share buyback and cancellation program alongside an increase in short-term borrowings on the same day. Shareholder return programs are constructive; the debt detail warrants close reading before drawing conclusions.

Today’s session was constructive for Korean semiconductor names and broadly positive for the headline index. The structural concerns — elevated currency pressure, elevated crude, and foreign outflows — have not resolved. Investors with existing exposure to Korean memory and AI infrastructure plays have reason to stay engaged. Those considering new positions would do well to wait for a session where institutional and foreign flows confirm, rather than contradict, the price action.

Korea Quality Re-Rating Watch 2026-05-14: APR, TES and TLB — Earnings Lead While Retail Carries the Tape

Thu, 14 May 2026 16:30:00 +0900

Section 1 — Macro Dashboard

Indicator	Level	5-Day Change	Signal
KOSPI	7,981.4	+6.5%	Bullish
KOSDAQ	1,191.1	−1.4%	Flat
USD/KRW	1,492	+2.6%	KRW weak
VIX	17.9	+3.8%	Stable
US 10Y	4.48%	+0.09pp	Flat
Brent	$105.4	+1.2%	Stable

Regime: Korea = Neutral | US = Bull. The 7.8-percentage-point divergence between KOSPI’s 5-day breadth (50MA: 46%) and KOSDAQ’s suggests large-cap defense is doing the heavy lifting. The KRW at 1,492 and Brent holding above $100 remain structural headwinds for import-heavy industries and overall market quality.

Section 2 — Market Wrap

Korea’s tape on May 14 delivered a risk-on result with a foreign-selling asterisk. KOSPI closed at 7,981.41 (+1.75%) and KOSDAQ at 1,191.09 (+1.20%), but the engine was retail — foreigners sold a net ₩2.2 trillion on KOSPI and ₩170 billion on KOSDAQ. Individual buyers absorbed the supply. That keeps today’s session one grade below a clean institutional-led advance.

Sector leadership was narrow. Semiconductors and memory were the clear frontrunner, driven by ongoing US-China trade optimism and Micron strength overnight. Food and consumer names held up well, with the inflation/staples bid staying intact at Brent $105. Insurance and securities saw technical bounces after recent underperformance.

Weakness clustered in cyclical capex plays. Shipbuilding, heavy machinery, and utilities underperformed — a notable divergence from recent weeks, when those sectors attracted some of the strongest institutional flow. The rotation suggests the market is trimming winners in rate-sensitive industrials as the 10-year US Treasury holds at 4.48%.

Flow quality is the key caveat. The most important single data point today was Samsung Electronics (005930.KS) drawing simultaneous foreign, institutional, and program buying — a rare clean-sweep setup — while SK Hynix (000660.KS) saw ₩1.6 trillion in net foreign selling. Within the same semiconductor theme, differentiated flow signals that the market is still stock-picking rather than sector-rotating in bulk.

Semiconductor equipment and components continued to attract institutional interest. Duksung Electronics (000150.KS) remained among the strongest relative strength names, up 5-day +19.8%, with capex announcement momentum intact.

Outside the semiconductor complex, the food sector held a quiet bid. Samyang Foods (003230.KS) continued to see combined institutional and foreign inflows — confirmed in today’s Smart Money screeners — as global demand for Korean food exports shows no sign of slowing.

In gaming, intra-sector rotation was visible. Mirae Asset’s morning note flagged NC Soft (036570.KS) as a Q1 earnings beat with fresh catalyst, pulling attention toward proven earnings names and away from titles still awaiting new game launches.

Bottom line on market character: direction positive, but the foreign-selling overhang, KRW near 1,500, and oil above $100 mean the risk-on quality remains one notch below a conviction-grade rally. New money should be selective; trailing momentum stays with semiconductor and consumer quality.

Section 3 — Today’s Quality Re-Rating Candidates

The meta-screener processed 125 tickers and ranked 20 article candidates. Seven names hit 3+ screener overlap — the threshold for leading the discussion.

Top Candidates Table

Meta Rank	Ticker	Name	Meta Score	Screeners Hit	Overlap	Key Metrics
1	278470.KS	APR (에이피알)	85.3	QC + SMQ + SME + PEAD	4	ROE 65%, OP YoY +198%, F+QI 5d +₩91.7B
2	187870.KQ	Device (디바이스)	63.8	QC + CR + PEAD	3	ROE 11%, OP YoY +312%, margin +11.7pp
3	095610.KQ	TES (테스)	59.9	QC + SMQ + SME	3	ROE 14.5%, OP YoY +50%, F+QI 5d +₩56.4B
4	368770.KQ	FiberPro (파이버프로)	58.1	QC + CR + PEAD	3	ROE 22%, OP YoY +40%, margin +1.4pp
5	098120.KQ	Microcontact Sol (마이크로컨텍솔)	52.4	QC + CR + PEAD	3	ROE 23%, OP YoY +74%, margin +3.0pp
6	356860.KQ	TLB (티엘비)	51.9	CR + SMQ + SME	3	OP YoY +665%, margin +8.2pp, F+QI 5d +₩21.7B
7	322000.KS	HD Hyundai Energy Solution	45.3	SMQ + SME	2	OP YoY +1077%, F+QI 5d +₩60.6B

Top 3 Deep-Dive

APR (278470.KS) — Meta Rank #1, Score 85.3

APR manufactures skincare devices and beauty consumer products, with a brand portfolio that has scaled aggressively into Asia and beyond. The reason it leads every screener today is structural: ROE of 65%, operating profit YoY growth of +198%, and revenue nearly doubling at +111% YoY. That combination clears the Quality Compounder filter at the highest tier. Smart Money screeners confirm the story isn’t hypothetical — foreign and quality-institutional flows combined for +₩91.7 billion over the past five sessions. The PEAD signal adds a timing layer: the May 7 preliminary earnings release (confirmed via DART) triggered the drift window, and the stock remains within that window. One caution: short interest at 11.1% and retail-net-selling flow (금투수급 negative) suggest some distribution at the margin. The next check is whether the institutional bid holds above the May 7 print after short sellers test it.

Device (디바이스, 187870.KQ) — Meta Rank #2, Score 63.8

Device is a specialty machinery manufacturer, and the numbers are stark: operating income grew +312% YoY on revenue growth of +78%, with margin expansion of 11.7 percentage points. That is textbook operating leverage — fixed-cost absorption amplifying revenue gains into disproportionate profit. The Cycle Rerating screener picked it up exactly for this pattern. The PEAD signal confirms a recent beat, and the Quality Compounder screen validates that the underlying business economics are durable rather than a one-period spike. A April bonus share (권리락) event was filed via DART, which may be contributing to recent volume. The key variable to track: whether the margin expansion is capacity-constrained or structural. If next quarter shows a second margin step-up, the re-rating argument strengthens considerably.

TES (테스, 095610.KQ) — Meta Rank #3, Score 59.9

TES designs and manufactures semiconductor deposition and etch equipment, placing it directly in the capex cycle tied to domestic fab investment. Operating profit grew +50% YoY with ROE at 14.5% and revenue +46% — solid, if less explosive than the top two names. The Smart Money confirmation is the decisive factor here: quality-institutional net buying of +₩56.4 billion over five sessions, while foreign flow was net negative. That institutional-without-foreign pattern sometimes precedes a broader re-rating once foreign conviction catches up. Short interest at 12.2% and negative program flows add caution — this is a name where the institutional thesis needs to hold for another quarter of data before the short position fully capitulates. The Q1 2026 quarterly report was filed May 13 (DART), making now the earliest point to verify whether the earnings trajectory held into the most recent period.

All screener data as of 2026-05-14. This is not investment advice. Screener signals are candidates for further research, not trade recommendations.

KOSPI May 13 V-Reversal — All-Time High 7,844 (+2.63%) Through ₩3.76tn Foreign Selling, on a Day That Should Have Been Risk-Off

Thu, 14 May 2026 11:00:00 +0900

🔗 Related: KOSPI May 12 selloff / rebound — AI national-dividend / windfall-tax debate · Why Korea Part 4 — $6.7B inflow vs. value-trap setup · Korean Semis May 6 Rally Synthesis · Korea Daily Market Hub · Samsung Electronics Citi TP ₩460,000

The KOSPI opened May 13 at -1.69%, slid toward 7,400, then closed at 7,844.01 (+2.63%) — a fresh all-time high. Foreign investors net-sold ₩3.76tn, and the index made a new high anyway. The KOSDAQ closed -0.20% and local-DB breadth was 955 advancers vs. 1,455 decliners — 500 more stocks fell than rose. “Index hits new high, market feels weak” is the actual signature of the session.

TL;DR

The biggest oddity is the intraday V-reversal. Open -1.69%, slide toward 7,400, close 7,844.01 (+2.63%) at a new all-time high.
But the KOSDAQ closed -0.20%, and local-DB breadth was 955 up vs. 1,455 down — this is a compressed large-cap rally, not a broad rally.
Foreigners net-sold ₩3.76tn but retail (+₩1.89tn) and institutions (+₩1.69tn) absorbed it. Domestic liquidity pulled the index up.
The leadership axis was semis large-caps + substrates + autos. Conversely, batteries, parts of nuclear / power, gaming, and individual growth stocks lagged.
The macro backdrop was outright unfriendly, yet the index rallied. Hot US CPI, WTI back above $100, Middle East tension, soft prior-day SOX. Macro lost to flow and large-cap news flow.
The three things to verify today (May 14) — Samsung Electronics / SK hynix foreign selling stopping, Samsung Electro-Mechanics / Daeduck Electronics institutional buying continuing, KOSDAQ catching up. If any one fails, May 13 was a “large-cap short-cover + flow-reversal rally” rather than the start of a new leg.

1. What happened — the intraday picture

1.1 Time-of-day KOSPI movement

09:00 open: -1.69% (vs. prior close 7,643)
Morning: slide toward \~7,400 (estimated low \~ -2.5%)
Afternoon: rebound despite continued foreign selling, on retail + institutional buying
14:00+: acceleration
15:30 close: 7,844.01 (+2.63%, all-time high)

Intraday range: \~4.3% from low to close

Closing at a new all-time high is the strongest move since the May 6 +6.5% session (when KOSPI cleared 7,000). The critical difference: May 6 saw foreign net buying of ₩3.13tn; May 13 saw foreign net selling of ₩3.76tn.

1.2 May 13 market-level flow

Market	Close	Daily change	Retail	Institutions	Foreigners
KOSPI	7,844.01	+2.63%	+₩1,890bn	+₩1,690bn	-₩3,760bn
KOSDAQ	1,176.93	-0.20%	—	—	—

Combined domestic net buying (retail + institutional) ≈ ₩3.58tn vs. foreign net selling ₩3.76tn — almost 1:1 absorption.

1.3 Index-vs-breadth divergence

Local-DB breadth May 13:
Total listed: \~2,827
Advancers: 955 (33.8%)
Decliners: 1,455 (51.5%)
Flat: \~417 (14.7%)

→ 500 more stocks fell than rose
→ Index +2.63% all-time high, but more than half the market was down

This is the operational definition of “compressed large-cap rally.” The top 10-20 names by market cap pulled the index, while most mid- and small-caps stayed out.

2. Leadership stocks — where the money went

2.1 The six names that led

Name	Close	Daily	Flow read
SK hynix	₩1,976,000	+7.68%	Heavy foreign selling, absorbed by institutions +₩907.7bn and retail
Samsung Electronics	₩284,000	+1.79%	Foreign -₩1.55tn, institutions +₩456.1bn, retail +₩1.01tn
Samsung Electro-Mechanics	₩1,029,000	+7.41%	Foreign + institutional combined ~+₩188bn. Cleanest flow signature
Daeduck Electronics	₩142,900	+11.64%	Institutional +₩40.1bn. AI substrate diffusion beneficiary
Hyundai Motor	₩710,000	+9.91%	Foreign + institutional combined ~+₩428.8bn, retail -₩450bn
Hyundai Mobis	₩649,000	+18.43%	Auto-parts large-cap absorbing rotation

The read is direct:

Samsung Electronics + SK hynix = index-defense axis. The #1 and #2 market-cap names had to move together for KOSPI to print a new all-time high.
Samsung Electro-Mechanics + Daeduck Electronics = AI-component / substrate diffusion axis. The names from Korea AI PCB Ecosystem followed semis higher.
Hyundai Motor + Hyundai Mobis = non-semis large-cap rotation axis. Foreigners did not sell Korea wholesale; they sold parts of memory and rotated some into auto large-caps.

2.2 What didn’t participate — the underside of breadth

Sector / Group	May 13 behavior	Probable reason
Battery (Ecopro, LG Energy Solution, etc.)	Weak or flat	US IRA-policy uncertainty, Chinese LFP competition
Nuclear / power (selected)	Weak	Profit-taking after early-May rally
Gaming (Pearl Abyss, etc.)	Weak	Profit-taking after Pearl Abyss 1Q26 guidance disclosure
Individual growth / biotech	Mixed	Pre-ASCO/EASL wait-and-see

Those laggards are what produced KOSDAQ -0.20% and the 1,455 down-tickers.

3. The macro backdrop was unfriendly — why did this rally happen?

3.1 The risk-off inputs going into May 13

Variable	Detail	Pre-Korea-open impact
US April CPI	+0.4% MoM, above expectations	Fed-cut-delay concerns
WTI oil	Back above $100/bbl	Inflation / cost-side pressure
Middle East	Iran / Israel tension persists, Hormuz throughput constrained	Energy cost + logistics risk
US semis index (SOX)	Prior day -1.8%	Korea semis follow-through risk
FX	KRW/USD stuck at 1,470s	Foreign-flow burden

In a normal regime, this combination produces a -2% to -3% Korea open. Open was -1.69%. The intraday V-reversal that followed → all-time-high close. Macro variables didn’t set the price in this session.

3.2 Why was macro ignored — three hypotheses

Hypothesis A — self-reinforcing flow. With KOSPI in record-high territory after the May 6 +6.5% session, dry-powder waiting to “buy any pullback” was thick. -1.69% open → bid trigger → foreign supply absorbed.

Hypothesis B — large-cap news flow. Samsung Electronics Citi TP raise to ₩460,000 on May 11; multiple foreign-broker memory-cycle re-rating reports across May 12-13; refreshed SK hynix HBM share work — concentrated upgrade flow in the same week. Single-name news flow outweighed macro.

Hypothesis C — Hyundai Motor 1Q earnings + autonomy momentum. Hyundai Motor +9.91% and Hyundai Mobis +18.43% are hard to explain by simple sector rotation alone. Coverage on autonomous driving / robotaxi, plus a reframing of Hyundai Motor’s US-plant IRA exposure, likely added single-name catalysts.

The most defensible read is that all three operated in parallel. None of the three alone explains the -1.69% → +2.63% (4-point reversal).

4. The ₩3.76tn foreign selling — dissected

4.1 What foreigners actually sold and bought

Group	Foreign direction	Est. scale
Samsung Electronics	Net sell	-₩1,550bn
SK hynix	Net sell	Large
Samsung Electro-Mechanics	Net buy	+
Daeduck Electronics	Partial buy	Small
Hyundai Motor	Net buy	+₩100s of bn
LG Display	Net buy	+
Battery (selected)	Net sell	-
KOSDAQ broadly	Net sell	-

Foreigners didn’t sell Korea wholesale. Memory mega-caps were sold; AI components (Samsung Electro-Mechanics, Daeduck Electronics) and non-semis large-caps (Hyundai Motor) saw some buying. The absolute net-sell number (-₩3.76tn) was large, but direction wasn’t one-way.

4.2 Linking to Why Korea Part 4

Why Korea Part 4 framed the paradox of $6.7bn YTD ETF inflow to Korea with forward PER still below the 10-year average. “Earnings outran price” was the resolution. May 13’s foreign selling can be read as short-term profit-taking inside that cumulative inflow context.

YTD 2026 Korea ETF net inflow: \~$6.7bn (\~₩9.5tn)
May 13 foreign KOSPI net sell: -₩3.76tn

→ May 13 sell ≈ 40% of cumulative YTD inflow
→ Large as a single-session print, but partial-profit-take in cumulative terms
→ Single-session data is insufficient to declare a trend reversal

The key is whether foreign selling stops or turns to buying today (5/14) and through the rest of the week. May 7 saw a record single-day ₩7.15tn foreign net-sell that flipped back to buying immediately after. Single-session prints don’t define trends.

5. Compressed rally vs. broad rally — what’s the difference?

5.1 Side-by-side

Feature	Broad rally	Compressed rally (May 13’s pattern)
Index move	+2-3%	+2.63%
Breadth	Advancers > decliners (60:40+)	Decliners > advancers (33:51)
KOSDAQ participation	Strong	-0.20%, lagging
Foreign direction	Net buy	-₩3.76tn net sell
Volatility profile	Stable ascent	Intraday V-reversal
Durability	1-2 weeks+	Short-term rebound

May 13 is clearly a compressed rally. A broad rally requires KOSDAQ participation and foreign buying — neither was present.

5.2 Signals that end a compressed rally

Signal 1: Large-caps stop adding gains, go flat
 → Top-10 names go sideways
 → Compression energy exhausted

Signal 2: KOSDAQ strong recovery
 → 1,176 → 1,200+
 → Capital diffuses to mid/small-caps (= broad-rally transition)

Signal 3: Foreign selling accelerates or stops
 → Acceleration → compressed rally also ends
 → Stop / reversal → sustainable mode

Which of these three surfaces first determines the next 1-2 weeks for KOSPI.

6. Analytical implications — “what’s working works harder”

6.1 Where the money is concentrating

The structure that produced the May 13 close:

Cluster	Names	Common feature
Semis flagships	Samsung Electronics, SK hynix	AI memory super-cycle + foreign-broker TP raises
AI substrate / SUMs	Samsung Electro-Mechanics, Daeduck Electronics, ISU Petasys	Memory → substrate (bottleneck diffusion)
Auto large-caps	Hyundai Motor, Hyundai Mobis	Autonomy / robotaxi momentum + US market
Display	LG Display	OLED + automotive display

The four clusters share: (a) top market-cap, (b) recent foreign-broker target-price upgrades, (c) high earnings visibility.

The names that lagged share the opposite features — (a) small / mid-cap, (b) thin foreign coverage, (c) priced more on expectation than delivery: battery (selected), gaming, biotech, parts of nuclear / power.

6.2 Signals from an analytical perspective

The 5/13 all-time high is the result of macro-ignored compression flow. Near-term momentum is strong, but the conditions to sustain it are demanding.
Whether foreign selling is a trend or a single-session print isn’t determinable from one day. 5/14 through 5/16 require additional observation.
Without KOSDAQ catching up, this can’t be called a “broad rally." Compressed rallies are inhospitable to chasing large-caps and unhelpful to mid/small-caps.

6.3 Three checks for today (5/14)

Does foreign selling on Samsung Electronics / SK hynix stop? Stop = continuation candidate; accelerate = 5/13 was one-off.
Does institutional buying on Samsung Electro-Mechanics / Daeduck Electronics persist? The AI-component diffusion axis verification.
Does KOSDAQ catch up? Capital diffusion / broad-rally transition signal.

If any of these three fails, May 13 reads as “large-cap short-cover + flow-reversal rally”. All three triggering = trend-shift candidate.

7. Bottom line

May 13’s KOSPI 7,844.01 all-time high was a V-reversal that ignored macro headwinds. Foreign -₩3.76tn was absorbed by retail +₩1.89tn and institutions +₩1.69tn. SK hynix +7.68%, Samsung Electro-Mechanics +7.41%, Daeduck Electronics +11.64%, Hyundai Mobis +18.43% — large-caps did the work. KOSDAQ closed -0.20% with 500 more decliners than advancers.

This is a compressed large-cap rally, not a broad rally. Macro lost to flow and single-name news flow. Structurally, this widens the gap between names that work and names that don’t.

The verification today (5/14) is simple. Does foreign selling stop, do institutions keep buying Samsung Electro-Mechanics / Daeduck Electronics, does KOSDAQ catch up. All three = the start of a trend shift. Any failing = single-session compression rally. The answer arrives within 1-2 sessions.

FAQ

Q: How does the KOSPI hit a new all-time high while foreigners sell ₩3.76tn? A: Retail (+₩1.89tn) and institutions (+₩1.69tn) absorbed it — combined ₩3.58tn vs. foreign ₩3.76tn = near-1:1 offset. The risk is that single-session foreign selling congealing into a sustained trend would be hard for domestics to defend alone. The foreign direction across the rest of this week is the variable that matters.

Q: Why did KOSDAQ close -0.20%? A: Capital concentrated in large-caps. In a compressed top-of-cap rally, mid/small KOSDAQ names don’t tag along. KOSDAQ recovery requires capital diffusion — that needs the KOSDAQ average daily turnover to recover and foreign + institutional buying to align.

Q: Hot CPI, WTI back above $100, Middle East tension — why did markets ignore them? A: Three hypotheses appear to operate in parallel: (1) dry-powder around record highs after the May 6 +6.5% session, (2) concentrated single-name upgrade flow across the same week (Samsung Electronics Citi TP raise, multiple memory re-rating reports), (3) Hyundai Motor / Hyundai Mobis autonomy / robotaxi catalysts. Single-name news flow outweighed macro.

Q: How does May 6 (+6.5%) compare to May 13 (+2.63%)? A: The most important difference is foreign direction. May 6 = foreigners +₩3.13tn; May 13 = foreigners -₩3.76tn. Same strong move, opposite flow signature. May 6 = foreign-driven cohesive advance; May 13 = domestic-absorbed compressed large-cap rally.

Q: What kinds of names do compressed rallies favor? A: Generalizing is hard, but May 13 data shows: top market-cap + foreign-broker coverage + earnings visibility tends to attract the flow. The four clusters (semis flagships, AI substrate / SUMs, auto large-caps, display) are examples. Mid/small-cap names with thin coverage and expectation-based valuation participate later.

Q: Are foreigners selling Korea wholesale? A: No. The May 13 mix shows Samsung Electronics / SK hynix selling alongside Samsung Electro-Mechanics / Hyundai Motor / LG Display buying — partial profit-take in memory + rotation into other large-caps. Against the YTD $6.7bn inflow (see Why Korea Part 4), May 13’s sell was ~40% of cumulative inflow — large for one day, modest as a profit-take in cumulative terms. Insufficient data to declare trend reversal.

Q: What does KOSDAQ need to participate? A: (a) Foreign selling to stop or reverse, (b) KOSDAQ average daily turnover to recover, (c) earnings momentum to diffuse into mid/small-caps. Currently item (1) is unverified, so KOSDAQ recovery may take a few more sessions.

Q: Is this compression a dangerous signal? A: Data is insufficient to call it dangerous. Compressed rallies can be the early signal of a strong trend or a late-stage signal. The distinguishing test is simple — does capital diffuse from large-caps to mid/small-caps over the next 1-2 weeks (trend start), or do large-caps deflate and the rally ends (late-stage).

This article is for research and informational purposes only and does not constitute investment advice. KOSPI / KOSDAQ closes, daily moves, and investor-segment flow are from KRX and media reports (Yonhap, AJU Press). Single-name flows reference local data (screener_kr.db / kr_market_surface_daily) and may differ from official tick-level data. Global variables (US CPI, WTI, Middle East, SOX) come from public reporting. Hypotheses A / B / C are analyst inferences and may differ from actual participant decisions. Whether foreign -₩3.76tn is a trend or a single-session print cannot be settled from one day of data; the May 14-16 pattern adds the needed confirmation. Analysis can be wrong. Data cut: May 13, 2026 KST (Korea close) and May 14, 2026 KST (analysis).

Korea Quality Re-Rating Watch 2026-05-13: Gigabis, TES, TLB — Earnings Inflection Where Institutional Money Is Arriving

Wed, 13 May 2026 16:30:00 +0900

Macro Dashboard — 2026-05-13

Indicator	Level	5-Day Change	Signal
KOSPI	7,844.0	+4.7%	Bullish
KOSDAQ	1,176.9	−1.9%	Flat
USD/KRW	1,490	+3.1%	KRW Weak
VIX	17.9	+5.0%	Stable
US 10Y	4.46%	+0.11pp	Rising
Brent	$108.1	+6.7%	Rising
DXY	98.5	+0.7%	Firm

Regime verdict: Korea — Neutral | US — Bull. The KOSPI has outperformed the KOSDAQ by 6.6 percentage points over five sessions, driven by large-cap defense. A strengthening won (USD/KRW +3.1% to 1,490) and elevated oil add macro friction. The regime divergence between a US Bull and a KR Neutral points toward selective, quality-focused exposure rather than broad-based risk-on.

Market Wrap — Session Character (Reference: 2026-05-08 close briefing, most recent available)

Korea’s session carried a Bull-leaning neutral character with selective risk-on pockets rather than a uniform broad advance. The large-cap index held better than it appeared on the surface — but under that, flows were sharply divergent by sector.

The leading theme was AI infrastructure hardware. Fiber-optic cable and power-transmission names dominated the tape, with the cable segment seeing breakout moves backed by strong institutional buying. 대한전선 (Korea Electric Wire) drew over ₩167.7bn in net foreign buying alongside ₩353bn institutional — a rare dual confirmation. 가온전선 (Gaon Cable) surged nearly 30%, though an RSI above 95 and a 231% gap from its 200-day MA placed it firmly in the blow-off zone.

Semiconductor flows were split. Earnings thesis for the sector remained intact — broker target upgrades and AI infrastructure narratives continued to circulate — but near-term foreign selling in Samsung Electronics (005930.KS) ran at roughly ₩2.55 trillion for the session, with program selling near ₩1.66 trillion. A price decline of just −1.1% was technically resilient given that magnitude of supply, but the immediate read is that the AI semiconductor rally is digesting rather than accelerating.

The PCB / electronics hardware cluster was the quality pocket. Daedeok Electronics (353200.KS) printed a clean day: +2.07%, with net foreign, institutional, and program buying all aligned. Samsung Electro-Mechanics (009150.KS) held its five-day gain of +10.5% with institutional defense offsetting persistent foreign and short-seller pressure (short interest ~13.3%).

Gaming and select defense names lagged. Some game names faced foreign and institutional selling simultaneously, compounding a weak relative-strength reading. Without a fresh earnings catalyst or short-cover trigger, the technical picture in those sub-sectors remained cautious.

Summary: Large-cap Korea held on AI-infra hardware and quality chip-adjacent names; KOSDAQ lagged as momentum in consumer-facing tech softened. Institutional money continued rotating toward infrastructure and earnings-inflection names.

Today’s Quality Re-Rating Candidates — 2026-05-13

Six names cleared three screeners simultaneously in today’s run. Meta screener order is used as the primary ranking, with overlap logic providing the quality confirmation layer.

Candidate Table

Meta Rank	Ticker	Name	Meta Score	Screeners Hit	Key Metrics
1	420770.KQ	기가비스 (Gigabis)	68.9	QC · SMQ · SME	ROE 7.3%, OP YoY +777%, 5d F+QI +₩15.6bn
2	095610.KQ	테스 (TES)	62.3	QC · SMQ · SME	ROE 14.5%, OP YoY +50%, 5d F+QI +₩51.5bn
3	080220.KQ	제주반도체 (Jeju Semiconductor)	62.1	QC · SMQ · SME	ROE 17.4%, OP YoY +274%, CoBuy+NetBuy
4	082920.KQ	비츠로셀 (Vitzrocell)	60.9	QC · SMQ · SME	ROE 17.2%, OP YoY +33%, foreign hold 30%
5	229640.KS	LS에코에너지 (LS Eco Energy)	55.8	SMQ · CR · SME	Consensus up revision, margin +1.8pp, 5d F+QI +₩8.7bn
6	356860.KQ	티엘비 (TLB)	54.4	CR · SMQ · SME	OP YoY +665%, margin +8.2pp, 5d F+QI +₩16.7bn

QC = Quality Compounder · SMQ = Smart Money Quality · SME = Smart Money Earnings · CR = Cycle Rerating

Top 3 Context

기가비스 (420770.KQ) — Meta Rank 1, Score 68.9 Gigabis makes automated optical inspection (AOI) equipment for PCBs and displays — a direct beneficiary of capacity expansion in AI-related printed circuit board manufacturing. It clears the Quality Compounder screen on the back of a 777% YoY operating income surge and revenue nearly doubling (+101%), with low debt (13.4% ratio) and the stock near its 52-week high. Smart Money screens confirm ₩15.6bn of net foreign plus quality-institutional buying over five days. One caution: the consensus revision screen flags a downward revision, which means the headline earnings beat is real but forward estimates are being trimmed — worth checking whether the Q1 result beat represents a pull-forward or a sustainable step change. DART shows a sales-contract filing (April 24) as the catalyst anchor.

테스 (095610.KQ) — Meta Rank 2, Score 62.3 TES manufactures CVD/ALD semiconductor deposition equipment, serving both memory and logic fabs. ROE of 14.5% and operating income growth of +50% YoY pass the Quality Compounder hurdle cleanly. What makes it stand out is the ₩51.5bn five-day smart-money flow — the largest among the top six — combined with strong RS at the 95.6th percentile. Caution: foreign holdings are just 12.9% and foreign net flow has been negative recently, with short interest at 8.7% and program selling at −9.2%. The quality-institutional flow is real but broad foreign conviction hasn’t confirmed yet. A quarterly report filed today (2026-05-13) is the immediate catalyst to check.

제주반도체 (080220.KQ) — Meta Rank 3, Score 62.1 Jeju Semiconductor designs low-power DRAM and specialty memory chips. ROE of 17.4% and operating income growth of +274% YoY deliver both quality and cycle-rerating characteristics. The stock ranks #2 on both the Smart Money Quality and Smart Money Earnings screens, with co-buying (institutional + foreign simultaneously) confirmed. Two DART risk filings in the lookback window warrant review before treating this as a high-conviction candidate — the filings relate to shareholder meeting notices rather than regulatory issues, but confirmation is recommended.

All screener data is from 2026-05-13. KR close briefing data references the most recent available session (2026-05-08). This post is market analysis only — screener names are candidates for further research, not buy or sell recommendations.

Humanoid Compare — Robotis (Actuators, 2026E PSR 79×) vs. Rainbow Robotics (Samsung Affiliate, 2025 PSR 476×). Both Expensive, but Verifiability Differs

Tue, 12 May 2026 23:30:00 +0900

📚 Korea robotics series. Part 1: Korea Robotics Value Chain Complete Map Part 2: SPG vs. Halla Cast — Robot Component Compare

🔗 Related: Humanoid / Robotics Investment Hub · Hyundai Mobis Deep Dive · Samsung Electro-Mechanics AI Infrastructure Re-rating

Part 1 mapped components broadly; Part 2 compared reducers (SPG) and structural parts (Halla Cast). This piece moves to the “top” of the value chain — humanoid finished-product makers. Robotis and Rainbow Robotics are both flagship Korean robotics names, but what they make, how they earn, and how much expectation is in the price differ. Robotis makes the “fingers” of a robot; Rainbow makes the “whole body.” Robotis is profitable; Rainbow is loss-making. Yet Rainbow’s market cap is 3× larger. Why — and which is more verifiable?

TL;DR

Robotis = the robot’s “fingers and joints." Dynamixel actuators are the core product. 2025 revenue ₩38.9bn, OP turn-positive ₩3.4bn, market cap ~₩5.1tn. 2026E PSR 79×, 2027E PSR 50× — multiple compression path is visible.
Rainbow Robotics = humanoid platform with Samsung Electronics as anchor shareholder. Collaborative robots (RB series), dual-arm mobile robot (RB-Y1). 2025 revenue ₩34.1bn, OP -₩2.5bn (loss), market cap ~₩16.2tn. 2025 PSR 476× — 2027 revenue must hit ~₩330bn and net income ~₩63bn to align with Robotis’s multiple.
Why Rainbow is 3× larger despite smaller revenue and losses. Samsung Electronics now owns 35% as anchor shareholder. The market is pricing in “Samsung’s robotics strategy = Rainbow’s future revenue.”
Comparative read. Robotis is more verifiable — earnings are already arriving, growth path through 2027 is in numbers, and multiple compression is feasible. Rainbow has the larger optionality but needs 10× revenue growth in 2 years to justify current cap, and those numbers are not visible yet. Neither is cheap.

1. First — why what they make is different

1.1 Robotis — the “fingers and joints” of a robot

Robotis’s core product is the Dynamixel actuator — the “motor + reducer + sensor + electronics in one module” that drives a robot’s joints, as covered in Part 1.

Why Robotis matters:

1. 20+ years of actuator focus
 - \~98% of revenue from actuators
 - In-house component ratio 90%+ (including reducers)

2. Builds the "robot hand"
 - HX5-D20: 20-DOF robotic hand
 - Senses touch and shape; precision-grips objects
 - Can crack-free hold an egg, turn a screw, grip a screwdriver

3. Small → large form-factor extension
 - Existing: small actuators (Dynamixel X / P / Y)
 - New: large actuators for humanoid lower-body (Dynamixel Q = QDD)
 - QDD targets 200K-unit production from 4Q26 at the new Uzbekistan plant

Why the “robot hand” matters: Part 1 listed “fingers” as Problem #2 of “why humanoids are hard.” To replace humans on a factory floor, grip-rotate-insert-press-assemble hands matter more than walking. Robotis sits directly on this bottleneck.

1.2 Rainbow Robotics — the “whole body” of a robot

Rainbow Robotics is a robot platform company that designs and manufactures whole robots. Spun out of the KAIST HUBO research group, it is Korea’s most recognizable humanoid robot brand.

Core products:

1. RB series (collaborative robots)
 - Robot arms that work alongside humans
 - Drives, encoders, brakes, controllers all in-house
 - International safety certification (TÜV SÜD)

2. RB-Y1 (dual-arm mobile robot)
 - Wheeled mobile base + dual-arm humanoid upper body
 - 3kg payload per arm, 24-DOF total
 - Logistics, services, manufacturing-floor applications

3. RB-Y2 (industrial dual-arm robot) — in development
 - Heavier payload, manufacturing-line targeted

4. Relationship with Samsung Electronics
 - Samsung exercised call option in late 2024, securing 35% stake (largest shareholder)
 - Future Robotics Office established within Samsung
 - Robots to be deployed in Samsung's semiconductor, display, appliance, and logistics operations

1.3 One-line summary

Robotis: robot "parts" — makes joints and fingers
 actuators can go into multiple other robot makers
 analogy: a parts supplier of engines and transmissions

Rainbow: robot "finished product" — designs and assembles whole robots
 has Samsung as the giant customer behind it
 analogy: an EV startup that Samsung has invested in

Parts from Robotis can go into many robots.
Robots from Rainbow depend on Rainbow's own success.

2. The numbers — how wide the gap is

May 12 basis.

Item	Robotis	Rainbow Robotics
Price	₩345,000	₩837,000
Market cap	~₩5.1tn	~₩16.2tn
2025 revenue	₩38.9bn	₩34.1bn
2025 OP	+₩3.4bn (profit)	-₩2.5bn (loss)
2025 OPM	~8.7%	~-7.3%
2025 PSR	~130×	~476×
2026E revenue	₩64.0bn	Unverified (est. ₩130-210bn needed)
2026E OP	₩9.0bn	₩32.0bn (consensus)
2026E PSR	79×	78-130× (depending on revenue assumption)
2026E PER	496×	~518×
2027E revenue	₩101.0bn	Unverified (est. ₩330bn needed)
2027E PSR	50×	Unverified
2027E PER	264×	Unverified
Value-chain position	Components (actuators)	Finished product (platform)
Core premium	Pure-play actuators	Samsung Electronics anchor shareholder
Major customers	LG Electronics (2nd-largest shareholder), global robot makers	Samsung Electronics (largest shareholder)
2025 revenue growth	+30%	+76%

Cross-checks:

Robotis PSR = ₩5.06tn / ₩38.9bn = ~130× ✓
Rainbow PSR = ₩16.24tn / ₩34.1bn = ~476× ✓

2.1 What the table shows

Rainbow’s cap is 3.2× larger, but revenue is smaller and the business is loss-making. Robotis revenue ₩38.9bn > Rainbow revenue ₩34.1bn. Robotis profitable, Rainbow loss. Yet market cap: Robotis ₩5.1tn < Rainbow ₩16.2tn.

The wedge that produces this is one word: Samsung. Samsung Electronics holds 35% as anchor shareholder, and the strategy of deploying Rainbow robots into Samsung’s semiconductor, display, appliance, and logistics operations is already in the price.

Simplified:

Robotis price = current earnings + future actuator-growth expectation
Rainbow price = expectation that Samsung succeeds in robotics

3. Robotis — earnings starting to materialize

3.1 Why turning profitable matters

Part 1 noted that “of 28 Korean robot-related listed companies, only 3-4 are profitable.” Robotis is one of them. 2025 revenue ₩38.9bn, OP ₩3.4bn, OPM ~8.7%.

A profitable robot company means “the business model can actually earn money." Most Korean robot names are still burning cash in R&D, while Robotis is selling Dynamixel actuators and generating real cash flow.

3.2 Growth trajectory ahead

Consensus from Hana Securities’s report (Consensus Data section):

Period	2025 (actual)	2026E	2027E
Revenue	₩38.9bn	₩64.0bn (+65%)	₩101.0bn (+58%)
OP	₩3.3bn	₩9.0bn	₩20.0bn
OPM	8.5%	14.1%	19.8%
EPS	~₩380	₩696	₩1,306

Inter-broker dispersion is wide. Samsung Securities at 2026 revenue ₩51bn (conservative); Daol Investment & Securities at ₩80bn (aggressive). Consensus mid ₩64bn sits between. The decisive variable is how fast QDD (Dynamixel Q) — the large-form-factor actuator — scales into mass production.

Why QDD matters:

Existing Dynamixel X / P / Y: small actuators
→ Education / research robots, robot hands, small joints
→ Already selling and generating cash

Dynamixel Q (QDD): large actuators
→ Humanoid waist, knee, shoulder — large joints
→ Target 200K units in 4Q26 from the Uzbekistan plant
→ When this ramps, revenue steps up materially

Analogy:
Existing = smartphone components (small but stable market)
QDD = EV components (large market but mass-production is the gate)

3.3 Demand exists; supply is the bottleneck

Per Hana Securities, 2026 actuator orders are estimated at 1M+ units. Actual 2025 production was 220K units. Demand is there; supply capacity isn’t.

2026 production targets:

Existing Dynamixel X / P / Y: 300K units
New QDD: 200K units (4Q26, Uzbekistan plant)
Total: ~500K units

500K units producible against 1M+ orders = demand sufficient, supply constrained. Not a bad analytical structure — if it can be built, it can be sold.

Caution: whether “1M orders” is confirmed POs (purchase orders) or inquiries / expressions of interest is not disclosed. That distinction matters.

3.4 Robotis — strengths and weaknesses

Strengths	Weaknesses
Turn-positive — earnings arriving	Market cap ₩5.1tn on revenue ₩38.9bn — PSR 130× is rich
Actuator purity 98% — direct exposure to robot demand	QDD mass-production hasn’t started
Robot hand (HX5-D20) — directly addresses humanoid bottleneck	Possible price competition from Chinese QDD makers
LG Electronics 2nd-largest shareholder — large-corporate customer secured	External motor sourcing — rare-earth supply risk
In-house component ratio 90%+ — cost competitiveness	DataFactory new business still pre-revenue

4. Rainbow Robotics — the giant Samsung option

4.1 Why Samsung bought in

Samsung Electronics exercised its call option in December 2024 to secure a 35% stake in Rainbow Robotics — making it the largest shareholder. Samsung also established a Future Robotics Office internally.

The reason Samsung is in robotics was covered in Part 1 — “Physical AI.” Samsung intends to automate manufacturing (semiconductor, display) first, then expand to home and retail.

Rainbow can be the core executor of that strategy:

Samsung robotics roadmap (estimated):

Phase 1: Samsung-factory automation
 → RB-Y1/Y2 deployed in semiconductor back-end and display lines
 → Rainbow robots secure Samsung as internal customer first

Phase 2: Samsung logistics + services
 → Service robots deployed in Samsung logistics centers and stores

Phase 3: External market expansion
 → Samsung brand + Rainbow tech enters the global robot market

Rainbow revenue could scale into the trillions of won if this roadmap executes.
Current ₩16tn market cap embeds a substantial portion of that scenario.

Through 3Q25, Samsung Electronics’s purchases from Rainbow Robotics totaled ~₩6.9bn. Annualized that’s ~₩10bn of Samsung-related revenue.

For a ₩16.2tn market-cap company, ~₩10bn from its largest-shareholder customer = still test / pilot stage. The phase where Samsung issues large-volume orders for hundreds-to-thousands of robots into actual factories has not arrived.

Rainbow's actual position:

Market cap: ₩16.2tn → priced on "Samsung will succeed in robotics"
Revenue: ₩34.1bn → still in "initial-delivery stage"
Samsung-related: \~₩10bn → "started but small"

To close the gap:
→ Samsung-related revenue must scale to hundreds-to-thousands of ₩bn annually
→ External customers also need to come on
→ OP must turn positive

All three are currently "possibility," not "confirmed."

4.3 Rainbow — strengths and weaknesses

Strengths	Weaknesses
Samsung Electronics anchor shareholder (35%) — strategic customer secured	Market cap ₩16.2tn on revenue ₩34.1bn — PSR 476× is extreme
KAIST HUBO heritage — locomotion and control algorithms	OP losses persist — not yet making money
In-house core component development — claimed cost edge	Samsung-related revenue ~₩10bn — pilot stage
RB-Y1 at field-deployable level	Uncertain external customer expansion
Sejong manufacturing facility in preparation	Sejong yield / utilization still to be validated

5. Side-by-side — which is more verifiable

5.1 Core comparison

Item	Robotis	Rainbow	Edge
Earnings	Profitable, OPM 8.7%	Loss	Robotis
Revenue growth	+30%	+76%	Rainbow
2025 PSR	130×	476×	Robotis
2026E PSR	79×	78-130× (est.)	Robotis (firm number)
2027E PSR	50×	Unverified (needs ₩330bn revenue)	Robotis
2026E PER	496×	~518×	similar; Robotis has better visibility
2027E PER	264×	Unverified (needs ~₩63bn NI)	Robotis
Core customer	LG Electronics (2nd shareholder)	Samsung Electronics (largest shareholder)	Rainbow
Verifiability	High (orders, production, margin trackable)	Low (Samsung-internal-strategy-dependent)	Robotis
Industry leverage	Components → into many robots	Finished product → dependent on Rainbow’s own success	Robotis
Option size	Mid (actuator market)	Very large (Samsung whole-of-robotics strategy)	Rainbow
Technology moat	20 years actuators, robot hand, component integration	Locomotion algos, in-house components, Samsung captive	Similar

5.2 Two completely different analytical setups

Watching Robotis = 
testing "actuator demand growth → revenue follow-through"
→ Verifiable: order intake, production, ASP, margin — quarterly
→ Diversifiable across customers (LG, Tesla, Unitree, etc.)

Watching Rainbow = 
testing "Samsung succeeds in robotics"
→ Hard to verify: Samsung-internal strategy is selectively disclosed
→ Concentrated on Samsung
→ Big if it works; long wait if delayed

Auto analogy: Robotis is the Bosch-style supplier of engines and transmissions to multiple OEMs — grows with the industry. Rainbow is an EV startup that Samsung has invested in — huge if it works, but failure means losing the capital.

5.3 Relative read — Robotis has higher verifiability

Three reasons:

One: earnings already arriving. Robotis turned positive; per sell-side estimates, 2026 OPM could reach ~15%. Rainbow remains loss-making.

Two: verification is possible. Robotis publishes “of 1M orders, how many were built, what ASP, what margin” quarter by quarter. Rainbow’s “how many robots did Samsung actually order” is hard to determine from outside.

Three: price burden lower with a visible compression path. Robotis 2026E PSR 79× → 2027E 50× shows a multiple-compression path. Rainbow’s 2026E PSR requires a revenue assumption to even calculate; 2027E has no public consensus. Visibility differs.

Note: “Robotis is more verifiable” ≠ “Robotis is cheap." Both are expensive. Robotis at 2027E PSR 50×, PER 264× still requires revenue to reach ₩101bn and OPM to reach 20% to make sense.

6. Fair value — what the numbers show

6.1 Robotis — consensus-based, numbers visible through 2027

Sell-side consensus (Hana Securities’s report, Consensus Data section):

Period	2025 (actual)	2026E	2027E
Revenue	₩38.9bn	₩64.0bn	₩101.0bn
Revenue growth	+30%	+65%	+58%
OP	₩3.3bn	₩9.0bn	₩20.0bn
OPM	8.5%	14.1%	19.8%
EPS	~₩380	₩696	₩1,306

At ₩345,000 current price:

Basis	PSR	PER
2025 (actual)	130×	~910×
2026E	79×	496×
2027E	50×	264×

Cross-checks: 2026E PER = 345,000 / 696 = 495.7× ✓; 2027E PER = 345,000 / 1,306 = 264.2× ✓

Read: 2026 PER 496× is extreme, but moving to 2027 PER 264× and PSR 50× shows a multiple-compression path. Revenue 640 → 1,010 won bn and OPM 14% → 20% supports a “high-growth components” framing.

Key check variables: 2026 revenue must reach ₩60bn+, OPM must hold above 14%.

Average sell-side target price ~₩317,000; high (Daol Investment & Securities) ₩435,000. The current ₩345,000 already exceeds the average target.

6.2 Rainbow Robotics — consensus is thin; reverse-engineer instead

Rainbow’s problem is lack of reliable consensus. What is confirmable is 2026E OP ₩32.0bn (FnGuide); 2026 revenue and 2027 figures lack public consensus.

So we must reverse-engineer what’s needed to justify current cap.

2026 reverse-engineering — what revenue is needed to produce OP ₩32.0bn:

Assumed OPM	Required 2026 revenue	Growth vs. 2025	Implied PSR at current cap
15%	₩213.3bn	+525%	78×
20%	₩160.0bn	+369%	104×
25%	₩128.0bn	+275%	130×

4-6× revenue growth in one year from ₩34.1bn (2025). Essentially impossible without large-scale Samsung-related revenue.

2026E PER from current cap is ~518× (₩16.6tn / estimated NI). Looks similar to Robotis’s 2026E PER (496×), but Robotis’s underlying revenue / OP estimates are firm; Rainbow’s revenue visibility is low.

6.3 The 2027 alignment math

Robotis’s 2027 multiples are PSR 50×, PER 264×. What Rainbow must produce in 2027 to align:

Standard	Rainbow’s required number
Robotis 2027E PSR (50×)	Revenue ~₩330bn
Robotis 2027E PER (264×)	NI ~₩63bn

Cross-checks: ₩16.6tn / 50 = ₩332bn revenue ✓; ₩16.6tn / 264 = ₩62.9bn NI ✓

Rainbow needs revenue ₩330bn and NI ₩63bn in 2027 to align with Robotis’s multiple. That’s ~10× revenue growth from ₩34.1bn in 2 years — annualized ~+211%.

Not impossible — Samsung would need to order thousands of robots. But that number is not yet visible anywhere.

6.4 Scenario price grid

Robotis:

Scenario	2027 revenue	Applied PSR	Implied price	Vs. current
Bear (production delays)	₩80bn	35×	~₩190,000	-45%
Base (consensus delivered)	₩101bn	45×	~₩310,000	-10%
Bull (QDD breakout success)	₩130bn	50×	~₩440,000	+28%

Rainbow Robotics:

Scenario	Base business value	Samsung-option value	Total	Implied price	Vs. current
Bear	₩2.0tn	₩3.0tn	₩5.0tn	~₩255,000	-70%
Base	₩4.8tn	₩6.0tn	₩10.8tn	~₩550,000	-34%
Bull	₩10.0tn	₩8.0tn	₩18.0tn	~₩920,000	+10%

Average sell-side target: Robotis ~₩317,000, Rainbow ~₩580,000. Both currently trade above the average target price.

6.5 Asymmetric structure — the core of the analysis

Robotis:
→ Base case: -10% (slightly rich)
→ Bull case: +28%
→ Bear case: -45%
→ Upside and downside similar — neutral asymmetry

Rainbow:
→ Even the bull case: only +10%
→ Reversion to base: -34%
→ Bear case: -70%
→ Downside 3-7× the upside — unfavorable asymmetry

Rainbow sits in a structure where the best-case Samsung-option realization still leaves only +10% upside, while reversion to base implies -34%. Numerically, this is unfavorable for new entry.

6.6 The one line that matters

Robotis: expensive, but a multiple-compression path is visible through 2027.
Rainbow: to justify current cap, needs 10× revenue and \~₩63bn NI in 2 years.
 Those numbers are not yet officially visible.

7. What to verify next

7.1 Robotis — verification by earnings

Item	Window	Why it matters
1Q earnings	this/next month	Are revenue growth + OPM holding?
QDD mass-production start	4Q 2026	Large-actuator mass-production = the revenue-step-up gate
Uzbekistan plant operation	October 2026~	Core of capacity expansion
Reality of the 1M-unit order	quarterly results	Confirmed PO or inquiry level?
Robot-hand (HX5-D20) revenue	quarterly results	Demo product or commercial product?

7.2 Rainbow Robotics — verification through Samsung

Item	Window	Why it matters
Samsung-related revenue scale	half-year report	Does ₩10bn level expand to hundreds of ₩bn?
RB-Y2 field deployment	2H 2026	Real factory line, or demo only?
OP turn-positive	quarterly results	Can the business actually earn?
Sejong manufacturing operation	2H 2026~	Mass-production capability validation
External-customer expansion	ongoing	Reducing Samsung concentration?

8. Bottom line

Robotis and Rainbow Robotics are both symbols of Korean robotics. But analytically, they’re entirely different structures.

Robotis is a “components supplier where earnings have started to arrive." Turn-positive on OP, actuator orders growing, QDD mass-production poised to step up revenue. Expensive but verifiable on numbers.

Rainbow Robotics is a “company priced for a giant Samsung option." Samsung’s robotics success would make Rainbow successful. But the current price has already embedded most of that success. Even the bull case offers only +10%; reversion to base implies -34%.

Of the two, Robotis has higher verifiability. But Robotis at ₩345,000 still trades above its base-case implied price (₩310,000). The most disciplined analytical approach is to track 1Q earnings for revenue growth + OPM, then QDD ramp progress.

Korea robotics is still an “industry of expectations.” The cleanest analytical window opens where expectation converts to earnings. And that window is more likely to open at Robotis first than at Rainbow.

FAQ

Q: Which is the better company, Robotis or Rainbow? A: Quality-of-business is a different question. Both are symbols of Korean robotics with credible technical capabilities. The “more verifiable” one is Robotis — turn-positive financials, quarterly-trackable orders / production / margin, visible multiple-compression path through 2027. Rainbow has the bigger option but lower verifiability.

Q: Why is Rainbow 3× Robotis in market cap? A: Samsung Electronics is the largest shareholder (35%). The market is pricing “Samsung’s robotics strategy = Rainbow’s future revenue.” Robotis currently has larger revenue (₩38.9bn vs ₩34.1bn) and profitability, but Rainbow’s cap (₩16.2tn) exceeds Robotis’s (₩5.1tn).

Q: PSR 79× and PER 496× — is that ever justifiable? A: Extreme by Korean-market norms. In high-growth themes (robotics, AI), fast revenue growth can naturally compress multiples. Robotis’s path to 2027 PSR 50× / PER 264× is visible if revenue ₩101bn and OPM 20% materialize.

Q: If Rainbow doesn’t reach ₩330bn revenue in 2027? A: The market repositions the multiple. Bear case (revenue <₩100bn) implies cap ~₩5tn — current-price downside ~-70%. Partial Samsung-option realization could pull it back to base case (cap ₩10.8tn, -34%).

Q: Is Robotis’s “1M-unit order” real? A: It’s a Hana Securities-cited estimate. Whether it represents confirmed POs or inquiry-level interest isn’t disclosed. Verification = whether quarterly shipment and revenue line items actually increase. 2025 actual production was 220K units.

Q: Why is QDD so important? A: Existing Dynamixel (X/P/Y) addresses small education / research markets. QDD addresses the large humanoid joints (waist, knee, shoulder) — directly exposed to the broader humanoid market, with materially higher ASP. Targets 200K units in 4Q26 at the Uzbekistan plant.

Q: Could Samsung outright acquire Rainbow? A: Possible but not officially confirmed. Samsung adding to its 35% stake or fully consolidating Rainbow would be a major catalyst. But the premium implied by a tender-offer scenario is hard for the market to anticipate.

Q: If one must pick a robot name, what’s the analytical guidance? A: Definitive recommendations aren’t appropriate. The value-chain Part 1 covered component suppliers (SPG, Robotis, HL Mando, Samsung Electro-Mechanics, LG Innotek, Hyundai Mobis) where a robust base business cushions robotics-option risk. Pure-play robotics names carry higher price burden. Analytically, “base business + robot option” structures are easier to verify than “robot-option-only” structures.

This article is for research and informational purposes only and does not constitute investment advice. Robotis data references Samsung Securities, Hana Securities, Daol Investment & Securities, KB Securities, IBK Investment & Securities, and Eugene Investment & Securities reports. Robotis 2026E/2027E figures (revenue ₩64.0/₩101.0bn, OP ₩9.0/₩20.0bn, EPS ₩696/₩1,306) reference Hana Securities’s report Consensus Data section. Rainbow Robotics 2026E OP ₩32.0bn references FnGuide consensus; 2026E revenue and 2027E figures lack public consensus. Rainbow’s “required-earnings reverse-engineering” is an analyst estimate that applies Robotis’s multiples to Rainbow’s current cap. Prices reflect Maeil Business Newspaper data (May 12). Samsung’s robotics strategy details are selectively disclosed. Both names are high-multiple thematic stocks with elevated volatility. Robotis received an investment-warning designation notice on May 12. Analysis can be wrong. Data cut: May 12, 2026 KST.

Pearl Abyss 5/21 IR Watch: After ₩212.1B Operating Profit, Capital Return, DLC and DokeV Will Decide the Re-Rating

Tue, 12 May 2026 23:30:00 +0900

📚 Pearl Abyss follow-up. Read first: Pearl Abyss 1Q26 earnings comprehensive: ₩212.1B OP, 64.6% margin, FY26 OP guidance of ₩487.6-572.6B Read next: What the DLC “exploring” comment really means — not the KRW 60B in sales, but the removal of the “2027 cliff” discount Hub: Pearl Abyss and Crimson Desert Research Hub

Pearl Abyss’s May 12 earnings release changed the center of the debate. The question is no longer whether Crimson Desert worked. The first-quarter numbers answered that. The next question is whether Pearl Abyss can be valued as a quality Korean AAA IP company, not just as a one-time package-game launch story. The answer now depends on what the company says at the May 21 IR about capital allocation, Crimson Desert’s expansion roadmap, and DokeV visibility.

TL;DR

The earnings proof is in. Pearl Abyss reported 1Q26 revenue of ₩328.5B, operating profit of ₩212.1B, and a 64.6% operating margin. Korea JoongAng Daily reported that Crimson Desert generated ₩266.5B in the quarter and that North America plus Europe accounted for 81% of revenue.
The 2Q cliff argument is weaker. The company guided to FY26 operating profit of ₩487.6-572.6B and 2Q26 operating profit of ₩129.6-176.7B. ZDNet Korea via Daum also reported 2Q revenue guidance of ₩271.3-324.7B.
The next issue is the multiple. Good earnings raise EPS. Good capital allocation and post-launch roadmap visibility raise the multiple. The May 21 IR is the first test of whether Pearl Abyss can connect the two.
Capital return is not just a dividend question. The deeper question is whether management can state a credible rule for balancing buybacks, dividends, next-IP investment, talent retention, and M&A after Crimson Desert and the CCP/Fenris divestiture improve cash flow.
DLC and platform expansion matter because they attach 2027 cash flow. If the story is only base-game sales, the market will treat 2026 as peak earnings. Paid DLC, expansions, new platforms, or repeatable endgame content can extend the revenue curve.
DokeV is still an option. But if development stage, disclosure timing, and BlackSpace Engine reuse become clearer, Pearl Abyss moves from single-hit upside into multi-IP repeatability.

1. This Is Not Another Earnings Review

The previous note covered the first-quarter numbers: ₩328.5B in revenue, ₩212.1B in operating profit, and a 64.6% operating margin. Pearl Abyss also guided to FY26 operating profit of ₩487.6-572.6B.

This follow-up is not trying to restate the same earnings table. The numbers are now out. The investor question has shifted.

Stage	Old Question	New Answer or Question
Before launch	Will Crimson Desert sell?	It sold.
Before earnings	Can quarterly OP reach the ₩200B level?	It did.
After earnings	Does 2Q collapse?	Company guidance does not assume a collapse.
Next stage	How much did Pearl Abyss earn?	What will it do with the cash, and how will it build 2027+ cash flow?

That is why the May 21 IR matters. It is not just an earnings explanation session. It is where institutional investors begin rewriting models. If management simply repeats the 1Q numbers, Pearl Abyss remains a game stock with a strong quarter. If it gives clearer answers on capital allocation, DLC, and DokeV, the market can start changing the company’s category.

The core line is simple:

The earnings release was proof. The May 21 IR is the first test of reclassification.

2. Preview Versus Actuals: Lower Revenue Recognition, Stronger Earnings Power

The earlier preview centered on 1Q26 revenue of ₩395.0B, operating profit of ₩205.0B, and an operating margin of 51.9%. The actual result was ₩328.5B in revenue and ₩212.1B in operating profit.

Item	Prior Base Case	Actual 1Q26	Difference
Revenue	₩395.0B	₩328.5B	-₩66.5B
Operating profit	₩205.0B	₩212.1B	+₩7.1B
Operating margin	51.9%	64.6%	+12.7ppt

This distinction matters. The original revenue-recognition assumption needs to be revised down.

Crimson Desert revenue was booked at ₩266.5B in 1Q. The game crossed 5M copies on April 15, after reaching 4M copies on April 1. Dot Esports summarized the same path: 2M copies on day one, 4M by April 1, and 5M by April 15.

Per-unit recognized revenue is therefore lower than a simple gross-price assumption. Console sales likely look closer to net recognition after platform fees, while PC sales reflect regional pricing, taxes, and platform mix. That part of the model should become more conservative.

But the more important point is profit. Revenue was lower than expected, yet operating profit was higher. The reason is the cost structure. Much of Crimson Desert’s development cost had already passed through historical P&L, so post-launch revenue flowed through at a very high incremental margin. Platform fees, marketing, and other launch costs also appear lighter than the bearish case assumed.

So the correct adjustment is this:

The revenue-recognition frame should become more conservative. The profitability frame should become stronger.

3. Why the 2Q Cliff Frame Is Weaker

The market’s biggest concern was a 2Q cliff. Package games usually have front-loaded sales. It was rational to worry that 1Q would be great and 2Q would drop sharply.

The company guidance weakens that frame.

Item	Company Guidance
FY26 revenue	₩879.0-975.4B
FY26 operating profit	₩487.6-572.6B
2Q26 revenue	₩271.3-324.7B
2Q26 operating profit	₩129.6-176.7B

The 2Q revenue range is especially important. Compared with ₩266.5B of 1Q Crimson Desert revenue, management is not describing a revenue cliff. It is describing limited decay or near-maintenance.

The 2Q operating margin can still decline from 1Q. Incentives, operations, extra marketing, and update costs can all show up. But the market’s main fear was not a normal cost step-up. It was revenue collapse. The guidance directly reduces that concern.

The investor implication is clear:

If 2Q results land inside company guidance, Pearl Abyss can start escaping the peak-earnings discount.

A company with one great launch quarter deserves a low multiple. A company whose launch revenue holds into 2Q deserves a different EPS model.

4. From Here, Capital Allocation Matters More Than the Earnings Beat

The first-quarter numbers proved that Pearl Abyss can make money. The harder question is now:

What will management do with the money?

Good earnings raise company value. Good capital allocation raises the multiple. Pearl Abyss has historically carried a discount not only because earnings were low, but because cash-flow durability, next-title visibility, and capital-allocation rules were unclear.

That changes after Crimson Desert. The game produces cash. The CCP/Fenris divestiture simplifies the asset base. The old Pearl Abyss had to survive a long development cycle. The new question is whether the company can allocate cash like a quality compounder.

The May 21 IR should answer these questions:

Question	Why It Matters
Is there a buyback or treasury-share cancellation plan?	It shows whether one-time earnings can become shareholder value.
Is a dividend policy under review?	It signals confidence in recurring cash flow.
How will CCP/Fenris proceeds be used?	It separates asset clean-up from true capital efficiency.
How does management think about large M&A?	Undefined M&A risk can keep the discount alive.
What is the priority among DokeV, Plan8, and Crimson Desert operations?	It shows whether reinvestment can raise ROIC or only raise cost.

Capital return is not a simple demand for dividends. The key is a rule. If management says that Crimson Desert cash flow will be balanced among next-title development, talent retention, and shareholder return, and then gives a concrete framework, the discount rate comes down.

If the only answer is generic growth investment, the governance discount stays. Growth matters, but growth without a clear rule can become a discount.

5. DLC Roadmap Matters More for Depreciation Defense Than Incremental Revenue

Crimson Desert has already crossed 5M copies. The next focus is the path to 6M, 7M, and 8M. But unit sales alone are not enough. The more important question is how much incremental cash flow Pearl Abyss can attach to the installed base.

DLC and platform expansion matter for three reasons.

First, they reduce base-game decay. Whether the content is free updates or paid DLC, the goal is to keep players talking about the game. GamesRadar+ wrote that Pearl Abyss moved through much of its early roadmap quickly and that the next post-launch territory is less defined. That can be positive or negative. It proves the studio moves fast, but it also means the next roadmap needs to be shown.

Second, DLC reactivates base-game demand. DLC is not only sold to existing players. It tells non-buyers that the game is alive. When combined with sales windows, streaming content, and new platforms, it can re-accelerate base-game purchases.

Third, DLC changes the multiple. If Pearl Abyss only has base-game sales, the market treats 2026 as peak earnings. If it has DLC, expansions, multiplayer, or new platforms, the market attaches 2027+ cash flow.

So the market does not need to hear only that DLC is being considered. It needs specifics.

IR Item	What Investors Need to Hear
Free updates	What friction points will be fixed, and when?
Paid DLC	Is there a 2H26 or 2027 plan?
Expansion packs	How will the world and narrative expand?
Platform expansion	Are new platforms or regions possible?
Repeatable content	How will the endgame loop stay active?

The point of DLC is not one extra revenue line. It is evidence that Pearl Abyss can still earn money from Crimson Desert in 2027.

6. DokeV Is Still an Option, But Visibility Would Change the Company

The most important long-term variable for Pearl Abyss’s quality-company case is DokeV.

Crimson Desert’s success is already meaningful. But one hit game is a great event. A repeatable pipeline is a great company.

DokeV matters because it is different from Crimson Desert. Crimson Desert is a Western-facing open-world action RPG. DokeV has the potential for a broader age group, a more casual audience, stronger character and collection mechanics, and a lifestyle-IP angle. If it works, Pearl Abyss is no longer only a company that made one AAA action game. It becomes a studio that can use a proprietary engine to build multiple global IPs.

That said, DokeV should not carry heavy valuation weight yet. The market needs visibility, not hope. ZDNet Korea via Daum reported that DokeV is currently in pre-production and that Pearl Abyss will disclose more when development progress warrants it.

The May 21 IR questions are straightforward:

Question	Investment Interpretation
What is DokeV’s current development stage?	Distinguishes pre-production from a playable build path.
How is the core team allocated?	Shows whether Crimson Desert operations and DokeV development can coexist.
When is the next disclosure event?	Turns expectation into a calendar event.
Is there a target release window?	Helps measure the 2027-2028 revenue gap.
Can BlackSpace Engine be reused?	Points to shorter development cycles and better margins.
What is the business model?	Package, live service, or hybrid models deserve different multiples.

If DokeV becomes more concrete, Pearl Abyss’s multiple changes. Markets assign low multiples to one-time earnings. They assign higher multiples to repeatable IP creation.

Crimson Desert is proof. DokeV is repeatability.

7. Valuation: The KRW 50,000s Still Embed a Peak-Earnings Discount

Based on 1Q earnings and FY26 guidance, a low-KRW-50,000 stock price still looks like the market is treating 2026 earnings as peak earnings. Pearl Abyss closed regular trading on May 12 at KRW 52,800. Yonhap reported that the stock jumped more than 10% in after-hours trading after the earnings announcement.

A simplified FY26 EPS frame looks like this:

Scenario	FY26 OP	Implied EPS	PER at KRW 52,800
Low	₩487.6B	~KRW 5,300	~10.0x
Mid	₩530.1B	~KRW 6,000	~8.8x
High	₩572.6B	~KRW 6,700	~7.9x

This is not an action guide. It is a way to see how the market is interpreting 2026 guidance.

Using the mid-case EPS of roughly KRW 6,000, different market frames imply different price zones:

Market Frame	Multiple	Arithmetic Price
Peak-earnings discount	9x	KRW 54,000
Guidance properly reflected	12x	KRW 72,000
2Q holds and DLC becomes visible	14x	KRW 84,000
Multi-IP quality-company candidate	15x	KRW 90,000

The precision of the numbers is not the point. The point is what conditions justify each multiple.

The KRW 70,000 zone can be explained by a simple earnings reset. Above KRW 90,000, earnings alone are not enough. The market would need to view Pearl Abyss as a quality-company candidate rather than a one-hit launch story.

Three conditions matter:

2Q results land inside company guidance.
DLC and platform expansion attach visible 2027 cash flow.
DokeV development becomes more visible.

8. Red Team: This Is Not Yet a Proven Quality Company

The first-quarter result was excellent, but Pearl Abyss does not automatically become the best quality company on KOSDAQ.

There are four risks.

First, 2Q results could miss even the low end of guidance. In that case, the market will call 1Q peak earnings and reapply a discount.

Second, cash use may remain unclear. If there is no shareholder-return framework and no capital-allocation rule, the multiple stays low even if earnings are strong.

Third, the DLC roadmap may remain vague. If the message is only that the company is considering options, 2027 cash flow will not be capitalized.

Fourth, DokeV may stay an undated option. If it remains unclear for another two to three years, the post-Crimson-Desert gap risk returns.

So the conclusion is not that Pearl Abyss is already the highest-quality company on KOSDAQ. The more precise conclusion is:

Pearl Abyss has, for the first time, the conditions to become a KOSDAQ quality-company candidate. Those conditions only turn into a higher multiple if the May 21 IR provides concrete capital-allocation and pipeline visibility.

Final Note

Pearl Abyss’s 1Q26 earnings ended the first debate. Crimson Desert worked, and the company produced an operating margin above 60%. FY26 operating-profit guidance of ₩487.6-572.6B is more than enough to reset earnings.

What remains is the multiple.

Pearl Abyss trades at a low multiple not because the 1Q earnings were weak, but because investors still question durability, capital allocation, and pipeline visibility. If the May 21 IR reduces those three uncertainties, the market will need to reclassify the company.

Capital return lowers the capital-allocation discount. DLC roadmap visibility adds 2027 cash flow. DokeV visibility turns a single-hit story into a multi-IP developer story.

If those three arrive together, Pearl Abyss is not just another game stock. It becomes a KOSDAQ quality-company candidate with global revenue, proprietary technology, high margins, net cash, and pipeline option value.

That is why the May 21 IR matters. The earnings release was the proof. The next IR is the first test of reclassification.

Korea Robot Component Compare — SPG (Reducers, PER 110×) vs. Halla Cast (Lightweight Structural Parts, PER 62×). Both Expensive — Which Carries Less Risk?

Tue, 12 May 2026 22:00:00 +0900

📚 Korea robotics series — Part 2. Part 1: Korea Robotics Value Chain — Samsung / Hyundai Motor / LG Moving in Parallel Part 3: Robotis vs. Rainbow Robotics — Humanoid Whole-Robot Compare Hub: Korea Humanoid / Robotics Investment Hub

SPG appeared 5 times in the Korea-robotics value-chain piece — supplying reducers to Boston Dynamics, Rainbow Robotics, Samsung, and LG. Market cap ₩3.0tn, PER ~110×. Halla Cast more recently won a humanoid-component order from a “global AI automaker.” Market cap ₩620bn, PER ~62×. Both are tagged as “robot names” — but the position in the value chain, the price, and the shape of the risk differ. Which is the better setup — more precisely, which carries less risk?

TL;DR

SPG = the robot’s “joint." Makes reducers and actuators — the most core motion components in the value chain. Only Korean company producing harmonic, planetary, and RV reducers all in-house. Market cap ~₩3.0tn, 2026E PER ~110×.
Halla Cast = the robot’s “skeleton." Magnesium / aluminum die-casting for lightweight structural parts and thermal-management components. In 2025 won a ₩12.4bn humanoid-component order from a “global AI automaker.” Market cap ~₩620bn, 2026E PER ~62×.
Read. SPG is the more core robot asset, but the current price has already absorbed that core status aggressively. Halla Cast is less core, but its price burden is lower and there is still verification ahead. Both are expensive, but the reason and the risk shape differ.

1. First — why the parts they make are different

1.1 SPG reducers — the robot joint’s core

As explained in the Korea robotics value-chain piece, reducers convert fast motor rotation into slow, strong torque. Every robotic joint (shoulder, elbow, wrist, hip, knee, ankle) uses one. A humanoid contains dozens of reducers.

Why reducers are expensive and hard:

Precision: gears must mesh with sub-micron tolerance
Durability: joints must move tens of thousands of cycles without play
Variety: harmonic, RV, planetary — each application uses different types

Global #1: Japan's Harmonic Drive (decades of effective monopoly)
Korea's only in-house challenger: SPG (produces all three types)

SPG additionally produces actuators (motor + reducer + sensor integrated module). Its proprietary SDD (Smart Direct Drive) actuator targets a 5,000-unit sales goal in 1H 2026 — a strategy to step up from “reducer supplier” to “reducer + actuator integrated-solution supplier.”

1.2 Halla Cast structural parts — the robot skeleton’s lightweighting

Halla Cast makes precision metal parts by die-casting molten magnesium and aluminum. Originally focused on auto sensor housings, display frames, and EV inverter cases, the company is now expanding into humanoid structural frames, drive-mechanism peripherals, and thermal-management components.

Why a light skeleton matters:

Heavy arms, legs, torso for a robot:
→ require bigger motors (more power)
→ drain batteries faster (shorter runtime)
→ generate more heat (shorter component life)

Magnesium is 33% lighter than aluminum:
Magnesium density: 1.8 g/cm³
Aluminum density: 2.7 g/cm³
→ identical volume, significantly lighter part

Magnesium is also harder to work with:
→ oxidation / ignition risk, corrosion susceptibility, machining difficulty
→ holding yield in mass production is the core capability

Halla Cast’s technical edge is high-vacuum die-casting (eliminates porosity) and thin-wall die-casting (large-area parts at minimal thickness). The platform was validated in auto electronics; humanoid structural parts are now the extension.

1.3 In one line

SPG: the part that makes a robot move (joints = reducers)
Halla Cast: the part that keeps a robot light (skeleton = structural frame)

Without SPG, a robot does not move.
Without Halla Cast, a robot becomes heavier.

Both are needed, but "absolutely necessary" applies more strongly to SPG.
Hence SPG trades more expensively.

2. The numbers — how far apart

May 12 close basis.

Item	SPG	Halla Cast
Price	₩135,100	₩16,900
Market cap	~₩3.0tn	~₩620bn
2025 revenue	₩341.7bn	₩155.9bn
2025 OP	₩17.9bn	₩10.3bn
2025 OPM	5.2%	6.6%
2026E revenue	~₩418.6bn	~₩181.4bn
2026E OP	~₩27.1bn	~₩14.6bn
2026E OPM	~6.5%	~8.0%
2026E PER	~110×	~62×
2026E PSR	~7.2×	~3.4×
Value-chain position	Joints (reducers / actuators)	Skeleton (structure / thermal / lightweighting)
Mentions in value-chain piece	5× (most cited)	1-2×
Robot share of backlog	Undisclosed (assumed small)	₩12.4bn (1.1% of total)

2.1 What the table shows

One: SPG is 4.8× more expensive than Halla Cast. ₩3.0tn vs. ₩620bn. PER 110× vs. 62×.

Two: Halla Cast has the higher operating margin. SPG 5.2-6.5% vs. Halla Cast 6.6-8.0%. SPG’s mix blends fan motors (OPM 2-3%) and geared motors (OPM ~10%), pulling the consolidated number down. Precision robot reducers are still only ~3% of revenue.

Three: SPG’s premium is paid for “core-ness." Reducers account for 30-40% of robot bill-of-materials, and SPG is the only Korean producer of all three reducer types. Halla Cast’s structural parts matter, but the level of “irreplaceable” is lower than reducers.

3. SPG — better robot asset, but a price problem

3.1 Why it’s core

SPG is the only company that appeared 5 times across the value-chain piece — Boston Dynamics (Spot, Atlas), Rainbow Robotics, Samsung Electronics, LG Science Park, all source reducers from SPG. Practically every major Korean robot platform.

The SDD-actuator launch (5,000-unit 1H26 target) is the strategic move from “reducer supplier” to “integrated drive-solution supplier.”

3.2 Why the price is heavy

The numbers:

SPG valuation:
Market cap: \~₩3.0tn
2025 OP: ₩17.9bn
2026E OP: \~₩27.1bn
PER: \~110×
PBR: \~11.6×

Cross-check:
Market cap ₩3.0tn ÷ OP ₩27.1bn = \~111×
→ \~111 years of current operating profit pre-paid

PER 110× embeds “Korean robot industry will explode and SPG’s earnings will multiply by tens.” That expectation has to deliver. If it doesn’t, the multiple compresses sharply.

Specifically, precision robot reducers are still ~3% of SPG’s revenue. The other 97% is industrial motors / reducers. The name “robot core-part play” is the marketing frame; the financials still read like an “industrial motor company.”

3.3 Analytical position

Good company, not good price. The reducer-core thesis is real. But the price has substantially front-run the “global humanoid orders → earnings explosion” scenario. Multiple compression resolves through one of two paths: (a) SDD-actuator real sales cleanly clearing the 5,000-unit guidance, or (b) a major global humanoid order officially confirmed. Without one, PER 110× is “supported by expectation alone.”

4. Halla Cast — less core, lighter price, still has verification

4.1 What’s driving the attention

In May 2025, Halla Cast won a humanoid-robot-component order for 3 SKUs totaling ₩12.4bn from a “global AI automaker." An additional ~10 SKUs are reportedly under discussion, with full mass-production targeted for 2H 2026.

The market interprets “global AI automaker” as Tesla. But no public source confirms it is Tesla. The conservative analytical framing: not a “confirmed Tesla supplier” but a “humanoid-component option play tied to a global AI automaker.”

4.2 The backlog mix

Halla Cast’s end-2025 backlog is ₩1,134.6bn — large. But the composition is:

Item	Amount	Share
Display	₩641.9bn	56.6%
Autonomous driving	₩350.6bn	30.9%
EV inverter	₩95.6bn	8.4%
Robot	₩12.4bn	1.1%
Battery	₩11.2bn	1.0%
Other	₩22.9bn	2.0%

Robot is 1.1% of total backlog. Halla Cast is still a display / autonomous-driving auto-electronics supplier that has just begun adding humanoid parts. Robotics is small but optionality.

The option’s value depends on whether the ~10 SKUs under discussion convert. ₩12.4bn growing to ₩50.0bn+ would reclassify the company as a “robot component play.” Without that conversion, it stays “expensive auto-electronics supplier.”

4.3 Where Halla Cast scores better than SPG

Lower price burden. PER 62× is still rich but half SPG’s 110×.

Higher and improving operating margin. Halla Cast’s OPM trajectory: 6.6% (2025) → 7.8% (2026E) → 8.4% (2027F). SPG progresses more slowly from 5.2% to 6.5%.

Defensive base business. Halla Cast’s non-robot backlog of >₩1tn protects against scenarios where robotics doesn’t deliver. SPG also has industrial-motor base business, but the high stock price means “robotics not materializing → share-price drawdown.”

4.4 Where Halla Cast scores worse

Lower value-chain core-ness. Reducers are mandatory for a robot to move. Structural die-cast parts are important but more substitutable — another die-caster could in principle provide them.

No customer confirmation. SPG’s Rainbow / Boston Dynamics relationships are documented. Halla Cast’s “global AI automaker” designation has not been officially identified.

5. Side-by-side — which carries less risk

Item	SPG	Halla Cast	Edge
Robot core-ness	Reducers / actuators (joints)	Structure / thermal / lightweighting (skeleton)	SPG
PER	~110×	~62×	Halla Cast
PSR	~7.2×	~3.4×	Halla Cast
OPM	5.2-6.5%	6.6-8.0%	Halla Cast
Earnings growth (2026E)	+70%	+112%	Halla Cast
Value-chain mentions	5×	1-2×	SPG
Customer verification	Rainbow / BD confirmed	“Global AI automaker” (unconfirmed)	SPG
Base-business safety net	Industrial motors (exists but price-protective effect weaker)	Display / autonomous backlog >₩1tn	Halla Cast
Pricing already-in	Very heavy	Heavy but relatively less	Halla Cast

Relatively, Halla Cast carries less risk. SPG is the better robot asset, but its current price has absorbed that quality too aggressively. Halla Cast is less core but priced less full, with a thicker base-business safety net.

That said, “less risky” ≠ “cheap." Halla Cast at PER 62× is rich for an auto-parts supplier. Without further robot-order expansion, the multiple is hard to defend.

6. Fair value — what’s adequate for each

6.1 SPG

Scenario	Logic	Applied PER	Implied price range
Current (over-anticipation)	Robot revenue 3%, multiple driven by expectation	110×	₩135,100 (now)
Reasonable entry	Confirmed global order + multiple compresses	70-80×	~₩86,000-₩98,000
Sharp-decline interest	SDD volumes + major order delays	50-60×	~₩61,000-₩74,000

At ₩135,100 the price embeds a meaningful share of the “global humanoid order → revenue explosion” scenario. Multiple normalization requires fundamental evidence — SDD-actuator real volumes or a confirmed major global customer order.

6.2 Halla Cast

12-month forward EPS (average of 2026E and 2027F) ≈ ₩333 basis:

Scenario	Applied PER	Implied price	Vs. current
Bear (robot orders stall)	35×	~₩11,700	-31%
Base (current order holds + base business grows)	55×	~₩18,300	+8%
Bull (robot SKU expansion + mass production)	70×	~₩23,300	+38%

Current ₩16,900 sits slightly below base ₩18,300. If robot backlog expands to ₩50.0bn+, the bull-case ₩23,300 opens.

7. What to verify next

7.1 SPG

SDD-actuator actual unit sales (target 5,000)
Confirmed global humanoid customer orders (currently Rainbow / BD anchored)
Precision robot-reducer revenue share (currently ~3% → potential 10%+)
2H 2026 results showing recognized robotics revenue

7.2 Halla Cast

Robot-backlog evolution (₩12.4bn → ₩50.0bn+ would shift classification)
~10 additional SKUs under discussion converting to firm orders
2H 2026 robot mass-production revenue recognition
Margin restoration (2025 6.6% → 2026 8%+)
New plant yield / utilization stabilization

7.3 Joint

Tesla Optimus mass-production timeline (Fremont line currently in “Construction”)
Samsung Electro-Mechanics 2H humanoid camera-module mass production → ecosystem-wide signal
Hyundai Motor Group 3Q RMAC operation → Boston Dynamics Atlas supply-chain acceleration

8. Bottom line

SPG is the “crossroads” of Korea’s robot value chain — the only supplier of reducers to Boston Dynamics, Rainbow Robotics, Samsung, and LG. But ₩3.0tn at PER 110× — for a company where robot revenue is still 3% — requires “everything to go right” for that price to feel justified.

Halla Cast is less core than SPG — structure / thermal / lightweighting rather than reducers. But ₩620bn at PER 62× is half the multiple, with a >₩1tn non-robot base-business safety net. If the robot backlog expands from ₩12.4bn to ₩50.0bn+, re-rating room opens.

Both are expensive. The reason for being expensive differs. SPG: “core part, so expensive.” Halla Cast: “option attached, so expensive.” SPG risk: “if expectations don’t realize, sharp drawdown.” Halla Cast risk: “if option doesn’t realize, reverts to auto-parts multiple.”

FAQ

Q: What’s the most important difference between SPG and Halla Cast? A: Their position in the robot value chain. SPG is the “joint” (reducers); Halla Cast is the “skeleton” (structural frame). Reducers are non-substitutable; structural parts can be sourced from other die-casters. That asymmetry is why SPG carries a higher multiple.

Q: Can SPG’s PER 110× normalize? A: Two paths. (1) Earnings grow fast enough that the multiple compresses naturally — current OP ₩27.1bn rising to ₩80bn+ by 2027-28 would bring PER toward 30-40× — possible if global humanoid mass-production accelerates. (2) Price drops to compress the multiple — if SDD or the major order is delayed. The timing of which path materializes first is the variable.

Q: Is the “global AI automaker” Tesla? A: The market reads it that way, but no public source confirms it. The company filing only says “global AI automaker.” Analytically: treat it as a “global-AI-automaker-tied option play,” not a “confirmed Tesla supplier.” Tesla-confirmation would add a re-rating leg on top of the current premium.

Q: If robotics is only 3% of SPG’s revenue, what’s the other 97%? A: Industrial motors and reducers — appliance motors (air-con / washer), industrial geared motors, fan motors. Stable but single-digit-growth base business. Robotics is the “growth accelerant” on top.

Q: Is Halla Cast’s ₩1tn backlog a real safety net? A: Partially. Display (56.6%) + autonomous (30.9%) = 87.5% of backlog is auto-electronics base business. That means “if robotics doesn’t deliver, the stock reverts to an auto-parts multiple.” Useful protection. But defending PER 62× requires the robotics option to work.

Q: Should one avoid both and look at other robot names? A: Reasonable alternative. The robotics value-chain piece covers other component suppliers (HL Mando, Samsung Electro-Mechanics, LG Innotek, Hyundai Mobis) and pure-play robotics names (Rainbow Robotics, Doosan Robotics). Pure-plays carry heavier price burden; large component suppliers carry tiny robot-revenue share, so robotics catalysts have limited share-price effect.

Q: What signals would justify owning both? A: Three simultaneously: (1) Tesla Optimus or Boston Dynamics Atlas formally entering mass production; (2) Korean component suppliers showing recognized robot revenue in their P&L (quarterly robot revenue ≥₩10bn); (3) global humanoid market-size guidance revised higher. All three together would re-rate Korea robot components broadly.

This article is for research and informational purposes only and does not constitute investment advice. SPG data references Samsung Securities, KB Securities, and Hana Securities reports. Halla Cast data references NH Investment & Securities, Korea Investment & Securities, and KB Securities reports. Prices reflect Maeil Business Newspaper data (May 12). Portions of SPG 2026 consensus and Halla Cast’s actual customer identity are not publicly verified. The market’s “global AI automaker = Tesla” inference is unofficial. Fair-value scenarios are analyst estimates and may be wrong. Korean robotics remains industry-early with uncertain mass-production timelines. Analysis can be wrong. Data cut: May 12, 2026 KST.

Northern Sea Route and Korea: Busan-Rotterdam in 20 Days, Arctic Shipping and the Shipbuilding Beneficiaries

Tue, 12 May 2026 18:00:00 +0900

The easiest mistake in discussing Arctic shipping is to make one of two extreme claims. The first is that the Northern Sea Route will immediately replace the Suez Canal. The second is that Arctic shipping is a distant northern issue with little relevance to Korea. Both are wrong.

The Northern Sea Route is not a finished global highway. It is seasonal, ice-sensitive and dependent on Russian permissions, icebreaker support, insurance and geopolitics. Yet it is also a real industrial option for Korea. Korea is an export economy, it has Busan Port, it has world-class shipyards, and it is starting to align HMM, the Ministry of Oceans and Fisheries, ship finance and regional policy around a single maritime-cluster agenda.

The investment question, therefore, is not simply whether HMM will send a ship through the Arctic. The better question is this: can Korea turn Arctic shipping into a cluster of polar-capable vessels, Busan port logistics, marine finance, insurance, legal services, green fuel and maritime data infrastructure?

TL;DR

The Northern Sea Route is not a year-round replacement for Suez. The most realistic near-term route is the Northeast Passage along Russia’s Arctic coast, commonly discussed as the Northern Sea Route, and it remains a summer-autumn seasonal corridor.

For Busan-Rotterdam, the commercial logic is easy to understand. The conventional Suez route is often framed at about 20,000 km and roughly 30 days, while the Arctic route could reduce that to about 13,000 km and 20 days. That is roughly 35% less distance and about one-third less time. But this is a gross shipping advantage before icebreaker fees, insurance, Russian permissions, polar vessel capex and sanctions risk.

Korea is moving from policy language to execution. The Ministry of Oceans and Fisheries has said a domestic private carrier will pursue a Busan-Rotterdam container trial voyage in the second half of 2026. Korea has also launched a public-private Arctic shipping council, passed Arctic-route legislation and approved HMM’s headquarters relocation to Busan.

The structural beneficiaries are more likely to be shipbuilding and polar vessels, Busan port and hinterland logistics, marine finance, insurance and maritime law, green fuel and bunkering, and maritime data, satellite and navigation services. HMM has policy optionality, but it does not control the core bottlenecks of the route.

The realistic timeline is a 2026 trial voyage, repeated voyages and polar vessel orders in 2027-2029, and a test of seasonal commercial service around 2030. Until regular cargo contracts exist, this is more a policy and capex theme than a direct earnings theme.

1. What Is the Northern Sea Route?

Arctic shipping is an attempt to connect Asia and Europe through the Arctic Ocean rather than through Southeast Asia, the Indian Ocean, Suez and the Mediterranean. A Korea-Europe cargo ship today usually heads south through Singapore, the Indian Ocean and the Suez Canal. The Arctic route heads north and then west across northern Russia or, in another variant, through the Canadian Arctic.

There are three broad routes.

Route	Location	Commercial Reality
Northeast Passage / Northern Sea Route	Russia’s northern coast, from the Bering Strait toward the Kara Sea	The most realistic near-term commercial route
Northwest Passage	Canadian Arctic archipelago	Difficult due to ice, narrow channels and navigation complexity
Transpolar Route	More direct route across the central Arctic Ocean	A future concept, not a practical commercial route today

Korea’s current focus is the Northeast Passage. The reason is practical: Russia’s Arctic coast already has navigation rules, icebreaker support, operational data and a limited but real shipping system. Rosatom’s Northern Sea Route administration site lists navigation rules, route boundaries, icebreaker escort tariffs and communication recommendations. This is not a free ocean highway. It is a controlled, seasonal corridor that depends heavily on Russian infrastructure.

That distinction matters. A shorter route is not automatically a better business if someone else controls the permissions, icebreaker support and emergency infrastructure. Arctic shipping is therefore not just a freight-rate story. It is a combined story of shipbuilding, insurance, sanctions compliance, port strategy, maritime data and diplomacy.

2. Why It Matters: Busan-Rotterdam in 20 Days

The commercial case starts with distance and time. Maritime Executive summarized Korea’s planned Busan-Rotterdam trial voyage by noting that the Arctic route could reduce distance by about 35%, from roughly 20,000 km to 13,000 km, and cut transit time from about 30 days to 20 days.

Route	Conventional Suez Route	Arctic Route
Distance	About 20,000 km	About 13,000 km
Transit time	About 30 days	About 20 days
Reduction	-	About 35% distance, about 33% time

The math is simple.

Distance reduction = (20,000 - 13,000) / 20,000 = 35.0%
Time reduction = (30 - 20) / 30 = 33.3%

For a non-specialist, 10 days is meaningful. Cargo owners can hold less inventory. Carriers can potentially turn ships faster. Fuel use can fall per voyage. The case is especially relevant for cargo categories where inventory and delivery timing matter: electronics, auto parts, batteries, machinery and high-value industrial goods.

But the gross saving is not the net saving. Arctic shipping adds costs. A ship may need ice-class specifications or escort support. Insurance premiums can rise. Russian permits and navigation support matter. Ice can force delays, rerouting or waiting time. A delay on a container liner service can disturb the wider network schedule.

The real economic question is therefore not whether the route is shorter. The question is whether the distance advantage exceeds the polar-navigation premium. Korea’s 2026 trial voyage is important because it will begin to produce real operating data for that question.

3. Why Now: Ice, Chokepoints and China

The Arctic route is not new. What has changed is the combination of a longer seasonal navigation window and a stronger geopolitical need for alternatives.

First, Arctic sea ice is thinner and less extensive than it used to be. NOAA’s 2025 Arctic Report Card says that at the end of summer 2025, Arctic sea ice was younger, thinner and 28% less extensive than in 2005. This does not mean the Arctic is open year-round. It means the summer-autumn probability of navigation is rising. Korea’s September-October trial window is designed around that seasonal reality.

Second, Suez, the Red Sea and Hormuz have become visible geopolitical chokepoints. Asia-Europe shipping normally passes through Suez, the Red Sea, Bab el-Mandeb or nearby strategic waters. When Middle East risk rises, major carriers consider or implement Cape of Good Hope diversions. In 2026, reporting around Maersk, Hapag-Lloyd and CMA CGM again highlighted how quickly the Suez-Red Sea corridor can become a security problem. Arctic shipping is therefore not only a shorter route. It is a logistics-security option.

Third, China and Russia are moving early. China views Arctic shipping as part of a broader “Polar Silk Road” logic. Chinese-backed Sea Legend pursued a 2025 Arctic container service connecting Ningbo-Zhoushan with Felixstowe, while Chinese state media emphasized the route’s time advantage. Russia sees the NSR as a strategic system linked to Arctic resources, LNG, oil exports, military presence and control over northern sea lanes.

Put together, Arctic shipping is not just climate news. It is a strategic industry where shipping, energy, security, shipbuilding, port policy and satellite data overlap.

4. Where Things Stand

Globally, the Northern Sea Route is still not a mainstream liner route. High North News, citing Centre for High North Logistics data, reported that the 2025 NSR season lasted about four and a half months, with 103 transit voyages and roughly 3.2 million tons of cargo. Container traffic increased, but it remains tiny relative to global container trade. Tankers, bulk carriers and LNG-related movements are still central.

Korea is now entering the execution phase. The Ministry of Oceans and Fisheries said in its 2026 work plan that a domestic private carrier would pursue a Busan-Rotterdam container trial through the Arctic route in the second half of 2026. The ministry also laid out incentives: up to KRW 11 billion for polar-navigation vessel construction, 50-100% port facility fee reductions, a 1 percentage point reduction in ship-finance investment rates, and an increase in loan-to-value recognition up to 90%.

In January 2026, Korea launched a public-private council in Busan to promote Arctic-route activation. In May 2026, the National Assembly passed special legislation for the promotion of Arctic-route use and related industries. The framework includes five-year plans, an Arctic Route Committee under the prime minister, an implementation headquarters inside the Ministry of Oceans and Fisheries, fiscal and financial support, R&D, professional training and a comprehensive support center.

HMM’s relocation to Busan fits the same pattern. Yonhap reported on May 8, 2026, that shareholders approved HMM’s move from Seoul to Busan. This follows the relocation of the Ministry of Oceans and Fisheries to Busan and the government’s effort to develop the city as a Northeast Asian maritime hub.

Korea is not merely trying to send one ship through the Arctic. The broader plan is to make Busan a cluster for shipping headquarters, ports, logistics, marine finance, maritime law, green fuel and Arctic-route data.

5. Geopolitics: A Commercial Route Inside a Strategic Space

The Northeast Passage follows Russia’s Arctic coast. That makes Russia the physical gatekeeper. Permits, icebreaker support, ice navigation, search-and-rescue infrastructure and route data are all tied to Russia. In a sanctions environment, ship finance, insurance, chartering, payments and equipment procurement become complicated.

Korea’s government recognizes this. Its 2026 work plan explicitly treats sanctions as a variable. If sanctions ease, Korea can expand container and LNG/resource transport through the Northeast Passage. If sanctions persist, Korea may test alternatives such as the Northwest Passage. That is a realistic stance. The map may be shorter, but the compliance path is longer.

NATO also treats the Arctic as a strategic region. NATO says Finland and Sweden’s accession has strengthened its posture in the Arctic and High North, while Russia’s military activity and China’s growing Arctic interest have made the region more strategically competitive. The Arctic is now a space where energy, minerals, undersea cables, submarine routes, satellite surveillance and military logistics intersect.

Korea is not an Arctic state. But it became an observer to the Arctic Council in 2013. It is a shipbuilding power, an export economy and a U.S. ally. Korea’s Arctic role is not territorial. It is industrial, logistical and alliance-based.

6. Where Korea Can Benefit

The simplest market headline is “HMM beneficiary.” That is partly true. HMM is Korea’s largest carrier and is directly tied to the Busan relocation and trial-voyage policy narrative. But HMM is not the deepest bottleneck. HMM can operate ships. It does not control polar vessel technology, Russian permissions, icebreaker availability, insurance markets or Arctic route data.

The layers of potential benefit look more like this.

Layer	Benefit Strength	Timing	Why It Matters
Shipbuilding and polar vessels	Very high	2026-2030	Vessel orders and technology development precede route commercialization
Shipping operators	Medium	2026-2030	Operators benefit, but cargo, insurance, permissions and seasonality are constraints
Busan port, terminals and hinterland logistics	Medium-high	2027-2032	Seasonal services could lift transshipment and port-linked logistics
Marine finance, insurance and maritime law	Medium-high	2026-2030	High-risk routes require financing, insurance and compliance structuring
Green fuel and bunkering	Medium-high	2027-2035	HFO restrictions and polar ESG standards alter vessel fuel systems
Maritime data, satellite and navigation	Medium-high	2026-2035	Ice, weather, communications and route optimization become critical infrastructure
Energy and resource logistics	High	Already visible	Current NSR cargo remains heavily linked to oil, LNG and resources
General container liner service	Low-medium	After 2030	Regular liner economics remain unproven

The first hard-capex category is shipbuilding. Arctic routes require vessels with ice-class capability, low-temperature operation, reinforced hulls, special navigation systems and cleaner propulsion.

HD Hyundai Heavy Industries said in April 2026 that it had won a $348.9 million order from the Swedish Maritime Administration for a dedicated icebreaker, the first such overseas order for a Korean shipbuilder. Hanwha Ocean was selected as preferred bidder for Korea’s next-generation icebreaking research vessel, designed with LNG dual-fuel electric propulsion, two-way icebreaking through 1.5 meters of ice, operation at minus 45 degrees Celsius and Polar Class 3 capability.

This connects directly to our broader Korean shipbuilding work. Hanwha Ocean has LNG and naval exposure, with polar vessels adding another special-purpose ship option. Samsung Heavy Industries is tied to LNG and FLNG infrastructure, which matters if Arctic energy logistics expand. Hanwha Engine sits in the propulsion and marine-equipment layer, where cleaner fuels and special-purpose vessel orders can create second-order exposure.

7. Five Common Misreadings

The first misreading is that Arctic shipping will immediately replace Suez. It will not. The NSR is still a seasonal alternative, constrained by ice, weather, search-and-rescue infrastructure, insurance and Russian permissions.

The second misreading is that a shorter route is automatically cheaper. Distance and fuel use may fall, but Arctic voyages add icebreaker support, higher insurance, special vessel capex, Russian navigation support and delay risk.

The third misreading is that HMM is the biggest beneficiary. HMM has policy optionality, but it does not own the chokepoints. The deeper bottlenecks are polar vessel technology, icebreaker capacity, insurance, data and permissions.

The fourth misreading is that Arctic shipping is automatically green. Shorter voyages can reduce fuel use, but Arctic ecosystems are highly fragile. Heavy fuel oil spill risk, black carbon, marine-mammal disruption and weak emergency infrastructure all matter. The IMO’s Arctic heavy-fuel-oil restrictions began in July 2024, but exemptions and delays remain a point of criticism from environmental groups.

The fifth misreading is that Korea is irrelevant because it is not an Arctic state. Korea has no Arctic territory, but it participates through shipbuilding, shipping, ports, export logistics and alliance networks. In this market, industrial capacity matters almost as much as geography.

8. What To Track

For 2026, the key event is the Busan-Rotterdam trial voyage. The important details are the actual carrier, vessel name, Russian permission status, insurance terms, cargo owners, route economics, actual transit time and delay experience. A successful trial will not instantly create an earnings theme. But it will create operating data for ship design, insurance pricing, port work and cargo marketing.

For 2027-2029, the key question is repetition. One trial voyage is an event. Repeated voyages are an industry. Orders for ice-class container ships, icebreaker support vessels, polar research vessels or icebreaking LNG carriers would make the capex theme more tangible.

Around 2030, the question becomes whether the route can support seasonal commercial service. Occasional September-October transits are one thing. A repeatable seasonal service is another. The latter would matter much more for Busan Port, terminals, feeder networks, ship finance and insurance earnings.

The final variable is sanctions and Arctic security. If sanctions persist, Northeast Passage commercialization is constrained. If sanctions ease or Arctic energy logistics accelerate, LNG, oil and minerals could move before container liners do.

Final Note

The Northern Sea Route is not a typical short-term equity theme. The headline that can move stocks may be HMM’s relocation or Korea’s 2026 trial voyage. But the deeper money is likely to flow into polar-capable vessels, ice-class designs, port automation, marine finance, insurance, maritime law, green fuel systems and satellite-based maritime data.

The right question is therefore not “who sends the ship?” but “who reduces the risk and solves the bottleneck?” Korea has some of those bottleneck capabilities already. It has world-class shipyards, Busan as a Northeast Asian transshipment hub, and experience in LNG carriers, special-purpose ships and cleaner propulsion systems.

The weaknesses are just as clear. Korea does not control the route. The Northeast Passage is tied to Russia and sanctions. Container-liner economics remain unproven. That is why the conclusion should stay conservative.

Before a successful trial, this is a policy theme. Before repeated voyages, it is a capex theme. Before regular cargo contracts, it is not yet a direct earnings theme. But if the policy theme becomes a capex theme, the first industries to move are likely to be shipbuilding, polar vessels and the infrastructure around them.

Sources Reviewed

Sources reviewed include the Ministry of Oceans and Fisheries 2026 work plan, the ministry’s Arctic-route public-private council release, Maritime Executive’s report on Korea’s Busan-Rotterdam trial voyage, NOAA’s Arctic Report Card 2025, High North News on the 2025 NSR season, Maritime Korea on the Arctic-route special law, Yonhap on HMM’s relocation to Busan, Arctic Council observer information for Korea, NATO’s Arctic security brief, HD Hyundai’s icebreaker order release, Hanwha Ocean’s next-generation icebreaking research vessel release, ISDP’s analysis of Korea’s Arctic role, and Reuters, Maersk and Global Times reporting on shipping-route and Arctic-container developments.

Pearl Abyss 1Q26 Comprehensive — OP ₩212.1bn (+48% Above Consensus), 64.6% Margin. Company's First-Ever Guidance Sets FY26 OP at ₩487.6-572.6bn

Tue, 12 May 2026 18:00:00 +0900

📚 Pearl Abyss Series — 1Q26 Comprehensive Edition. Earlier: Earnings Preview (4/21) — Revenue ₩395.0bn / OP ₩205.0bn Central Estimate Earlier: Earnings-Week Setup (5/11) — Consensus ₩143.5bn but Likely ₩250bn+ Actual

🔗 Related: Pearl Abyss Hub — 263750 Price / Target / Crimson Desert Analysis · CCP / EVE Divestment · Shinhan Target-Price Gap Analysis Follow-up: Pearl Abyss May 21 IR Watch — Capital Return, DLC and DokeV Decide the Next Re-Rating

Pearl Abyss released 1Q26 earnings together with its first-ever quantitative annual guidance. The headline: OP ₩212.1bn beat consensus by +48%, and the company set FY26 OP at ₩487.6-572.6bn. 2Q Crimson Desert revenue guidance of ₩224.2-276.5bn (-16% to +4% vs. 1Q) directly rebuts the market’s “1Q boom → 2Q cliff” fear. The arithmetic that has to land for the guidance to be confirmed is still ahead.

TL;DR

1Q26 print. Revenue ₩328.5bn, OP ₩212.1bn, NI ₩158.0bn. OPM 64.6%. OP +48% above consensus (₩143.5bn).
FY26 company guidance. Operating revenue ₩879.0-975.4bn, OP ₩487.6-572.6bn, OPM 55.5-58.7%. Materially above the implicit pre-print market expectation (~₩410bn).
2Q guidance. Crimson Desert revenue ₩224.2-276.5bn — -16% to +4% vs. 1Q’s ₩266.5bn. Plateau, not cliff.
Why the surprise. Not primarily a revenue beat — it’s a light cost base. R&D was already expensed in prior quarters, marketing was efficient, and the revenue-recognition method (net / blended) deducted platform fees from revenue rather than booking them as expense.
Vs. prior predictions. Preview’s ₩395.0bn revenue ran 17% too high; OP ₩205.0bn ran 3% too low. Revenue-mix-recognition assumption was wrong, but the OP direction was correct.
What to watch next. May 21-22 institutional NDR, the 6M-unit announcement timing, whether 2Q lands inside the company range, and DLC / platform-extension scheduling.

1. Today’s disclosed numbers — exact figures

1.1 1Q26 preliminary earnings

Item	Amount
Revenue (operating revenue)	₩328.499bn
Operating profit	₩212.096bn
Continuing-operations net income	₩170.0bn
Final net income (incl. discontinued-operations loss)	₩157.969bn
Operating margin	64.6%
Net margin	48.1%
Implied operating expenses	₩116.4bn

Cross-check: OPM = 212.096 / 328.499 = 64.6% ✓

Why two NI figures: Pearl Abyss completed the divestiture of CCP Games (developer of EVE Online) on May 6. The CCP-related loss (₩-12.0bn) was reclassified into “discontinued operations.” For the operating economics of the business that remains, use continuing-operations NI ₩170.0bn; for shareholder-attributable NI, use ₩158.0bn.

1.2 Revenue by game

Game	1Q revenue	Share
Crimson Desert	₩266.5bn	81%
Black Desert	₩61.6bn	19%
Total	₩328.1bn	100%

Crimson Desert alone produced 81% of total revenue — from just 12 days of sales (March 20-31).

1.3 Revenue by region and platform

Region	Share
Americas / Europe	81%
Asia	13%
Korea	6%

Platform	Share
PC	59%
Console	38%
Mobile	3%

Confirmed by the geographic distribution: Crimson Desert is a Western-PC-and-console global hit. 81% from the Americas / Europe is unusual for a Korean game studio — most Korean titles lean heavily Asia.

1.4 Vs. consensus

Item	Consensus	Actual	Diff
Revenue	₩311.9bn	₩328.5bn	+5.3%
Operating profit	₩143.5bn	₩212.1bn	+47.8%
Net income	₩110.4bn	₩158.0bn	+43.1%

Small revenue beat (+5%), very large OP beat (+48%). The asymmetry is the story.

1.5 Vs. comparable periods

Comparison	Revenue	OP
YoY (1Q25)	+293%	Loss ₩-5.2bn → Profit ₩212.1bn
QoQ (4Q25)	+244%	Loss → Profit

One title repositioned the entire company in a single quarter.

2. What earlier predictions got right and wrong

2.1 Vs. the Earnings Preview (April 21)

Item	Preview	Actual	Diff
Revenue	₩395.0bn	₩328.5bn	-16.8%
OP	₩205.0bn	₩212.1bn	+3.5%
OPM	51.9%	64.6%	+12.7pp

Revenue was overpredicted by 17%; OP was under-predicted by 3%; OPM gap was 13 percentage points.

The source of the contradiction: revenue-recognition method.

2.2 Revenue recognition — the preview’s key assumption was wrong

Game-revenue accounting has two methods. Per company disclosure:

Console (PlayStation / Xbox): platform fee is deducted first, with the residual booked as revenue
PC (Steam, etc.): user payment is booked net of sales tax
Offline packaging: contract terms produce booked revenue below actual unit sales

The preview leaned more heavily on gross-recognition assumptions (platform fees in revenue, then out as expense). Console turned out closer to net (fee out first). Result: lower revenue line, but fewer fees in expenses → higher reported margin.

2.3 Per-unit recognized revenue — revised

Reverse-engineering Crimson Desert’s 1Q revenue ₩266.5bn against the official ~4M unit count:

Per-unit recognized revenue = ₩266.5bn ÷ 4M = \~₩66,625/unit

Cross-check: 66,625 × 4M = ~₩266.5bn ✓

Lower than the preview’s gross assumption of ₩80,000/unit and higher than the Earnings-Week Setup post’s working estimate of ₩58,000/unit (which had baked in EVE-inclusive legacy revenue, producing some drift).

Revised conclusion: per-unit recognized revenue ~₩66,625. Neither full gross nor full net — a blended structure.

2.4 Did the Earnings-Week Setup’s read hold?

The week-of post said: “Consensus is ₩143.5bn but actual could be ₩250bn+.” Actual was ₩212.1bn — well above consensus, but below the optimistic broker estimates (Shinhan ₩254.7bn, Meritz ₩275.2bn).

The “highest surprise probability of the week” call was correct in direction, lower in magnitude than the upper-end broker estimates.

3. Why the OP surprise — by cost line

3.1 Operating expense detail

From the company’s 1Q26 IR Letter:

Item	Amount	vs. revenue	vs. prior quarter
Labor	₩38.4bn	11.7%	+29%
Platform fees / payments	₩42.5bn	12.9%	+193%
Marketing	₩23.4bn	7.1%	+152%
Depreciation	₩2.4bn	0.7%	-4%
Other	₩9.7bn	3.0%	-14%
Total	₩116.4bn	35.4%	+73%

3.2 Three reasons for OPM 64.6%

Reason 1: R&D already expensed in earlier quarters.

Crimson Desert was developed over multiple years. Most development costs were already booked to prior-period expense. After launch, revenue rises sharply while incremental development cost is small. Depreciation at just 0.7% of revenue reflects this.

Analogy: a factory built over 5 years has already absorbed its construction cost. When the factory starts production, revenue appears with no fresh construction expense → margin spikes.

Reason 2: Marketing was efficient.

Marketing spend ₩23.4bn against Crimson Desert revenue ₩266.5bn = 8.8%. Low for a launch quarter. Word-of-mouth pull was strong online; the company didn’t need to flood TV / outdoor with launch ads.

Reason 3: Revenue-recognition method shrinks the apparent fee rate.

Platform fees ₩42.5bn against revenue ₩266.5bn = 12.9%. That looks low, but it’s because console revenue is recognized net of platform fees. The total fees Pearl Abyss actually pays platforms is higher; accounting just doesn’t capture all of it on the expense line.

3.3 The core — the cash earned is real

Combining the three reasons: revenue-recognition produces a “higher-looking margin” optical effect, but the actual dollars earned (OP ₩212.1bn, NI ₩158.0bn) are strong regardless of recognition method. Accounting choice doesn’t change the cash itself.

4. Company guidance — why first-time numbers matter

4.1 Pearl Abyss issued quantitative guidance for the first time

Quantitative annual OP ranges from Korean game studios are rare. This is the first.

FY26 guidance:

Item	Range
Operating revenue	₩879.0-975.4bn
Black Desert	₩234.9-240.6bn
Crimson Desert	₩644.1-734.8bn
Operating expenses	₩391.4-402.8bn
Operating profit	₩487.6-572.6bn
Operating margin	55.5-58.7%

2Q guidance:

Item	Range
Operating revenue	₩271.3-324.7bn
Black Desert	₩47.1-48.2bn
Crimson Desert	₩224.2-276.5bn
Operating expenses	₩141.7-148.0bn
Operating profit	₩129.6-176.7bn
Operating margin	47.8-54.4%

4.2 Why these numbers matter

One: the company directly rebutted the 2Q-cliff concern.

The market’s biggest worry was the package-game pattern — concentrated revenue in launch quarter, sharp drop after.

Company’s 2Q Crimson Desert revenue: ₩224.2-276.5bn. 1Q Crimson Desert revenue: ₩266.5bn.

2Q change:
Low end: 224.2 / 266.5 - 1 = -15.9%
High end: 276.5 / 266.5 - 1 = +3.8%

-16% to +4% — plateau, not cliff. If the company is seeing this trajectory in internal sales data, the “1Q one-off” narrative weakens.

2Q OP is guided ₩129.6-176.7bn, below 1Q’s ₩212.1bn. The reason isn’t the business — it’s performance-bonus accrual. The company expects 2Q labor cost to roughly double vs. 1Q. Revenue holds; cost rises; OP falls. That’s expense structure, not a business deceleration.

Two: FY26 OP ₩487.6-572.6bn is well above prior market expectation.

Against the implicit pre-print expectation (~₩410bn):

Low end ₩487.6bn - prior ₩410bn = +₩77.6bn (+19%)
High end ₩572.6bn - prior ₩410bn = +₩162.6bn (+40%)

This isn’t just “good print” — it’s a signal that annual earnings estimates themselves need to lift. As sell-side updates models, the starting point of valuation work — including the earlier Shinhan target-price gap — shifts.

4.3 Can guidance be trusted at face value?

The company explicitly flagged: “This guidance is based on current estimates under an uncertain operating environment, and actual results may differ materially from the figures shown.”

For guidance to be realized:

Crimson Desert needs to keep selling at 500K+ units/month through the second half
Major updates / discount events need to sustain sales
Operating costs need to stay within the guided ₩391.4-402.8bn range

Any one breaking can pull the print below even the low end.

5. Lessons from the prediction series

5.1 What was right

Earlier analysis	Confirmed
1Q OP could exceed ₩200bn	✓ ₩212.1bn
OPM ≥45% was the key check line	✓ 64.6%
Revenue-recognition is the central variable	✓ Company confirmed blended recognition
Cost structure might be lighter than expected	✓ Operating expenses at 35.4% of revenue
Earnings-Week’s “highest surprise probability” call	✓ OP beat consensus by +48%

5.2 What needs revision

Earlier assumption	Revised
Revenue ₩395.0bn (gross-recognition lean)	→ Actual ₩328.5bn. Console net, PC net-of-tax. -17% gap
Legacy-game revenue ₩90-97bn (EVE included)	→ EVE classified as discontinued. Continuing-basis legacy = Black Desert ₩61.6bn
Per-unit recognized revenue ~₩80,000 (gross)	→ Actual ~₩66,625 (blended)
FY26 OP expectation ~₩410bn	→ Company guidance ₩487.6-572.6bn. Raise required

5.3 Lesson

Revenue-recognition method is the hardest-to-forecast variable in game-company analysis. Same units sold and same price can produce 30%+ revenue differences depending on accounting choice. OP, however, is much less variable across methods — that’s why the OP direction was right even when revenue was wrong.

6. Valuation — updated against company guidance

6.1 EPS estimates

Shares outstanding 64,247,855. Post-tax conversion ratio (OP → NI) assumed 70-75% based on 1Q actual (~74.5%) plus tax-rate and discontinued-operations effects.

Scenario	FY26 OP	Tax conversion	NI	EPS
Bear (guidance low end)	₩487.6bn	70%	₩341.3bn	₩5,313
Base (guidance midpoint)	₩530.1bn	72.5%	₩384.3bn	₩5,982
Bull (guidance high end)	₩572.6bn	75%	₩429.5bn	₩6,685

Cross-check: Base EPS = 384.3 / 64.248M = ₩5,982 ✓

6.2 What the current price implies

May 12 close ₩52,800:

Scenario	EPS	Current-price PER
Bear	₩5,313	9.9×
Base	₩5,982	8.8×
Bull	₩6,685	7.9×

Current price implies PER 8.8× on the guidance midpoint. That looks low. The market discounts the multiple because of one concern: repeatability — what if 2026 earnings are a one-off Crimson Desert launch year?

6.3 PER × scenario grid (reference)

EPS scenario	PER 10×	PER 12×	PER 14×	PER 15×
Bear ₩5,313	₩53,100	₩63,800	₩74,400	₩79,700
Base ₩5,982	₩59,800	₩71,800	₩83,700	₩89,700
Bull ₩6,685	₩66,800	₩80,200	₩93,600	₩100,300

This is “EPS scenario × PER multiple” arithmetic. PER 10× is closer to “single-package one-off” framing; PER 15× requires recognition as “a studio capable of running multiple successful titles.” Which multiple the market awards depends on 2Q guidance delivery, sales persistence, and DLC / platform-extension announcements.

7. Forward price triggers — by sequence

7.1 Near-term — sell-side EPS revisions (this week to next)

With print and guidance now public, sell-side models revise. Since company guidance (₩487.6-572.6bn) sits well above the prior market expectation (~₩410bn), EPS upgrades and target-price upgrades are likely.

Price tends to respond to FY26 EPS revision magnitude more than report headlines.

7.2 May 15 — 1Q quarterly filing

The full filing on DART includes cost-line detail, revenue-recognition footnotes, and balance-sheet detail. This is the key source for distinguishing “structurally light cost” from “temporarily deferred cost.”

7.3 May 21-22 — Meritz Securities NDR (Non-Deal Roadshow)

Yeouido-based 1:1 and group meetings with domestic institutions. First major opportunity for institutional models to update post-print. Post-NDR institutional flow can reset the share-price level.

7.4 May-June — 6M unit announcement

Official latest unit count is 5M (April 15). Acceleration to 6M strengthens guidance credibility. Hitting FY26 Crimson Desert guidance (₩644.1-734.8bn) requires sustained add-on sales.

7.5 August — 2Q earnings

The single most important trigger. Delivering 2Q guidance (OP ₩129.6-176.7bn) weakens the “1Q one-off” narrative significantly. Missing the low end damages guidance credibility.

7.6 TBD — DLC / platform expansion

Pearl Abyss stated: “Researching ways to extend the game into different dimensions, including DLC.” Not a confirmed plan yet, but on-record. A paid expansion materially extends sales-cycle longevity — CD Projekt Red’s Blood and Wine expansion (Witcher 3) launched ~18 months post-base and re-extended the title’s revenue curve as a comparable reference.

Pipeline titles (DokeV in pre-production, Plan 8 in concept iteration) remain long-dated options. Premature to load them into base-case valuation.

7.7 Trigger summary

Trigger	Timing	Importance
Sell-side EPS upgrade reports	This week to next	High
1Q quarterly filing (cost notes)	May 15	High
Meritz NDR (institutional model updates)	May 21-22	High
6M unit announcement	May-June	High
2Q earnings	August	Highest
DLC / platform-extension announcement	TBD	Long-term core

8. Investment-thesis update

8.1 Prior thesis

“Crimson Desert sells more than expected, and the market is slow to price it in.”

8.2 Updated thesis

“Pearl Abyss now has 1Q numbers that confirm Crimson Desert’s commercial trajectory and FY26 guidance that partially dismantles the 1Q-one-off concern. The current price is PER 8.8× on the guidance midpoint — still low. For the market’s residual doubt (‘is this earnings stream repeatable?’) to fully clear, 2Q sales persistence, the 6M unit milestone, and post-NDR institutional re-rating need to land. The real alpha resides in the window where that doubt resolves.”

8.3 Invalidation conditions

This thesis can break:

2Q Crimson Desert revenue undershoots the company-guidance low end (₩224.2bn) → cliff narrative returns
2Q OPM collapses below 45% → cost-control failure
6M-unit milestone meaningfully delays, Steam / console rankings deteriorate → sales-persistence damage
Pure single-package monetization with no DLC / update extension → no PER expansion path
DokeV / Plan 8 development schedule slips materially → multi-IP optionality fades

9. Bottom line

Pearl Abyss 1Q26 print is strong. OP ₩212.1bn, OPM 64.6%, +48% above consensus. The bigger development is FY26 guidance — OP ₩487.6-572.6bn, and 2Q Crimson Desert revenue at plateau (-16% to +4%) rather than cliff.

As the market absorbs the numbers, sell-side EPS upgrades, the May 21-22 NDR, and post-NDR institutional repositioning can compound. Current price ₩52,800 = guidance-midpoint PER 8.8× — arithmetically low.

But “looks low” becoming “confirmed undervaluation” requires verification — whether 2Q sales hold, whether 6M lands, whether costs stay within range. That sequence is what makes ₩70K-handle a reasonable level, and DLC / platform extension is what would unlock anything above. 1Q opened the door. What’s behind the door is determined by 2Q.

FAQ

Q: Is the 1Q OPM 64.6% sustainable? A: This margin mixes three effects: (1) prior-quarter R&D expense (structural, sustainable); (2) efficient launch-quarter marketing (partly one-off); (3) revenue-recognition method shrinking the apparent fee line (structural). The company’s 2Q guidance OPM is 47.8-54.4%, lower than 1Q because of performance-bonus accrual — not because the business decelerated.

Q: Should one fully trust the company guidance? A: The company explicitly flagged “estimates under uncertain operating environment.” For the guidance to be realized: (a) Crimson Desert needs to sustain 500K+ units/month through 2H, (b) major updates need to support sales, (c) costs need to stay within the guided range. That said, this is the company’s first quantitative guidance, and the 2Q range (-16% to +4% vs. 1Q) is conservative on the upper end, both of which add credibility.

Q: Why did consensus ₩143.5bn miss actual ₩212.1bn by so much? A: The consensus median was anchored in post-launch March estimates and hadn’t fully absorbed April sales data (5M units cleared) or the cost structure. Some sell-side estimates (Shinhan ₩254.7bn, Meritz ₩275.2bn) were closer to actual but the median update was slow.

Q: Revenue -17% from prior estimate but OP +3% — how? A: Revenue recognition. Console deducts platform fees from revenue first, so revenue is lower-looking but the fee line in expenses is also smaller. Margin lifts. The same business can show 30%+ revenue divergence depending on accounting choice.

Q: Why does 2Q OP guidance fall below 1Q despite revenue being held? A: Cost rises. The company expects 2Q labor cost to roughly double vs. 1Q — performance-bonus accrual on the 1Q print. Revenue holds at -16% to +4% but expense rises, so OP comes down. Not a business deceleration.

Q: Is PER 8.8× really cheap? A: It’s low against both the Korean-market average and global game-studio peers. But the market discounts the multiple for the “repeatability” question — if Crimson Desert becomes a single-package one-off, FY26 earnings won’t recur. PER 8.8× is the multiple the market awards under that doubt.

Q: Which trigger matters most? A: August 2Q earnings. Delivering the company guidance (OP ₩129.6-176.7bn) cleanly dismantles the “1Q one-off” narrative. Intermediate checkpoints are how May 21-22 NDR-driven institutional model updates land, and how the 6M-unit announcement paces between May and June.

Q: Among DokeV / Plan 8 / DLC, which matters first? A: Short term, DLC. The company explicitly stated on-record that “DLC and ways to extend the game into different dimensions” are under research. CD Projekt Red’s Witcher 3: Blood and Wine expansion, which extended the title’s revenue curve by 18+ months, is the comparable reference. DokeV and Plan 8 remain long-dated options not yet appropriate to load into base-case valuation.

This article is for research and informational purposes only and does not constitute investment advice. 1Q26 figures are from Pearl Abyss’s KIND (KRX) regulatory filing and 1Q26 Earnings Letter. Company guidance (FY26 and 2Q) is taken from the company’s IR materials, where the company explicitly states “based on uncertain estimates and actuals may differ materially.” Revenue-recognition method (console net / PC net-of-tax) is drawn from the company’s IR Letter narrative. Cost-line figures reference the IR Letter Appendix. CCP-divestment-related discontinued-operations loss is reflected in 1Q; cash-inflow effects flow more fully into 2Q. EPS / PER / scenario-grid in the valuation section are analyst estimates and may be incorrect. Analysis can be wrong. Data cut: May 12, 2026 KST.

Korea Quality Re-Rating Watch 2026-05-12: VM, Gigavis, TES — Smart Money on Earnings Inflection

Tue, 12 May 2026 16:30:00 +0900

Macro Dashboard

Both Korea and U.S. market regimes are rated Bull as of May 12, 2026.

Indicator	Level	5D Change	Signal
KOSPI	7,643.1	+3.5%	Bull
KOSDAQ	1,179.3	-2.5%	Lagging
USD/KRW	1,490	+1.1%	KRW softening
VIX	18.9	+8.7%	Stable
US 10Y	4.41%	-0.01pp	Flat
Brent	$107.5	+7.4%	Rising

KOSPI is outperforming KOSDAQ by 6 percentage points over five days — a selective, large-cap-led bull rather than a broad one. KRW softening and Brent’s sharp 5-day climb add macro tail-risk worth tracking even within a bull stance.

Market Wrap (Close data from 2026-05-08; May 12 official close pending)

The May 8 session is best described as selective risk-on: the broad breadth reading was firmly constructive — KR Discovery hit 100/100 with 290 stocks passing — but the character underneath was rotational, not indiscriminate.

Sector leaders: Power cable, fiber optic, and energy infrastructure stocks dominated. Gaon Cable (가온전선) surged nearly +30% with institutional buying of over +21 billion KRW. Daehan Electric Wire (대한전선, 001440.KS) logged coordinated foreign (+167.7B KRW), institutional (+35.3B KRW), and program (+171.5B KRW) net buying in a single session — the kind of tri-party flow that signals conviction rather than momentum chasing. AI datacenter and optical networking themes repeated across brokerage and channel sources, with Daehan Fiber Optics (대한광통신) reaching RS 99.9 after a sharp multi-day rally.

Sector laggards: Gaming and select defense names underperformed. Large-cap semiconductors absorbed heavy foreign outflows: Samsung Electronics (삼성전자, 005930.KS) saw approximately -2.55 trillion KRW in foreign net selling, partially offset by +293.4B KRW of institutional buying. The stock held (-1.1%), suggesting underlying institutional demand, but the foreign supply overhang remains a near-term headwind. MLCC and components names saw split signals — institutional defense paired with ongoing foreign and short pressure.

Breadth context: KR Discovery is on Day 19 of a confirmed Follow-Through Day signal. 43.9% of stocks are above their 50-day MA; 51.6% above their 200-day MA. Conditions support stock selection, not blanket buying. The premium in large-cap KOSPI over KOSDAQ suggests rotation toward quality defensibles rather than broad speculative expansion.

Today’s Quality Re-Rating Candidates

The May 12 meta-screener and overlap analysis flag five names clearing three screeners simultaneously. The editorial lens: good businesses where institutional money is entering and the market is beginning to re-price the earnings story.

Candidate Table

Meta Rank	Ticker	Name	Meta Score	Screeners	Key Metrics
1	089970.KQ	VM (브이엠)	79.0	Quality Compounder, Smart Money Quality, Smart Money Earnings, Consensus ↑	ROE 14.8%, OP YoY +387%, 5D F+QI +₩17.8B
2	420770.KQ	Gigavis (기가비스)	72.7	Quality Compounder, Smart Money Quality, Smart Money Earnings	ROE 7.3%, OP YoY +777%, DART supply contract Apr 24
3	095340.KQ	ISC	68.5	Quality Compounder, Smart Money Quality, Smart Money Earnings, PEAD	ROE 10.5%, OP YoY +34%, Short interest 12.8%
4	095610.KQ	TES (테스)	64.8	Quality Compounder, Smart Money Quality, Smart Money Earnings	ROE 14.5%, OP YoY +50%, 5D F+QI +₩49.4B
5	082920.KQ	Vitrocel (비츠로셀)	59.1	Quality Compounder, Smart Money Quality, Smart Money Earnings	ROE 17.2%, OP YoY +33%, Foreign ownership 30%
6	218410.KQ	RFHIC	54.4	Smart Money Quality, Smart Money Earnings, PEAD	OP YoY +1,960%, RS 99, 5D F+QI +₩27.6B
7	017510.KS	Semyung Electric (세명전기)	52.7	Quality Compounder, Cycle Rerating	ROE 15.1%, OP YoY +393%, OPM +24.8pp
8	098120.KQ	Microcontact Sol (마이크로컨텍솔)	52.4	Quality Compounder, Cycle Rerating, PEAD	ROE 23.1%, OP YoY +74%, OPM +3.0pp
9	147830.KQ	Jerong Industrial (제룡산업)	48.8	Quality Compounder, Cycle Rerating	ROE 13.9%, OP YoY +159%, OPM +13.9pp

Top 3 in Focus

#1 — VM (브이엠, 089970.KQ) makes specialized semiconductor and display process equipment. It clears Quality Compounder (ROE 14.8%, OP +387% YoY, revenue +105%), Smart Money Quality (5-day foreign + quality-institution net buy +17.8B KRW), and Smart Money Earnings. A consensus upgrade revision — composite z-score +0.83 — adds a forward-looking re-pricing signal on top of the backward-looking operational improvement. No DART risk filings. The 79.0 meta score leads today’s screened universe by a clear margin. What to check next: whether estimates are being revised higher across multiple brokers and whether the 5-day institutional flow is sustained beyond a single day.

#2 — Gigavis (기가비스, 420770.KQ) makes automated optical inspection (AOI) equipment for PCBs and panels — a direct beneficiary of PCB complexity upgrades driven by AI server demand. Operating income grew 777% YoY on 101% revenue growth. Both foreign and quality-institution buying were positive over five days (+9.7B KRW net). The stock recently moved to a higher exchange tier (DART filing, April 30) and disclosed a new supply contract (April 24) — tangible sequential catalysts. One flag: consensus revision is downward, meaning forward estimates are being cut even as trailing numbers look exceptional. Market may be pricing in near-term margin peak. Watch next quarterly guidance closely.

#3 — TES (테스, 095610.KQ) manufactures semiconductor CVD/ALD deposition equipment used in memory and logic fabs. ROE 14.5%, OP YoY +50%. The 5-day F+QI inflow of +49.4B KRW is the largest in today’s cluster — a meaningful signal of institutional conviction. Short interest is moderate at 6.7%, and foreign ownership at 13.3% leaves room for incremental accumulation. Caution: the Kiwoom surface flags foreign net outflow even as quality-institution flows are positive. That divergence warrants checking whether foreign selling has genuinely paused or is ongoing beneath the institutional bid.

Honorable Mention — RFHIC (218410.KQ)

Ranked #6 by meta score, RFHIC leads all three money-flow sub-screeners: Smart Money Quality #2, Smart Money Earnings #1, PEAD Tier A #1 (score +3.54). Operating income surged +1,960% YoY; RS percentile is 99. A confirmed DART earnings disclosure (잠정실적, April 27) anchors the signal with an official catalyst. The main gap: no Quality Compounder tag, which raises the question of whether ROE and margins are durable or one-cycle. The 5-day F+QI net buy of +27.6B KRW confirms active institutional accumulation. Worth monitoring closely for the quality validation step.

Data sources: KR Meta Screener, Kiwoom Market Surface, KR Quality Compounder, KR Smart Money Quality/Earnings, KR Cycle Rerating, KR PEAD — all as of 2026-05-12T16:24 KST. Market close briefing data from 2026-05-08. This is market analysis for informational purposes only, not investment advice.

Korean Biotech Sector Investment Thesis — Not a 'Buy Good Tech' Market. A 'Buy What the Market Has Underpriced' Market (2026-05-12)

Tue, 12 May 2026 13:30:00 +0900

🔗 Related reading: Pamicell Part 3 — 1Q26 Earnings Confirmation · Rejuran — PharmaResearch PN Skin Booster · Easy Bio Part 1 · Easy Bio Part 2 · ROE-25% Screen — 4 Names That Pass 9 Filters

Korean biotech is not a broad-beta buy regime right now. It’s a selection regime where only data-, BD-, royalty-, or CDMO-validated names work. End-of-May stacks ASCO (May 29-Jun 2), EASL (May 27-30), and EHA abstract releases (May 12 / Jun 2) into one window. The core principle is simple — you’re not buying clinical success probability. You’re buying the gap between market-implied success probability and your internal estimate.

TL;DR

The sector is in selection regime, not broad-buy regime. Conference catalysts, global regulatory change, and Korea-specific positioning reflexivity all interact in the same window.
End-of-May conference data is the first price-setting window. ASCO general abstracts release May 21 17:00 ET (May 22 06:00 KST); congress May 29-Jun 2. EASL May 27-30. EHA general abstracts May 12 22:30 KST, late-breakers June 2 22:30 KST.
Five concurrent theses. (1) Oncology BD — LegoChem Bio, Voronoi, GI Innovation, Lunit; (2) MASH 48-week histology — D&D Pharmatech DD01; (3) SC-conversion royalty — Alteogen; (4) CDMO + biosimilars — Samsung Biologics, Celltrion; (5) FDA regulatory-innovation long option.
The market’s biggest mispricing is “abstract acceptance ≠ data quality." ORR, PFS, OS, DoR, Grade 3/4 AEs, discontinuation rates, and biomarker-defined subgroups determine post-release price action.
The avoid list is clear. Names already running on abstract acceptance alone, names with only preclinical / platform narrative and no human data, names with short cash runway and large CB/BW overhang, names recycling “global big-pharma interest” without partners or upfronts, names entering hot biotech ETF flows.
Core discipline. Don’t believe the tech; believe the data and the cash runway.

1. Three forces operating simultaneously in Korean biotech right now

1.1 Conference data re-pricing

ASCO 2026 runs May 29-Jun 2. General abstracts release May 21 17:00 ET — May 22 06:00 KST. EASL Congress 2026 runs May 27-30 in Barcelona and online.

End-of-May is the window where oncology (ASCO) and liver-fibrosis / MASH (EASL) data hit the same week. The 1-2 weeks between abstract release and main-stage presentation is the first price-setting window for Korean biotech event names.

1.2 Global regulatory environment shift

FDA in March 2026 released a draft guidance on simplifying biosimilar development by reducing unnecessary pharmacokinetic (PK) testing. FDA stated this can cut developer PK-trial costs by up to 50%, or roughly $20M.

In April 2026, FDA announced the Real-Time Clinical Trials initiative — a proof-of-concept and pilot program for real-time submission of trial-data signals to FDA. The broader direction favors biosimilar and clinical-development efficiency.

1.3 Korean market positioning reflexivity

KRX Data Marketplace publishes short-sale, flow, and investor-segment data, and Korean short-selling has been re-enabled since March 31, 2025. Korean biotech event names see short-sale, securities-lending, ETF flow, and retail-overheating amplifying volatility. Data analysis and flow analysis cannot be separated for the asset class to make sense.

2. The five concurrent theses

2.1 In one line

Thesis	Mechanism	Representative names
Data → BD	Trial data re-prices big-pharma licensing optionality	LegoChem Bio, Voronoi, GI Innovation
Disease-specific binary	One readout shifts the entire equity value	D&D Pharmatech, Yuhan MASH read-through
Royalty infrastructure	Repeatable royalty structure, not single-trial success	Alteogen
Cash-flow bio	Real revenue / orders / utilization protect valuation	Samsung Biologics, Celltrion, Samsung Bioepis line
FDA regulatory-innovation option	Long-dated optionality on clinical-development time-value	Names with US clinical infrastructure

2.2 The avoid list

Avoid	Why
Names already running on abstract acceptance alone	Presentation format ≠ data quality
Preclinical / platform narrative names with no human data	BD optionality is low without human data
Short cash runway + large CB/BW overhang	Dilution destroys shareholder value before any good readout
Names recycling “global big-pharma interest” with no contracts	Without counterparty, upfront, or royalty terms, the claim is empty
Names entering hot biotech ETF flows	Look like alpha on the way up; forced sellers on the way down

3. The framework for looking at Korean biotech

[Capital flows]
Big-pharma patent cliff / FDA deregulation / Conference events / ETF + retail flow / Short-sale re-enabled
 ↓
[Value chain]
Basic tech → candidates → preclinical → Phase 1/2/3 → approval → manufacturing → sales / royalty
 ↓
[Choke points]
Human PoC / trial design / safety / manufacturing CMC / IP / cash runway / partnering
 ↓
[P×Q×C]
P = drug price / licensing $ / royalty rate
Q = patients / indication expansion / prescribing conversion / backlog
C = trial cost / cost of goods / dilution cost / time cost

The discipline: separate technology → trials → regulation → commercialization → public-market price. Good science doesn’t equal a good stock. The recurring failure mode in Korean biotech isn’t a failure to understand the science — it’s a failure to price data maturity and funding risk.

4. Thesis 1 — Oncology BD axis

4.1 Core logic

Oncology is where global big-pharma BD demand most directly applies. ADCs, bispecifics, immuno-oncology combinations, EGFR/CNS-penetrant targeted therapies, and AI biomarkers can all re-price licensing expectations through ASCO data.

4.2 Names

Name	Ticker	Catalyst	Read
LegoChem Biosciences	141080	ASCO ADC data + partner read-through	Korea’s flagship ADC platform; data quality + partner optionality drives value
Voronoi	310210	VRN11 brain-mets / CNS data	EGFR-mutant NSCLC + CNS-penetrance verification event
GI Innovation	358570	GI-101A ASCO Rapid Oral Abstract	Phase 1 — format weight matters less than efficacy / safety read
Lunit	328130	ASCO AI-biomarker studies	Companion-diagnostics / AI-biomarker validation rather than NCE licensing

Voronoi disclosed in pipeline materials that VRN11 brain-metastasis clinical data is scheduled at ASCO 2026 May, and noted a DCR of 100% in 11 brain-mets / leptomeningeal-mets patients in the ≥160mg cohort. The number is interesting but the patient count is small; final judgment requires the full ASCO dataset and follow-up duration.

GI Innovation disclosed that GI-101A Phase 1 data was accepted as an ASCO 2026 Rapid Oral Abstract. Phase 1 = safety / dose / early efficacy signal stage — the oral-presentation format itself shouldn’t trigger a valuation re-rating.

Lunit is reportedly presenting 5 abstracts based on the Lunit SCOPE platform at ASCO 2026, with one biliary-tract cancer study selected as a Rapid Oral Presentation. This is AI-biomarker clinical-utility validation, not NCE licensing.

4.3 Market mispricing variable

The market overweights “abstract acceptance” and underweights data quality. What actually matters: ORR, PFS, OS, DoR, Grade 3/4 AEs, discontinuation rates, dose-response, biomarker-defined subgroups.

4.4 Red-team

ADCs — linker / payload / toxicity profile can dominate the target choice.
Phase 1 single-arm data leans heavily on historical controls.
Small patient counts + short follow-up can produce post-print rallies without real BD substance.
Competitor compounds advancing faster into Phase 2/3 can leave Korean assets in the second tier even with good data.

5. Thesis 2 — MASH / obesity axis

5.1 Core logic

MASH (metabolic dysfunction-associated steatohepatitis) is the follow-on alpha after obesity / GLP-1. In MASH, the value driver is not weight loss or liver fat — it’s biopsy-based fibrosis improvement and MASH resolution.

EASL Congress 2026 runs May 27-30. For Korean investors, the highest-impact event is D&D Pharmatech DD01.

5.2 Names

Name	Ticker	Catalyst	Read
D&D Pharmatech	347850	DD01 Phase 2 48-week EASL Late-Breaking Abstract	Largest May binary in Korean biotech
Yuhan	000100	MASH candidate read-through	Re-development possibility of returned asset
Hanmi Pharm	128940	efinopegdutide presentation delay possibility	May conference momentum weakened
Olix Pharmaceuticals	226950	RNAi-based obesity / MASH option	Read-through play more than direct data

D&D Pharmatech disclosed that the DD01 Phase 2 48-week patient dataset is database-locked and will be presented as a Late-Breaking Abstract at EASL Congress 2026. The company indicated key data including biopsy-confirmed liver-fibrosis improvement will be disclosed to accelerate global partnering.

5.3 Market mispricing variable

The market over-reacts to “GLP-1” as a category. The actual investment point for DD01 is not the GLP-1 theme — it’s whether MASH histological improvement materializes.

5.4 Red-team

If 48-week data shows good liver-fat / weight but weak fibrosis improvement, valuation re-rating is constrained.
MASH endpoints are demanding; even a strong Phase 2 still faces large Phase 3 cost and duration.
If pre-print sentiment over-runs, even a good readout can produce sell-the-news.

6. Thesis 3 — Alteogen and SC-conversion royalty infrastructure

6.1 Core logic

The Alteogen-style SC (subcutaneous) conversion platform differs from conventional clinical biotech. Value is not from one drug’s clinical-success probability — it’s from capturing as royalty the economic value of administration convenience, clinic throughput, and patient chair-time savings when IV formulations convert to SC.

6.2 Market mispricing

The market tends to read Alteogen as a single-drug (Keytruda SC) theme. The variable that actually matters is whether the SC-conversion platform is repeatable. As prescription-conversion data accumulates, Alteogen re-rates from “single-event bio” to “royalty DCF.”

6.3 Red-team

Low real-world prescription-conversion damages royalty expectations.
Competitor SC platforms or in-house formulation extensions erode platform moat.
High market recognition already; “good company” and “good entry price” are separate questions.

7. Thesis 4 — CDMO + biosimilars cash-flow axis

7.1 Core logic

CDMO (contract development and manufacturing) and biosimilars are more defensive than event biotech. But they aren’t just defensive — they’re a structural growth axis combining global biopharma supply chain, FDA deregulation, and manufacturing quality trust.

Samsung Biologics’ 1Q26 CDMO revenue was ₩1,257.1bn with OP ₩580.8bn — full utilization on plants 1-4 plus project execution drove the print.

FDA’s March 2026 biosimilar-simplification draft guidance directionally lowers biosimilar development cost and duration. Long-term constructive for Celltrion, Samsung Bioepis, and the rest of the biosimilar field.

7.2 Market mispricing

The market often reads CDMO as a “defensive biotech” trade. The actual industry is quality + delivery + capacity + regulatory trust — customers buy “won’t fail manufacturing,” not “cheapest price.”

7.3 Red-team

Samsung Biologics is high-quality, but labor / quality / production disruption risks discount new-order trust.
Biosimilar deregulation lifts the field but can accelerate ASP compression.
FDA deregulation isn’t an automatic Korean-corporate win — execution matters.

8. Thesis 5 — FDA regulatory innovation as long-dated option

The FDA Real-Time Clinical Trials initiative aims to enable real-time trial-data submission and reduce administrative delay. FDA released the PoC and pilot on April 28, 2026.

If clinical-development dead-time falls, rNPV time-value discount can soften. But this is still early-stage regulation. Actual beneficiaries are most likely names with US clinical infrastructure, data-management capability, and regulatory bandwidth.

Red-team: pilot scope may be narrow; Korean small-to-mid-cap biotech likely won’t be early adopters in the FDA real-time pipeline; the regime change lowers time cost but doesn’t raise approval probability itself.

9. May-June conference calendar

Date (KST)	Event	Field	Impact	Names
2026-05-11 ~ 05-15	ASGCT 2026	Cell + gene + RNA	3/5	Argonomix, CGT / RNA platforms
2026-05-12 22:30	EHA general abstracts	Hematology	3/5	Hematology, CAR-T, ADC
2026-05-22 06:00	ASCO general abstracts	Oncology	5/5	LegoChem Bio, Voronoi, GI Innovation, Lunit
2026-05-27 ~ 05-30	EASL 2026	MASH / liver	5/5	D&D Pharmatech, Yuhan, Hanmi read-through
2026-05-29 ~ 06-02	ASCO 2026	Oncology	5/5	Oncology platforms broadly
2026-06-02 22:30	EHA late-breaking	Hematology	3/5	Hematology / cell therapy
2026-06-05 onward	ADA 2026	Diabetes / obesity	3/5	Hanmi, Ildong, Olix, D&D read-through

ASGCT 2026 — Boston, major cell + gene therapy congress. EHA 2026 main congress runs Jun 11-14 but the investment events (abstract releases) fall within the next month.

9.1 Event-trading principles

Window	Principle
3-4 weeks pre-event	Build event list; check if price has already moved
Pre-abstract	Don’t chase pre-running names; engage only under-positioned ones
Immediately post-abstract	Verify endpoints, patient counts, safety, comparators
Main presentation	Read discussant tone, subgroups, durability, AEs
1 week post-presentation	Confirm sell-side and flow follow-through
1 month post-event	Without licensing / partnering / follow-up papers, event premium compresses

10. Valuation — market-implied success probability

10.1 Clinical-bio rNPV basic form

rNPV =
 Σ [future cash flow × success probability / (1 + discount rate)^t]
 + upfront expectation
 + milestone expectation
 - future R&D
 - SG&A
 - dilution cost
 + net cash

10.2 Market-implied success probability

Implied SuccProb =
 (Current EV - net cash - legacy business value - other pipeline value)
 ÷ unrisked NPV of the asset

The reason this matters is simple. Biotech equity is not “high success probability” — it’s the gap between market-implied success probability and your internal estimate.

Example:

Current EV: ₩1,000bn
Net cash: ₩200bn
Legacy business: ₩100bn
Other pipeline: ₩100bn

Implied core-asset value
= ₩1,000bn - ₩200bn - ₩100bn - ₩100bn
= ₩600bn

Core-asset unrisked NPV: ₩2,000bn

Implied success probability
= ₩600bn / ₩2,000bn
= 30%

Cross-check: ₩2,000bn × 30% = ₩600bn. Rounded to ₩100bn.

If internal estimate is 45%, undervalued. 15%, overvalued. That gap is the alpha source.

11. Research methodology — primary-source priority

Priority	Source	Key fields
1	DART	Cash, CB/BW, top shareholder, licensing contracts, material disclosures
1	ClinicalTrials.gov / CRIS / MFDS	Phase, enrollment status, endpoints, primary completion date
1	Official conference abstract	Patient count, endpoints, safety, presentation format
1	Patent database	Composition / use patents, platform scope
1	KRX	Short-sale, securities-lending, investor-segment flow, turnover
2	Papers / guidelines	Competitor benchmarks, indication-specific endpoints
2	Sell-side reports	Consensus and market narrative
3	News / IR / community	Sentiment / overheating

ClinicalTrials.gov primary-completion-date = date of final data collection for primary outcome (last participant examined / intervened). In practice, registry primary-completion-date drift matters more than company-stated “expected readout dates”.

12. Name priority

Priority	Name	Classification	Core reasoning	Core risk
1	D&D Pharmatech	Event-driven watchlist	DD01 EASL 48-week biopsy data	Sharp drop if fibrosis weak
2	LegoChem Bio	Conditional watchlist	ADC platform, ASCO + partner optionality	Pre-priced expectations, toxicity / competition
3	Alteogen	Core watchlist	SC-conversion royalty platform	Real-world conversion / valuation
4	Voronoi	Wait	VRN11 CNS data	Patient count / follow-up limits
5	Samsung Biologics	Wait / quality on weakness	1Q26 print + CDMO utilization	Valuation / labor / quality risk
6	GI Innovation	Speculative watchlist	ASCO Rapid Oral Abstract	Phase 1 limits
7	Lunit	Watchlist	AI-biomarker ASCO validation	Lag vs. NCE licensing revenue
8	Yuhan / Hanmi / Olix	Read-through watchlist	MASH / obesity theme	No direct data or delay

13. Invalidation conditions

When each thesis breaks, the signal is specific.

Thesis	Invalidation
ADC / oncology BD	ASCO data shows no differentiation vs. competitors, toxicity events
MASH	Insufficient fibrosis improvement or MASH resolution
SC-conversion platform	Weak real-world conversion, unclear royalty economics
CDMO	Utilization decline, new-order slowdown, quality / delivery issues
Biosimilars	Regulatory deregulation outweighed by price-competition intensification
FDA real-time clinical	Pilot scope narrow, no Korean-company application
Undervaluation logic	Market-implied probability rises above internal estimate

14. Bottom line

Korean biotech right now is not a “pick good names” market — it’s a “pick names whose good data the market has under-priced” market. ASCO and EASL are powerful catalysts, but buying on abstract acceptance alone is low-grade event trading.

D&D Pharmatech is the cleanest binary; LegoChem Bio is the textbook BD-optionality name; Alteogen is the highest-quality royalty platform. Conversely, story-only names in obesity / ADC / cell-gene therapy without actual data are the explicit avoid list.

The single discipline that matters in this sector: don’t believe the technology; believe the data and the cash runway. Late-May congresses are the first verification gate; the June ADA is the follow-on. The licensing / partnering / follow-up papers in the month after the events set the final price.

FAQ

Q: Why is D&D Pharmatech #1? A: At EASL Congress 2026 (May 27-30), DD01 Phase 2 48-week data will be presented as a Late-Breaking Abstract. The MASH endpoint that matters most — biopsy-confirmed liver-fibrosis improvement — is disclosed. Good data triggers re-rating against global MASH comparables; weak data triggers a sharp drop. The cleanest binary structure of the cohort.

Q: Why isn’t ASCO abstract acceptance enough to buy? A: Acceptance is not a guarantee of data quality. What actually matters: ORR, PFS, OS, DoR, Grade 3/4 AEs, discontinuation, dose-response. Buying purely on acceptance leads to sell-the-news when post-release data turns out average.

Q: How is Alteogen different from clinical biotech? A: Alteogen’s value is not in one drug’s clinical success — it’s in capturing as royalty the economic value of administration convenience and clinic-throughput improvements when IV formulations convert to SC. As real-world prescription-conversion data accumulates, the valuation can re-rate from “single-event bio” to “royalty DCF.” Binary-event risk is structurally lower.

Q: Why isn’t Samsung Biologics #1? A: 1Q26 print is strong (revenue ₩1,257.1bn, OP ₩580.8bn). But unlike event biotech, short-term conference catalysts are limited. Valuation, plus labor / quality risk, has to be priced. CDMO is a “on weakness” candidate, not a chase.

Q: Is FDA deregulation an immediate tailwind? A: Long-dated option, limited short-term effect. PK-trial cost cuts (up to 50%, ~$20M) and the real-time clinical-trials initiative are sector-positive but Korean small-to-mid-cap biotech is unlikely to be an early beneficiary. Names with US clinical infrastructure benefit first.

Q: How do you calculate market-implied success probability? A: (Current EV − net cash − legacy business − other pipeline) ÷ core-asset unrisked NPV. If that ratio is below your internal estimate, undervalued. The Section 10 worked example shows EV ₩1tn + core-asset NPV ₩2tn → implied 30%.

Q: Which sources to read first? A: DART (disclosures), ClinicalTrials.gov (phase, endpoints, completion dates), official conference abstracts (patient count, safety, format), patent databases, KRX (short / lending / segment flow). Sell-side reports for consensus; news / IR for sentiment temperature. Registry primary-completion-date drift is a more reliable signal than company-stated “expected readout” timing.

This article is for research and informational purposes only and does not constitute investment advice. Conference dates and abstract-release timing follow ASCO / EASL / EHA / ASGCT / ADA official sources. FDA biosimilar draft guidance and real-time clinical-trials initiative reference FDA press materials (March / April 2026). Samsung Biologics 1Q26 figures are from official disclosure. D&D Pharmatech DD01 / Voronoi VRN11 / GI Innovation GI-101A / Lunit SCOPE references are from each company’s official releases and conference-acceptance announcements. The market-implied-probability example uses a generic framework; actual name-level values require live price, net cash, and pipeline-specific assumptions. Event-impact scores (3/5, 5/5) are analytical estimates and can diverge from actual outcomes. Analysis can be wrong. Data cut: May 12, 2026 KST.

Dashboard link	Connected blog page	How to use together
HBM Hub	HBM / KOSPI Investment Hub	Memory signal in dashboard → thesis depth in hub
Daily Market	Korea Daily Market Hub	Daily market-flow analysis collection
Telegram	@korea_invest_insights	New-post notifications immediately
Substack	Subscribe	Curated analysis by email