Hanmi Semiconductor on Korea Invest Insights

Korea Theme ETF Rebalance Flow: Semicap Redistribution Buy Pressure, Megacap Cap-Trim Risk

Fri, 12 Jun 2026 17:40:00 +0900

Context
This follows Have Foreign Investors Returned?, Real Money Flow Framework, Korea Has Liquidity, But Breadth Has Broken, Samsung C&T as a Samsung Electronics proxy and the Korea Semiconductor / HBM Hub.

TL;DR

Thesis OS added KR Theme ETF Rebalance Flow v1, a monitor for potential mechanical flow from Korea theme ETF periodic rebalances and constituent cap adjustments.
The first run scanned 31 ETFs, found 30 valid ETFs, processed 291 constituent rows, and produced 69 candidates: 60 redistribution-buy and 9 cap-trim candidates.
The strongest signal was not nuclear/SMR. It was semiconductor equipment and materials redistribution.
Leeno Industrial, EO Technics, Soulbrain, DB HiTek, Hanmi Semiconductor, ISC and Wonik IPS screened as the strongest redistribution-buy candidates.
SK Hynix, Samsung Electronics, Samsung Electro-Mechanics, LS ELECTRIC and Doosan Enerbility screened as cap-trim pressure candidates.

Core Point

This is not an automatic buy list. It is a way to identify where theme ETFs may cut oversized leaders and redistribute weight into second-line constituents. As of June 12, the clearest pocket was Korean semiconductor equipment and materials.

1. What The Monitor Measures

The monitor estimates a cap-redistribution proxy:

ETF-level flow = ETF NAV × target weight delta
Ticker-level flow = sum of ETF-level flows
Flow intensity = estimated flow / 20-day average trading value

It uses Naver mobile ETF surface data, including constituentList and totalNav, plus local prices_daily ADV20. If a constituent is above an assumed cap, the model cuts it back to cap and redistributes the excess to under-cap constituents in proportion to current weights.

This is not official PCF data. Before treating it as a trade signal, investors need issuer PDFs/PCF, index methodology, actual cap rules, and closing-auction evidence.

Item	Value
Date	2026-06-12
ETFs scanned	31
Valid ETFs	30
Constituent rows	291
Total candidates	69
Redistribution-buy candidates	60
Cap-trim candidates	9
Confidence	medium-low

2. Top Redistribution-Buy Candidates

Rank	Name	Signal	Est. flow	Flow/ADV20	Day return	Main ETFs
1	Leeno Industrial	Redistribution buy	KRW 270.4bn	+2.68x	+4.7%	TIGER Semiconductor TOP10, KODEX AI Semiconductor TOP2 Plus
2	EO Technics	Redistribution buy	KRW 197.8bn	+2.49x	+21.4%	TIGER Semiconductor TOP10
3	Soulbrain	Redistribution buy	KRW 69.4bn	+2.44x	+24.4%	TIGER Semiconductor TOP10
4	DB HiTek	Redistribution buy	KRW 264.3bn	+2.02x	+12.3%	TIGER Semiconductor TOP10, KODEX AI Semiconductor TOP2 Plus
5	Hanmi Semiconductor	Redistribution buy	KRW 721.0bn	+1.81x	+24.1%	TIGER Semiconductor TOP10, KODEX AI Semiconductor TOP2 Plus
6	ISC	Redistribution buy	KRW 92.0bn	+1.76x	+20.7%	TIGER Semiconductor TOP10, SOL AI Semiconductor TOP2 Plus
7	Wonik IPS	Redistribution buy	KRW 211.1bn	+1.63x	+30.0%	TIGER Semiconductor TOP10
8	HPSP	Redistribution buy	KRW 143.7bn	+0.66x	+30.0%	TIGER Semiconductor TOP10, KODEX AI Semiconductor TOP2 Plus

The important field is not only absolute flow. It is flow intensity versus liquidity. Leeno, EO Technics, Soulbrain, ISC and Wonik IPS screen as names where estimated ETF redistribution may be large relative to normal trading value.

3. Cap-Trim Watch

Rank	Name	Signal	Est. flow	Flow/ADV20	Day return
1	SK Hynix	Cap trim	-KRW 1.76tn	-0.15x	+2.3%
2	Samsung Electronics	Cap trim	-KRW 49.4bn	-0.00x	+7.9%
3	Samsung Electro-Mechanics	Cap trim	-KRW 514.9bn	-0.21x	-5.0%
4	LS ELECTRIC	Cap trim	-KRW 47.0bn	-0.15x	-3.0%
5	HD Hyundai Heavy Industries	Cap trim	-KRW 21.8bn	-0.06x	+0.6%
6	Hanwha Ocean	Cap trim	-KRW 7.8bn	-0.04x	+7.8%
7	Doosan Enerbility	Cap trim	-KRW 26.5bn	-0.07x	+5.1%

Cap-trim does not mean short. Large names can absorb technical ETF selling if foreign or institutional demand is strong. The more useful interpretation is: do not chase cap-trim names unless price action proves real demand is absorbing the mechanical supply.

4. Nuclear / SMR Was Watch, Not The Main Signal

Nuclear names appeared in Priority Watch, but the signal strength was lower than semicap redistribution.

Name	Signal	Est. flow	Flow/ADV20	Day return
KEPCO Engineering	Redistribution buy	KRW 2.85bn	+0.13x	+30.0%
Doosan Enerbility	Cap trim	-KRW 26.5bn	-0.07x	+5.1%
Hyundai Engineering & Construction	Redistribution buy	KRW 8.75bn	+0.06x	+28.4%
BHI	Redistribution buy	KRW 1.57bn	+0.06x	+30.0%
Woori Technology	Redistribution buy	KRW 2.91bn	+0.04x	+30.0%

5. Trade Discipline

Treat this as an event screen, not a final signal.

Step	Required confirmation
1	Check issuer ETF PDF/PCF and index methodology
2	Confirm closing-auction and end-of-day volume
3	Check T+1/T+3 relative strength
4	Confirm program, foreign or institutional flow alignment
5	Upgrade only the strongest names to watchlist or small event trade

Final View

The first KR Theme ETF Rebalance Flow run says the strongest June 12 passive-flow setup was semiconductor second-line redistribution, not broad Korea beta and not nuclear/SMR as the primary signal.

The signal is useful because it quantifies where ETF mechanics may redirect weight after megacap concentration. It remains a medium-low confidence proxy until official PCF/PDF and actual close-auction flow confirm it.

Evidence Ledger

Item	Source	Detail
Monitor	Thesis OS / Alpha	KR Theme ETF Rebalance Flow v1
Local report	Vault `2026-06-12_theme_etf_rebalance_flow.md`	31 ETFs scanned, 30 valid, 69 candidates
External Signal Judge input	Vault candidate file	Candidate score, flow, intensity and confidence
Data	Naver ETF surface + local DB	`constituentList`, `totalNav`, `prices_daily` ADV20

Caveat

This is not investment advice and not a confirmed PCF-based trade instruction. It is a cap-redistribution proxy that requires issuer confirmation and post-event validation.

The Market Missed Jensen Huang’s Bigger HBM4 Comment: All Three Vendors Are Qualified

Fri, 05 Jun 2026 16:30:00 +0900

Follow-up context: this note follows Sam-Ha-Ma parity follow-up, Samsung and SK hynix are cheap versus Micron again, Samsung HBM4E 12-high sample, and Hanmi Semiconductor TC-bonder TAM expansion. Related hubs: AI HBM hub, Korea semiconductor value-chain hub, and Korea Daily Market Hub.

TL;DR

Jensen Huang’s reported comment means Samsung Electronics, SK hynix and Micron have all passed qualification and entered production competition for NVIDIA Vera Rubin HBM4. Korean media quoted Huang as saying, “All three vendors have been qualified,” and that all three are in production and racing to support NVIDIA. (Daum/JoongAng)

The largest incremental beneficiary is Samsung Electronics because the market’s main doubt was whether Samsung could enter the NVIDIA HBM socket. For SK hynix, the read-through is mixed: monopoly premium dilution, but stronger total HBM demand.

But qualified does not mean equal allocation, equal margin or guaranteed high-volume supply. The investment question has moved from qualification to HBM4 allocation, yield, ASP and margin.

Core Takeaway

The HBM4 supply chain is moving from a single-leader narrative toward a three-vendor qualified competition for Vera Rubin. Samsung gets the largest discount-compression event. SK hynix loses some monopoly premium, but remains the leadership name in a larger TAM.

The Comment

According to Korean media coverage on June 5, 2026, Huang told reporters after arriving at Seoul Gimpo Business Aviation Center that all three HBM vendors were qualified and in production. The three vendors are Samsung Electronics, SK hynix and Micron. (Daum/JoongAng)

This matters because the wording is different from the 2024 discussion, when Samsung’s HBM qualification was still in progress and Huang said more engineering work was needed. (Yonhap)

Phrase	Investor interpretation	What not to over-read
All three vendors have been qualified	NVIDIA qualification has been passed	Not equal allocation
All three vendors are in production	HBM4 ramp has started	Not proof of mass yield or margin
Racing to support Vera Rubin	Three-vendor competition for Vera Rubin supply	Not the end of SK hynix leadership

Why Vera Rubin Matters

NVIDIA says the Vera Rubin platform has entered full production ramp across a broad supply chain. (NVIDIA Newsroom) The official Vera Rubin NVL72 spec shows why HBM4 matters: 72 Rubin GPUs, 36 Vera CPUs, 20.7TB of HBM4 and 1,580TB/s of GPU-memory bandwidth. Each Rubin GPU carries 288GB of HBM4 and 22TB/s of bandwidth. (NVIDIA Vera Rubin NVL72)

The arithmetic is simple:

288GB HBM4 per Rubin GPU
× 72 Rubin GPUs per NVL72 rack
= 20,736GB ≈ 20.7TB HBM4 per rack

This is not a small product refresh. Vera Rubin is a rack-scale AI factory platform that can pull all three HBM vendors into production competition.

Samsung Electronics: Biggest Incremental Positive

Samsung is the largest surprise beneficiary. The market doubt was whether Samsung HBM could pass NVIDIA qualification. That doubt is now lower.

Samsung has also officially begun shipping 12-layer HBM4E samples. The company says the part reaches up to 16Gbps, 48GB capacity and up to 3.6TB/s per stack. (Samsung Semiconductor)

The key checks now are:

Checkpoint	Why it matters
HBM4 shipment start	Turns narrative into revenue
NVIDIA allocation	Separates qualification from real share
DS gross margin / OPM	Shows whether HBM mix lifts earnings
Base-die and packaging yield	Determines the true profitability of HBM4

Samsung’s thesis is now HBM discount compression. It is not yet equal allocation or equal profitability versus SK hynix.

SK hynix: From Monopoly Premium To Leadership Premium

The read-through for SK hynix is mixed.

The negative is obvious: a three-vendor qualified supply chain lowers the “NVIDIA monopoly” premium. But Vera Rubin’s HBM intensity expands the total market.

Reuters reported that SK Group Chairman Chey Tae-won said SK hynix aims to double wafer capacity over the next five years and repeated his view that memory bottlenecks could persist through 2030. (MarketScreener/Reuters)

The thesis moves from:

NVIDIA HBM monopoly premium

to:

leader allocation + yield + customer trust + next-generation HBM roadmap

So “Samsung qualified = short SK hynix” is too simple. The cleaner interpretation is that SK hynix shifts from monopoly premium to leadership premium.

Micron: Benchmark For Korea Memory Discount

Micron is also confirmed as part of the Vera Rubin HBM4 race. For Korean investors, that is important because Micron remains the U.S.-listed AI-memory proxy.

As discussed in the Sam-Ha-Ma parity follow-up, Micron’s premium is not simply a technology premium. It is a U.S. listing premium, dollar-asset premium, AI-memory scarcity premium and HBM4/SOCAMM2/data-center SSD narrative.

That keeps Micron relevant. It also strengthens the case for Samsung and SK hynix discount compression if their EPS estimates hold.

Hanmi Semiconductor Read-Through

Hanmi Semiconductor is a positive second-order read-through. If all three memory vendors are racing to support Vera Rubin, HBM4 capacity expansion can move beyond a single SK hynix-centered narrative.

But this is not an order confirmation. Hanmi’s direct upside depends on whether TC-bonder orders expand from SK hynix to Samsung, Micron, OSATs or foundry-related advanced packaging customers.

The revenue logic is:

Hanmi HBM equipment revenue
= memory-vendor HBM capex
× TC bonders per line
× ASP
× replacement and upgrade demand

The first variable has improved. The customer-order variable is still unconfirmed.

Customer	Current read	Hanmi implication	Confidence
SK hynix	Direct order history	Core direct customer	High
Micron	Relationship optionality	Overseas expansion option	Medium
Samsung	HBM ramp positive, direct order unconfirmed	Internal SEMES / alternative vendor risk	Low-Medium
OSAT / foundry	Advanced packaging TAM expansion	2.5D and AI ASIC equipment option	Medium

What Not To Over-Read

Still unconfirmed:

NVIDIA HBM4 allocation by vendor
Samsung HBM4 mass-production yield
ASP and gross margin by product
12-high / 16-high HBM4 or HBM4E split
Supply scope outside Vera Rubin
Hanmi TC-bonder orders from Samsung or Micron

The accurate wording is:

NVIDIA’s Vera Rubin HBM4 supply chain appears to have moved into a three-vendor qualified structure. The game now shifts from qualification to allocation, yield, ASP and margin.

Bottom Line

This is positive for all three memory vendors, but the largest incremental positive is Samsung Electronics. SK hynix remains the quality leader, but the monopoly-premium language should be toned down. Micron remains the U.S. benchmark. Hanmi Semiconductor is a high-beta equipment read-through, but needs actual TC-bonder orders to turn narrative into earnings.

Practical priority:

1. Samsung Electronics
- largest HBM discount-compression event

2. SK hynix
- still the quality leader, but not a monopoly story

3. HBM equipment / inspection / packaging chain
- benefits if three-vendor production competition extends capex

4. Micron
- benchmark for the Korean memory discount

Sources

Broadcom Reconfirmed a $100B+ AI Semiconductor Run-Rate for 2027: How Korea Should Read the Value Chain

Fri, 05 Jun 2026 09:30:00 +0900

📚 Connected context This is a follow-up to the Marvell/Broadcom Korea bottleneck preview, the Jensen Huang-Marvell read-through, the laser-source follow-up, and the Samsung-Hynix-Micron parity note. Relevant hubs are the Korea semiconductor value chain hub, AI HBM hub, and AI PCB/substrate hub.

TL;DR

Broadcom’s Q2 FY2026 result is not negative for Korean semiconductors. The better read is that AI custom silicon demand continues to flow into HBM, package substrates, networking, and power-integrity components.

The stock fell because expectations were even higher, not because the underlying AI demand collapsed.

Core View

Broadcom reconfirmed the FY2027 AI semiconductor revenue frame of more than $100B. The market wanted an upward revision, not just confirmation. Korean investors should treat the selloff as an expectation reset and re-check the durability of HBM, FC-BGA, AI networking, silicon capacitors and test/packaging bottlenecks.

The Korean translation is:

Priority	Layer	Korean names to watch	View
1	HBM / server DRAM / eSSD	Samsung Electronics, SK hynix	Core layer. Custom ASIC broadens the HBM customer base beyond NVIDIA GPUs.
2	FC-BGA / Si-Cap / MLCC	Samsung Electro-Mechanics	Most direct non-memory read-through, but valuation discipline matters.
3	AI networking PCB / high-speed boards	Isu Petasys, Korea Circuit and peers	Second-order beta to AI Ethernet and switch ASIC demand.
4	TC bonders / sockets / inspection	Hanmi Semiconductor, ISC, Leeno and peers	Follow-through depends on real orders.

One-line conclusion: Broadcom’s fall is not a bearish signal for Korean AI infrastructure. It is a reset in over-heated expectations. The first name to revisit is Samsung Electronics, followed by SK hynix and Samsung Electro-Mechanics.

What Broadcom Actually Confirmed

[Fact] Broadcom reported Q2 FY2026 revenue of $22.187B, up 48% year over year. GAAP net income was $9.310B, non-GAAP net income was $12.074B, and free cash flow was $10.262B. (Broadcom Q2 FY2026)

[Fact] AI semiconductor revenue was the key. Q2 AI semiconductor revenue was $10.8B, up 143% year over year, and Q3 AI semiconductor revenue is guided to $16.0B, implying more than 200% year-over-year growth. Q3 total revenue guidance is $29.4B. (Broadcom Q2 FY2026)

Q2 AI semiconductor revenue mix = $10.8B / $22.187B = about 48.7%
Q3 AI semiconductor guidance mix = $16.0B / $29.4B = about 54.4%

Broadcom is no longer just a diversified semiconductor/software company with VMware attached. AI custom silicon is approaching half of revenue.

[Fact] Earnings-call summaries indicate Broadcom still frames FY2026 AI semiconductor revenue at about $56B and FY2027 at more than $100B. It also cited Q2 AI semiconductor bookings of more than $30B. (MarketBeat)

The important word is “still.” Broadcom did not introduce a new, higher 2027 frame. It maintained the existing one. That is where the market was disappointed.

Why the Stock Fell

The absolute numbers were strong. The problem was expectation.

First, investors expected more than the Q3 AI semiconductor guide of $16B. Some reports put market expectations around $16.3B to $17.2B. (Reuters)

Second, the FY2027 $100B+ frame was a confirmation, not an upgrade.

Third, margin mix became a debate. The AI custom silicon mix is rising, and investors questioned whether that could pressure gross margins even as operating leverage remains strong.

So the message is not “AI semiconductor demand is rolling over.” It is: AI ASIC leaders are now priced for very large upside, and anything short of another step-up can trigger a reset.

Why It Matters for Korea

Broadcom’s AI revenue is not only custom XPU. It also includes AI networking. More custom AI chips mean more HBM, high-performance packages, power-integrity parts, high-speed boards, test sockets and optical/networking hardware.

[Inference] The custom AI compute stack is widening from NVIDIA GPUs into Google TPU, Broadcom XPU, OpenAI accelerators, Meta MTIA, Anthropic compute and other hyperscaler designs. Korea participates mostly through the physical bottlenecks: memory, substrates, passive components, test and high-speed interconnect.

Korean Stock Read-Through

Samsung Electronics is the balanced read-through. It has HBM catch-up, server DRAM, eSSD/NAND and a long-term foundry/base-die option. The key check is HBM4E/HBM4 customer approval and whether DS earnings estimates keep rising.

SK hynix remains the cleanest HBM exposure. But the investment question is no longer simply “HBM is good.” It is whether the stock is attractive versus Micron and Samsung Electronics after the recent move. That relative-value angle is covered in the Samsung-Hynix-Micron parity note.

Samsung Electro-Mechanics is the strongest non-memory translation. FC-BGA, MLCC and silicon capacitors sit directly under AI ASIC and AI networking growth. The thesis is right, but the stock already reflects a large part of the re-rating, so repeated orders and margin proof matter.

Hanmi Semiconductor, ISC, Leeno and testing/socket names benefit only when the package/test capex turns into orders. Here the right posture is watchlist, not automatic buying.

Execution

The event supports Korean AI infrastructure, but it does not say “buy everything.”

Layer	What to verify
Samsung Electronics	HBM4E/HBM4 approval, DS earnings revision, foreign flow stabilization
SK hynix	Relative PER vs Micron, HBM price/mix, foreign selling pressure
Samsung Electro-Mechanics	Si-Cap repeat orders, package-solution margin, AI customer diversification
High-speed PCB	AI networking orders, ASP, capacity allocation
Packaging/test	Real orders, not only TAM expansion narratives

Final View

Broadcom reconfirmed a large AI ASIC market. The market sold the stock because it wanted an even larger upgrade. For Korea, that distinction matters. The demand signal remains positive, but the investable alpha should be sorted by physical bottleneck and valuation.

The cleanest path is Samsung Electronics as the core, SK hynix on relative-value pullbacks, and Samsung Electro-Mechanics only with valuation discipline and repeated order evidence.

Sources

Hanmi Semiconductor IR Notice: Not an Order, but a TC-Bonder TAM Expansion Signal

Tue, 02 Jun 2026 09:20:00 +0900

Context This is a follow-up to the June 1 Hanmi Semiconductor TC-bonder flow check. The prior note asked whether one day of foreign buying was enough to confirm a flow turn. This note asks what Hanmi’s June 1 IR notice changes about its addressable market. Related hubs: Korea Semiconductor Value Chain Hub and AI HBM Hub.

TL;DR

The June 1 filing is not an order announcement. It is an IR meeting notice. There is no confirmed revenue number. Instead, the company is re-presenting its 2H26 business-expansion narrative to domestic institutional investors under the title “2H26 outlook and overseas new-market entry.” (DART)

The key is TC-bonder TAM expansion. Hanmi is trying to move the market’s frame from HBM memory TC bonders into 2.5D package / AI system semiconductors / OSAT / HBF / U.S.-led AI semiconductor alliance. The direction is constructive, but the disclosure still lacks customer names, order size, production timing and margin.

The investment stance is Watchlist / enter only after flow confirmation. The stock is down about 28% from its 60-day high, so the overheating has cooled. But recent 5-10 day flows still show retail absorbing supply while institutions, program trading and short selling remain pressure points.

CEO insider buying is a constructive secondary signal. Based on the DART filing history provided by the user, CEO Kwak Dong Shin bought 692,398 shares in 29 open-market purchases from July 26, 2023 to April 27, 2026, spending about KRW 56.53bn at a weighted-average price of KRW 81,643. But the most relevant number now is the latest purchase: 9,576 shares at KRW 315,407. At the June 1 close of KRW 293,000, that latest lot is still underwater. The practical question is whether the KRW 290,000-315,000 zone becomes an insider-supported defense line. (DART)

1. What the Filing Says

[Fact] Hanmi Semiconductor filed an IR meeting notice on June 1, 2026 at 09:09:15 KST. The meeting date is June 4, 2026; location is Seoul; target attendees are major domestic institutional investors; format is group meetings; sponsor is LS Securities. DART receipt number is 20260601800176. (DART)

The stated purpose is “2H26 outlook and overseas new-market entry.” IRGO also lists the same latest IR schedule. (IRGO)

The filing summary is:

Item	Filing Content	Investment Read
1	Entry into AI system semiconductor, foundry and OSAT markets	Reduce dependence on memory HBM and expand into logic / ASIC packaging equipment
2	Launch of 2.5D Package TC bonders, 40 and 120	Attempt to attach to the 2.5D packaging ecosystem
3	New-concept TC bonder for HBF, High Bandwidth Flash	Option in high-bandwidth memory / storage packaging beyond HBM
4	Participation in a U.S.-led AI semiconductor alliance	Respond to AI package equipment demand from big tech, memory and foundry ecosystems

The important boundary is:

This is not an order. It is a scheduled investor meeting to explain where Hanmi wants to expand in 2H26.

2. What It Means: From HBM Equipment to an AI Packaging Equipment Platform

The old Hanmi thesis was relatively simple:

HBM growth
→ more demand for HBM TC bonders
→ high margin
→ premium for scarce equipment leadership

This filing broadens the thesis:

HBM TC bonder
→ 2.5D package TC bonder
→ AI ASIC / foundry / OSAT packaging equipment
→ equipment vendor inside the U.S.-led AI semiconductor supply chain

In other words, Hanmi is arguing that the market should not value it only as an HBM memory equipment stock. The key question is whether the TC bonder can move from an HBM-specific tool into a broader advanced-packaging process tool.

That matters because the HBM cycle is powerful but concentrated. If the battlefield expands into 2.5D packaging, AI system semiconductors, OSAT and HBF, Hanmi could gain exposure to a wider AI packaging CAPEX cycle rather than one memory product cycle.

What is still missing:

Missing Data	Why It Matters
Customer names	Confirms design-in quality and repeatability
Order value	Converts narrative into revenue size
Production timing	Determines whether it affects 2026 or 2027 numbers
Margin	Tests whether the new market is as profitable as the existing tool franchise

So this is positive strategic direction, not confirmed earnings.

3. Why 2.5D Package and HBF Matter

3.1 2.5D package: the door into AI ASIC packaging

2.5D packaging connects logic dies, memory, bridges, interposers and substrates inside one package. As AI accelerators get larger, the bottleneck is not just one chip. It is internal package connectivity, thermals, power and bonding precision.

When Hanmi talks about 2.5D Package TC bonders, it is saying that it wants to move beyond HBM stacking and into the process flow for AI system semiconductor packaging.

In investor language:

The company is trying to expand from HBM volume beta into AI advanced-packaging CAPEX beta.

3.2 OSAT: a broader customer language

OSATs are outsourced semiconductor assembly and test companies. Hanmi’s OSAT-market language means it wants to expand the equipment customer base toward the back-end ecosystem that actually assembles advanced packages.

The upside is a broader customer base. The trade-off is higher qualification difficulty. OSATs care about package-specific requirements, equipment uptime, yield, delivery and service response. For Hanmi’s new tools to become real revenue, investors need to see line adoption and repeat orders.

3.3 HBF: an option in high-bandwidth memory and storage packaging

HBF, High Bandwidth Flash, is not yet as standardized as HBM. But as AI infrastructure scales, the boundary between memory and storage becomes more complex, and high-bandwidth, low-latency storage tiers matter more.

Hanmi’s HBF TC-bonder language suggests an attempt to build equipment options for high-bandwidth memory / storage packaging beyond HBM.

This part remains closer to [Speculation]. Market size, standardization, customer adoption and production timing all need proof. But as a long-term option, it is worth tracking.

4. Price and Flow: Overheating Has Cooled, but the Turn Is Not Confirmed

Data basis: user-provided Research OS local DB and Kiwoom REST, June 1, 2026 close.

Item	Value
Close	KRW 293,000
1D	+3.9%
5D	-8.6%
10D	-20.6%
20D	-18.0%
60D	+13.1%
YTD	+130.0%
60-day high	KRW 409,500
Drawdown from high	-28.4%
RSI 14	42.7
20-day moving average	KRW 347,050
50-day moving average	KRW 313,200

The overheating has cooled meaningfully. A -28% drawdown from the 60-day high lowers chase risk. But the trend has not recovered. The stock is below both the 20-day and 50-day moving averages.

Flows

Unit: KRW 100mn.

Period	Foreigners	Institutions	Retail	Read
1D	+261.9	-502.4	+228.5	Foreign rebound, institutions still selling
5D	-775.0	-5,234.3	+5,940.3	Retail absorbed supply
10D	-3,332.9	-5,692.8	+8,916.2	Flow quality still weak
20D	-3,954.0	-1,276.9	+5,121.2	Foreign and institutional exit, retail buying

The key source of recent institutional selling is financial-investment accounts at about -KRW 476.4bn over five sessions. Investment trusts, private funds and pension funds are not yet a clean positive. The June 1 foreign buying is constructive, but it is too early to call it a flow turn.

Short Selling and Program Trading

Item	Recent State
Jun 1 short-selling ratio	25.8%
May 27 short-selling ratio	41.8%
May 28 short-selling ratio	37.1%
Jun 1 program trading	-KRW 4.48bn
May 29 stock-loan balance	About 16.727mn shares, about KRW 4.7tn value

The short-selling ratio has fallen from the peak, but it remains high. The stock-loan balance is also large. A short-covering setup is possible, but the current question is whether pressure is easing, not whether it has disappeared.

5. Insider Buying History: Strong Secondary Signal, but the Latest Purchase Price Matters

[Fact] Based on the user-provided DART filing aggregation for Hanmi Semiconductor corp code 00161383, submitter Kwak Dong Shin, and the filing type report on ownership of specific securities by executives / major shareholders from Jan. 1, 2020 to Jun. 2, 2026, Kwak’s open-market purchases total 29 events from Jul. 26, 2023 to Apr. 27, 2026. The aggregate purchase size is 692,398 shares, about KRW 56.53bn, with a weighted-average purchase price of KRW 81,643.

The latest DART report confirms that Kwak bought 9,576 common shares in the open market on Apr. 27, 2026 at KRW 315,407 per share. The report’s base date is Apr. 27, 2026, and the actual settlement / change date is Apr. 29, 2026. The prior major purchase of 47,715 shares at KRW 129,910 on Jan. 5, 2026 is confirmed in the Dec. 30, 2025 DART report. (DART) (DART)

Annual purchase summary:

Year	Open-Market Purchases	Shares Bought	Amount Invested	Weighted Avg. Price	Versus Jun. 1 Close
2023	16	388,800	KRW 20.98bn	KRW 53,965	+442.9%
2024	8	139,254	KRW 16.32bn	KRW 117,181	+150.0%
2025	3	107,053	KRW 10.01bn	KRW 93,515	+213.3%
2026	2	57,291	KRW 9.22bn	KRW 160,915	+82.1%
Total	29	692,398	KRW 56.53bn	KRW 81,643	+258.9%

Key purchase prices:

Change Date	Shares	Avg. Purchase Price	Amount	Same-Day Close	Versus Current Price
2026-04-27	9,576	KRW 315,407	KRW 3.02bn	KRW 373,500	-7.1%
2026-01-05	47,715	KRW 129,910	KRW 6.20bn	KRW 167,300	+125.5%
2025-11-28	41,103	KRW 121,757	KRW 5.00bn	KRW 123,600	+140.6%
2025-04-17	43,692	KRW 68,821	KRW 3.01bn	KRW 68,000	+325.7%
2025-03-14	22,258	KRW 89,834	KRW 2.00bn	KRW 87,800	+226.2%
2024-06-05	20,958	KRW 143,124	KRW 3.00bn	KRW 155,800	+104.7%
2024-02-27	28,000	KRW 73,589	KRW 2.06bn	KRW 82,700	+298.2%
2023-11-15	50,000	KRW 61,436	KRW 3.07bn	KRW 62,300	+376.9%
2023-07-26	24,000	KRW 42,452	KRW 1.02bn	KRW 43,800	+590.2%

[Inference] This is not a symbolic KRW 100-200mn purchase pattern. Since 2023, Kwak has repeatedly bought in KRW 2-6bn increments. As an insider signal, that is meaningful.

But the key for today’s trade is not the full-period weighted-average price of KRW 81,643. That lot is already deeply profitable. The key is the latest high-price purchase at KRW 315,407. At the Jun. 1 close of KRW 293,000, that latest purchase is about 7.1% underwater.

The read:

Interpretation	Meaning
Positive	The current price is below the latest CEO purchase price, creating a gap between management conviction and market pricing
Negative	If the stock cannot recover the KRW 315,000 area, insider buying alone is not enough to offset institutional / foreign selling and short pressure
Practical trigger	Defend KRW 290,000, regain KRW 315,000, rising turnover, and easing foreign / institutional selling

If the stock breaks into the KRW 270,000-280,000 area, the latest insider purchase signal should be treated as a short-term failure. In that case, the IR notice and insider buying remain long-term support, but near-term flow pressure takes priority.

6. Investment Read

Bull Case

First, the filing points to AI package equipment TAM expansion, not just HBM.

Second, if the 2.5D package TC bonder converts into real customer or OSAT adoption, Hanmi can be re-rated from an HBM equipment name into an AI advanced-packaging equipment platform.

Third, the stock has corrected about 28% from the 60-day high.

Fourth, the June 1 foreign net buying is at least the first rebound signal after the selloff.

Bear Case

First, this is not an order.

Second, the disclosure does not include customers, volume, ASP or revenue-recognition timing.

Third, recent 5-10 day flows show retail buying while institutions and foreigners have been exiting.

Fourth, short-selling ratio and stock-loan balance remain high.

Fifth, with FY2026 consensus P/E around 80x, delayed numerical proof can compress the multiple.

7. Entry Strategy

Stance: Watchlist / conditional entry only.

Level	Action
KRW 285,000-295,000	Observation zone. Only small probing size
Regain KRW 300,000 + foreigners net buy for 2-3 sessions	First entry review
Regain KRW 313,000-315,000	50-day MA and latest CEO purchase-price recovery zone
Regain KRW 347,000 / 20-day MA	Trend recovery confirmation
Break below KRW 280,000	Reassess. IR-notice effect and latest insider defense line may be weakening

The key signal is flow, not just price. Even if the stock regains KRW 300,000, upside may remain heavy if financial-investment selling continues and short-selling ratio stays in the mid-to-high 20s. If foreigners keep buying for 2-3 sessions, financial-investment selling slows, and the stock regains the KRW 315,000 area, the IR notice and insider buying can become the start of a narrative reset rather than just another schedule filing and a secondary signal.

Final Take

The IR notice is constructive. The direction is right. Hanmi is telling the market that it wants to expand from an HBM-only equipment story into a broader AI packaging TC-bonder platform.

The insider buying history is also constructive. Kwak’s cumulative open-market purchases since 2023 are not small, and most of them were very good prices in hindsight. But the number that matters now is not the old weighted-average price. It is the latest purchase price of KRW 315,407.

But the investment action is different from the narrative. The filing expands the story; insider buying is a secondary support signal; the flow is still weak. One day of foreign buying on June 1 is not enough.

The practical trigger is KRW 300,000 regained + continued foreign buying + financial-investment selling easing + recovery of the KRW 315,000 area. For now, this is a good disclosure and a strong insider signal to track while waiting for a real flow turn.

Evidence Classification

[Fact]

Hanmi’s June 1, 2026 filing is an IR meeting notice. The meeting date is June 4, 2026 and the purpose is “2H26 outlook and overseas new-market entry.” (DART)
The filing summary includes AI system semiconductors, foundry, OSAT, 2.5D Package TC bonders, HBF and participation in a U.S.-led AI semiconductor alliance. (DART)
IRGO also lists the same Hanmi IR schedule. (IRGO)
The latest DART insider report confirms that Kwak Dong Shin bought 9,576 common shares on Apr. 27, 2026 at KRW 315,407. (DART)
Based on the user-provided DART aggregation, Kwak’s open-market purchases from Jul. 26, 2023 to Apr. 27, 2026 total 29 events, 692,398 shares, about KRW 56.53bn, with a weighted-average price of KRW 81,643.
As of June 1 close, the stock was KRW 293,000 and down 28.4% from its 60-day high. Price and flow data are based on the user’s Research OS local DB / Kiwoom REST.

[Inference]

The filing is a strategy message about expanding the TC-bonder TAM from HBM into AI advanced packaging, not an order.
If 2.5D package, OSAT and HBF convert into customer adoption and repeat orders, Hanmi’s multiple logic can shift from HBM equipment to an AI packaging equipment platform.
The stock has cooled, but flow confirmation has not arrived yet.
Insider buying is a strong secondary signal, but for the current trade the latest high-price purchase at KRW 315,407 matters more than the full-period average price.

[Speculation]

Whether HBF TC bonders become a meaningful production market is still unclear.
The revenue path from the U.S.-led AI semiconductor alliance language is not yet disclosed.
A short-covering setup is possible, but not confirmed.
Whether the CEO continues buying after the latest purchase is unknown.

[Blocked]

Customer names, order values, ASP, volume, revenue-recognition timing.
Whether the 2.5D Package TC bonder is adopted in production lines.
HBF market standardization and Hanmi’s tool margin.
The identity and motive of stock-loan / short-selling positions.
A complete pre-2020 review of older / non-standard filings was not performed.

Samsung HBM4E 12-High Samples: What Is the Market Really Watching?

Mon, 01 Jun 2026 23:58:00 +0900

Context This note follows the Samsung vs SK Hynix forward P/E inversion, the Samsung stock-bonus / memory-supercycle note, and the Samsung C&T proxy / NAV gap trade. The related hubs are the AI HBM Hub and the Korea Semiconductor Value Chain Hub.

TL;DR

Samsung Electronics’ HBM4E 12-high sample shipment is positive, but it is not yet proof of a completed HBM share recovery. A sample is the start of the path, not the end.

The market is watching three things: whether Samsung’s HBM laggard discount can narrow, whether the sample converts into customer qualification and mass production, and whether the high-stack HBM ramp translates into TC bonder orders for equipment names such as Hanmi Semiconductor.

For Hanmi, the signal is constructive but not confirmed. On June 1, 2026, Hanmi rose 3.9% and foreigners bought KRW 26.2bn, but institutions sold KRW 50.2bn and financial-investment accounts sold KRW 57.6bn. Over the last five sessions, foreigners were still net sellers. The right stance is Watch / wait for confirmation.

1. What Was Confirmed: HBM4E 12-High Samples

[Fact] Samsung Electronics said it supplied HBM4E 12-high samples to global customers. The official specs include up to 16Gbps per pin, 3.6TB/s per stack, 48GB capacity, 1c DRAM plus a 4nm logic base die, 16% better energy efficiency and 14% better thermal resistance versus HBM4. (Samsung Semiconductor)

Item	Samsung HBM4E 12-high
Speed	Up to 16Gbps per pin
Bandwidth	3.6TB/s per stack
Capacity	48GB, 12-high
Process	1c DRAM + 4nm logic base die
Energy efficiency	+16% vs prior generation
Thermal resistance	+14% vs prior generation
Planned lineup	32GB 8-high, 64GB 16-high

The valuation path is:

HBM4E 12-high sample
→ customer qualification
→ mass-production yield and schedule
→ HBM share recovery
→ DS earnings revision
→ Samsung HBM laggard discount compression

The sample matters, but the market ultimately buys qualification, production and earnings revisions.

2. What the Market Is Watching in Samsung

2.1 HBM laggard discount

Samsung’s memory scale is enormous, but it has carried a laggard discount in HBM versus SK Hynix. The HBM4E sample shipment is therefore not just a product announcement. It is a relative-value event.

The precise sentence is:

Samsung’s HBM4E sample proves the catch-up path is alive. It does not prove catch-up is complete.

2.2 1c DRAM + 4nm base-die integration

Samsung’s HBM4E is not just a DRAM stack. It combines advanced DRAM, a logic base die, packaging, power and thermals. That matters because Samsung can tell a broader story than a pure memory stack: memory plus foundry plus system LSI plus packaging.

The market is asking whether Samsung can become an AI-memory turnkey supplier, not just another DRAM vendor.

2.3 2027 EPS revisions

The stock impact is less about immediate 2026 revenue and more about 2027 revisions. If qualification, volume supply and pricing are confirmed, sell-side DS earnings estimates can move again.

Checkpoint	Pass Signal	Fail Signal
Customer qualification	Global AI customer approval or guidance	Sample stage drags on
Mass production	Supply begins on customer schedule	Yield, thermal or power issues delay ramp
EPS revision	2027E DS estimates rise	News flow improves but numbers do not

3. Why Hanmi Semiconductor Is Connected

The Hanmi link is not the HBM4E product itself. It is the high-precision bonding equipment required to mass-produce high-stack HBM.

HBM stacks DRAM dies vertically. Moving from 8-high to 12-high and 16-high makes alignment, micro-bump bonding, thermal compression uniformity, warpage and yield more difficult. HBM4E is therefore a memory-product event and a back-end equipment event.

Hanmi’s HBM TC bonder sits in this stacking and bonding process. Computex Taipei describes HBM TC BONDER 4 as equipment for HBM die stacking and bonding. (Computex Taipei)

Step	Event	Hanmi Read-Through
1	Samsung ships HBM4E 12-high samples	Samsung HBM credibility improves
2	Customer qualification probability rises	Mass-production ramp probability rises
3	HBM4E capacity needs to expand	Stacking / bonding equipment demand rises
4	High-stack yield bottleneck intensifies	TC bonder value rises
5	Samsung may diversify equipment vendors	Hanmi’s Samsung-order option rises

The important boundary:

Samsung HBM4E samples do not equal direct Hanmi orders. They raise the option value of a future Samsung TC-bonder order.

Samsung-Hanmi TC-bonder discussions have been reported by local media, but there is no official order disclosure yet. (Daum News)

4. Hanmi Flow: First Positive Piece, Not a Confirmed Turn

Based on user-provided Research OS local DB / Kiwoom REST data for June 1, 2026:

Item	Value
Price	KRW 293,000
Return	+3.90%
Foreigners	+KRW 26.2bn / +90,254 shares
Institutions	-KRW 50.2bn / -173,433 shares
Retail	+KRW 22.8bn / +79,057 shares
Financial investment	-KRW 57.6bn
Investment trusts	-KRW 7.4bn
Private funds	+KRW 17.7bn
Pension funds	-KRW 1.8bn
Program trading	-KRW 4.5bn
Short-selling ratio	25.8%

One day looks better: foreigners bought, and the stock bounced. But institutional selling was much larger. Financial-investment selling overwhelmed foreign buying.

Recent cumulative flow is still cautious.

Period	Stock Return	Foreigners	Institutions	Retail
1D	+3.9%	+KRW 26.2bn	-KRW 50.2bn	+KRW 22.8bn
5D	-8.6%	-KRW 77.5bn	-KRW 523.4bn	+KRW 594.0bn
10D	-20.6%	-KRW 333.3bn	-KRW 569.3bn	+KRW 891.6bn
20D	-18.0%	-KRW 395.4bn	-KRW 127.7bn	+KRW 512.1bn

The right read:

June 1 foreign buying is the first piece of a bottoming signal. It is not enough to reverse the five-day and ten-day selling trend.

5. Short Selling Has Eased, But It Still Matters

Hanmi’s short-selling ratio was still 25.8% on June 1. That is below 41.8% on May 27, 37.1% on May 28 and 27.1% on May 29.

That is positive: the most aggressive short-pressure window may have passed. But a 25% short-selling ratio is still high. Foreign buying alone may not carry the stock if short selling and institutional selling remain heavy.

Confirmation signals:

Condition	Meaning
Reclaim and hold KRW 300,000	Sentiment repair
Foreign net buying for 2-3 consecutive days	June 1 was not one-off
Institutional / financial-investment selling fades	Flow pressure eases
Short-selling ratio below 20%	Short pressure normalizes
Reclaim KRW 319,000	Upper end of recent breakdown box

If KRW 282,000 breaks again, June 1 buying should be treated as a technical bounce.

6. Samsung and Hanmi Are Different Instruments

Item	Samsung Electronics	Hanmi Semiconductor
Essence	Direct HBM4E / memory earnings exposure	HBM stacking-equipment option
News path	Sample → qualification → mass production → DS earnings	Sample → capacity ramp → TC bonder orders
What to verify	Customer approval, volume, EPS revisions	Samsung orders, 2H26 sales recovery, OPM recovery
Key risk	Sample does not become volume	No Samsung order, short selling, multi-vendor pressure
Investment type	Core HBM catch-up	High-beta equipment option

Samsung common stock is the direct tool for HBM4E catch-up. Hanmi is the option that can move harder if Samsung’s HBM ramp turns into real equipment orders.

Final Take

Samsung’s HBM4E 12-high sample shipment strengthens the Samsung catch-up trade. The market can now assign a higher probability to Samsung re-entering the HBM leadership conversation.

But this is still a sample-stage event. Customer qualification and mass production must follow. For equipment names, the hurdle is even higher: Samsung HBM4E must lead to capacity expansion, and capacity expansion must lead to actual TC-bonder orders.

My view:

Samsung Electronics: Core Hold / watch HBM4E qualification and 2027 EPS revisions.
Hanmi Semiconductor: Watch / potential turn. Need KRW 300,000 support, continued foreign buying, easing financial-investment selling, and short-selling ratio below 20%.
Korean HBM chain: The market is not buying samples. It is buying qualification, production, orders and earnings.

Evidence Classification

[Fact]

Samsung said it supplied HBM4E 12-high samples to global customers, with up to 16Gbps speed, 3.6TB/s bandwidth per stack, 48GB capacity, 1c DRAM + 4nm logic base die, 16% energy-efficiency improvement and 14% thermal-resistance improvement. (Samsung Semiconductor)
Hanmi reported KRW 576.7bn of 2025 revenue and KRW 251.4bn of operating profit according to Korean media coverage. (Seoul)
Hanmi’s 1Q26 revenue and operating profit fell sharply, according to Korean media coverage. (Korea Economic Daily)
Hanmi June 1 flow figures are based on user-provided Research OS local DB / Kiwoom REST data.

[Inference]

Samsung HBM4E samples are an early signal for HBM laggard-discount compression.
The Hanmi impact is an option on future Samsung TC-bonder orders, not direct confirmed revenue.
June 1 foreign buying is positive but not sufficient to confirm a flow turn.

[Blocked]

Samsung’s HBM4E customer names, qualification status and mass-production volume are not public.
No official Samsung-Hanmi TC-bonder order disclosure has been confirmed.
Hanmi customer-level revenue mix and Samsung qualification stage are not public.