Exclusive Analysis on Korea Invest Insights

Are Samsung Electronics and SK Hynix Really Oversold Versus the 2027 Consensus? The Worst Case Is Already the Price

Fri, 10 Jul 2026 11:00:00 +0900

Context This piece follows Samsung and SK Hynix 2028E Profit Valuation. Where that post worked through 2028E point estimates and a cliff scenario, this one uses consensus dispersion itself as a reverse-engineering tool. It reads well alongside Big Tech’s Late-July Earnings Calls and Memory Thesis Scenarios, Who Pays for the 2027 Semiconductor Consensus, and Samsung and SK Hynix Through the NVIDIA Inflection Point. Related hubs: AI HBM Hub and Exclusive Analysis Hub.

TL;DR

On average analyst estimates, the current share prices look cheap. But the same math held at the 2018 cycle peak, and it was wrong then.
So we reverse-engineered it instead of trusting the average. Samsung’s 285,000 won divided by the lowest 2027 EPS estimate of 24,323 won gives 11.72x. SK Hynix’s 2,180,000 won divided by its lowest estimate of 186,357 won gives 11.70x. For both names, the current price equals the street’s worst-case scenario multiplied by a normal mid-cycle multiple.
In other words, the market isn’t “discounting the average,” it has already adopted the worst-case scenario as its default. The remaining question isn’t whether the consensus is right, but whether that worst case is right.
The split in views originates in the post-2028 window, but the visible symptom shows up in 2027 estimates. The high/low ratio on 2027 EPS has widened to 3.87x for Samsung and 3.83x for SK Hynix, while 2028 is largely a data void with almost no coverage at all.
Execution frame: a discount to the average consensus is not used as a buy signal. It’s read as a re-rating signal only when the lowest estimate rises for two consecutive months and the high/low ratio narrows below 3x. The documents that will settle this are big tech’s 2027 capex commentary on July 28-30 and the long-term contract price floor disclosures in SK Hynix’s Q2 earnings.

Key Framing

Samsung and SK Hynix currently trade almost exactly at the most pessimistic 2027 street estimate multiplied by 11.7x, not at a multiple of the consensus average. The market has already priced in the worst case. What's now on trial isn't the consensus, it's that worst-case scenario.

1. 2027 Profit Estimates: Same Companies, Same Year, Numbers That Diverge Nearly Fourfold

Start with what analysts are projecting for fiscal 2027 earnings per share (Samsung and SK Hynix close their books in December; Micron’s fiscal year ends August 2027; SanDisk’s ends June 2027). [Fact: consensus aggregation]

Ticker	2026 Avg EPS	2027 Avg EPS	Growth	2027 Low	2027 High	High/Low Ratio
Samsung Electronics	47,693 won	65,100 won	+36.5%	24,323 won (-49%)	94,114 won (+97%)	3.87x
SK Hynix	317,254 won	445,531 won	+40.4%	186,357 won (-41%)	713,786 won (+125%)	3.83x
Micron	$73.32	$149.64	+104%	$70.77 (flat)	$221.27	3.13x
SanDisk	$66.41	$204.47	+208%	$137.87 (+108%)	-	-

That estimates for the same company, in the same year, can spread 3.8-3.9x between the high and low case is itself unusual. SanDisk stands out in particular: even its most bearish estimate assumes 108% earnings growth, meaning not a single analyst is yet modeling an earnings decline.

Price targets tell the same story. Samsung’s targets range from 210,000 won to 850,000 won, and SK Hynix’s from 1,030,000 won to 4,700,000 won. The gap between the current price and the average target is 42% for Samsung and 32% for SK Hynix, versus just 9% for SanDisk. Measured by the gap between consensus and price, the two Korean names are stretched far wider than the pure-play US NAND producer. [Fact: price target aggregation]

2. Should We Even Trust These Numbers? The Low-P/E Trap

This is where we need to pause. The logic that “earnings are rising while the stock isn’t, so it must be undervalued” is the exact same logic that applied at the 2018 semiconductor cycle peak. Forward P/E was similarly low then, at 4-6x, and consensus was calling for earnings growth. What actually happened: Samsung’s semiconductor operating profit collapsed 69%, from 44.6 trillion won to 14.0 trillion won, in 2019, and SK Hynix’s fell 87%, from 20.8 trillion won to 2.7 trillion won. Something similar recurred after the 2022 peak, with both companies swinging to losses in 2023. [Fact: historical results]

In the memory sector, a low P/E is often not evidence of undervaluation. It can instead be the market pre-discounting the lesson that this consensus has historically been wrong at every peak.

There’s a second warning sign. Over the past 90 days, 2027 earnings estimates have jumped +73% for Samsung, +93% for SK Hynix, +52% for Micron, and +127% for SanDisk. [Fact: estimate revisions] Estimates chasing price moves after the fact is a pattern that typically shows up near cycle peaks. [Inference: based on historical cycle patterns]

3. Reverse-Engineering: What Scenario Is the Current Price Actually Pricing?

Instead of anchoring to the average estimate, we asked the question backward: what scenario is the current price consistent with?

Ticker	Current Price	2027 Low EPS	P/E on Low Estimate
Samsung Electronics	285,000 won	24,323 won	11.72x
SK Hynix	2,180,000 won	186,357 won	11.70x
Micron / SanDisk	-	on low estimate	13.5-14.0x

The two Korean numbers match to the second decimal place. There’s one reading of that. The current price is the street’s worst 2027 scenario, multiplied by a normal mid-cycle multiple of 11.7x for the semiconductor sector. The market isn’t weighing whether to discount the average estimate, it has already selected the worst estimate among several and priced off it as the effective base case. [Inference: interpretation of reverse-engineered result]

Once this framing is confirmed, the risk-reward picture sorts itself out.

If the worst-case scenario becomes reality: the current price is roughly fair, and further downside would mainly come from the multiple compressing further.

If even just the average scenario plays out: the math implies upside of +47% for Samsung and +71% for SK Hynix to the average target price.

If a cliff scenario worse than the worst case (assigned a 20% probability in the prior analysis) materializes: the market would likely start pricing off P/B rather than earnings multiples. Current estimated P/B is roughly 3.5x for Samsung and 10x for SK Hynix, far above the 1.1-2.5x seen at past cycle troughs. In other words, actual downside could run deeper than the earnings-based math suggests. [Inference: cliff probability and P/B estimates are unconfirmed estimates]

4. The Real Reason Views Diverge: The Cause Is 2028, the Symptom Is 2027

The intuition that “views diverge because 2028 estimates are all over the place” points in the right direction but needs a sharper mechanism. The visible 3.8-3.9x dispersion is already observed within the 2027 estimates. 2028 barely has any coverage to begin with, so it’s a data void, not a spread you could actually measure.

The causal chain runs like this. People hold different beliefs about what happens after 2028. One camp believes this cycle has structurally reshaped the industry around AI infrastructure demand; the other believes it’s just a boom-bust cycle with a larger amplitude. That difference in belief flows upstream into assumptions about second-half 2027 pricing, specifically how much oversupply from new capacity coming online should be baked in, and how much credibility to assign to price-floor clauses in long-term contracts. The result is that the disagreement surfaces as a nearly fourfold spread in 2027 estimates. It isn’t that 2028 splits opinion; it’s that opinion has already split, and that split shows up expressed through 2027 numbers.

The price structure confirms this framing directly. Here’s the share of the current price explained by the next two years (2026-2027) of combined earnings estimates. [Fact: calculation]

Ticker	Share of Price Explained by 2-Year Earnings	Share Attributed to Post-2028 Expectations
Samsung Electronics	39.6%	60.4%
SK Hynix	35.0%	65.0%
Micron	22.5%	77.5%
SanDisk	14.6%	85.4%

60-85% of the share price is the present value of the post-2028 window, for which almost no published numbers exist. Flip that around: recovering 35-40% of market cap from just the two visible years of earnings implies an annualized earnings yield of 18-20%. For that yield to make sense, one of two things has to be true: the market is bracing for a sharp post-2028 earnings decline, or the stock is meaningfully undervalued today. A comfortable middle ground doesn’t hold up mathematically.

5. What the Overlay Reveals

First, a low P/E has been the lowest-hit-rate buy thesis at this point in the cycle. The reliability of memory-sector consensus shifts by phase. During upswings it lags price and comes in too low; near peaks it turns excessively optimistic (2018, 2021); at troughs it underestimates the recovery. The current pattern of estimates jumping +52% to +127% over 90 days is a textbook near-peak signature. But there’s a countervailing implication too. If the market has already made the worst case its default, then fresh bad news going forward may not have as much power to push the price down further as it once did.

Second, Korea’s core names are carrying a disproportionate share of the dispersion burden. The pure-play US memory producer trades at 13.5-14.0x even on its lowest estimate, while the two Korean names trade at 11.7x on the same math. Even assuming the worst, Korea is still cheaper. The target-price gap tells the same story: 9% for SanDisk versus 42% for Samsung. This also suggests there’s still unspent energy in the valuation-convergence channel that SK Hynix’s ADR listing newly opened up. [Inference]

Third, the dispersion itself is a trackable signal. Once long-term contract terms are disclosed following late-July big tech earnings and SK Hynix’s Q2 results, three paths diverge.

Movement in Dispersion	Interpretation	Response
Low estimate rises, high/low ratio narrows below 3x	Worst-case scenario being rejected	Re-rating signal, phased-buy condition met
High estimate falls, dispersion narrows	The average itself is breaking down	Reduce exposure signal
Dispersion holds or widens	Debate unresolved	Stay on the sidelines

Rather than gauging vague market sentiment, directly tracking which direction Samsung’s low estimate (currently 24,323 won) and SK Hynix’s low estimate (currently 186,357 won) move next month is a far more precise decision tool.

Fourth, SanDisk is structurally different. Even its lowest estimate assumes 108% earnings growth, so the shock of a first genuinely bearish model has not yet arrived, and its 2-year earnings share of price, at just 14.6%, is the lowest of the four names, meaning it leans most heavily on distant expectations. The prior entry-wait band of $1,240-1,490 stands unchanged.

6. The Practical Frame: What to Watch and When to Move

The base-case positioning is unchanged. SK Hynix waits for conditions to be met, Samsung holds with new buying paused, and SanDisk waits for its band. What this analysis adds is the decision tool.

Replace the buy logic. “Cheaper than the average consensus” is no longer used as a justification. “Evidence that the market’s already-priced worst case is wrong” is the only acceptable buy trigger going forward. That evidence will come from two documents: big tech’s first commentary on 2027 investment plans, due July 28-30, and whether SK Hynix’s Q2 earnings release discloses the price-floor terms in its long-term contracts.

Add to the tracking list. Track the monthly direction of Samsung’s low estimate (24,323 won) and SK Hynix’s low estimate (186,357 won). If the low estimate rises for two consecutive months and the high/low ratio narrows below 3x, add that to the existing flow-based conditions for starting a phased buy into SK Hynix. Conversely, if the high estimate starts falling first, treat that as a signal to trim Samsung.

Invalidation condition. If you entered on an “overreaction” thesis, exit immediately the moment two or more credit-risk warning signs light up simultaneously, or big tech cuts this year’s investment plans. Conversely, if you’re holding cash on a “concerns are justified” view, drop that view once the rise in the low estimates is confirmed.

Closing: Two Facts Worth Acting On

“Concerns are overdone relative to consensus” is a true statement. But it was also true in 2018. So the statement itself is not, on its own, useful for a decision.

Two facts are useful. First, the current price is already set against the worst scenario the street has produced. Second, whether that worst-case scenario is right or wrong will be confirmed on paper within the next three weeks. The right move isn’t to buy ahead of time on faith in the average estimate, it’s to buy after actually confirming the worst case has been rejected. Until then, the 3.8-3.9x dispersion in estimates isn’t an opportunity, it’s an unresolved verdict still pending.

This post is an analytical piece based on public consensus aggregates and independent reverse-engineering. The stocks mentioned are framework examples, not investment recommendations. EPS and price-target estimates can change depending on the aggregation date, and the cliff-scenario probability and P/B estimates are unconfirmed estimates. Investment decisions and responsibility rest with the investor.

Samsung and SK Hynix 2028E Profit Valuation: Cheap-Looking Numbers and Cycle Tests

Thu, 09 Jul 2026 14:20:00 +0900

Related context: this is a follow-up to AI token value and memory value added, Samsung 2Q26 signals and noise, late-July Big Tech earnings and the memory thesis, and Micron FY3Q26 AI memory review. Relevant hubs are the Exclusive Analysis Hub and the AI HBM Hub.

TL;DR

Samsung Electronics and SK Hynix look very cheap on 2028E net income. If we use MarketScreener aggregate estimates, 2028E net income is about KRW 439.8tn for Samsung and KRW 346.6tn for SK Hynix. Applying a 9 to 12x P/E gives fair market caps of KRW 3,958tn to 5,277tn for Samsung and KRW 3,119tn to 4,159tn for SK Hynix.

Against current market caps, that implies roughly +120 to +193% for Samsung and +104 to +172% for SK Hynix. The market is valuing them at only about 4.1x and 4.4x 2028E P/E, respectively.

But the bottleneck is not the multiple. It is the 2028E profit number. If 2028 earnings are normalized, both stocks are cheap. If 2028 is the cycle peak, a 4 to 5x P/E may not be cheap.

Item	Samsung Electronics	SK Hynix
MarketScreener 2028E net income	KRW 439.8tn	KRW 346.6tn
9 to 12x fair market cap	KRW 3,958tn to 5,277tn	KRW 3,119tn to 4,159tn
Current implied 2028E P/E	About 4.1x	About 4.4x
Scenario-weighted 2028E net income	About KRW 272tn	About KRW 208tn
Scenario-weighted 9 to 12x fair market cap	KRW 2,444tn to 3,258tn	KRW 1,876tn to 2,502tn
Scenario-weighted upside	About +36 to +81%	About +23 to +64%

The stronger relative setup is Samsung. SK Hynix has higher-quality HBM earnings, but more of that quality is already embedded in the price. Samsung has HBM catch-up, commodity DRAM/NAND leverage and a lower implied 2028E P/E.

1. Current Coordinates

Samsung’s net income belongs economically to both common and preferred shares, so the right denominator is total equity value, not common-share market cap alone. Depending on the public screen, Samsung market cap appears around KRW 1,753tn, while a common-plus-preferred equity value framework is closer to KRW 1,800tn. SK Hynix is around KRW 1,530tn.

Item	Samsung Electronics	SK Hynix
Reference share price	KRW 274,000 to 274,500	KRW 2,151,000 to 2,160,000
Analysis market cap	About KRW 1,800tn	About KRW 1,530tn
2026E OP reference	KRW 371tn to 381tn	KRW 280tn to 291tn
2027E OP reference	KRW 500tn to 574tn	KRW 379tn to 432tn

Samsung guided to 2Q26 revenue of KRW 171tn and operating profit of KRW 89.4tn. SK Hynix reported 1Q26 revenue of KRW 52.576tn, operating profit of KRW 37.610tn and net income of KRW 40.346tn, with operating margin of 72%.

Using net income as roughly 75% of operating profit, current prices imply 2026E P/E near 6 to 7x and 2027E P/E near 4 to 5x. The market is not treating these companies as secular growth stocks. It is treating them as memory companies at or near peak earnings.

2. The Point-Estimate Valuation

MarketScreener aggregate estimates put 2028E net income at KRW 439.781tn for Samsung and KRW 346.562tn for SK Hynix. The same dataset puts 2028E EBIT at KRW 543.635tn and KRW 423.866tn, respectively.

KRW tn	2028E net income	9x fair cap	10x fair cap	12x fair cap	Current cap	Upside
Samsung Electronics	439.8	3,958	4,398	5,277	About 1,800	+120% / +144% / +193%
SK Hynix	346.6	3,119	3,466	4,159	About 1,530	+104% / +127% / +172%

This simple calculation says both stocks are cheap. It also says Samsung is slightly more asymmetric because its implied 2028E P/E is lower.

The danger is that 2028E is still far away. Memory cycles have repeatedly produced large peak profits that later collapsed. Therefore, the real question is whether 2028E is a normalized profit base or a peak.

3. How to Estimate 2028E

Extending 2027E mechanically is not enough. A top-down “AI capex keeps rising, so memory profit rises” approach is also too thin. Memory profits are often set by supply, not demand. The 2028 supply base is already being shaped by 2026 to 2027 capex.

The better process is:

Step	What to do	Why it matters
1	Split profits into contracted HBM, non-contracted HBM/server DRAM, and spot DRAM/NAND	Each bucket has a different earnings floor
2	Track supply instead of forecasting it abstractly	2027 capex determines 2028 bit supply
3	Segment demand into AI servers, general servers and consumer devices	Price elasticity differs sharply
4	Force memory-cycle base rates into the model	Peak-year earnings historically fall hard
5	Use scenario distributions and reverse implied earnings	Current price already embeds a view

The single most useful future disclosure would be the 2028 HBM long-term agreement coverage ratio. If a large share of 2028 shipment is already contracted, the 2028 profit floor is much higher than in historical cycles.

The main supply variables are 2027 capex guidance, equipment orders from ASML/AMAT/TEL, Hynix’s Yongin and Cheongju investments, Samsung’s HBM and packaging catch-up spending, CXMT and YMTC expansion, and whether HBM bottlenecks remain in stacking and packaging rather than wafers alone.

Demand must also be segmented. AI server demand is relatively inelastic. General server demand is cyclical. PC and smartphone memory demand is price-sensitive and can already show demand destruction when memory prices rise too quickly.

Historical base rates are harsh.

Cycle	Operating profit change
Samsung consolidated 2018 to 2019	-53%
Samsung DS 2018 to 2019	-69%
SK Hynix 2018 to 2019	-87%
2023 memory downcycle	Loss-making episodes

This cycle deserves an upward adjustment because of HBM contracts, oligopoly structure, wafer cannibalization by HBM and low AI-server price elasticity. But it also deserves caution because of Chinese supply, consumer demand destruction, policy-driven capacity expansion and the statistical extremity of 70%+ operating margins.

4. Scenario Valuation

Scenario	Probability	Samsung 2028E NI	9x	12x	Upside	SK Hynix 2028E NI	9x	12x	Upside
S1 shortage persists	25%	KRW 420tn	3,780	5,040	+110 to +180%	KRW 315tn	2,835	3,780	+85 to +147%
S2 soft landing	45%	KRW 280tn	2,520	3,360	+40 to +87%	KRW 215tn	1,935	2,580	+26 to +69%
S3 hard landing	30%	KRW 135tn	1,215	1,620	-32 to -10%	KRW 110tn	990	1,320	-35 to -14%
Probability-weighted	-	KRW 272tn	2,444	3,258	+36 to +81%	KRW 208tn	1,876	2,502	+23 to +64%

MarketScreener’s estimate sits closer to S1. The probability-weighted model blends S2 and S3 risks. These two approaches do not conflict. One shows how cheap the stocks are if consensus is right. The other shows how much risk discount should be applied to that consensus.

5. What the Current Price Already Embeds

Reverse the current market caps by a 9 to 12x P/E and we get the 2028E net income implied by today’s prices.

Item	Current cap	Implied 2028E NI at 9x	Implied 2028E NI at 12x
Samsung Electronics	About KRW 1,800tn	About KRW 200tn	About KRW 150tn
SK Hynix	About KRW 1,530tn	About KRW 170tn	About KRW 127tn

Against rough 2027E net income estimates, current prices already imply a meaningful drop.

Item	Rough 2027E NI	Implied 2028E NI	Implied decline
Samsung Electronics	About KRW 400tn	KRW 150tn to 200tn	-50 to -63%
SK Hynix	About KRW 310tn	KRW 127tn to 170tn	-45 to -59%

This is the key reframing. The current bet is not necessarily that 2028 earnings will be spectacular. It may simply be that the decline from 2027 will be shallower than the 45 to 63% decline already embedded in prices.

6. Samsung Versus SK Hynix

Company	Core thesis	Risk
Samsung Electronics	Lower implied 2028E P/E, HBM catch-up, commodity DRAM/NAND leverage	HBM4 delay, foundry/system LSI losses, weaker reported-to-core profit conversion
SK Hynix	Highest-quality HBM profit pool, customer reference, strong margin	More optimism embedded, ADR/event flow, need to prove 2028 LTA coverage

Samsung is not the HBM leader, but that is partly the point. Expectations are lower. If HBM4 qualification, yield and allocation improve, the stock gets both earnings revision and multiple repair. Samsung also has more leverage to conventional DRAM and NAND pricing.

SK Hynix has the cleaner HBM exposure and better current profit quality. But the market already gives it more credit for structural HBM change. The stock needs clearer evidence that 2028 profit is contracted and not just a peak-cycle extrapolation.

7. Practical Stance

Stock	Stance	Reason	What to confirm
Samsung Electronics	Hold-biased, pullback candidate	More asymmetric 2028E valuation and catch-up optionality	July 30 segment data, HBM4 progress, NAND/DRAM pricing
SK Hynix	Watchlist, no event-chasing	Best HBM quality, but more expectation embedded	ADR absorption, 2Q call LTA coverage, HBM4 premium

For Samsung, selling aggressively here requires assigning a very high probability to a 2028 hard landing. The data does not yet justify that. For SK Hynix, the company quality is clear, but ADR, new-share absorption and 2028 contract visibility matter before chasing.

8. Invalidation

The thesis weakens if two or more of the following happen.

Invalidation signal	Meaning
2027 capex across the three major memory firms rises more than 50% YoY	Supply discipline breaks
CXMT qualifies high-end server DDR5 with major customers	Commodity DRAM ceiling falls
Chinese suppliers make visible HBM progress	Premium durability weakens
Big Tech capex guidance slows sharply	AI server demand downside
No disclosure or evidence of 2028 HBM LTA coverage	Peak-profit discount remains
YMTC NAND ramp accelerates	NAND pricing risk

Final View

Samsung and SK Hynix both look cheap on 2028E earnings. But cheap-looking is not the same as cheap.

Using MarketScreener point estimates, both have roughly double-like potential. Using a supply, contract and base-rate scenario model, the upside narrows to +36 to +81% for Samsung and +23 to +64% for SK Hynix. That is still attractive, but the margin of safety is smaller.

The key is 2028E. It cannot be built from demand stories alone. It depends on HBM LTA coverage, 2027 capex discipline, DRAM/NAND pricing, Chinese supply and Big Tech capex.

The relative preference is Samsung. SK Hynix has better HBM earnings, but requires more belief. Samsung has catch-up optionality, broader DRAM/NAND leverage and a lower implied 2028E P/E. For both names, the next checks are late-July Big Tech capex, Samsung’s July 30 detailed results, SK Hynix ADR absorption and the 2Q call.

Evidence and Limits

Classification	Content
Fact	Samsung 2Q26 preliminary revenue KRW 171tn and OP KRW 89.4tn. SK Hynix 1Q26 revenue KRW 52.576tn, OP KRW 37.610tn and net income KRW 40.346tn. MarketScreener 2028E net income of KRW 439.781tn for Samsung and KRW 346.562tn for SK Hynix
Inference	Net income approximated at 75% of OP. Current prices imply 2028E net income of KRW 150-200tn for Samsung and KRW 127-170tn for SK Hynix
Speculation	Scenario probabilities and 2028E net income assumptions: S1 25%, S2 45%, S3 30%
Blocked	2028 HBM LTA coverage, customer allocations, actual contract prices and detailed analyst-by-analyst 2028E assumptions are not fully available from public data

Sources

AI Token Value Today and Tomorrow: Value Added for Memory Companies

Thu, 09 Jul 2026 09:00:00 +0900

Related context: this article follows our work on late July big tech earnings and the memory thesis, Micron’s FY3Q26 AI memory earnings, and the 1H26 AI infrastructure bottleneck review. Related hubs are the Exclusive Analysis Hub and the AI HBM Hub. The question is not simply whether AI demand is strong. The question is who keeps the economic value created by AI tokens.

TL;DR

If we decompose one current dollar of AI token revenue, the largest share of value added still sits at the model and application layer. The equity market, however, is being driven by data center investment that is being pulled forward before final application revenue has fully arrived. That is why today’s AI value chain looks less like a normal pyramid and more like an inverted one.

The current value-added split of one dollar of AI token revenue is roughly as follows.

Layer	Current value-added estimate
AI model company	45 to 55 cents
Cloud and data center	10 to 16 cents
GPU supplier	About 13 cents
HBM, server DRAM, SSD	About 2.3 cents
Other infrastructure software and storage	3 to 5 cents
Foundry	About 1 cent
Power	About 1.3 cents

The market capitalization and capex cycle cannot be explained by this table alone. Current AI service revenue is estimated around 60 to 100 billion dollars per year, while the leading GPU supplier’s data center revenue run rate has reached around 300 billion dollars, and the big four cloud companies’ 2026 capex is estimated near 725 billion dollars. The industry is building future demand before current revenue catches up.

Memory is a real bottleneck in this structure. But if memory captures only about 2 to 5 cents of value added per token dollar, the central question for memory stocks is not “is the P/E low?” The real question is whether the industry can sustain mid-cycle net income of roughly 100 to 140 trillion won.

Question	View
Where does AI token value accrue?	Near term: models and cloud. Long term: workflow applications and physical bottlenecks
Is memory a structural beneficiary?	Yes, but its value-added share is small and the stocks already price in a strong mid-cycle profit base
What should investors monitor?	HBM4 contract pricing, customer allocation, blended memory ASP, cloud backlog, and application revenue conversion

1. The Real Question: Who Keeps the Dollar?

The most common mistake in AI investing is mixing up revenue attribution with value-added attribution. If a customer pays one dollar for an AI service, that dollar may first appear as revenue at a model company. But inside that dollar are cloud costs, GPU depreciation, memory, foundry capacity, power, networking, servers, cooling, labor, software, taxes, and financing costs.

So “who books the revenue?” is not the same question as “who keeps the surplus?”

This article uses five rules.

Rule	Meaning
Revenue attribution	The first seller that invoices the customer
Value-added attribution	The economic value left after removing double counting
Token dollar	A dollar of API or model usage revenue
Application dollar	A dollar of final workflow or agent software spend
Depreciation basis	Data center capex is annualized over a five-year asset life

The capex treatment matters. AI data center spending does not disappear as an immediate cost. It is capitalized into GPUs, servers, power gear, buildings, cooling, and networks, then depreciated over several years. Therefore, the right cost in a token-dollar model is annual depreciation, not total capex.

Time also matters. Inference cost in 2025 is not the same as inference cost in 2028. Cost per unit of performance is falling quickly.

Period	Inference cost as share of revenue	Model gross margin
2025	50 to 67%	33 to 50%
2026	40 to 55%	45 to 60%
2027 to 2028	30 to 40%	60 to 70%

This leads to an important point. Token prices can fall while model gross margins rise, if inference costs fall faster than prices. Much of that margin is then reinvested into training, compute reservations, and customer acquisition. The value does not simply stay on the income statement. It flows back into GPUs, memory, power, and foundry capacity.

2. The Current Value Waterfall

The table below decomposes one current AI token dollar. Revenue attribution is the amount a layer effectively invoices or absorbs. Value added is the economic share after removing double counting.

Layer	Revenue attribution	Value-added estimate	Interpretation
AI model company	100 cents	45 to 55 cents	Direct customer layer. Falling inference cost is the largest lever
Cloud and data center	About 50 cents	10 to 16 cents	Operating layer that bundles GPUs, networks, power, and buildings
GPU supplier	About 18 cents	About 13 cents	Main physical bottleneck today
HBM, server DRAM, SSD	About 3 cents	About 2.3 cents	Critical to system performance, but small as a token-dollar share
Infrastructure software, storage, security, observability	About 5 cents	About 3 cents	More important as operations become more complex
Foundry	About 1.5 cents	About 1 cent	Common bottleneck across AI chip designs
Server OEM and ODM	About 3 cents	About 1.4 cents	High throughput, limited margins
Third-party networking	About 1.3 cents	About 0.7 cents	Important at cluster scale, but supplier power varies
Power	About 2.5 cents	About 1.3 cents	Small cost line, large site bottleneck
CPU and IP	About 0.5 cents	About 0.3 cents	Control and support compute
Construction, components, materials, residual	About 9 cents	Distributed	Lower layers of the data center stack

Two observations matter.

First, power is not the largest cost item inside a token dollar. But it is one of the largest physical bottlenecks. Power matters through grid access, substations, transmission, permits, and long-term power contracts.

Second, memory’s value-added share per token dollar is still small. This does not mean HBM is unimportant. HBM is indispensable. But the economic share retained by memory companies is far smaller than the share retained by GPUs or model/application layers. If the equity market ignores that difference, valuation errors appear.

3. The Inverted Pyramid

The current AI stack looks like this.

Item	Estimated scale
AI model and application revenue	60 to 100 billion dollars per year
GPU data center revenue run rate	About 300 billion dollars
2026 big four cloud capex	About 725 billion dollars
AI revenue needed to normalize 700 billion dollars of capex	About 2.8 to 3.0 trillion dollars

If AI service revenue is currently 60 to 100 billion dollars, and memory revenue is 3 to 7 cents per token dollar, AI memory flow revenue should be only about 2 to 7 billion dollars per year. Yet the leading memory companies are producing revenue and profit numbers far above what current token revenue alone can explain.

For example, SK Hynix reported 97.1 trillion won of revenue in 2025 and 52.6 trillion won of revenue in the first quarter of 2026. That scale is hard to explain only from current AI token revenue. Much of today’s memory revenue is therefore tied to capex being pulled forward for future AI demand.

This is not automatically wrong. Data centers cannot be built instantly after demand appears. Power, land, substations, GPUs, servers, cooling, and networking must be secured years in advance. Pre-investment is necessary.

The question is speed. If application and cloud revenue catch up, today’s capex is normal growth investment. If final demand lags, the correction begins upstream: memory and server components first, then GPUs, then cloud capex plans.

4. Three Convergence Paths

AI revenue is not determined by token price alone.

AI revenue = token volume × token price

Token price per unit of performance is falling quickly. It may decline by roughly 10x per year through the early phase, then slow to 1.5 to 2x per year after 2027. But volume can explode at the same time. A simple question may use a few hundred tokens. A workflow, document process, coding task, or agent job can use tens of thousands, hundreds of thousands, or even more than a million tokens.

That means price declines do not automatically mean revenue declines. If price falls by 90% but volume rises 100x, revenue still rises 10x.

Scenario	Probability sense	Description	Market implication
Demand convergence	45%	Application revenue scales quickly and 1.0 to 1.6 trillion dollars of capex normalizes	Limited upstream drawdown, bottleneck premium remains
Digestion then reacceleration	40%	2027 to 2028 capex pauses by 20 to 30%, then workflow AI demand returns	Memory ASP could correct 40 to 60%, then recover
Structural oversupply	15%	Usage elasticity disappoints and data center buildout proves excessive	Capex cuts of 50% or more and upstream recession

On-device AI changes the distribution, not the whole demand picture. More inference on phones, PCs, cars, robots, and industrial devices can reduce some cloud token revenue. But silicon demand does not disappear. LPDDR, mobile SoCs, edge NPUs, embedded storage, and local servers may all benefit.

5. Long-Run Equilibrium: Workflows Capture the Dollar

Over the long run, tokens are unlikely to remain scarce. As token prices fall and model performance gaps narrow, value moves from token production toward workflow ownership.

On a pure API dollar, the long-run value split may look like this.

Layer	Long-run value-added estimate
Model and API company	45 to 55 cents
Cloud and data center	12 to 15 cents
GPU supplier	8 to 10 cents
Custom ASIC	4 to 5 cents
Memory	4 to 5 cents
Foundry	3 to 4 cents
Power and data center infrastructure	4 to 5 cents
CPU and IP	About 1.5 cents
Server OEM	About 1.5 cents
Other infrastructure software	4 to 6 cents
Residual	About 5 cents

But customers do not really buy tokens. They buy outcomes: customer support, coding productivity, claims processing, contract review, factory quality control, and logistics automation. On a final application dollar, the distribution changes.

Layer	Long-run revenue attribution	Long-run value added
Workflow applications, agents, distribution	100 cents	45 to 50 cents
Model and API	35 to 40 cents	18 to 24 cents
Cloud and data center	13 to 15 cents	4 to 6 cents
Total silicon	9 to 12 cents	6 to 8 cents
Power, land, cooling	1.5 to 2 cents	About 1 cent
Server OEM and generic infrastructure	About 1 cent	Low
Other software	3 to 5 cents	Medium

The long-run conclusion is simple: tokens commoditize, while workflow ownership and physical bottlenecks retain value.

6. Capture Tiers

The best AI businesses are not simply the ones with the most revenue. The best businesses have three traits.

Customers cannot switch easily.
Cost declines become margin expansion.
Supply cannot expand quickly when demand rises.

Tier	Layer	Why it matters
Tier 1	Foundry, workflow applications, integrated cloud platforms	Bottleneck supply, customer lock-in, full-stack control
Tier 2	GPU, platform AI suppliers, HBM leaders, custom ASIC	Current bottlenecks and high growth, but with price and competition risk
Tier 3	Server OEM, neocloud, commodity memory, regulated utilities	Volume grows, but rent capture is weaker

Foundry belongs in Tier 1 because AI wafer demand exists whether the winning chip is a GPU or a custom ASIC. Workflow applications belong there because model costs can fall while end-user value remains tied to business outcomes.

Integrated cloud platforms need case-by-case judgment. Pure GPU rental can become a capital-heavy utility. But a company that owns search, ads, productivity software, cloud, chips, models, and data centers can turn capex into margin.

7. Multiples: Cheap Looking Is Not the Same as Cheap

AI value-chain multiples should differ by layer.

Layer	Appropriate valuation lens
Foundry	EV/S 8 to 11x, P/E 20 to 26x if growth persists
GPU supplier	Normalized EV/S 9 to 13x, scarcity 15 to 18x, P/E 18 to 24x
Integrated cloud	EV/S 5 to 8x, full-stack 8 to 10x, P/E 20 to 28x
Platform AI supplier	Hypergrowth EV/ARR 10 to 18x, deceleration 5 to 8x
Workflow AI software	Simple tools 2 to 4x, workflow owners 8 to 12x, regulated data owners 12 to 20x
HBM and memory	EV/S 3 to 5x, leaders 5 to 7x, mid-cycle P/E 8 to 14x
Storage	EV/S 2 to 3.5x, platform 4 to 7x, normalized P/E 8 to 14x
CPU and IP	CPU 12 to 18x, IP conditionally 30 to 50x
Server OEM	EV/S 0.5 to 1.5x, P/E 8 to 15x
Power and data center infrastructure	Contracted assets at EV/EBITDA 10 to 18x or project IRR

Memory is the layer where investors should be most careful. A 5 to 7x P/E can signal cheapness. At a cycle peak, it can also simply mean that the earnings denominator is temporarily inflated.

Useful valuation formulas include:

Formula	Use
EV/NOPAT = (1 - g/ROIC) / (WACC - g)	Normalized multiple with growth and capital efficiency
Justified PBR = (sustainable ROE - g) / (COE - g)	Long-run capital-intensive industry PBR
Mid-cycle EPS = (peak + trough + 2×normal) / 4	Cycle-aware earnings
Pre-profit value = terminal NOPAT × terminal multiple / (1 + discount rate)^t	Early-stage or loss-making firms

Capital costs also differ.

Layer	WACC or discount rate
Hyperscalers	8.5 to 9.5%
GPU and foundry	9 to 10.5%
Memory	10.5 to 12%
Vertical AI software	12 to 18%
Loss-making model layer	14 to 20%

8. The Memory Question: Can Mid-Cycle Net Income Reach 100 to 140 Trillion Won?

For memory companies, the key question is no longer whether HBM is good. The market already knows that. The key question is whether the current market cap requires a realistic or unrealistic mid-cycle profit base.

Suppose a leading memory company is valued near 1,400 trillion won. To sustain that value after one or two peak years, mid-cycle net income must be very large.

Item	Assumption
Peak net income	240 trillion won
Trough net income	50 trillion won
Normalized net income	60 trillion won
Mid-cycle net income	(240 + 50 + 2×60) / 4 = about 102 trillion won

Applying an 8 to 14x mid-cycle multiple gives a fair-value range of about 820 to 1,430 trillion won. If the current market cap is already near the top of that range, the stock is not automatically expensive, but the margin of safety is thin.

Further upside requires one or more of the following.

Requirement	Meaning
Evidence that mid-cycle net income can exceed 140 trillion won	HBM premium and high-margin mix remain after the peak
HBM4 custom premium is defended	Product customization protects pricing
ASP holds after Samsung’s supply entry	More supply does not crush pricing
General DRAM and NAND pricing also improves	The cycle broadens beyond HBM
Long-term customer agreements and volume commitments	Downcycle earnings risk falls

9. Samsung, SK Hynix, and Micron Are Not the Same Trade

Company	Core logic	Risk
Samsung Electronics	HBM share recovery and memory mix improvement. Expectations are lower than for the leader	HBM qualification, yield, foundry and system LSI losses
SK Hynix	Highest HBM profit quality and customer reference	Much of the good news is already in the price
Micron	Leading indicator for AI memory contracts, pricing, FCF, and capex	Volatility, short interest, and supply-cycle sensitivity

Samsung has more catch-up optionality. SK Hynix has the highest-quality memory profit pool, but expectations are embedded. Micron is the canary because its pricing, long-term agreements, data center revenue, and capex guide influence Korean memory estimates directly.

10. Cloud, Power, and Vertical AI

Cloud companies split into two groups: integrated platforms and capital-heavy rental platforms. The most important numbers are backlog, revenue conversion, operating margin, internal silicon mix, free cash flow, and bond issuance.

Power is a small token-dollar cost line but a large site bottleneck. Grid access, substations, transmission, PPAs, gas turbines, nuclear, SMRs, water, land, cooling, and permits matter more than the simple power bill.

Vertical AI companies can capture the largest long-run share if they own the workflow. The winners will have regulated or physical domains, proprietary data, action authority, measurable ROI, model replaceability, a path to 70%+ gross margin, and compliance moats. Generic chat wrappers, basic retrieval integration, API resale, data-less automation, GPU brokerage, and server assembly should be treated with caution.

11. What to Monitor

Metric	Why it matters
Combined cloud backlog	Whether capex is tied to future revenue
Self-built capex ratio	How much demand is moving outside public cloud
On-device inference share	Cloud token revenue substitution risk
HBM4 contract pricing	Durability of memory premium
Customer allocation by HBM supplier	Supplier bargaining power
Blended memory ASP	Whether the cycle is broadening
GPU lead times	Persistence of scarcity
Cloud operating margin	Whether capex turns into profit
AI application revenue growth	Whether final demand catches infrastructure spending

The thesis weakens if open-weight systems compress model pricing faster than expected, on-device AI reduces cloud demand more quickly, 2027 cloud capex slows sharply, HBM4 premiums compress, or AI application revenue fails to scale.

Final View

The AI token economy can be summarized in one sentence:

Tokens get cheaper, while value moves to workflow ownership and hard-to-expand physical bottlenecks.

Memory is one of those bottlenecks. HBM, high-performance DRAM, and SSDs are essential to AI servers. But memory’s value-added share per token dollar remains limited. That is why memory stocks require more than the statement that HBM is strong. Current prices already demand a strong mid-cycle profit base.

The four key questions are:

Does HBM4 preserve custom pricing?
Do leading suppliers hold ASP and margins after Samsung’s entry?
Do cloud backlog and AI application revenue catch up to capex?
Can memory companies sustain mid-cycle net income of 100 to 140 trillion won?

If those four conditions are met, today’s large memory market caps can be justified. If not, low P/E ratios may prove to be a cycle-peak illusion rather than a margin of safety.

Samsung Electronics recorded the world's largest quarterly operating profit. Signals and noise poured in today.

Wed, 08 Jul 2026 22:35:00 +0900

Context This is a follow-up to the Samsung 2Q26 preview, Micron FY3Q26 review, NVIDIA earnings analogue work, late-July Big Tech earnings-call scenario map, and Big Tech financing relay note. Related hubs are the Exclusive Analysis Hub and the AI HBM Hub.

TL;DR

Samsung Electronics presented provisional sales of 171 trillion won and operating profit of 89.4 trillion won for the second quarter of 2026. Based on Samsung Electronics’ official newsroom, this is a very large number compared to the operating profit of 57.23 trillion won in the previous quarter and 4.68 trillion won in the same period last year. Looking at memory prices, AI server demand, hyperscaler investment, and HBM4 ramp-up together, the quality of Samsung Electronics’ investment argument has actually become stronger.

However, the stock price moved in the opposite direction. Based on local DB, Samsung Electronics closed at 296,000 won on July 7 and 277,500 won on July 8. On July 7, foreigners net sold about 1.82 trillion won, and on July 8, they sold about 874.1 billion won. Individuals made net purchases of approximately 2.32 trillion won and 584.7 billion won, respectively. Although good performance was achieved, foreign and programmatic supply and demand saw peak-out, interest rates, exchange rates, and overcrowding of positions before profit levels.

The conclusion of this article is as follows.

Item	Judgment
Samsung Electronics Investment Argument	enforce. DRAM/NAND prices, AI servers, and hyperscaler CapEx are all favorable.
Short term timing	damage. Despite good performance, there was selling pressure for two days in a row.
supply and demand	danger. Foreign selling and private absorption are conflicting, and the Korean market is weak.
Sell decision	no. The current data is closer to digesting expectations and supply and demand pressures than to damaging the thesis.
Additional purchase decision	Not yet. You must check ASML, TSMC, Samsung Electronics IR, foreign supply and demand, and exchange rates and interest rates.
action	Strong holding, freeze on additional purchases, make a decision after confirming the next gateway.

To sum it up in one line, the information released today is not “Samsung Electronics has worsened,” but is closer to “the market has already had too many good numbers, and now demands the next number and the quality of supply and demand.”

1. Why should we separate signal and noise?

Today’s data is not read in only one direction. Numbers are good. However, stock prices and supply and demand are bad. Memory price comments are strong. However, spot trading takes a breather on some items. Hyperscaler CapEx continues to grow. However, long-term interest rates in the United States are also high.

On days like this, looking at only one side is wrong. “It’s the world’s largest quarterly operating profit, so you must buy it” is also hasty, and “the cycle is over because it fell behind despite good performance” is also hasty.

Therefore, this article divides the data into four layers.

floor	Content	Role in Investment Decisions
Official Performance	Samsung Electronics 2Q26 provisional sales and operating profit	Check profit level
industrial signals	DRAM·NAND price, AI server, hyperscaler CapEx	Check the quality of the thesis
market noise	Sluggish spot trading, Morgan Stanley peak logic, ASML·TSMC high bar	Check short-term timing
Supply/Demand/Macro	Foreign selling, exchange rate, interest rate, Korean market breadth	Position Sizing

The key is to separate the profit direction and the stock price direction. Samsung Electronics’ main business profit direction has improved. However, the drivers of short-term stock prices were not profits, but supply and demand and the discount rate.

2. Officially confirmed values: Sales 171 trillion won, operating profit 89.4 trillion won

[Fact] Samsung Electronics announced its second quarter provisional results through its official newsroom on July 7, 2026. Consolidated sales are approximately 171 trillion won and operating profit is approximately 89.4 trillion won. According to Korean disclosure regulations, the company presented a median value rather than a range, and explained that the actual estimated range was KRW 170 trillion to KRW 172 trillion in sales and KRW 89.3 trillion to KRW 89.5 trillion in operating profit. Source: Samsung Electronics Official Newsroom

Item	2Q26 provisional figures	1Q26	2Q25	Compared to previous quarter	Year-on-year comparison
Sales	171.0 trillion won	133.87 trillion won	74.57 trillion won	+27.7%	+129.3%
operating profit	89.4 trillion won	57.23 trillion won	4.68 trillion won	+56.2%	+1,810.3%

This number is extremely unusual for a single quarter’s operating profit. The expression “world’s largest” may be subject to consideration depending on the comparison target and accounting standards, but it is clear that it is an overwhelmingly large number, at least compared to the normal quarterly profits of global mega-caps.

What is important is not the operating profit itself of 89.4 trillion won, but the quality within it. The interpretation presented in the attached research is as follows.

Item	Interpretation
reported operating profit	89.4 trillion won
Adjustment of one-time provisions such as DS performance bonus	The logic that profits from main business should be considered higher
Core OP estimation	Possibility of over 100 trillion won excluding provisions
DRAM ASP	+47% compared to previous quarter
NAND ASP	+66% compared to previous quarter
HBM4	Maintain ramp-up expectations

[Inference] Just looking at the reported OP of KRW 89.4 trillion is strong, but if DRAM and NAND prices rose significantly at the same time and there was an effect of provisioning, the strength of the main business may be stronger than the reported number. This point is the most important signal of Samsung Electronics’ thesis.

3. Stock price and supply and demand: Foreigners sold after good numbers

Despite good performance, the stock price fell. The prices and supply and demand based on local DB are as follows.

date	closing price	Daily fluctuations	Foreign net purchase estimates	Estimated institutional net purchase	Individual net purchase estimate
2026-07-03	309,500 won		-KRW 387.2 billion	+1.3063 trillion won	-KRW 918.4 billion
2026-07-06	318,000 won	+2.75%	-640.9 billion won	+5.8 billion won	+596.4 billion won
2026-07-07	296,000 won	-6.92%	-KRW 1.8207 trillion	-KRW 547.5 billion	+2.3232 trillion won
2026-07-08	277,500 won	-6.25%	-KRW 874.1 billion	+282.7 billion won	+584.7 billion won

The source is the local Kiwoom-based prices_daily and investor_flow_daily tables. On July 8, the foreign ownership rate was recorded at 46.58%.

The message of this table is clear. The market did not believe that Samsung Electronics’ profits had weakened, but that it was a structure in which foreigners reduced their risk and individuals accepted it in stocks that had already risen. On July 7, foreigners and institutions sold together, and individuals bought significantly. Institutions received some on July 8, but foreign selling continued.

Therefore, today’s judgment is “performance is good, but supply and demand are bad.” It is not a basis for selling, but a basis for freezing additional purchases.

4. Four axes of Samsung Electronics thesis

The argument for investing in Samsung Electronics can be organized into four axes.

axis	Content	Current judgment
AI server DRAM·HBM	Demand for high-bandwidth and high-capacity memory for AI servers and GPU/ASIC systems	Strengthening
eSSD·NAND mix	AI server storage and enterprise SSD demand	Strengthening
Hyperscaler CapEx	Long-term investment in cloud and AI data centers	Strengthening
Foundry turnaround	Edge logic·packaging·customer recovery options	Not yet an option

Among these, the key ones are numbers 1 to 3. Foundry is a valuation upside option. For the stock price to hold up now, the foundry does not need to improve immediately. Investments in memory, eSSD, and hyperscaler alone explain the strength of profits. However, additional rerating requires reducing the foundry deficit and confirming HBM4 and HBM4E customers.

5. Signal 1: Legacy DRAM and NAND prices are stronger than expected

The most important bullish signal in the attached research is the memory price trend summarized in the comments of Meritz Securities Commissioner Kim Seon-woo. Since the full text of the original data was not directly compared, this article classifies it as “market comment-based interpretation.” Still, numbers and direction are important in interpreting Samsung Electronics’ investment argument.

The key point is that seasonal demand for the second half of the year began to arrive in July, and while B2B server demand was absorbing supply, B2C smartphone and PC customers also began to feel pressure to secure supplies.

Legacy DRAM

Item	Figures and Content
Nanya 3Q26 DRAM price proposal	DDR4·LPDDR4 +50~90% QoQ
Consumer DRAM price	Above $3/Gb, upward acceleration
supply expansion	Judgment that meaningful supply expansion will be difficult before 2027
Cause	Decrease in investment between 2023 and 2025 after loss in 2023, lack of cleanroom

Legacy NAND

Item	Figures and Content
Windbond·MXIC SLC NAND proposal	3Q26 +70~100% QoQ
price range	$1.9~2.0/Gb
Demand changes	SLC NAND orders increase due to MLC NAND shortage
Interpretation	Price pressure spreads not only to high-performance AI NAND but also to legacy NAND

Supply Structure

Item	Content
DRAM total supply growth rate in 2027	Expected 20~25%
Server demand growth rate next year	50~70% expected
Bottleneck causes	Low-end/legacy shortage occurs due to concentration on high-end memory

[Inference] If this data is correct, Samsung Electronics’ bull case is wider than SK Hynix’s HBM pure exposure. Samsung Electronics has a discount factor for being behind in HBM, but it also receives price increases for general DRAM, server DRAM, eSSD, NAND, and legacy products. In other words, “the frame of disparaging Samsung Electronics based solely on HBM” is weakened.

6. Signal 2: It is difficult to say that AI CapEx has decreased yet

Hyperscaler investment also supports Samsung Electronics’ thesis. According to numbers shared by Goldman Sachs charts, hyperscaler CapEx grows as follows:

Year	Hyperscaler CapEx
2024	$227 billion
2025	$406 billion
2026E	$740 billion
2027E	$996 billion

The forecast for 2027 is about 4.4 times that of 2024. This number is also in line with the bond and capital increase flows covered in Big Tech Financing Relay. Amazon, Alphabet, and Meta are even using corporate bonds and stock issuances to invest in AI infrastructure at a level that is difficult to cover with internal cash flow alone.

What is important for Samsung Electronics is that “AI CapEx does not end at GPU.” Investing in an AI data center is not just about purchasing GPUs. Server DRAM, HBM, eSSD, NAND, substrate, power, network, and cooling will also increase.

AI CapEx Items	Connection with Samsung Electronics
GPU·ASIC server	HBM, server DRAM
data center storage device	eSSD, NAND
High-density server proliferation	High capacity DDR5, LPDDR series, storage
Rack-level AI factory	Increased memory payload
Long-term investment cycle	Strengthening visibility in 2027 rather than short-term performance in 2026

[Inference] If hyperscaler CapEx increases by 2027, it is difficult to conclude that Samsung Electronics’ memory profits will immediately decline from the peak in one quarter. The problem is not whether there is no demand, but how much the market has already reflected that demand in the price.

7. Signal 3: AI server supply chain is not yet broken

Taiwan AI server supply chain data included in the attached research is also important.

Enterprise	Latest Signals	Interpretation
Wistron	June sales +10.9% MoM, +53.8% YoY	GB300 NVL72, B300 shipment increases
Wiwynn	June sales +32.5% MoM, +29.8% YoY	AMD Helios enters, prepares for mass shipment of 3Q ASIC AI servers
Unimicron	ABF·HDI substrate revenue record high	ABF expansion CapEx increase mentioned

This data does not fit the claim that “demand for AI servers has declined.” If server OEMs, ODMs, and substrate companies still show strong shipments and sales, memory demand is unlikely to disappear easily.

물론 이 데이터만으로 삼성전자 실적을 바로 산출할 수는 없다. But the direction is clear. AI 서버 공급망은 아직 멈추지 않았다. 오히려 GPU, ASIC, 서버, 기판, 메모리, storage가 함께 움직이는 구조다.

8. 소음 1: TrendForce spot 데이터는 약하지만 붕괴는 아니다

There are also opposing signs. TrendForce spot market comments were summarized as an atmosphere where trading was quiet and buyers were waiting.

Item	spot change
DDR5 16Gb	+0.50%
DDR5 eTT	0.00%
DDR4 16Gb 3200	+3.42%
DDR4 eTT	-1.51%
DDR4 8Gb 3200	+2.88%

Looking at this table, it is excessive to say, “Memory prices have fallen.” 대부분은 상승 또는 보합이고, 일부 eTT 품목만 하락했다. 더 정확한 표현은 “거래는 한산하고, spot 시장은 가격 발견 중”이다.

The important distinction is spot and contract.

Category	meaning	Relevance to Samsung Electronics thesis
Spot	Short-term transaction prices in distribution and module markets	Quickly reflects sentiment and short-term inventory
Contract	Large customer long-term contract price	Server DRAM, HBM, eSSD more important to profits

삼성전자 thesis의 핵심은 spot이 아니다. 핵심은 contract 가격, server DRAM, HBM, eSSD 주문, hyperscaler 수요다. 진짜 위험 신호는 spot 거래 부진 자체가 아니라 다음 네 가지가 함께 나오는 경우다.

Contract price cut
Slowdown in server DRAM orders
HBM·eSSD order slowdown
주요 고객의 inventory correction 코멘트

Currently, all four have not been confirmed. Therefore, it is correct to view TrendForce data not as bear confirmation but as a signal of anticipation digestion and price discovery.

9. Noise 2: Morgan Stanley peak logic is only partially correct

Morgan Stanley-type peak thesis should not be ignored either. The key point is that the speed of improvement in memory indicators is approaching its peak, and inventory normalization and price increases may slow down. AI trade 안에서는 반도체보다 Alphabet, Amazon 같은 AI cloud 쪽을 선호한다는 관점도 제시됐다.

This claim is partly correct. This is especially important for short-term stock prices.

Morgan Stanley-style warning	Areas to acknowledge
Improvement speed peak possibility	Stock prices are more sensitive to slope than level.
Price increases slowing	3Q 이후 상승률이 낮아지면 멀티플 압박 가능
Semiconductor overcrowding position	Sell-the-news is possible even with good performance
AI cloud preferred	Possibility of rotation towards consumers rather than semiconductors

However, extending this to the logic of structural selling is weak. There is too much opposing data.

Opposite logic	Evidence
Continued expansion of AI investment	Hyperscaler CapEx expected to increase by 2027
Prolonged supply shortage	Legacy shortage intensifies due to transition to cleanroom and high value-added products
Server demand growth	Server demand expected to increase by 50~70% next year
Domestic securities firms continue to buy	Target prices maintained or raised, including KB 600,000 won, Daol 585,000 won, NH 530,000 won, and Hana/iM 480,000 won

[Inference] The Morgan Stanley peak thesis is valid as a warning, “Don’t chase now.” However, the basis for “discarding Samsung Electronics’ thesis” is insufficient.

10. Noise 3: ASML and TSMC are high bar events

The next gateway is ASML and TSMC. ASML’s official financial calendar displays Q2 2026 financial results on July 15, 2026. TSMC’s official IR page announces the 2Q26 earnings conference on July 16, 2026. Source: ASML financial calendar, TSMC 2Q26 quarterly results

The problem is that even if good performance is achieved, it is impossible to know whether the stock price will rise. Samsung Electronics’ 2Q26 provisional results are already a preview. If the high bar is high, even good numbers seem lacking.

Here’s what to look for in ASML and TSMC:

Event	checkpoint	Connection with Samsung Electronics
ASML 2Q26	bookings, EUV·High-NA demand, Chinese regulatory comments	Confirmation of semiconductor capex expectations and equipment demand
TSMC 2Q26	AI/HPC revenue, CoWoS·advanced packaging, CapEx guidance	Identifying AI accelerator and HBM packaging bottlenecks
Market reaction	SOXX, NVDA, MU, TSM, ASML stock price reaction	Check if it is sold or relieved with good performance

As Mizuho commented, the fact that Samsung Electronics’ stock price fell despite its good preliminary results shows the possibility of a global IT momentum shift. So ASML and TSMC are events that see “how the market digests good numbers” rather than “confirming the numbers.”

11. Noise 4: Press multiple for interest rates and exchange rates

Even if Samsung Electronics’ thesis is good, the stock price may be depressed if the discount rate increases. The interest rate environment based on the attached research is as follows.

Item	level
US 10-year note	About 4.58%
US 30-year note	About 5.07%
Japanese 10-year note	About 2.88%
Japanese 40-year note	About 4.01%

These numbers do not undermine the demand thesis. However, PER can be lowered. A typical section where stock prices fall despite strong profits is “the section where EPS rises but the multiple falls faster.”

The same goes for exchange rates. A weak won may be beneficial to Samsung Electronics’ performance. However, from a foreigner’s perspective, it increases foreign exchange losses, foreign exchange hedging costs, and the risk of Korea’s share. The foreign ownership rate of 46.58% and continuous selling on July 8 are signals that this part needs to be checked.

12. Noise 5: Foreign outflow and leverage piping

The attached research includes comments on foreign outflow and leverage piping based on Bloomberg and Telegram. The figures presented here, such as foreign outflow in the Korean market since the beginning of the year -$100.16 billion, foreign securities outflow in March -$36.55 billion, and USD/KRW 1,554 won, need to be compared with official statistics by collection range. Therefore, in this article, only directional signals are used.

Still, the message is important.

Observation	Interpretation
Foreigners reduced Korea	Rather than avoiding Korea, it may be a memory peak-out bet.
Domestic institutions and individuals absorbed the rebound	It is a domestic liquidity absorption market rather than a foreign-led rally.
Single stock leverage/short gamma structure	Forced selling and hedging flows during declines can exaggerate prices.
Exchange rate rise	It is a barrier to additional purchases by foreigners.

[Inference] The short-term plunge is more likely to be the result of memory peak-out betting, liquidation of leverage positions, and reduction of foreign risks, rather than the collapse of Samsung Electronics’ main business. However, we cannot conclude that this is not a structural risk. We need to check whether foreign spot sales stop.

13. Korean market performance is also not good

This is not just a problem for Samsung Electronics. As of 06:10 on July 8, the local KR integrated screener classified the Korean market as NEUTRAL.

Item	figures
Proportion of stocks above the 50-day line	15.6%
Proportion of stocks above the 200-day line	21.5%
Passed items	17
New Passed	5
Breakaway	7
Breakout	0

Newly passed stocks were APR, Handsome, GS Retail, Hana Financial Group, and KT&G. Stocks that left included TSE, YG-1, and Samsung Electronics.

13 passed the quality compounder screen, and at the top were Gigabis, SK Hynix, VM, AP, and PSK. Only three companies, APR, Seegene, and Kolmar Korea, passed the smart money + quality condition.

The meaning of this data is simple. Market breadth is weak. Even if Samsung Electronics’ thesis is good, it is not widely accepted by the entire Korean market. In this environment, additional purchases should not be made by reporting only profit data. We need to see whether supply and demand and market width return.

14. Then what should I do?

The current conclusion is “strong holding, freeze on additional purchases.”

action	Judgment
holding	possible. The thesis is intact.
Additional Purchase	Still waiting. Need to confirm supply/demand, ASML, TSMC, IR
Sell	It’s not just data now. The profit direction actually improved.
Short term trading	After confirming the slowdown in foreign selling and the global semiconductor response from July 15th to 16th

There are four additional purchase conditions:

Foreign spot sales of Samsung Electronics are slowing down, and programmatic selling pressure is decreasing.
ASML bookings and TSMC AI/HPC·CoWoS·CapEx comments do not indicate a slowdown in AI investment.
DS core profit, DRAM·NAND contract price, HBM4·HBM4E customer progress, and eSSD orders are confirmed in Samsung Electronics IR.
The sluggish spot market does not spread to contract price cuts or a slowdown in server DRAM orders.

The conditions for invalidation are as follows:

Invalidation Conditions	meaning
Contract DRAM/NAND price decline	price thesis undermined
Server DRAM·HBM·eSSD order slowdown	AI memory demand thesis undermined
ASML·TSMC strongly talks about CapEx slowdown	Hyperscaler investment thesis weakened
Foreign selling and won weakening intensify simultaneously	Pressure to reduce Korea’s proportion
Market breadth further worsens	Even if Samsung Electronics is good, supply and demand absorption is weakened

15. Conclusion

Samsung Electronics’ 2Q26 provisional performance is not a number that weakens the thesis. Rather, it is the opposite. Operating profit of KRW 89.4 trillion, DRAM ASP +47%, NAND ASP +66%, soaring legacy memory prices, hyperscaler CapEx continuing until 2027, and AI server supply chain sales all strengthen the case for investing in Samsung Electronics.

However, today’s stock prices and supply and demand have damaged the timing. Foreigners sold, and individuals received. Interest rates are high, exchange rates are a burden, and high bar events such as ASML and TSMC remain. Market width is also weak.

So the conclusion is clear.

Samsung Electronics has not become worse. However, today’s market called for “the next good number and supply and demand to believe in that number” rather than “a good number.” Now is not the time to abandon the thesis, but rather to protect the position, stop making additional purchases, and check for the next gateway.

Basis classification

Category	Content
Fact	Samsung Electronics’ official 2Q26 provisional sales of 171 trillion won and operating profit of 89.4 trillion won. Comparative figures for 1Q26 and 2Q25. ASML July 15th, TSMC July 16th earnings schedule. Samsung Electronics price, supply and demand, and foreign ownership rate in local DB.
Inference	Samsung Electronics’ thesis was strengthened, but the short-term timing was judged to be damaged. TrendForce spot data is interpreted as price discovery, not bear confirmation. Morgan Stanley’s peak thesis is a catch-up warning and not a basis for structural selling.
Speculation	A scenario in which hyperscaler CapEx increases to approximately $1 trillion by 2027 and can support Samsung Electronics’ memory demand for a longer period of time.
Blocked	Comparison of Bloomberg foreign outflow figures with official statistics, confirmation of the full text report by Meritz·TrendForce, 2Q26 actual profit by Samsung Electronics DS division, and volume and margin by HBM4·HBM4E customer.

Evidence Ledger

Item	Source
Samsung Electronics 2Q26 provisional performance	Samsung Electronics Official Newsroom
ASML 2Q26 Schedule	ASML financial calendar
TSMC 2Q26 Schedule	TSMC 2Q26 quarterly results
Samsung Electronics price/supply	Local Research DB: `prices_daily`, `investor_flow_daily`, `foreign_ownership_daily`, as of 2026-07-08
Memory price·AI server·foreign outflow comments	Attached research packet. Uncollated items in the full text are separated from the main text as market comments or directional signals
Related prior analysis	Korea Invest Insights’ Samsung Electronics 2Q26 preview, Micron FY3Q26, Big Tech earnings call, Big Tech financing, NVIDIA similar quarter analysis

Which NVIDIA Quarter Does Samsung's Earnings Selloff Resemble?

Wed, 08 Jul 2026 14:30:00 +0900

This is a follow-up to Samsung 2Q26 preview, Micron FY3Q26 earnings, NVIDIA elasticity and the Korea HBM cycle, late-July Big Tech calls and the memory thesis, and hyperscaler financing and memory bottlenecks. Related hubs: Exclusive Analysis and AI HBM.

TL;DR

Samsung Electronics’ 2Q26 preliminary earnings were strong on the surface. Revenue was about KRW 171 trillion and operating profit was about KRW 89.4 trillion. Sales rose 27.74% quarter over quarter and operating profit rose 56.21%. Operating profit was up 1,810.26% year over year. These figures were disclosed by Samsung Electronics.

Yet the stock fell 6.92% on July 7 and another 6.25% on July 8. The two-day decline was 12.7%.

That is not the same as the famous NVIDIA August 2024 sell-on reaction. NVIDIA fell 6.4% on the first reaction day in August 2024, but rebounded 1.5% the next day. Samsung did not bounce. It fell again.

Across 17 NVIDIA post-earnings reactions, the closest analogue is NVIDIA’s Q4 FY26 report on February 25, 2026. NVIDIA fell 5.5% on the first reaction day and 4.2% the next day, for a two-day decline of 9.4%. The reaction-vector distance from Samsung’s path is 2.55, the smallest in the sample.

The key point is not that a strong earnings print is useless. The point is that a strong print alone was not enough once the market started asking whether AI infrastructure growth was slowing. NVIDIA’s large post-print drawdowns eventually recovered, but the recovery came only after a new catalyst directly killed the fear that had caused the drawdown. For Samsung, the equivalent catalyst is late-July Big Tech CapEx guidance.

The practical conclusion is Wait. This is not a mechanical dip-buy setup yet. The next gates are late-July hyperscaler CapEx commentary, DRAM/NAND price momentum, and Samsung’s clean 3Q operating-profit path.

1. What was strange about Samsung’s earnings reaction

The numbers were strong.

Samsung Electronics announced 2Q26 preliminary earnings on July 7, 2026. The company disclosed the following consolidated figures.

Metric	2Q26 preliminary	1Q26	2Q25
Revenue	about KRW 171T	KRW 133.87T	KRW 74.57T
Operating profit	about KRW 89.4T	KRW 57.23T	KRW 4.68T
Revenue growth		+27.74% QoQ	+129.31% YoY
Operating profit growth		+56.21% QoQ	+1,810.26% YoY

Samsung also disclosed estimate ranges of KRW 170T to KRW 172T for sales and KRW 89.3T to KRW 89.5T for operating profit. Korean disclosure rules required the company to publish a single median figure.

The stock reaction was the problem. Samsung closed at KRW 318,000 on July 6, KRW 296,000 on July 7, and KRW 277,500 on July 8. Based on yfinance daily close data:

Date	Close	Daily return	Read-through
2026-07-06	KRW 318,000	+2.75%	pre-announcement base
2026-07-07	KRW 296,000	-6.92%	announcement-day selloff
2026-07-08	KRW 277,500	-6.25%	no bounce
Two-day cumulative		-12.74%	post-print path

Compared with the June 18 high of KRW 362,500, the July 8 close was already 23.45% lower. Samsung was not falling from a calm setup. It was already volatile before the earnings print, and the good number failed to stop the selling.

The key is that the market was no longer asking only how much Samsung earned. It was asking whether the pace of earnings improvement could keep accelerating.

2. Why compare Samsung with NVIDIA

Samsung is a memory company. NVIDIA is a GPU platform company. Their businesses are different. But both have been core price signals for the AI infrastructure cycle.

NVIDIA moved first when the AI infrastructure boom began. As bottlenecks moved downstream into HBM, DRAM, eSSD, substrates, and power infrastructure, Samsung and SK hynix began facing the same market question: is the next leg still accelerating?

Question	Meaning
Did earnings improve?	confirmed result
Did earnings beat the bar?	surprise versus expectations
Will the next quarter be better?	guidance and pricing
Is the growth rate still rising?	the part stocks care about most
Will hyperscalers keep spending?	the final payer for AI memory demand

Samsung’s two-day selloff came from that last layer. The print was strong, but expectations were already high. Some investors had expected operating profit above KRW 90T. At the same time, the market was worried about possible hyperscaler CapEx moderation. Korean single-stock leverage products, futures, and ETF flows amplified the move.

This analysis therefore does not compare Samsung’s business model with NVIDIA’s business model. It compares post-earnings price paths and the fear behind those paths.

3. Timing adjustment: NVIDIA reports after the close

NVIDIA usually reports after the US market close. The earnings information is not reflected in the announcement-day close. The real reaction appears in the next regular trading session.

Samsung’s guidance was reflected in the Korean session on the announcement day. So the timing has to be aligned as follows.

Company	Reporting style	First trading session with full reaction
Samsung Electronics	reflected in the Korean session	announcement day
NVIDIA	after-market release	next trading day

In this note, NVIDIA’s reaction D0 means the first trading day after the after-market release. NVIDIA’s D+1 means the following trading day. This aligns NVIDIA’s D0 with Samsung’s July 7 and NVIDIA’s D+1 with Samsung’s July 8.

Without this adjustment, the comparison would be wrong.

4. NVIDIA’s 17-quarter post-earnings reaction table

The table below combines NVIDIA quarterly revenue, growth rates, and post-earnings stock reactions. Revenue and report timing are sourced from NVIDIA IR and SEC 8-K releases. Stock returns are calculated from yfinance daily close data.

Quarter	Report date	Revenue	QoQ	YoY	Reaction D0	D+1	Two-day	T+20
Q1 FY23	2022-05-25	$8.3B	n/a	n/a	+5.2%	+5.4%	+10.8%	+0.9%
Q2 FY23	2022-08-24	$6.7B	-19.3%	n/a	+4.0%	-9.2%	-5.6%	-27.1%
Q3 FY23	2022-11-16	$5.9B	-11.9%	n/a	-1.5%	-1.7%	-3.1%	+6.5%
Q4 FY23	2023-02-22	$6.1B	+3.4%	n/a	+14.0%	-1.6%	+12.2%	+27.5%
Q1 FY24	2023-05-24	$7.2B	+18.0%	-13.3%	+24.4%	+2.5%	+27.5%	+38.2%
Q2 FY24	2023-08-23	$13.5B	+87.5%	+101.5%	+0.1%	-2.4%	-2.3%	-12.9%
Q3 FY24	2023-11-21	$18.1B	+34.1%	+206.8%	-2.5%	-1.9%	-4.3%	-3.7%
Q4 FY24	2024-02-21	$22.1B	+22.1%	+262.3%	+16.4%	+0.4%	+16.8%	+33.9%
Q1 FY25	2024-05-22	$26.0B	+17.6%	+261.1%	+9.3%	+2.6%	+12.1%	+33.3%
Q2 FY25	2024-08-28	$30.0B	+15.4%	+122.2%	-6.4%	+1.5%	-5.0%	-1.2%
Q3 FY25	2024-11-20	$35.1B	+17.0%	+93.9%	+0.5%	-3.2%	-2.7%	-10.4%
Q4 FY25	2025-02-26	$39.3B	+12.0%	+77.8%	-8.5%	+4.0%	-4.8%	-13.3%
Q1 FY26	2025-05-28	$44.1B	+12.2%	+69.6%	+3.2%	-2.9%	+0.2%	+15.0%
Q2 FY26	2025-08-27	$46.7B	+5.9%	+55.7%	-0.8%	-3.3%	-4.1%	-2.2%
Q3 FY26	2025-11-19	$57.0B	+22.1%	+62.4%	-3.2%	-1.0%	-4.1%	-6.6%
Q4 FY26	2026-02-25	$68.1B	+19.5%	+73.3%	-5.5%	-4.2%	-9.4%	-8.6%
Q1 FY27	2026-05-20	$81.6B	+19.8%	+85.0%	-1.8%	-1.9%	-3.6%	-5.7%

The key message is clear. Strong results did not always lift the stock. As the AI cycle matured, the market became more sensitive to the change in growth rate than to the level of revenue.

5. The closest analogue is NVIDIA Q4 FY26

Samsung’s path was:

Company/event	D0	D+1	Two-day
Samsung 2Q26 preliminary earnings	-6.92%	-6.25%	-12.74%

We compare this two-day vector with all 17 NVIDIA quarters.

Rank	NVIDIA quarter	Report date	D0	D+1	Two-day	Distance from Samsung
1	Q4 FY26	2026-02-25	-5.46%	-4.16%	-9.39%	2.55
2	Q3 FY24	2023-11-21	-2.46%	-1.93%	-4.34%	6.21
3	Q3 FY26	2025-11-19	-3.15%	-0.97%	-4.10%	6.48
4	Q1 FY27	2026-05-20	-1.77%	-1.90%	-3.64%	6.74
5	Q2 FY26	2025-08-27	-0.79%	-3.32%	-4.09%	6.79

Q4 FY26 is the clear match. It is not just the numbers. The drivers were similar.

NVIDIA’s Q4 FY26 revenue was $68.1B, up 73% year over year and 20% quarter over quarter. The next-quarter revenue guide was $78B. The print was strong. But the stock fell 5.5% on the first reaction day and 4.2% on the next day.

The market was asking whether the next step could still get better. Customer concentration, hyperscaler spending durability, AI CapEx peak risk, and competition were all in focus.

Samsung faced the same structure. KRW 89.4T of operating profit was strong, but investors had already moved the bar higher. The market wanted proof that the growth rate was still accelerating, not only proof that the current profit level was high.

6. Why the August 2024 NVIDIA sell-on is the wrong template

The intuitive comparison is NVIDIA Q2 FY25 in August 2024. NVIDIA fell 6.4% on the first reaction day. That resembles Samsung’s July 7 decline.

But the next day changes the conclusion.

Case	D0	D+1	Read-through
NVIDIA Q2 FY25, August 2024	-6.4%	+1.5%	one-day selloff, then bounce
NVIDIA Q4 FY25, February 2025	-8.5%	+4.0%	one-day selloff, then strong bounce
NVIDIA Q4 FY26, February 2026	-5.5%	-4.2%	two-day selloff
Samsung 2Q26	-6.92%	-6.25%	two-day selloff

The August 2024 and February 2025 NVIDIA reactions were dip-and-bounce. Samsung was dip-and-continue. That distinction matters. A one-day sell-on can be a dip-buy setup. A two-day selloff can mean the market has not finished digesting the fear.

7. Did NVIDIA recover immediately after large post-print selloffs?

No.

There were three NVIDIA quarters with a first-reaction-day decline worse than 5%: Q2 FY25, Q4 FY25, and Q4 FY26. All three were still below the pre-announcement price after 20 trading days.

NVIDIA quarter	D0	T+5	T+10	T+20
Q2 FY25, August 2024	-6.4%	-14.6%	-5.1%	-1.2%
Q4 FY25, February 2025	-8.5%	-10.6%	-11.8%	-13.3%
Q4 FY26, February 2026	-5.5%	-6.4%	-4.9%	-8.6%

All three fell further by T+5. All three failed to recover the pre-announcement price by T+20.

The sample is small, so this is not a deterministic rule. But it is enough to reject the simple idea that strong earnings plus a large drop automatically equals an immediate buying opportunity.

8. The real low and recovery took weeks

The second attached analysis goes further. It asks when NVIDIA actually bottomed and when it recovered.

Anchor	Pre-print base	Low	Max drawdown	Time to low	Breakeven	New high	Low-to-later-high
Q2 FY25, August 2024	$125.55	$102.78 on 2024-09-06	-18.1%	6 sessions	2024-10-07, 5.5 weeks	2024-10-14, 6.5 weeks	+45.3% to 2025-01-06
Q4 FY25, February 2025	$131.24	$94.29 around 2025-04-04 to 2025-04-08	-28.2%	27 sessions	2025-05-14, 11 weeks	2025-06-25, 16 weeks	+119.6% to 2025-10-29
Q4 FY26, February 2026	$195.55	$165.17 on 2026-03-30	-15.5%	23 sessions	2026-04-14, 6.5 weeks	2026-04-24, 8 weeks	+42.7% to 2026-05-14

The pattern is important.

First, the low did not arrive immediately. Even the fastest case took six trading sessions. The slower cases took more than a month.

Second, breakeven took 5.5 to 11 weeks.

Third, the rebound was eventually large. But it did not come from cheapness alone. It came after a catalyst removed the fear behind the selloff.

9. The rebound trigger had to kill the exact fear

9-1. August 2024: the Fed addressed macro fear

The August 2024 drawdown was about macro fear. Weak labor-market data raised recession concerns. Regulatory noise also hurt sentiment.

The rebound trigger was the Fed’s 50bp cut on September 18, 2024. The fear was rates and recession. The Fed directly addressed that fear.

9-2. February 2025: tariff fear required a policy reversal

The February 2025 post-print decline later deepened because of tariff shock. NVIDIA moved toward a 12-month low in early April.

The rebound trigger was the 90-day tariff pause. A policy shock created the fear; a policy reversal removed it.

9-3. February 2026: GTC addressed AI CapEx peak fear

This is the most relevant anchor for Samsung.

NVIDIA’s Q4 FY26 print was strong, but the market worried that AI investment might peak in 2026. Hyperscaler concentration, AI return-on-investment questions, and customer spending durability all mattered.

The rebound catalyst was GTC 2026. Jensen Huang said NVIDIA had visibility into at least $1T of demand or orders through 2027, compared with the prior year’s $500B figure through 2026. Data Center Dynamics and Constellation Research reported the comments.

This directly addressed the fear. The market worried that AI CapEx was peaking; NVIDIA answered that demand through 2027 was much larger.

Even then, the bottom came after GTC, not on GTC day. The stock bottomed on March 30, about two weeks later.

10. Applying the lesson to Samsung

Samsung’s selloff fear has two parts.

First, memory pricing and HBM demand may have been priced too quickly.

Second, if hyperscalers slow AI infrastructure spending, memory pricing and volume could both weaken.

Therefore, the rebound does not need another reminder that 2Q earnings were good. The market already knows that. The rebound needs evidence that the fear is wrong.

Samsung selloff fear	Required catalyst	Timing
Big Tech AI CapEx slowdown	Alphabet, Microsoft, Meta, and Amazon maintaining or raising 2026 to 2027 CapEx	late-July 2026 earnings calls
Memory price-growth slowdown	DRAM/NAND spot and contract-price momentum staying firm	weekly checks and 3Q negotiations
HBM catch-up doubt	HBM4/HBM4E qualification, production shipments, NVIDIA mix	2H26 IR and earnings calls
2Q one-off cost confusion	clean 3Q operating-profit path after bonus/provision adjustments	3Q preview and 3Q earnings
flow-driven liquidation	foreign and program-selling slowdown, lower leverage-product volatility	daily flow checks

The most important gate is late-July Big Tech earnings. If Alphabet, Microsoft, Meta, and Amazon keep or raise AI infrastructure spending plans, Samsung’s selloff can be reclassified as a flow correction rather than demand damage.

If two or more hyperscalers talk about slowing the pace of AI infrastructure investment, the NVIDIA Q4 FY26 template becomes more relevant and the correction can last longer.

11. Why the Fed put is weaker this time

Two of NVIDIA’s prior recoveries had macro or policy support.

In August 2024, the Fed cut rates by 50bp.

In early 2025, a tariff pause reversed a policy shock.

In February 2026, GTC provided a company-driven demand reset.

Samsung now has less help from the first two channels. Since June 2026, strong US labor data and inflation concern have made rates a constraint again. The market is not confidently pricing a near-term Fed rescue.

That makes the third channel more important: direct demand confirmation. Late-July hyperscaler CapEx commentary can play the same role for Samsung that GTC played for NVIDIA in March 2026.

12. Practical view: wait for proof, then scale in

The conclusion is not a strong sell call. It is also not an automatic buy call.

The stance is Wait.

A strong print followed by a two-day selloff can be a signal that the market has moved from current profit level to next growth rate. That is a different regime.

The entry gates are:

Gate	Meaning	If confirmed
late-July Big Tech CapEx maintained or raised	final demand for AI memory remains intact	selloff fear weakens
DRAM/NAND price momentum holds	2Q pricing logic continues into 3Q	3Q earnings support improves
Samsung clean 3Q QoQ remains healthy	underlying earnings path survives provision noise	quality of earnings improves

If two of these three gates pass, the case for staged entry improves. If all three weaken, the correction can extend.

The main mistake would be to assume that record earnings automatically lead to a bounce. NVIDIA’s history says otherwise. The rebound required a catalyst that killed the specific fear behind the selloff.

13. Samsung versus SK hynix

This note focuses on Samsung, but the read-through also matters for SK hynix.

Samsung has a huge earnings base, a lower multiple, and HBM4/HBM4E catch-up optionality. But investors also have to deal with HBM purity, foundry and S.LSI losses, bonus provisions, and foreign-flow pressure.

SK hynix has cleaner HBM exposure and a stronger position in NVIDIA’s supply chain. But it is already priced as the winner. If Big Tech CapEx remains strong, SK hynix is the trend-reacceleration candidate, while Samsung is the re-rating candidate where the Micron/NVIDIA read-through has been partly erased.

Lens	Samsung Electronics	SK hynix
Strength	earnings level, low multiple, HBM catch-up option	HBM purity, customer trust, leadership position
Burden	HBM certification, foundry/S.LSI, provision interpretation	high expectations, crowded-winner risk
Rebound condition	clean 3Q earnings and HBM customer proof	Big Tech CapEx durability and HBM allocation
Style	under-reflected alpha candidate	trend re-acceleration candidate

Both depend on the same final question: will hyperscalers keep spending?

14. Conclusion: Samsung needs better disproof, not just better earnings

Samsung’s 2Q26 preliminary earnings were strong. But the stock fell sharply for two straight days. Compared with 17 NVIDIA post-earnings reactions, Samsung’s path looks much closer to NVIDIA Q4 FY26 in February 2026 than to the August 2024 one-day sell-on.

In that NVIDIA case, the company posted strong numbers but failed to recover the pre-announcement price for about a month. The recovery began only after GTC 2026 directly challenged the AI CapEx peak fear with a $1T through-2027 demand statement.

Samsung’s equivalent proof point is not another reminder that 2Q earnings were strong. It is hyperscaler confirmation that AI infrastructure spending remains firm into 2027.

That is why the late-July Big Tech earnings calls matter. Until then, the right discipline is to wait for evidence rather than buying only because the stock has fallen after a record print.

Evidence classification

Fact

Item	Evidence
Samsung 2Q26 preliminary earnings	Revenue about KRW 171T and operating profit about KRW 89.4T, Samsung official newsroom, 2026-07-07
Samsung comparison periods	1Q26 revenue KRW 133.87T and operating profit KRW 57.23T; 2Q25 revenue KRW 74.57T and operating profit KRW 4.68T, Samsung official newsroom
Samsung price path	July 6 KRW 318,000, July 7 KRW 296,000, July 8 KRW 277,500, yfinance daily close calculation
Samsung two-day reaction	July 7 -6.92%, July 8 -6.25%, two-day -12.74%
NVIDIA Q4 FY26 event	February 25, 2026 event confirmed on NVIDIA IR
NVIDIA Q1 FY27	Revenue $81.6B, up 85% YoY and 20% QoQ, NVIDIA IR financial reports
GTC 2026 $1T comment	Demand/order visibility through 2027 reported by Data Center Dynamics and Constellation Research
NVIDIA price-path calculations	yfinance daily closes, 17 quarters aligned with next-day reaction for after-market releases

Inference

Item	Interpretation
Samsung/NVIDIA comparison	This is a price-path and fear comparison, not a business-model comparison.
Q4 FY26 similarity	Two-day path is the closest, and the fear set also overlaps: expectations, CapEx durability, and customer concentration.
Late-July Big Tech calls	If AI CapEx concern drove the Samsung selloff, hyperscaler guidance is the most direct disproof.
Wait stance	The print was strong, but the stock is asking about growth-rate durability. Confirmation should come before aggressive buying.

Speculation

Item	Possible path
Positive path	Big Tech CapEx holds or rises, DRAM/NAND price momentum holds, and Samsung’s clean 3Q earnings path improves.
Negative path	Two or more hyperscalers slow AI infrastructure spending or memory price momentum weakens.
Recovery timing	NVIDIA analogues suggest breakeven can take weeks, not days. For Samsung, mid-August to September is more realistic if catalysts turn positive.

Blocked

Item	Why it is blocked
Samsung 2Q26 segment detail	Preliminary earnings do not yet disclose DS, foundry, or S.LSI detail.
Late-July hyperscaler CapEx guidance	Not yet reported.
Samsung clean 3Q QoQ	Requires later earnings and IR detail after provision adjustments.
Exact consensus beat for all 17 NVIDIA quarters	Some quarters are verified, but a uniform consensus data set for every quarter was not fully obtained.

Sources and calculation notes

Samsung Electronics, Earnings Guidance for Second Quarter 2026
NVIDIA Investor Relations, Financial Reports
NVIDIA Investor Relations, 4th Quarter FY26 Financial Results
Data Center Dynamics, Jensen Huang says NVIDIA has one trillion dollars in orders through 2027
Constellation Research, Nvidia GTC 2026 and $1 trillion order flow
Stock calculations: yfinance daily closes for 005930.KS and NVDA, from 2022-05-20 to 2026-07-08. For NVIDIA, after-market releases are aligned so the next regular trading day is reaction D0.

The Big Tech Financing Relay: AI Capex Has Not Been Cut, And The Memory Bottleneck Looks Stronger

Tue, 07 Jul 2026 21:45:00 +0900

This is a follow-up to the late-July Big Tech earnings-call memory-thesis preview, the Micron FY3Q26 review, the 2027 semiconductor consensus-payability work, and the AI data-center capex bottleneck series.

TL;DR

The Big Tech financing relay says one thing clearly: AI infrastructure capex has not been cut. If anything, it has outgrown the old assumption that elite hyperscalers could fund everything internally.

Top hyperscaler 2026 capex guidance now sits around $695bn to $750bn. That compares with about $410bn in 2025 and about $245bn in 2024, implying roughly three consecutive years of 60% plus growth. The more important change is funding mix. The path has moved from internal cash to bonds, multi-currency bonds, off-balance-sheet structures, and now equity issuance.

For memory, the implication is cleaner than for hyperscaler equities. Microsoft has tied roughly $25bn of its 2026 capex to component-price inflation. Meta has cited component pricing, especially memory, as one reason for higher 2026 capex. Amazon has also warned about component and memory-cost pressure. The buyers are absorbing the price burden. Memory suppliers receive it.

The public-equity read-through is therefore not “own hyperscalers simply because they are spending.” It is more specific: watch the memory and HBM bottleneck nodes that can turn that spending into ASP, margin, and long-term contract visibility.

The Financing Timeline

Timing	Issuer	Type	Size	Interpretation
2025-10-30	Meta	Bonds	$30bn	Long-term AI and data-center funding
2025-11	Amazon	Bonds	about $15bn	First major return to US dollar bonds in years
2025-11	Alphabet	Bonds	about $17.5bn	Pre-funding AI infrastructure
2026-02	Alphabet	Bonds	about $32bn	Multi-currency deal including a 100-year tech bond
2026-02	Oracle	Debt-led funding	$25bn	Part of a larger cloud-infrastructure funding plan
2026-03-10/11	Amazon	Bonds	about $54bn	Dollar plus euro record-scale funding
2026-04-30	Meta	Bonds	$25bn	Issued just after Q1 results and higher capex guide
2026-05-12	Amazon	CHF bonds	CHF 2.82bn	Currency diversification
2026-06-01/03	Alphabet	Equity	$84.75bn	The decisive signal: dilution accepted for AI compute
2026-07-07	Amazon	Reported bond launch	$25bn plus	Reported eight-tranche US IG deal. Final terms not yet confirmed

The July 7 Amazon deal should not be mixed with the March record financing. As of publication, the final July 7 size, tranche spreads, new-issue concession and order book were not confirmed in public sources. It is therefore treated as a reported latest financing step, not a fully settled fact.

Capex Is Still Rising

Company	2026E capex	YoY	Key point
Amazon	about $200bn	about +50%	AWS, custom chips and data-center buildout
Microsoft	about $190bn	+61%	Roughly $25bn attributed to component-price inflation
Alphabet	$180bn to $190bn	about 2x	2027 expected to rise substantially again
Meta	$125bn to $145bn	about 2x	Memory and component pricing cited as pressure points
Oracle	volatile	high	Cloud buildout is highly capital-intensive
Top 5	$695bn to $750bn	about +67%	Versus about $410bn in 2025 and $245bn in 2024

The absolute dollar amount is still rising. The risk is the second derivative: even if 2027 capex reaches $1tn, the growth rate will likely slow from the 2025-2026 pace. That matters more for hyperscaler equity multiples than for memory suppliers. If memory keeps taking a larger share of the server bill of materials, memory revenue can still grow faster than total capex.

Why Funding Mix Matters

The escalation path is visible:

Internal cash.
Straight bonds.
Multi-currency bonds.
SPVs, private credit and off-balance-sheet structures.
Equity issuance.

Alphabet’s June 2026 equity raise is the key tell. This is not a weak company. It is one of the strongest cash-flow machines in the world. If it chooses dilution to fund AI compute, the capital intensity of the AI buildout has moved beyond the old internal-cash model.

Microsoft looks different. Its visible public benchmark funding has been less aggressive than Alphabet’s or Amazon’s, which may point to stronger internal funding capacity rather than lower AI capex ambition.

The Memory Read-Through

The key change is not simply that memory prices are up. It is that hyperscalers are not cutting demand in response. They are increasing budgets, signing long-term contracts and raising capital.

Signal	Memory implication
Microsoft component-inflation comment	More dollars are needed for similar capacity
Meta memory-price comment	Memory is moving the whole capex budget
Amazon component-cost warning	Costs are higher, but capex continues
Micron SCAs	Customers are locking in long-term supply
HBM4/HBM4E transition	Higher value per accelerator and more wafer intensity

This is why hyperscaler equities and memory suppliers should be separated. Hyperscalers bear the cost, FCF drag and payback risk. Memory suppliers receive the pricing and allocation benefit.

Public Equity Checklist

For Samsung Electronics, the questions are DS core profit, HBM4/HBM4E customer qualification, 3Q DRAM and NAND pricing, and the scale of foundry/System LSI losses.

For SK Hynix, the questions are HBM4 long-term contract structure, 2027 allocation, and whether US-listed access through the SKHY ADR broadens the investor base without distorting common-share pricing.

For Micron, the question is whether the SCA structure becomes the benchmark for how investors value memory-cycle durability.

For hyperscalers, the clean stance is more cautious: demand is real, but FCF, debt, dilution and payback period risk are also real.

Scenarios

Scenario	Condition	Memory read-through
Bull	Amazon order book strong, late-July calls maintain or lift 2027 capex, memory costs remain absorbed	HBM and server DRAM pricing durability strengthens
Base	2026 capex maintained, 2027 growth continues but slows, memory-price increases moderate	Memory thesis remains intact but stock reactions stay selective
Bear	Capex guides cut, payback concerns rise, component inflation causes deployment delays	ASP thesis weakens until contract pricing is confirmed
Structural damage	Customers reduce memory content per server, HBM contract prices fall, bond demand weakens	Memory thesis needs full re-underwriting

Final View

The Big Tech financing relay confirms that AI infrastructure demand remains real. But it is not a simple hyperscaler-equity bullish signal. The more precise read-through is that memory is one of the bottlenecks taking a larger share of AI infrastructure dollars.

The next tests are the final terms of Amazon’s July 7 deal, the late-July Big Tech calls, Samsung’s July 30 earnings call, SK Hynix’s 2Q call, and 3Q/4Q memory contract-price data.

Sources

Late-July Big Tech Earnings Calls and the Memory Thesis Scenario Map

Tue, 07 Jul 2026 20:59:50 +0900

Context This is a follow-up to Micron FY3Q26 earnings, Samsung Electronics 2Q26 preview, Who pays for 2027 semiconductor consensus?, NVIDIA’s elasticity as a memory-cycle guide and the 1H26 AI infrastructure bottleneck review. Related hubs: Exclusive Analysis and AI HBM Hub.

TL;DR

The late-July earnings calls from Alphabet, Microsoft, Meta and Amazon are the next stress test for the memory thesis. My probability map is: strengthening 50%, unchanged 35%, weakening 15%. Within the weakening bucket, true structural damage is only around 4%.

The key question is not whether Big Tech cuts capex. The four companies are still constrained by compute, power and component availability. The real questions are narrower: whether 2027 capex falls meaningfully short of the market’s roughly $1 trillion expectation, and whether memory pricing plus memory’s share of server BOM stay intact.

The execution rule is simple: do not chase before the calls. Wait for the checklist. Korean memory stocks have already risen sharply over the last year, so the upside from good news can be smaller than the downside from a bad surprise.

Samsung’s July 7 Update

Samsung Electronics released its 2Q26 preliminary result on July 7. It guided to revenue of about KRW 171 trillion and operating profit of about KRW 89.4 trillion. The official ranges were KRW 170-172 trillion for revenue and KRW 89.3-89.5 trillion for operating profit. Source: Samsung Electronics 2Q26 earnings guidance.

That is broadly in line with the earlier reported OP range of KRW 87-92 trillion. It does not settle the debate. The important event is the July 30 Samsung earnings call, where investors need to separate reported OP from core semiconductor profit, check 3Q DRAM and NAND pricing tone, and listen for HBM4/HBM4E customer qualification and allocation comments.

What the Event Judges

Local timing	Event	Status
July 7	Samsung Electronics 2Q26 preliminary result	Officially released
Around July 22-23 US	Alphabet 2Q26 call	Aggregator estimate, official IR date still needs confirmation
Around July 28 US	Microsoft FY4Q26 call	Aggregator estimate, official IR page still needs confirmation
Late July	SK Hynix 2Q26 result	Aggregators conflict, official page needs confirmation
Around July 29 US	Meta 2Q26 call	Aggregator estimate
Around July 30 US	Amazon 2Q26 call	Aggregator estimate
August	SK Hynix Nasdaq ADR listing	Pending SEC process

The headline question is not “will Big Tech cut spending?” The real question is whether 2027 spending growth is still steep enough to support the memory earnings path.

Big Tech Capex Baseline

Company	2025 actual	2026 plan	Growth	2027 commentary	Demand evidence
Alphabet	$91.4B	$180-190B	About 2x	CFO said 2027 spending should rise materially above 2026	Cloud backlog of $462B, roughly doubled sequentially
Microsoft	About $118B, inferred	About $190B	+61%	Double-digit revenue and operating income growth, but no explicit capex figure	RPO $627B, +99% YoY
Meta	$72B	$125-145B	+87.5% at midpoint	No number, flexible language	Contractual commitments rose $107B in 1Q
Amazon	$125B	About $200B	+60%	No number	AWS backlog $364B, Anthropic contract separate
Total	About $410B	About $710B	+73%	Market expectation above $1T, +41%

Alphabet is the strongest setup. It raised 2026 capex guidance, said 2027 would be meaningfully higher, and later announced an $80B capital raise to expand AI infrastructure and compute. Microsoft is the key swing factor: RPO is huge, but sequential RPO growth needs to reaccelerate. Meta is two-sided: it raised guidance because of component and memory costs, but it also kept flexibility language for 2027. Amazon has the largest absolute capex plan, but its free cash flow pressure makes its tone worth watching.

Memory Price Facts

Quarter	DRAM	NAND	Interpretation
1Q26	+55-60%	+33-38%	Server DRAM was even stronger
2Q26	+58-63%	+70-75%	Peak rate of increase
3Q26 forecast	+13-18%	+10-15%	Still rising, but at a slower pace
4Q26 reference	PC DRAM +3-8%	Not specified	A consumer-side reference point

Source: TrendForce 3Q26 price outlook.

The 3Q deceleration is not a demand collapse. It reflects base effects, consumer affordability limits and a rising share of long-term contracts. The more important point is that server memory remains tight. If Big Tech capex grows more slowly while memory takes a larger share of server BOM, memory revenue can still grow faster than aggregate capex.

Micron SCA Matters

Micron disclosed 16 strategic customer agreements in its FY3Q26 materials. The contracts run from 2026 through 2030, include take-or-pay structures, and cover about 20% of DRAM volume and roughly one-third of NAND volume. Minimum contracted revenue is around $100B, and customer deposits plus commitments are around $22B. Source: Micron prepared remarks.

This is the most important underappreciated change in the cycle. Big Tech is not merely accepting higher memory prices. It is signing multi-year, non-cancelable supply assurance contracts. That lowers the historical cyclicality discount for memory earnings. It does not eliminate risk, because price ceilings may cap upside on contracted volume, but it changes the downside profile.

For SK Hynix, the key question is whether its long-term contracts have similar floors without the same upside caps. That detail is not officially confirmed. It should be the first question on the SK Hynix 2Q call.

Scenario Map

Scenario	Probability	Meaning for memory stocks
Strengthening	50%	Earnings revision cycle extends. Upside depends on how explicit 2027 capex becomes.
Neutral	35%	2026 plans remain intact, but 2027 language stays cautious. Short-term headline noise likely.
Weakening	15%	Split into narrative weakening and structural damage.
Narrative weakening	About 11%	Stocks correct, but price and contract data stay healthy. Could become an entry window.
Structural damage	About 4%	Thesis reset. Examples: 2026 capex cut, memory-content redesign, digestion language from multiple companies.

Call-by-Call Update Rules

Sequence	Result	Strengthening probability update
Alphabet	Plan maintained, 2027 reaffirmed, backlog up	50 → 55
Alphabet	2027 language retreats or backlog stalls	50 → below 40
Microsoft	$190B plan maintained, supply tightness, 2027 growth tone, RPO growth resumes	55 → 62
Microsoft	Normalization or single-digit growth tone	55 → 45
Meta	Plan maintained, commitments rise, cloud monetization becomes official	62 → 66
Meta	April-style flexibility language returns with numbers	62 → 50
Amazon	$200B maintained, memory price absorbed	66 → 68-70
Amazon	Amazon-only pacing comments	66 → 60, not enough to kill the thesis

If two or more companies cut plans or use digestion language, the table no longer applies. That becomes a thesis reset exercise.

Korean Memory Checklist

This section is a public research checklist, not a trading instruction. The late-July Big Tech calls matter here because they help test the Korean memory thesis.

Name	Role	What to verify
Samsung Electronics	Korean memory core with DRAM and NAND volume leverage	DS core profit, 3Q memory price assumptions and HBM4 customer qualification.
SK Hynix	HBM purity and long-term contract structure	HBM shipment, ASP durability, contract pricing structure and HBM4 allocation.

Phrases To Listen For

Direction	Phrases
Positive	supply cannot keep up with demand / customer commitments rising / backlog rising / component or memory price pressure absorbed by budgets / 2027 capex expansion
Negative	digestion phase / lower utilization / investment pacing / optimization / delayed customer commitments / easing component price pressure / memory supply improving earlier than expected / memory content redesign

Kill Conditions

One item triggers reassessment. Two or more start the thesis-reset process.

Any of the four companies lowers 2026 capex plans.
Microsoft RPO or Alphabet Cloud backlog turns negative sequentially.
Companies use “digestion” or “pacing purchases to power availability” language.
A concrete server redesign reduces memory content per server because of price.
TrendForce cuts 3Q memory price growth back to single digits while Samsung DS core profit disappoints.

The fourth item is the real kill condition. Supply-driven forced reductions are different from price-driven voluntary memory-content cuts.

Evidence Classification

Facts: Samsung’s 2Q26 preliminary result, Samsung’s July 30 call schedule, Alphabet’s 2026 capex range and AI infrastructure capital raise, Microsoft’s $190B capex run-rate and $627B RPO, Meta’s $125-145B plan and $107B commitment increase, Amazon’s roughly $200B plan and AWS backlog, TrendForce’s 3Q price outlook, Micron’s FY3Q26 SCA disclosures.

Inferences: the 50/35/15 scenario map, the idea that +73% to +41% capex deceleration is already embedded, the view that Meta’s cloud report is more monetization than capex-cut signal, and the view that long-term contracts reduce memory cyclicality.

Speculation: SK Hynix contract upside without price ceilings, possible Microsoft financing, the exact timing of a 2027 digestion phase, and Meta’s cloud economics.

Blocked: verified 2027 consensus multiples for Samsung and SK Hynix, Marvell’s exact share in Trainium 3 and 4, SK Hynix contract details, company-level 2027 capex consensus distributions, memory supplier allocation by hyperscaler and some official late-July earnings dates.

The conclusion is practical: the odds favor a stronger memory thesis, but price already reflects part of that. The right way to trade the event is to update probabilities as evidence arrives, not to make a blind prediction before the calls.

Dukovany And The Westinghouse Settlement: Contract Risk Has Fallen, Margin Risk Remains

Sat, 04 Jul 2026 22:25:00 +0900

Context This memo follows Team Korea’s nuclear opportunity under the US investment package, Hyundai E&C and Woojin in US nuclear expansion and construction EPC rerating.

TL;DR

The Westinghouse IP risk that could have blocked the Czech Dukovany 5 and 6 project has fallen meaningfully. Westinghouse, KEPCO and KHNP announced a global settlement on January 16, 2025, and the Dukovany contract was signed in June 2025.

That does not mean Korean nuclear exports are now free to expand across Europe without constraints. The remaining issue is economic: how much Westinghouse can take as a toll, how broad the reported market-access restrictions are, how much local content is required, and who carries fixed-price or schedule risk.

The confirmed project numbers are large. World Nuclear News reported the Dukovany 5 and 6 EPC contract for two APR1000 units at about CZK 407 billion, or USD 18.6 billion. KEPCO E&C announced an A/E services contract worth EUR 725 million, about KRW 1.2 trillion. Doosan Enerbility announced a KRW 320 billion steam turbine and turbine-control contract through Doosan Skoda Power.

The investment conclusion is simple: contract survival risk has fallen, but margin risk remains. The question has moved from “can Korea win and sign the project?” to “how much profit remains for Korea’s listed value chain after Westinghouse, localization and project risk?”

1. What Has Been Resolved

In August 2024, Westinghouse protested CEZ’s selection of KHNP as preferred bidder for the Dukovany project. Westinghouse argued that KHNP’s APR1000 and APR1400 designs used Westinghouse-licensed System 80 technology and that KHNP lacked the right to sublicense or retransfer that technology to CEZ and local suppliers. Westinghouse also raised US export-control issues under DOE Part 810.

The landscape changed on January 16, 2025. Westinghouse announced a global settlement with KEPCO and KHNP to resolve the IP dispute. The company said the agreement would allow both parties to move forward with new nuclear reactor deployment, while the terms remained confidential.

The US Department of Energy welcomed the settlement on the same date. CEZ’s Dukovany project page then confirmed that the contract for two new Dukovany units was signed in June 2025 after the legal injunction was lifted.

Issue	Current state	Investment reading
Westinghouse IP dispute	Global settlement announced in January 2025	The main legal obstacle to signing has fallen
Settlement terms	Confidential	Cost, market access and technology-independence risks remain
Dukovany 5 and 6	Contract signed in June 2025	The project is alive
EU foreign subsidy review	EC did not open an in-depth FSR investigation, according to KHNP/WNN	One regulatory risk has eased
EDF Czech litigation	EDF lost and said it would not pursue further Czech legal action	Tender-procedure risk has eased

2. Why Westinghouse Did Not Simply Lose

The US court dismissal of Westinghouse’s Part 810 declaratory-judgment pathway should not be read as a full defeat on the economics. Finnegan’s legal analysis notes that the court found Part 810 did not give Westinghouse an independent judicially remediable right for that claim. It did not decide that Westinghouse’s broader contractual or IP economics were worthless.

This is why the settlement matters. Westinghouse may have moved from a legal blocker to an economic toll collector.

Simplistic view	More accurate view
Westinghouse lost	The Part 810 court route was limited, but settlement economics may remain
Korea can export freely	Dukovany is open, but regional market access may still be constrained
IP risk disappeared	signing risk fell, but cost and margin risk remain

3. The Reported Toll

Yonhap reported, citing industry sources, that KHNP may be restricted from new nuclear bids in North America, the EU, Britain, Japan and Ukraine, with the Czech Republic carved out as an exception. The same report said the agreement may require about USD 650 million of Westinghouse goods and services per exported reactor and about USD 175 million of technology license fees per reactor.

Korea Times also reported that the presidential office ordered a review of whether the KHNP, KEPCO and Westinghouse agreement followed proper legal basis and procedure.

These reported terms are not public settlement terms. They should be treated as reported but unconfirmed. Still, the direction is important.

Reported condition	Meaning	Investment implication
USD 650 million of Westinghouse goods and services per reactor	Part of the headline order may flow to the US supply chain	Korean revenue capture is lower than headline EPC value
USD 175 million technology fee per reactor	IP toll	Project margin is reduced
Reported limits in North America, EU, UK, Japan and Ukraine	Czech project may be an exception, not a broad European beachhead	Do not extrapolate Dukovany into all-Europe TAM
Technology independence verification	New reactor or SMR exports may also face constraints	The premium for independent Korean exports should be moderated

Using only the reported figures, two reactors would imply USD 1.65 billion of Westinghouse goods, services and license fees. Against a USD 18.6 billion project, that is roughly 8.9%. This is not a confirmed contract calculation, but it is a useful way to think about margin leakage.

4. Project Size Is Not Profit

World Nuclear News reported that the Dukovany 5 and 6 EPC contract covers two APR1000 units at a projected cost of CZK 407 billion, or USD 18.6 billion, with construction targeted to start in 2029.

That is a large project. But nuclear EPC value is not the same as shareholder value. The key variables are price escalation, localization, financing, schedule guarantees, delay penalties, equipment scope and Westinghouse-related costs.

Driver	What to watch
Price	EPC value, escalation clauses, FX and inflation treatment
Quantity	Dukovany 5 and 6 plus possible Temelin options
Cost	Westinghouse toll, local content, fixed-price risk, financing and schedule guarantees

The most common mistake is to buy the headline order while ignoring the cost stack.

5. Value Chain Implications

KEPCO E&C

KEPCO E&C is the cleanest direct exposure. It announced a EUR 725 million A/E services contract with KHNP for Dukovany 5 and 6, equivalent to about KRW 1.2 trillion. The scope includes A/E services, procurement support for auxiliary equipment and licensing activities for two 1,000-MW-class APR1000 units.

This gives KEPCO E&C direct design and licensing exposure. The next question is revenue recognition, staffing intensity, change-order scope and operating margin.

Doosan Enerbility

Doosan Enerbility announced a KRW 320 billion contract with Doosan Skoda Power for steam turbines and turbine-control systems for Dukovany 5 and 6. This is important because it combines Korean nuclear expertise with Czech local manufacturing through Doosan Skoda Power.

Doosan is the strongest beta to Czech nuclear equipment and potential Temelin follow-on work. But the stock already carries nuclear, SMR, gas turbine and AI-power narratives. Margin and follow-on contracts matter more than the headline.

Daewoo E&C

Daewoo E&C can be a construction exposure, but it needs more caution. Construction and EPC have large revenue potential, but also carry schedule, cost and delay-penalty risk. The key question is contract structure, not just participation.

Company	View	Key check
KEPCO E&C	Watchlist	A/E revenue recognition, margin, change orders
Doosan Enerbility	Watchlist / Wait	Equipment margin, Temelin option, additional nuclear scope
Daewoo E&C	Watchlist	Fixed-price exposure, schedule risk and local partner split
KEPCO KPS, Woori Technology, Woojin	Follow-up candidates	Actual O&M, I&C or instrumentation contracts

6. Catalysts And Invalidation

Upside catalysts:

Catalyst	Meaning
Additional subcontract announcements	Korean listed value-chain capture expands
Clarity on Czech state-aid and financing structure	Financing risk falls
Temelin 3 and 4 options	Dukovany becomes a repeat project, not a one-off
KHNP-Westinghouse cooperation model	Market-access uncertainty may fall
Pre-2029 permitting, site and supply-chain milestones	Project execution risk falls

Invalidation risks:

Risk	Impact
Settlement terms are worse than reported	Korean nuclear export economics deteriorate
Fixed-price and schedule risk fall heavily on Korean parties	EPC and construction stocks de-rate
Temelin and broader Europe remain closed	Multiple expansion based on Europe TAM fades
Key subcontracting shifts to Czech or Westinghouse supply chains	Korean listed-company capture shrinks
Reported margins disappoint in quarterly earnings	Headline-order premium reverses

Bottom Line

Dukovany is alive. The legal and procedural risks that could have blocked the contract have fallen. But the equity question has changed. Investors should no longer ask only whether Korea won the project. They should ask how much of the project becomes high-quality profit after Westinghouse tolls, localization, fixed-price risk and financing terms.

The best framing is:

Dukovany confirms Korea’s nuclear export survival. The next rerating depends on margin and repeatability.

SK Square Investment Memo: When Does SK Hynix Cash Flow Become SK Square Cash?

Sat, 04 Jul 2026 21:40:00 +0900

Context This memo follows SK Hynix F-1/A and F-6, SK Hynix ADR price path, SK Hynix HBM market share 2026 and Micron FY3Q26 earnings.

TL;DR

SK Square should not be framed as the investment vehicle for the entire SK Group. SK Inc. is the official group holding company and portfolio coordinator. For public-market investors, SK Square is better defined as:

A listed AI and semiconductor capital allocator built on SK Hynix equity value, NAV discount compression and a call option on future SK Hynix special cash dividends.

The key variable is not SK Hynix net income by itself. Higher SK Hynix earnings lift SK Square’s equity-method income and NAV, but they do not automatically become SK Square cash. Cash reaches SK Square through SK Hynix dividends, stake sales, collateralization or SK Square’s own portfolio exits.

Current view: Wait / Conditional Buy. The buy case strengthens when SK Hynix free cash flow converts into cash dividends, SK Square receives that cash, and the company allocates it to buyback cancellation or Hynix-linked AI and semiconductor investments.

1. Correct Framing

SK Inc. is the official holding company for SK Group. It manages the broader group portfolio across energy, AI, semiconductors and other businesses. SK Square is different. It is an ICT and semiconductor investment company whose NAV is overwhelmingly driven by SK Hynix.

Item	SK Inc.	SK Square
Role	Group holding company and portfolio manager	ICT and semiconductor investment company
Main exposure	Broad SK Group portfolio	SK Hynix stake and AI/semi portfolio
Investor thesis	Group restructuring and holding-company discount	SK Hynix NAV, cash conversion, NAV discount compression
Main risk	Conglomerate complexity	Hynix dividends may not convert into SK Square cash

This distinction matters. Calling SK Square the group’s treasury overstates the case. The cleaner thesis is a listed AI and semiconductor allocator tied to SK Hynix cash-flow conversion.

2. NAV Is Mostly SK Hynix

SK Square’s official NAV page showed the following as of July 2, 2026.

Item	Value
Total NAV	KRW 324.7T
SK Hynix stake value	KRW 319.52T
Net cash	KRW 0.79T
NAV per share	KRW 2,460,288
SK Square share price	KRW 1,525,000
NAV discount	About 38.0%

The discount is calculated as 1 - 1,525,000 / 2,460,288. SK Square holds 146,124,531 SK Hynix shares, or 20.5%.

3. Earnings Are Not Cash

SK Hynix earnings affect SK Square through accounting income and NAV. But SK Square needs actual cash for buybacks, cancellations, dividends or new investments.

SK Hynix’s fixed dividend for 2025-2027 is KRW 1,500 per share. Applied to SK Square’s 146,124,531 shares, this equals KRW 219.2B, or about KRW 0.219T per year. That is about 28% of SK Square’s KRW 0.79T net cash. It is meaningful, but not enough to re-rate a holding company with KRW 319.52T of SK Hynix stake value.

The real option is special or additional cash dividends.

Total SK Hynix cash dividend	SK Square pre-tax receipt	Estimated after-tax receipt
KRW 1T	KRW 0.205T	About KRW 0.194T
KRW 5T	KRW 1.025T	About KRW 0.969T
KRW 10T	KRW 2.050T	About KRW 1.937T
KRW 20T	KRW 4.100T	About KRW 3.875T

The after-tax estimates use a simplified Korean dividend received deduction framework. A domestic parent owning at least 20% and less than 50% of a domestic subsidiary may exclude 80% of dividends from taxable income. Applying a 27.5% tax rate to the remaining 20% implies about 5.5% leakage.

4. NAV Discount Math

Holding NAV per share constant at KRW 2,460,288, the implied share price changes materially as the discount narrows.

NAV discount	Implied price
50%	KRW 1,230,144
45%	KRW 1,353,158
40%	KRW 1,476,173
38%	KRW 1,525,379
35%	KRW 1,599,187
30%	KRW 1,722,202
25%	KRW 1,845,216

SK Square has stated an objective of reducing its NAV discount to 30% or less by 2028. That is a target, not a guarantee. The market needs evidence: Hynix cash dividends, SK Square buyback cancellations, non-core exits and disciplined AI/semi investments.

5. Shareholder Return And Capital Allocation

SK Square’s 2023-2025 shareholder-return policy uses at least 30% of recurring dividend income and part of meaningful investment gains for shareholder returns. Methods include buybacks, cancellations and cash dividends.

At a high NAV discount, buyback cancellation is powerful. If a company trades at a 38% NAV discount, buying back stock is economically similar to buying KRW 1 of NAV for KRW 0.62. That means low-return investments are hard to justify unless they have clear Hynix-linked strategic value.

6. U.S. Cash Is Not Automatically Trapped Cash

SK Hynix’s U.S. customer revenue is large. Reported 2025 U.S. customer revenue was KRW 66.885T, about 69% of total revenue. SK hynix America reportedly generated KRW 58.693T of sales and KRW 235.9B of net income, implying a net margin of about 0.40%.

That low margin suggests the U.S. sales entity may be a limited-margin distributor rather than the place where most economic profit accumulates. Cash can return to the Korean parent through product payment, transfer-pricing settlement, intercompany payments, dividends, royalties or other structures.

The important question for SK Square is not whether U.S. revenue is large. It is whether SK Hynix parent-level free cash flow grows and becomes cash dividends.

7. SKHY ADR Read-Through

SK Hynix is pursuing a U.S. ADR listing. Reuters reported that the company may raise up to $29.4B, with proceeds for Korean semiconductor factories and EUV equipment.

For SK Square, the ADR is positive for investor access and potentially for SK Hynix’s valuation. But it is not a direct cash-in event for SK Square. The proceeds go to SK Hynix. SK Square benefits through NAV, valuation and future dividends, while also facing potential ownership dilution from new share issuance.

8. Investment Ideas

Idea	Thesis	Required proof
Hynix FCF-to-cash option	Hynix earnings matter only when they become SK Square cash	FCF after capex and meaningful cash dividends
Listed AI/semi allocator	SK Square becomes a disciplined Hynix-linked investment platform	Investments tied to Hynix ecosystem, not generic AI hype
NAV discount compression	Discount narrows from about 38% toward 30%	Buyback cancellation, non-core exits, capital discipline
U.S. trapped-cash misunderstanding	Large U.S. revenue does not mean large trapped U.S. profit	Parent-level FCF and tax-efficient cash flow

9. Practical View

Security	View	Reason
SK Square	Wait / Conditional Buy	NAV discount is meaningful, but cash conversion and capital allocation need confirmation
SK Hynix	Watchlist	Source of cash flow and NAV
SK Inc.	Separate thesis	Official holding company, but less direct Hynix cash-flow exposure

Entry conditions:

Condition	What to see
Wider discount	Updated NAV discount at or above 45% while the Hynix HBM thesis remains intact
Cash conversion	Hynix special cash dividend or early additional shareholder return, followed by SK Square buyback cancellation or Hynix-synergy investment
Allocator trust	Non-core exits, AI/semi discipline, no group-support capital allocation

Invalidation:

Risk	Meaning
Hynix earnings are strong but FCF is absorbed by capex and working capital	No SK Square cash conversion
Hynix returns capital mainly through buybacks rather than cash dividends	Limited SK Square cash inflow
SK Square uses Hynix dividends for group support	Holding-company discount widens
Low-return investments take priority over buyback cancellation	Capital allocation trust declines
ADR dilution outweighs lower cost of capital	SK Square’s Hynix stake economics weaken

Final View

SK Square is not cash-rich today. Official net cash is KRW 0.79T. But it owns KRW 319.52T of SK Hynix stake value. If SK Hynix AI-memory earnings convert into free cash flow and cash dividends, SK Square can receive trillions of won over time.

The strongest thesis is not “SK Hynix is good, therefore SK Square is good.” It is more specific: SK Hynix FCF must become cash dividends, SK Square must allocate that cash to buyback cancellation and Hynix-linked investments, and the NAV discount must compress structurally.

Sources

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

SK Gas LNG Cold Energy: How Much Of The AI Data-Center Cooling Option Can Become Earnings?

Sat, 04 Jul 2026 14:48:00 +0900

Related context This follows SK Gas vs SK Oceanplant, AI data-center power bottleneck map, and AI data-center CapEx bottlenecks. The related hubs are the Exclusive Analysis Hub and the Korean Semiconductor Value Chain Hub.

TL;DR

The right wording matters. The confirmed fact is not that SK Gas has disclosed a commercial AI data-center cooling project. The confirmed fact is that an SK Gas engineer published the technical logic for connecting LNG cold energy with AI data-center cooling through SK Group’s DevOcean channel.

The January 2, 2025 DevOcean post, authored by Byeon Jae-jun of SK Gas, says he works on a task force linking LNG cold-energy utilization technology with AI data centers. The post explains that LNG produces about 0.2 MWh of cold energy per ton when warmed from about -162°C liquid to about 5°C gas, and that most of this cold energy is currently discarded through seawater heat exchange.¹

The investment conclusion is precise:

Item	View
Confirmed	SK Gas-linked technical logic exists. Ulsan LNG infrastructure and the SK-AWS AI data-center direction are publicly visible.
Investment logic	LNG cold energy is not new cooling energy. It is the reallocation of otherwise wasted low-temperature energy from LNG vaporization into data-center cooling load.
Unconfirmed	Actual commercial contract, target PUE, realized PUE improvement, tariff, capex, payback, AWS/SK economics and SK Gas margin attribution remain undisclosed.

So SK Gas is not a pure “cold-energy theme stock.” The cleaner thesis is undervalued core energy-infrastructure earnings plus an under-modeled AI data-center cooling option.

1. Why Ulsan Matters

LNG cold energy is hard to move over long distances. The key condition is physical proximity between the LNG terminal and the data center. Ulsan may satisfy that condition.

KET, a joint venture between Korea National Oil Corporation and SK Gas, operates an energy storage terminal in Ulsan North Port and began commercial operation of its oil and LNG terminal in 2024.² SK Telecom has also announced that it is working with AWS on an AI data center in Ulsan, using SK Group land and energy supply capabilities.³

SK Gas also has Ulsan GPS, an LNG/LPG dual-fuel combined-cycle power plant. SK Gas sold a 49% stake in Ulsan GPS for KRW 1.2242 trillion while retaining 51% control.⁴ This makes Ulsan a combined power, LNG terminal, cooling and AI data-center campus candidate rather than a generic LNG story.

Cooling is important, but the common claim that cooling is 30-40% of all data-center power should not be applied to all modern hyperscale facilities. IEA data show that cooling and environmental control may be around 7% in efficient hyperscale data centers, while it can exceed 30% in less efficient enterprise data centers.⁵

That distinction matters. LNG cold energy may lower electric chiller load, but the actual PUE benefit depends on facility design, rack density, cooling loop efficiency, pumps, heat exchangers, backup chillers and utilization matching.

3. How The System Would Work

The simplified chain is:

Step	Meaning
LNG storage	LNG is stored near -162°C.
Vaporization	LNG must be warmed and vaporized before use. Cold energy is released in the process.
Recovery	Instead of discarding that cold energy into seawater, a heat-exchange system captures it.
Coolant loop	Refrigerant or brine loops move the cooling effect safely toward the data-center cooling plant.
Data-center cooling	Electric chiller load can be reduced.
Redundancy	Mission-critical data centers still need backup chillers, pumps and control systems.

The value is real only if cold recovery, reliability, capex and contractual economics work together.

4. Investment Idea One: A Cold-Energy Toll Road

If commercialized, SK Gas and KET could turn otherwise wasted cold energy into a cooling service.

Variable	Driver
Price	Avoided chiller electricity cost, water cost and carbon cost can partly become a cold-energy tariff.
Quantity	Ulsan AI data-center MW scale, rack density, LNG vaporization volume and cold-energy utilization rate.
Cost	Heat exchangers, coolant or brine loops, pipes, pumps, control systems, backup chillers and maintenance.

The upside is a potential infrastructure-style toll road without directly buying GPUs. The missing proof is a long-term cold-energy supply contract, minimum volume, tariff, capex payback, target PUE and SK Gas margin attribution.

5. Investment Idea Two: Firm Power Call Option

The cold-energy idea sits on top of a more basic thesis. AI data centers need firm power. They do not only need cheap power. They need power that does not stop.

Ulsan GPS gives SK Gas a firm-power option. The key variables are power-market price, capacity payments, LNG/LPG fuel spreads, plant utilization and SK Gas’s retained 51% exposure after the stake sale.

The KRW 1.2242 trillion price for 49% of Ulsan GPS implies a simple 100% value of roughly KRW 2.5 trillion and a retained 51% value near KRW 1.27 trillion. This is only a simple cross-check, not a full valuation, but it supports the idea that the market may still be applying a large discount to SK Gas’s energy infrastructure assets.

6. What Is Still Missing

Item	Status
Target PUE for the Ulsan AI data center	Not disclosed
LNG cold-energy share of total cooling load	Not disclosed
Cold-energy supply tariff	Not disclosed
Long-term contract structure	Not disclosed
Cold-recovery capex and payback	Not disclosed
AWS/SK economic allocation	Not disclosed
SK Gas EBITDA or net-income contribution	Not disclosed
Proprietary technology or patent scope	Not disclosed

This table is the core of the analysis. The technical logic is credible, but the numbers required for an EPS model are not public yet.

Final View

SK Gas’s LNG cold-energy option is not empty. It has technical logic, a relevant Ulsan location and a broader SK-AWS AI data-center context. But it is not yet a confirmed earnings driver.

The right investment stance is Watchlist / conditional buy. SK Gas should be read as an undervalued energy infrastructure company with an AI data-center cooling option. The cold-energy thesis becomes investable only when the market sees a supply contract, PUE improvement, capex payback and SK Gas margin capture.

AI Data Center Power Bottleneck Map: After GnCenergy, VinaTech and Seojin System, Who Is Next?

Sat, 04 Jul 2026 13:40:00 +0900

Related context This post follows AI data-center CapEx and Korean bottlenecks, GnCenergy and AI data-center backup power, VinaTech and Bloom Energy power buffering, SK Gas vs SK Oceanplant, and the H1 2026 AI infrastructure postmortem. The relevant hubs are the Exclusive Analysis Hub and the Korean Semiconductor Value Chain Hub.

TL;DR

AI data-center power is not one trade. It breaks into several layers: generation and backup power, fuel cells and distributed power, ESS and UPS, power conversion and transformers, cables and bus ducts, cooling and EPC. Each layer maps to a different listed Korean supplier.

The first screen is straightforward.

Conclusion	Read-through
Strong direct evidence	GnCenergy, VinaTech and Seojin System have the clearest supply evidence. GnCenergy maps to Korean data-center backup generators. VinaTech maps to Bloom Energy-linked supercapacitors. Seojin System maps to ESS enclosures and racks.
Evidence still weak	HD Hyundai Marine Engine and Hanwha Engine have optionality in data-center engine power, but listed-entity-level supply evidence remains weak. Separate group-level orders from the listed company’s earnings capture.
Underpriced sub-layers	Backup generators, supercapacitors, ESS enclosures, internal power conversion, cable bus and bus ducts may still be less systematically priced than large transformer winners.

The first diligence priority is GnCenergy, VinaTech, Seojin System, Sanil Electric, Iljin Electric, Gaon Cable and SK Gas. Large electrical-equipment names are high quality, but a lot of the rerating is already known. The question now is which sub-bottlenecks turn into repeat revenue.

1. Why Power Is The Bottleneck

The IEA’s Energy and AI report sees global data-center electricity consumption rising from about 415 TWh in 2024 to around 945 TWh by 2030. In the United States, data centers may account for nearly half of electricity-demand growth through 2030.¹

The issue is not only generation. Grid equipment has become a bottleneck. The IEA notes that procurement can now take two to three years for cables and up to four years for large power transformers, about twice as long as in 2021.²

For investors, this means the opportunity set is wider than transformer makers. Backup power, ESS, UPS, internal distribution, cable systems and cooling can all become bottlenecks when AI data centers scale.

2. Value Chain Map

Layer	Products and services	Bottleneck	Listed Korean candidates
Backup and onsite generation	Diesel and gas generators, medium-speed engines, LNG/LPG power	Reliability, tier certification, delivery, fuel supply	GnCenergy, SK Gas, Doosan Enerbility, Hanwha Engine, HD Hyundai Marine Engine, SGC Energy
Fuel cells and distributed power	SOFC, PEMFC, fuel-cell BOP, supercapacitors	Load response, heat use, PPA structure, auxiliary power	VinaTech, Doosan Fuel Cell, SK Eternix, S-Fuelcell, Bumhan Fuel Cell
ESS, UPS and BBU	LFP/NCA batteries, racks, enclosures, UPS, supercapacitors	Fire safety, customer qualification, U.S. manufacturing, container design	Samsung SDI, LG Energy Solution, SK Innovation/SK On, Seojin System, VinaTech
Power conversion and distribution	PCS, DC solutions, switchgear, breakers, special transformers	Hyperscaler approval, density, efficiency	LS ELECTRIC, Sanil Electric, HD Hyundai Electric, Hyosung Heavy Industries
Grid equipment	Power transformers, GIS, reactors, breakers	Production slots, North American certification, lead times	HD Hyundai Electric, Hyosung Heavy Industries, LS ELECTRIC, Iljin Electric, Cheryong Electric
Cables and bus ducts	High-voltage cables, cable bus, bus duct, internal wiring	Copper, certification, installation	Gaon Cable, LS Eco Energy, Taihan Cable, Iljin Electric
Cooling and mechanical	HVAC, immersion cooling, CDU, cleanroom	PUE, heat density, coolant and refrigerant	Shinsung E&G, GST, LG Electronics, S-Oil
EPC and operations	Data-center construction, electrical work, power operations	Permitting, schedule, integrated design	Samsung C&T, HanmiGlobal, Hyundai E&C, DL E&C

3. Evidence-Based Ranking

Rank	Company	Ticker	View	Why
1	GnCenergy	119850	Watchlist / conditional Buy candidate	The most direct Korean data-center backup-generator supplier. Evidence includes Naver, kt cloud, LG CNS and other IDC projects.³
2	VinaTech	126340	Watchlist	A direct Bloom Energy AI data-center supercapacitor supply contract, with overseas revenue around 90% according to the company.⁴
3	Seojin System	178320	Watchlist	ESS enclosures, racks and containers. Samsung SDI America and U.S. ESS supply-chain exposure are the key evidence points.⁵
4	Sanil Electric	062040	Wait / Watchlist	Strong Bloom Energy special-transformer evidence, but valuation may already reflect the theme.⁶
5	LS ELECTRIC	010120	Wait	A complete power-distribution, DC, ESS and transformer platform. The issue is price.⁷
6	HD Hyundai Electric	267260	Wait	Very strong North American data-center equipment evidence, but the large-cap rerating is already advanced.⁸
7	Iljin Electric	103590	Watchlist	Both heavy electrical equipment and cables. The Malaysia data-center cable order is direct evidence.⁹
8	Gaon Cable	000500	Watchlist	U.S. AI data-center cable supply and U.S. capacity expansion are direct signals.¹⁰
9	SK Gas	018670	Watchlist	Ulsan GPS and SK’s Ulsan AI data-center power architecture make it a low-valuation infrastructure candidate.¹¹
10	Samsung SDI	006400	Watchlist	Data-center ESS, UPS and BBU exposure is real, but EV batteries still dominate the equity story.¹²

4. Main Ideas

GnCenergy: Backup power pure-play

GnCenergy is the clearest direct supplier in this screen. AI data centers cannot tolerate downtime, so backup generation is not an old-economy afterthought. It is the last reliability layer behind compute capacity. The risk is project lumpiness, margin volatility and potential environmental regulation against diesel backup generators.

VinaTech and Sanil Electric: Bloom Energy supply chain

The Bloom Energy supply chain is not only about fuel cells. If onsite SOFC power becomes more common for AI data centers, power smoothing and conversion hardware also matter. VinaTech supplies supercapacitors for AI data centers through Bloom Energy, while Sanil Electric has Bloom-linked special-transformer evidence. The key test is whether these become repeat orders rather than one-off projects.

Seojin System: ESS picks-and-shovels

AI data centers need ESS, UPS and BBU systems, but the cell supplier is not the only beneficiary. Enclosures, racks, containers, thermal design and fire-safety structures can be manufacturing bottlenecks. Seojin System is interesting for that reason. The issue is not revenue growth alone. Working capital, net debt and cash-flow quality must also improve.

Cables and bus ducts

Iljin Electric, Gaon Cable and LS Eco Energy sit in a less glamorous but important layer. Data centers need not only external grid connections but also dense internal power distribution. Cable bus, bus ducts and certified high-voltage cables can matter more as rack density rises. Copper pass-through and margin discipline are the main risks.

Energy campus

SK Gas, SK Eternix and Doosan Fuel Cell map to a broader energy-campus thesis. Data centers may value secured power more than cheap power. LNG/LPG generation, renewables, fuel cells and ESS may be bundled into site-level power solutions. SK Gas looks the most investable in this group because Ulsan GPS and the SK Ulsan AI data-center structure are more concrete, while SK Eternix and Doosan Fuel Cell still need direct PPA or commercial-order evidence.

Engine optionality

HD Hyundai Heavy Industries’ U.S. data-center HiMSEN engine order proves that medium-speed engines can be part of the onsite-power solution.¹³ But that does not automatically make HD Hyundai Marine Engine or Hanwha Engine direct beneficiaries. Listed-entity-level orders, production roles and margin capture must be confirmed first.

5. Entry, Catalysts and Invalidation

Group	Entry condition	Catalyst	Invalidation
GnCenergy	Data-center backlog grows, margins hold, valuation stays reasonable	LG Uplus Paju AIDC follow-on, Naver or telecom data-center expansion	Two quarters of slowing orders, margin collapse, diesel backup demand weakens
VinaTech	Bloom revenue recognition, module supply, visible path to profitability	Bloom data-center power expansion, follow-on purchase orders	One-off Bloom order, ongoing losses, no customer diversification
Seojin System	ESS enclosure growth and operating cash-flow improvement together	Samsung SDI America, Fluence or Powin-linked ESS expansion	Working-capital deterioration, debt pressure, low-margin OEM trap
Sanil Electric	New customers beyond Bloom, margin stability, valuation reset	Follow-on special-transformer orders	One-off Bloom order, margin compression, customer concentration
Iljin Electric and Gaon Cable	Repeat data-center cable orders and copper pass-through	U.S. and Southeast Asia data-center cable wins	Copper spike, certification or installation delays, commodity cable mix
SK Gas	More detail on SK/AWS Ulsan power economics	Ulsan AI data-center power contract, UGPS stable operation	Gas spread pressure, power-market regulation, weak PPA economics
Hanwha Engine and HD Hyundai Marine Engine	Listed-entity data-center engine order	North American onsite-generation orders	No order, unclear group margin capture, stock rises only on expectation

Final View

The strongest screen today is GnCenergy, VinaTech and Seojin System. They are not the obvious large transformer winners. They sit in lower-level bottlenecks inside data-center power infrastructure.

Sanil Electric, Iljin Electric and Gaon Cable have direct order evidence, but valuation and margin proof matter. SK Gas is not a pure-play, but Ulsan GPS and SK’s AI data-center power structure make it a low-valuation infrastructure candidate. Samsung SDI and LG Energy Solution have data-center ESS exposure, but EV battery cycles still dominate their equities.

Hanwha Engine and HD Hyundai Marine Engine remain stories ahead of evidence. Engine-based onsite power is a real theme, but listed-company revenue and margin capture must be proven.

The practical stance is not “buy every power stock.” It is to monitor repeat contracts, backlog quality, margin capture and cash flow, then enter selectively when the evidence improves.

VinaTech and Bloom Energy: Who Buffers AI Data Center Power Shocks?

Sat, 04 Jul 2026 10:30:00 +0900

Related context This post follows AI Data Center CapEx and Korean bottlenecks, MLCC and silicon capacitors, AI server passive-component bottlenecks, the expensive-money AI infrastructure frame, and the H1 2026 AI infrastructure postmortem. The relevant hubs are the Exclusive Analysis Hub and the Korean Semiconductor Value Chain Hub.

TL;DR

VinaTech is not primarily a “company that stores a lot of electricity.” The better frame is a company that can absorb transient power shocks between AI data centers and Bloom Energy’s SOFC systems. If Bloom is the on-site generator, VinaTech’s supercapacitor system is the power buffer between that generator and AI servers.

VinaTech signed a data-center supercapacitor-system supply contract with Bloom Energy. The contract value is KRW 41.215B, equal to 50.12% of VinaTech’s 2025 consolidated revenue of KRW 82.229B. The contract period runs from June 30, 2026 to April 10, 2027.¹ For a small company, that is large enough to change the revenue base.

The alpha is not the KRW 41.2B contract itself. The real question is whether this is a one-off Bloom project or the beginning of repeat standard content inside Bloom’s SOFC data-center package. If follow-on purchase orders and system margins appear, VinaTech can be reclassified from a supercapacitor-cell supplier into an AI data-center power-stability system supplier.

The technology moat is medium-high. Supercapacitor cells have global competitors. But customer qualification inside Bloom’s data-center power architecture, plus the shift from cell supply to module, control and software packaging, is a higher barrier than selling a discrete component.

The current stance is Watch, with a possible upgrade to conditional Buy. At the July 3, 2026 close of KRW 84,400, implied market capitalization is about KRW 606B. Volatility is high: the stock hit limit-up on July 1, fell 20.1% on July 2, then fell another 1.9% on July 3. Chasing the first contract is less attractive than waiting for Bloom repeat orders, system-margin proof and customer diversification.

Bottom line

VinaTech is not an AI data-center power-generation stock. It is a possible power-quality and transient-load-buffer stock. If Bloom's SOFC units become more common in AI data centers, the key question is whether VinaTech's supercapacitor system becomes repeat standard content.

0. Analysis Setup

Item	Detail
Company	VinaTech
Ticker	126340 / KOSDAQ
Analysis date	July 4, 2026 KST
Price basis	July 3, 2026 close of KRW 84,400, pykrx and local Kiwoom DB
Core event	KRW 41.2B Bloom Energy data-center supercapacitor-system supply contract
Core question	Can VinaTech become a repeat bottleneck supplier in the AI data-center power chain?
Stance	Watch, with possible upgrade to conditional Buy

This post starts with business position rather than price. Reading VinaTech as an AI data-center power-generation stock is wrong. Bloom Energy’s SOFC, or solid oxide fuel cell, generates the power. VinaTech’s job is to buffer the shock when AI servers suddenly pull or release power.

1. The Bottleneck Is Power Speed, Not Just Power Volume

AI data-center power has two different problems.

The first is power volume. The data center must secure enough electricity through grid interconnection, power plants, PPAs, nuclear, gas, fuel cells or BESS.

The second is power speed and quality. Voltage and current need to remain stable when AI servers abruptly pull more power or drop load. More generation alone does not solve this.

NVIDIA explains that AI training differs from traditional data-center workloads because thousands of GPUs operate in lockstep. The load does not naturally average out. Workloads shift rapidly between idle and high-power states, creating grid-level fluctuations. NVIDIA says the GB300 NVL72 power shelf includes energy storage that can smooth power spikes and reduce peak grid demand by up to 30%.²

In plain language, a traditional data center is like an apartment block where many residents use electricity at different times. AI training is closer to a building where thousands of heavy elevators start and stop together. The grid problem is not only that the building uses a lot of power. It is that it pulls and releases power abruptly.

VinaTech sits in that second bottleneck: transient load, power quality, voltage movement and current shock.

2. Bloom SOFC and VinaTech Supercapacitors Do Different Jobs

Bloom Energy announced a strategic partnership with CoreWeave to deploy fuel cells for on-site power at a high-performance AI data center in Volo, Illinois.³ The point is that AI data centers are trying to secure on-site power rather than wait indefinitely for grid interconnection.

But SOFC is closer to a steady generator than an infinitely fast load-following device.

Research on hybrid SOFC-based DC microgrids shows the issue. The RSC paper explains that a stand-alone SOFC system has difficulty with fast load following, and that battery plus supercapacitor storage can help cover transient power and reduce fuel-starvation risk. Supercapacitors provide high current during sudden load increases, reducing stress on the SOFC system.⁴

The division of labor is simple.

Layer	Simple analogy	Power-system role
Bloom SOFC	On-site generator	Stable long-duration power supply
Battery or BESS	Large water tank	Minute-to-hour energy storage and backup
VinaTech supercapacitor	Shock absorber	Millisecond-to-second peak buffering
PCS, BMS, control software	Valves and automatic controls	Voltage, current, charge-discharge and protection logic

Supercapacitors store less total energy than batteries. But they charge and discharge quickly and have strong cycle life. They are not best understood as long-duration backup power. They are short-duration shock absorbers.

For Bloom SOFC power to support AI data centers, the output must be smoothed into a form AI servers can consume reliably. That is where VinaTech matters.

3. Did VinaTech Actually Enter the Bloom Chain?

The confirmed fact is strong.

VinaTech signed a data-center supercapacitor-system supply contract with Bloom Energy. The contract is KRW 41.215B, equal to 50.12% of 2025 consolidated revenue. The term runs from June 30, 2026 to April 10, 2027. There is no advance payment, and supply will be handled through outsourced production via an overseas subsidiary.¹

The important change is from “cell supply” to “system supply.” The Bell reported that this is VinaTech’s first case of directly supplying an entire supercapacitor system for data centers; prior data-center business was mainly supercapacitor cells.⁵

VinaTech’s own materials point in the same direction. The company says it signed an AI data-center supercapacitor supply contract with Bloom Energy and plans to shift in the first half of 2026 from individual cell sales to modular products. It also describes integrated packages that include PCB and software.⁶

The product-status change matters more than the headline revenue.

Old frame: supercapacitor-cell company
Possible new frame: AI data-center power-stability system supplier

However, this is not yet a long-duration monopoly. We do not know whether the Bloom contract is tied to one project or whether it is becoming standard content in Bloom’s data-center power package.

4. What Is and Is Not Confirmed About Meta

This point requires discipline.

There is no confirmed direct VinaTech-to-Meta supply contract. There is also no confirmed direct Meta-to-VinaTech relationship. The confirmed chain is indirect.

Meta → CoreWeave → Bloom Energy → VinaTech

CoreWeave announced a roughly $21B long-term AI cloud capacity expansion with Meta in April 2026.⁷ Bloom announced on-site fuel-cell deployment for CoreWeave’s AI data center.³ VinaTech signed a data-center supercapacitor-system contract with Bloom.¹

So VinaTech is not a “Meta direct supplier.” The cleaner investment phrase is:

a transient-power-stability supplier inside Bloom systems when AI data-center power bottlenecks lead to Bloom SOFC adoption.

Using Meta’s logo makes the story sound larger, but it weakens the investment analysis. The real alpha is whether VinaTech gets repeat revenue inside Bloom’s system.

5. VinaTech’s Real Bottleneck: Power Rate, Not Energy Amount

Investors need to separate energy from power.

Energy (kWh) is the size of the water tank. Power (kW) is the instantaneous force from the faucet. AI data centers need more than a large tank. They need pipes that do not shake when the faucet opens and closes abruptly.

VinaTech’s supercapacitor system is a transient pressure regulator. When AI servers suddenly need current, the supercapacitor discharges. When server load drops, it absorbs excess energy. Bloom’s SOFC then does not need to ramp aggressively up and down.

The P, Q and C framework looks like this.

Driver	Meaning	What to verify
P, price	System products should carry higher ASP than cells	Does PCB, software and control packaging lift ASP?
Q, volume	Bloom SOFC data-center projects must scale	Does VinaTech content attach to each new Bloom project?
C, cost	System supply adds outsourcing, quality and control cost	Does revenue growth translate into operating margin?

The market can miss the shift by seeing VinaTech only as a supercapacitor-cell company. The contract points to a subsystem supplier role. But the market can also over-read the news. One system contract does not prove standardization. Repeat orders must confirm it.

6. Technology Moat: Application Qualification Matters More Than The Cell

Supercapacitor cells are not a monopoly. The global market includes Maxwell, Skeleton Technologies, Panasonic, Murata, Eaton, Nippon Chemi-Con, LS Materials and VinaTech.⁸

Therefore, simply making supercapacitors is not a strong moat. VinaTech’s moat is elsewhere.

First, it has qualified a product for Bloom’s data-center power system and converted that into a contract. Power-infrastructure products are not swapped like phone accessories. Voltage, current, temperature, lifetime, failure mode, software control, warranty and certification all matter.

Second, VinaTech is moving from cell to module and system. Once cells are connected in series and parallel, balancing, thermal control, protection circuits and control software become critical.

Third, customer qualification itself becomes a switching barrier in data-center power. A power-quality problem can become a server outage. Customers do not casually replace verified suppliers with cheaper unproven parts.

Moat element	Strength	Explanation
Supercapacitor cell manufacturing	Medium-high	High-output, long-life, low-resistance know-how matters, but competitors exist.
Module and system design	High	Voltage balancing, thermal control, protection and software control raise difficulty.
Bloom qualification	High	A supply contract is a strong qualification signal, though exclusivity is not confirmed.
Mass production and quality control	Medium-high	Large customer supply requires delivery, defect control and traceability.
Proprietary source-material monopoly	Medium	Public sources do not support a claim of global source-material monopoly.

Overall technology moat score: 7/10. Cell technology alone looks closer to a six. Including system design adoption inside Bloom’s data-center power system pushes it toward the mid-sevens. This is not an ASML-style equipment monopoly or a TSMC-style process moat. It is an application-qualification moat.

7. Business Moat: Standard Content Would Change The Story

The business moat is more conditional than the technology moat.

The strongest confirmed fact is the KRW 41.2B supply contract. It equals 50.12% of 2025 revenue. For a small company, that can change the base line.¹

But one contract is not enough to call this a durable moat.

First, customer concentration matters. As Bloom revenue grows, VinaTech may become more dependent on Bloom. If Bloom dual-sources or pushes price, margins can be pressured.

Second, system margins are not yet public. System supply has higher revenue, but it also includes PCB, enclosure, outsourcing, quality assurance, logistics and support costs. Gross margin could be better or worse than cell supply.

Third, repeat orders are the key. If KRW 41.2B is a one-off project order, the multiple is limited. If VinaTech becomes standard content in Bloom’s SOFC data-center package, the company changes category.

Business moat item	Current score	What could lift it
Customer access	8/10	Direct Bloom reference is valuable.
Repeat revenue probability	6/10	Follow-on POs could lift this above 8.
Pricing power	5/10	Bloom concentration may cap bargaining power.
Switching cost	8/10	Design-in power systems are costly to replace.
Customer diversification	4/10	Needs proof beyond Bloom.

Overall business moat score: 6.5/10 today, with possible upgrade to 8/10 if repeat orders appear. This is an early strong entry signal, not yet a structural monopoly.

8. How VinaTech Complements Bloom

Bloom SOFC’s advantage is clear: it can provide on-site power before grid interconnection catches up. The CoreWeave project and Brookfield’s expanded $25B AI infrastructure power financing framework show Bloom’s link to the “time-to-power” problem.³⁹

But SOFC is a high-temperature steady operating system. It is strong in stable operation, but AI GPU loads are abrupt. SOFC research highlights a trade-off between fast load following and safe operation because fuel starvation and thermal safety constrain how fast SOFC current should move.⁴

VinaTech complements Bloom’s product.

Bloom does this	VinaTech helps with this
Generates power on site	Buffers transient load so AI servers can consume that power reliably
Operates SOFC stacks stably	Reduces the need for abrupt SOFC output changes
Shortens grid-interconnection dependence	Improves internal power quality and peak handling
Supplies long-duration power	Absorbs millisecond-to-second power shocks

NVIDIA also frames AI-factory power as more than a capacity problem. It says BESS and power controls can manage fast-changing load profiles and improve power quality.¹⁰ VinaTech is not a BESS provider, but it addresses the faster-response layer within the same problem.

9. Price, Flow and Consensus: A Good Story Is Not A Comfortable Entry

VinaTech closed at KRW 84,400 on July 3, 2026. Both pykrx and the local Kiwoom DB show that close. Foreign ownership was 15.54%. Using 1,115,794 foreign-held shares to infer total shares gives roughly 7.18M shares, implying market capitalization near KRW 606B.

The near-term chart is unstable.

Period	Return or status
July 1	KRW 107,700, limit-up
July 2	KRW 86,000, -20.1%
July 3	KRW 84,400, -1.9%
Last 5 trading days	+5.2%
Last 10 trading days	-25.9%
Last 20 trading days	-43.4%
20-day closing high	KRW 132,900 on June 1, 2026
20-day closing low	KRW 80,200 on June 26, 2026

Flow is mixed.

Period	Retail	Foreign	Institution	Real money
Last 3D	+KRW 10.58B	-KRW 14.49B	+KRW 4.13B	+KRW 3.84B
Last 5D	+KRW 11.17B	-KRW 8.32B	-KRW 2.10B	-KRW 2.78B
Last 10D	+KRW 6.44B	-KRW 12.81B	+KRW 7.27B	+KRW 5.22B
Last 20D	+KRW 16.59B	+KRW 11.93B	-KRW 26.82B	-KRW 33.06B

Real money here combines investment trusts, private funds, pensions and insurance. The last three days show some real-money inflow, but the 20-day number remains weak. The news reaction was strong; durable long-only accumulation is not yet confirmed.

Consensus also shows valuation pressure. Local DB Naver consensus shows FY2026 revenue of KRW 145.7B, operating profit of KRW 4.4B, net income of KRW 2.0B, PER of 306.6x and PBR of 8.0x. Whether those estimates fully incorporate the Bloom system contract needs separate verification. Still, the stock is not cheap on reported consensus. This is a repeat-order option, not a value stock.

10. Investment Stance: Watch To Conditional Buy Candidate

VinaTech is not just a theme stock. It has a real Bloom contract and a product-step change from cell to system. AI data-center power volatility is a real structural issue supported by technical sources.

But unconditional buying at the current price is weak.

The stock has already reacted to the first contract.
Follow-on Bloom orders are not yet confirmed.
System-product margins remain blocked.
Customer concentration and outsourced production are real risks.

Item	Stance
Company	VinaTech
Ticker	126340 / KOSDAQ
Idea type	Idiosyncratic Alpha
Current stance	Watch, with possible upgrade to conditional Buy
Core thesis	Possible reclassification into a supercapacitor-system supplier inside Bloom SOFC data-center packages
Core risks	One-off Bloom order, low system margins, customer concentration, dual sourcing, high valuation

11. Entry Conditions

11.1 Follow-on Bloom Purchase Orders

This is the most important condition. If additional Bloom supply contracts appear in H2 2026 or H1 2027, the KRW 41.2B contract can be read as the start of standard content rather than a one-off.

11.2 System-Margin Proof

Revenue growth without margin improvement weakens the thesis. System supply can carry higher ASP, but outsourcing, PCB, control and quality costs also rise. Gross margin and operating margin matter more than headline sales.

11.3 Customer Diversification

If VinaTech expands beyond Bloom into other SOFC providers, data-center power-infrastructure suppliers, PSU vendors or rack-power systems, concentration risk falls. That would move the thesis from “Bloom supplier option” to “AI data-center power-stability platform.”

11.4 Price Stabilization After The News Spike

The July 1 limit-up and July 2 selloff show very high volatility. Without repeat orders, buying the spike means buying theme beta more than alpha.

12. Catalysts And Invalidation

Catalyst	Checkpoint
Catalyst 1	Additional Bloom system supply contract
Catalyst 2	Bloom revenue recognition in Q4 2026 to Q1 2027
Catalyst 3	Bloom/Brookfield AI power projects converting from framework to project-level orders
Catalyst 4	ASP lift from cell to module/system conversion
Catalyst 5	AI data-center power quality becoming a standard requirement across PSU, BESS and supercapacitor layers

The invalidation list is more important.

Invalidation	Meaning
No follow-on Bloom PO by H1 2027	KRW 41.2B contract may be one-off
System sales rise but operating margin does not	High-value-system thesis weakens
Bloom dual-sources aggressively	VinaTech share and pricing power fall
SOFC data-center projects shift to other power solutions	Bloom chain expansion slows
Outsourcing, quality or delivery issues appear	High-reliability data-center trust weakens

13. Evidence Classification

Fact

AI data-center power is not just a capacity issue. NVIDIA says synchronized GPU training creates rapid power swings, and GB300 NVL72 energy storage can reduce peak grid demand by up to 30%.²
Bloom announced fuel-cell deployment for CoreWeave’s AI data center.³
SOFC has limits in fast load following; hybrid SOFC DC microgrid research uses batteries and supercapacitors to cover transient power.⁴
VinaTech signed a KRW 41.215B data-center supercapacitor-system contract with Bloom, equal to 50.12% of 2025 revenue.¹
The contract was reported as VinaTech’s first direct supply of an entire data-center supercapacitor system.⁵

Inference

VinaTech’s core role is transient power buffering for Bloom SOFC systems serving AI data centers.
Its moat is stronger in module/system design and Bloom qualification than in the standalone cell.
Repeat orders could reclassify VinaTech from a supercapacitor-cell company into an AI power-stability system supplier.
The current stock price already embeds some repeat-order expectation.

Speculation

VinaTech supercapacitor systems could become standard content in Bloom SOFC data-center packages.
VinaTech could expand beyond Bloom to other data-center power-infrastructure customers.
System ASP uplift could translate into higher operating margin.
The Meta, CoreWeave, Bloom and VinaTech chain could become project-specific and direct in future disclosures.

Blocked

Exact Bloom product specs: voltage, capacitance, ESR, cycle life and thermal spec.
Actual gross margin on the Bloom contract.
VinaTech’s share and exclusivity within Bloom.
VinaTech content per MW of Bloom SOFC data-center capacity.
Whether Meta workloads actually run in a Bloom-powered CoreWeave data center.
Exact power-stage location: SOFC output, DC bus, UPS auxiliary layer or rack-side buffer.

Final View

VinaTech is not an AI data-center power-generation stock. The cleaner description is a transient-load-buffer system supplier inside Bloom SOFC data-center power systems. Its moat comes from customer qualification, design-in status and systemization more than unique raw cell technology.

The first contract is already in the price. The next alpha must come from follow-on purchase orders and margins. If Bloom repeat orders arrive and system revenue lifts profitability, VinaTech can be reclassified. Without follow-on orders, chasing the stock means buying an expensive theme, not a confirmed compounder.

The practical conclusion is simple: the business position has improved; the price has already moved; the next test is repeat Bloom orders and system margins.

Source Ledger

CBC News Korea, VinaTech signs KRW 41.2B data-center supercapacitor contract with Bloom Energy, July 1, 2026. ↩︎ ↩︎ ↩︎ ↩︎ ↩︎
NVIDIA Technical Blog, How New GB300 NVL72 Features Provide Steady Power for AI, July 28, 2025. ↩︎ ↩︎
Bloom Energy, Bloom Energy and CoreWeave Partner to Revolutionize AI Data Center Power Solutions, July 17, 2024. ↩︎ ↩︎ ↩︎ ↩︎
Lin Zhang et al., Optimization of energy management in hybrid SOFC-based DC microgrid, Sustainable Energy & Fuels, 2023. ↩︎ ↩︎ ↩︎
The Bell, VinaTech targets AI data-center demand after breaking into Bloom Energy, July 1, 2026. ↩︎ ↩︎
VINA Tech, Target AI Data Center and Sales by 2030 Trillion, July 2, 2026. ↩︎
CoreWeave, CoreWeave and Meta Expand $21B AI Cloud Deal, April 30, 2026. ↩︎
Global Market Insights, Supercapacitor Market Size & Share 2026-2035, accessed July 4, 2026. ↩︎
Bloom Energy, Brookfield and Bloom Energy Expand AI Infrastructure Partnership to $25 Billion, June 30, 2026. ↩︎
NVIDIA Technical Blog, Designing Production-Ready Battery Energy Storage Systems for AI Factories, 2026. ↩︎

How SK Hynix's ADR Could Change 000660 Price Discovery And Leveraged ETF Flows

Sat, 04 Jul 2026 07:18:00 +0900

Context This is a follow-up to SK Hynix F-1/A and F-6: how the SKHY ADR packages HBM capex. The previous note focused on the filing mechanics. This note focuses on price transmission, ADR-common-share arbitrage and single-stock leveraged ETF flows.

TL;DR

The SK Hynix ADR is structurally positive over the medium term, but it is not a near-term volatility dampener. It widens U.S. investor access and may allow SK Hynix to be compared more directly with U.S.-listed AI memory peers. But right after listing, the key variables are the offer discount, the first week of ADR trading quality, the ADR premium or discount to the KRX common share and the impact on Korea’s single-stock leveraged ETF plumbing.

Question	View
Is the ADR positive for 000660?	Medium-term yes, because investor access and peer framing improve.
Is it an immediate buy signal?	No. Pricing, discount and initial premium or discount must be checked first.
What is the key mechanism?	ADR-common-share arbitrage. If the ADR trades rich to implied parity, it can create KRX common-share buying pressure. If it trades cheap, it can pressure the common share.
Can the ADR absorb Korea’s 2x ETF overheating?	Not in the short run. The bottleneck is KRX close rebalancing, not U.S. access.
Current stance	Watch the common share and the ADR. Avoid leveraged wrappers if the goal is HBM exposure.

1. What Is Confirmed

SK Hynix has filed an F-1/A for a planned Nasdaq Global Select Market ADS listing under SKHY.¹ The F-6 filed by Citibank states that each ADS represents the right to receive one-tenth of one common share.² Press reports indicate a maximum new-share issuance of 17.79 million common shares, around 2.5% of existing common shares.³

The final ADS count, offer price and net proceeds remain blank in the F-1/A. That matters. The investment signal is not merely that an ADR is coming. The signal will be in the final price, the discount to the Korean common share and the first few days of trading.

2. Access Premium

The first effect is access. U.S. investors can buy SK Hynix exposure in dollars, during U.S. hours, without using a Korea trading account or indirect ETF exposure. That can shift the comparison set from “Korean memory stock” to “U.S.-tradable AI memory and HBM supplier.”

This is positive, but not new information for the stock. Much of the obvious ADR rerating argument has already been anticipated. The remaining alpha is whether U.S. investors pay at or above implied KRX parity after the ADR starts trading.

3. New Shares And Dilution

The ADR is linked to new share issuance, not merely a secondary sale by existing holders. The headline maximum is about 2.5% of existing shares. That is dilution in ownership terms, but not necessarily large economic dilution if the offer price is close to market and proceeds are invested into HBM, eSSD, advanced packaging and EUV-related capacity.

Scenario	Read-through
Small discount and strong order book	U.S. investors validate the common-share price. Positive for 000660.
Large discount and weak demand	U.S. market rejects the prevailing valuation. Negative for 000660.
ADR premium after listing	U.S. demand can support KRX common shares.
Persistent ADR discount	U.S. trading becomes a new downside anchor.

With one ADS representing 0.1 common share, parity is simple:

Calculation	Formula
Implied ADR price from KRX common share	`KRX common price × 0.1 ÷ USD/KRW`
Implied common-share price from ADR	`ADR price × USD/KRW × 10`

If the ADR trades above implied parity, arbitrageurs may buy KRX common shares and sell or create ADR exposure. That creates common-share buying pressure. If the ADR trades below parity, arbitrageurs may buy ADRs and sell common-share exposure, pressuring the KRX line.

Therefore the first week is not about the absolute ADR move. It is about the premium or discount versus the Korean common share, after conversion costs, FX, settlement and stock-borrow constraints.

5. U.S. Price Discovery

The ADR extends SK Hynix price discovery into U.S. trading hours. This helps incorporate Nvidia, Micron, SOX, rates and AI capex news faster. But it also imports U.S. overnight gap risk into the Korean open. The ADR is a price-discovery channel, not automatically a volatility cushion.

6. Leveraged ETF Plumbing

The ADR does not directly remove Korea’s single-stock 2x ETF rebalancing pressure. Korean 2x ETFs must maintain daily target exposure through common shares, futures, swaps and liquidity provider hedging around the KRX close. The Korean Financial Services Commission described the daily 2x structure when single-stock leveraged products were introduced.⁴ The Korea Capital Market Institute estimated SK Hynix leveraged ETF AUM at KRW 9.15tn as of June 19 and highlighted how daily rebalancing can amplify volatility.⁵

The core issue is mechanical. When SK Hynix rises, 2x ETF exposure may need to increase. When it falls, exposure may need to decrease. That creates a short-gamma-like flow. An ADR only offsets this if U.S. long-only ADR demand is large enough to replace the domestic leveraged flow. That is possible later, but not proven now.

7. What To Watch

Watch item	Why it matters
Final offer price and discount	Tests U.S. demand quality.
First 3 to 5 days of ADR premium or discount	Determines whether U.S. trading supports or pressures 000660.
ADR trading value	Shows whether SKHY can become a real price-discovery line.
Korean leveraged ETF AUM	Shows whether the domestic short-gamma pipe is cooling.
HBM ASP, shipment and capex commentary	Keeps the ADR story grounded in earnings.

Final View

SKHY is a major capital-market event for SK Hynix. It can broaden the investor base and reframe the company as a U.S.-tradable AI memory supplier. But the ADR does not immediately solve the domestic leveraged ETF problem. The near-term stance is watchlist for the common share and ADR, and avoidance of leveraged wrappers if the goal is clean HBM exposure.

SK Gas vs SK Oceanplant: Which Second-Line AI Data-Center Power Candidate Is Cleaner?

Sat, 04 Jul 2026 05:54:00 +0900

Context This is a follow-up to AI data-center capex and Korea’s bottlenecks, VinaTech and Bloom Energy, GNC Energy and AI data-center backup power, construction and EPC rerating and H1 2026 AI infrastructure review.

TL;DR

After SK Group’s AI data-center investment announcement, the market did not treat every SK affiliate as a clean winner. It looked for the names that are closest to the actual power bottleneck. On that basis, SK Gas is the more direct second-line AI data-center power candidate than SK Oceanplant.

View	Summary
SK Gas	Directly linked to Ulsan GPS, a 1.2GW LNG/LPG combined-cycle power plant, the KET LNG terminal, fuel trading capability and SK Group’s Ulsan AI data-center power structure.
SK Oceanplant	Not a direct power supplier to AI data centers. It is an offshore-wind foundation manufacturer, and works better as a second-line candidate if the power-shortage theme broadens into renewables and offshore-wind infrastructure.
Current priority	SK Gas > SK Oceanplant. SK Gas has valuation and asset-revaluation support. SK Oceanplant has early flow-turn signals, but the chart and earnings visibility are still weaker.

0. Basis

Item	Detail
Companies	SK Gas, SK Oceanplant
Topic	Second-line power candidates after SK Group’s AI data-center plan
Price, consensus and flow basis	July 3, 2026 close and recent flow snapshot
Public-source check	July 4, 2026 KST
Key public events	SK Group AI data-center plan, Ulsan AI data-center project, SK Gas sale of 49% of Ulsan GPS, SK Oceanplant’s Taiwan offshore-wind foundation delivery
Conclusion	SK Gas first, SK Oceanplant conditional watchlist

This is not a claim that either company has a confirmed AI data-center power contract. The point is to rank exposure, valuation and flow quality.

1. What The Market Selected After SK Group’s Announcement

SK Group’s June 29, 2026 announcement framed AI infrastructure as a full-stack system that combines semiconductors, data centers, power and telecom. Local reporting said SK is already working with AWS on an Ulsan AI data center, combining SK Telecom, SK Broadband and SK Gas capabilities, and plans to expand similar models toward 15GW of AI data centers.¹

Al Jazeera also reported South Korea’s broader chip and AI data-center investment drive. The reported plan includes large-scale data-center investment involving SK Group, GS Group and Naver, and mentions a 10GW additional AI data-center target by 2035.²

But stock-market reaction was selective.

Item	Jun 30-Jul 3 return	Interpretation
KOSPI	-3.6%	Broad market correction
SK Inc.	-12.1%	Holding-company and funding discount
SK Hynix	-7.7%	Crowded memory leader pulled back
SK Telecom	-4.2%	AI data-center opportunity offset by cost concerns
SK Square	-3.1%	NAV discount partly cushioned the move
SK Gas	+2.6%	Power and gas infrastructure directness stood out
SK REITs	+3.6%	Infrastructure and asset-value defense

The signal is not the absolute gain, but the dispersion. The market did not simply buy the SK Group label. It preferred the affiliates tied to power, land, infrastructure and asset value.

2. Core Comparison

Item	SK Gas	SK Oceanplant
Price	KRW 218,500	KRW 15,320
Jun 30-Jul 3 return	+2.6%	-0.8%
1-week return	+7.6%	+6.5%
1-month return	-2.0%	-5.4%
2-month return	-24.4%	-37.6%
2026F PER	3.9x	21.6x
2026F PBR	0.52x	1.15x
2026F ROE	15.3%	5.4%
Consensus target price	KRW 367,000	KRW 26,167
Target-price upside	+68.0%	+70.8%
AI data-center directness	High	Low to medium
Investment character	Power and gas infrastructure plus undervaluation	Offshore-wind infrastructure plus turnaround option

The target-price upside looks similar. That is not enough. The better question is whether the company is close to the bottleneck, whether valuation is already stretched, and whether flow is turning from a damaged chart into real accumulation.

On those tests, SK Gas is ahead.

3. SK Gas: The More Direct Power Candidate

SK Gas has a clearer link to SK Group’s AI data-center power structure. The Ulsan project reportedly combines SK Telecom, SK Broadband and SK Gas capabilities.¹ That makes SK Gas more than a generic group affiliate.

The relevant assets are straightforward.

Asset or capability	Investor meaning
Ulsan GPS	1.2GW LNG/LPG dual-fuel combined-cycle power plant, the closest physical power asset in this comparison.
KET LNG terminal	LNG procurement, storage and supply infrastructure.
LPG/LNG trading	Fuel procurement and pricing capability that can affect generation economics.

SK Gas describes its LNG and power business on its official site.³ The Fact reported that Ulsan GPS started commercial operation in December 2024 as a 1.2GW LNG/LPG dual-fuel combined-cycle power plant. On June 30, 2026, SK Gas sold a 49% stake in Ulsan GPS for roughly KRW 1.2242tn while retaining a 51% controlling stake.⁴

That transaction matters because it confirms external asset value, releases capital and preserves strategic control.

SK Gas Investment Points

Point	Interpretation
AI data-center power directness	Potentially tied to SK Group’s Ulsan AI data-center power stack.
Valuation	2026F PER 3.9x and PBR 0.52x are not demanding.
Asset revaluation	The Ulsan GPS stake sale validated infrastructure value.
Shareholder return	2026F DPS KRW 10,500, roughly a 4% yield.
Flow	Latest four trading days: institutions +KRW 0.52bn, real-money-style flow +KRW 0.86bn.

The stock is not yet a strong momentum name. It is above the 20-day moving average, but still 8.9% below the 60-day line, with RSI around 40. This looks more like a low-valuation infrastructure name trying to base after a drawdown than a confirmed trend stock.

Level	Meaning
KRW 214,000-218,000	First support zone to monitor.
Reclaim of KRW 222,000	Initial strength confirmation.
Break above KRW 231,500	A stronger trend-reversal signal.
Break below KRW 203,000-209,000	Weakens the rebound thesis.

The key follow-ups are the second-half strategy update, actual power-supply structure for the Ulsan AI data center, use of the Ulsan GPS sale proceeds and the 2027-2029 shareholder-return policy.

4. SK Oceanplant: Offshore-Wind Infrastructure Option

SK Oceanplant should be framed differently. It does not directly sell electricity to AI data centers. It manufactures offshore-wind foundations. So the right thesis is not “direct AI data-center power exposure,” but “second-line exposure if the power-shortage theme broadens into renewables and offshore-wind infrastructure.”

Recent business news is constructive. Yonhap reported that SK Oceanplant delivered the first foundations for Taiwan’s Fengmiao 1 offshore-wind project on July 2, 2026. The project uses 33 15MW turbines for a 495MW wind farm, and SK Oceanplant signed a KRW 393.4bn contract in June 2024 to supply 21 foundations.⁵

The company is also investing KRW 1.153tn in what it describes as the world’s largest specialized offshore-wind foundation production base, targeting completion by the end of 2026.⁶

SK Oceanplant Investment Points

Point	Interpretation
Theme	Benefits if AI power shortages broaden into generation buildout and offshore wind.
Recent event	First Fengmiao 1 delivery gives visibility to the KRW 393.4bn contract.
Flow	Jun 30-Jul 3: foreigners +KRW 1.67bn, institutions +KRW 3.07bn, retail -KRW 5.02bn.
Program flow	Jun 30-Jul 3 program buying of KRW 3.00bn.
Burden	2026F PER 21.6x, ROE 5.4% and a still-damaged chart.

The recent four-day flow is constructive, but not enough to call structural accumulation. On a 20-day basis, foreigners plus institutions were still net sellers by KRW 18.69bn.

The chart is also still weak. The stock is 4.6% below its 20-day moving average, 21.8% below the 60-day line and 48.5% below its two-month high. The July 1-2 rebound faded again on July 3.

Level	Meaning
KRW 14,300-14,700	Aggressive support zone.
KRW 16,200-16,650	First strength-confirmation zone.
Break above KRW 17,800	Potential trend reversal if turnover holds.
Break below KRW 13,900	Invalidates the early turn.

For SK Oceanplant, the decisive variables are additional offshore-wind orders, the new yard schedule and upward revisions to 2027 earnings estimates.

5. Final View

SK Gas is the better second-line candidate today.

It is closer to the AI data-center power stack.
It trades at a lower 2026F PER and PBR.
The Ulsan GPS stake sale validated asset value and provided capital.

SK Oceanplant is still worth watching, but it is not the same kind of exposure. The Taiwan delivery and large production-base investment are meaningful. Still, this is an indirect offshore-wind infrastructure option, not a direct AI data-center power supplier.

Rank	Name	Stance	What to confirm
1	SK Gas	Watchlist / Pullback Buy	KRW 214,000-218,000 support, reclaim of KRW 222,000, Ulsan AI data-center power structure, shareholder-return policy.
2	SK Oceanplant	Watchlist	KRW 14,300-14,700 support, reclaim of KRW 16,200-16,650, new offshore-wind orders and 2027 earnings revisions.

The simple conclusion is this: in the AI data-center power trade, SK Gas is the closer infrastructure asset. SK Oceanplant is a higher-beta candidate for a later phase if the market expands the theme into offshore-wind generation buildout.

GNC Energy Rally Analysis: AI Data-Center Backup Power Stock Or Short-Term Overheat?

Sat, 04 Jul 2026 05:42:00 +0900

Context This is a follow-up to AI data-center capex and Korea’s bottlenecks, VinaTech and Bloom Energy, construction and EPC rerating, the era of expensive money and H1 2026 AI infrastructure review.

TL;DR

GNC Energy is not just a loose AI theme stock. It is a potential earnings-backed name in AI data-center backup power infrastructure. The company’s core business is generator EPC for emergency and commercial power systems. In AI data centers, that places it in the backup-power layer that keeps servers running when the grid or the power path fails.

The latest rally has four drivers.

Driver	Why it matters
LG Uplus AIDC cooperation	Media reports say cooperation started with emergency generators for the Paju AIDC and could broaden across AIDC power infrastructure.
Samsung SDS contract	A reported KRW 29.748bn generator supply contract for Samsung SDS’s Gumi AI data center, equal to 11.3% of FY2025 revenue.
AI data-center construction cycle	Data-center capex is moving from servers into power, cooling, backup power and construction.
Institutional flow	Latest five-day flow: institutions +KRW 7.66bn, including investment trusts +KRW 4.14bn and private funds +KRW 3.09bn.

The business thesis is valid, but the entry is not clean after a 45.5% five-day rally. Foreigners sold KRW 5.85bn over the same period, and foreign ownership fell from 5.12% on June 29 to 4.06% on July 3. On July 3, the stock traded as high as KRW 30,850 but closed at KRW 27,800, leaving a wide upper wick.

My stance is Watchlist / Buy on pullback. The first support zone is KRW 25,400-26,000. The more comfortable zone is KRW 23,400-22,700. A reclaim of KRW 28,400 would be a 60-day moving-average recovery. A breakout above KRW 30,850 with stronger turnover and broader buying would revive the momentum case.

0. Basis

Item	Detail
Company	GNC Energy
Ticker	119850 / KOSDAQ
As of	2026-07-04 KST
Price and flow basis	2026-07-03 close and recent flow snapshot
Key events	LG Uplus AIDC cooperation report, Samsung SDS Gumi AI data-center generator contract report, data-center backup-generator demand
Stance	Watchlist / Buy on pullback

The Samsung SDS KRW 29.748bn contract is treated based on media and public-disclosure reports. Contract text, revenue-recognition schedule and margin details still need confirmation. The right interpretation is not “all revenue immediately recognized,” but “order visibility improved.”

1. What GNC Energy Does

GNC Energy supplies diesel generators, gas turbine generators, gas generators and biogas power systems.¹ For AI data centers, the relevant business is emergency backup generation.

Data centers do not only need more electricity. They need uninterrupted electricity. That is why a data center power stack includes grid connection, transformers, UPS, batteries, emergency generators and control systems.

GNC Energy sits near the last line of defense. If Bloom Energy or fuel cells supply on-site generation, and VinaTech-type supercapacitor systems buffer transient power movement, GNC Energy is in the backup-power layer that allows a data center to keep operating when grid power is disrupted.

That distinction matters. This is not a generic “AI power” story. It is more precise to describe GNC Energy as an AI data-center backup-power EPC supplier.

2. Why The Stock Rallied

LG Uplus AIDC

ETNews reported that LG Uplus and GNC Energy started cooperation with emergency generators for the Paju AIDC and intend to broaden the cooperation across AIDC power infrastructure.²

This matters because data-center power infrastructure requires qualification, reliability, service support and delivery capability. Once a supplier is inside the customer reference base, repeat opportunities become more plausible.

Samsung SDS AI Data-Center Contract

According to Herald Economy reporting, GNC Energy signed a generator supply contract with Samsung SDS for the Gumi AI data center. The amount is KRW 29.748bn, equivalent to 11.3% of FY2025 revenue, with a contract period reportedly running until May 31, 2029.³

This is the most direct number behind the rally. It is a meaningful single project relative to annual revenue. The long contract period means investors should not pull the entire amount into a near-term quarter, but it does strengthen backlog and customer reference quality.

AI Data-Center Construction Cycle

Asia Business Daily discussed GNC Energy alongside construction names such as Hyundai Engineering & Construction, GS E&C and DL E&C in the context of AI data-center construction investment.⁴

The implication is that data centers are not only a construction or server story. The project stack includes power equipment, backup generation and delivery schedules. GNC Energy is smaller than the EPC primes, but it can carry higher beta to the backup-power node.

Institutional Buying

Investor	5-day net flow
Retail	-KRW 2.47bn
Foreign	-KRW 5.85bn
Institutions	+KRW 7.66bn
Investment trusts	+KRW 4.14bn
Private funds	+KRW 3.09bn

The constructive signal is that domestic institutions absorbed retail supply. The weak point is that foreigners were sellers. This is not yet clean foreign-plus-institution accumulation.

3. Earnings Level-Up And Margin Risk

GNC Energy already produced real earnings in FY2025.

Item	FY2025
Revenue	KRW 262.6bn
Operating profit	KRW 49.4bn
Net income	KRW 40.0bn
Operating margin	18.8%
ROE	21.2%

At the July 3 close, market cap was roughly KRW 458bn. Against FY2025 net income, that is about 11.5x trailing earnings. For an AI power-infrastructure theme stock with realized earnings, that is not obviously excessive.

Samsung Securities argued in February 2026 that the company’s revenue step-up into the KRW 200bn-plus range was driven by a sharp increase in domestic data-center emergency-generator backlog. It also highlighted the move toward 80-100MW data centers and higher order value per MW.⁵

But 1Q26 was weaker.

Item	1Q26
Revenue	KRW 55.4bn
Operating profit	KRW 4.4bn
Net income	KRW 12.3bn
Operating margin	7.95%

Electimes reported that FX-driven cost pressure on older projects weighed on operating profit.⁶ This is the key risk. The next proof point is not only revenue growth. The company needs to recover double-digit operating margin in 2Q or 3Q.

4. Price And Flow Setup

Item	Value
Close	KRW 27,800
Daily return	+2.02%
Intraday high	KRW 30,850
Intraday low	KRW 25,400
Volume	625,004 shares
5-day return	+45.5%
20-day return	+34.3%
60-day return	-8.9%
Drawdown from 60-day high	-29.6%
RSI14	61.3
5-day MA	KRW 25,590
20-day MA	KRW 22,326
60-day MA	KRW 28,410

The chart is strong but not fully repaired. The stock rallied sharply, but it still closed below the 60-day moving average after failing to hold the intraday move above it.

Period	Foreign	Institutions	Retail
3 days	-KRW 4.15bn	+KRW 5.43bn	-KRW 2.01bn
5 days	-KRW 5.85bn	+KRW 7.66bn	-KRW 2.47bn
10 days	-KRW 6.39bn	+KRW 8.85bn	-KRW 3.03bn
20 days	-KRW 1.95bn	+KRW 5.22bn	-KRW 4.01bn

This looks like domestic institutional discovery, not a global long-only accumulation story yet.

5. What Is Durable And What Is Fragile

The durable part is the AI data-center power bottleneck. Higher power density, stricter uptime requirements and longer equipment lead times all support demand for backup power infrastructure.

The company also has evidence. FY2025 showed a real earnings step-up, and the Samsung SDS contract is large enough to matter relative to revenue.

The fragile part is timing and margin. The stock is up 45.5% in five sessions. Foreigners are selling. 1Q26 margin fell to 7.95%. The Samsung SDS contract extends to 2029, so it is not a near-term revenue lump.

6. Investment View

Level	Interpretation	Action
KRW 25,400-26,000	First pullback zone, near July 3 low and 5-day MA	Watch for pilot entry if institutional flow holds
KRW 23,400-22,700	More comfortable pullback zone, near the breakout base and 20-day MA	Better risk-reward
KRW 28,400 reclaim	60-day MA recovery	Trend repair signal
KRW 30,850 breakout	Recent intraday high	Momentum entry only with volume and broader buying
Below KRW 23,400	Breakout failure risk	Stand aside or cut

Final View

GNC Energy belongs on the AI infrastructure watchlist. It is not a pure narrative stock: FY2025 earnings, data-center generator backlog, LG Uplus cooperation and the Samsung SDS contract all point in the same direction.

The stock, however, has already moved. After a 45.5% five-day rally, foreign selling and a large upper wick argue against chasing. Watch the KRW 25,400-26,000 support zone first, and the KRW 23,400-22,700 zone for a more comfortable entry. A reclaim of KRW 28,400 and a breakout above KRW 30,850 would reopen the momentum case.

GNC Energy, generator business page, company website. ↩︎
ETNews, LG Uplus AIDC cooperation report, June 28, 2026, article. ↩︎
Herald Economy, Samsung SDS Gumi AI data-center generator contract report, July 2, 2026, article. ↩︎
Asia Business Daily, AI data-center construction cycle, July 1, 2026, article. ↩︎
Samsung Securities, GNC Energy report, February 3, 2026, PDF. ↩︎
Electimes, GNC Energy 1Q26 report, May 15, 2026, article. ↩︎

SK Hynix F-1/A and F-6: What The SKHY ADR Says About HBM Capex

Sat, 04 Jul 2026 05:20:00 +0900

Context This is a follow-up to SK Hynix: the HBM leader in the AI revolution, SK Hynix HBM market share 2026, SK Hynix vs Micron and Micron FY3Q26 earnings.

TL;DR

SK Hynix’s SEC filings are not just an ADR wrapper. The sequence from the June 24 F-1, the June 30 F-1/A and the July 1 F-6 shows the company packaging its HBM leadership, eSSD, Custom HBM and Korea capex roadmap for U.S. AI-memory investors.

Item	Confirmed	Investor interpretation
F-1/A	Filed June 30, 2026. Nasdaq Global Select Market application under `SKHY`. ADS count, price and net proceeds are still blank.	Not final terms yet. A further amendment or final prospectus is required.
F-6	Filed July 1, 2026. Citibank is depositary. Each ADS represents one-tenth of one common share.	This fills the key ADR ratio that remained blank in the F-1/A cover.
New-share ceiling	Board authorization for up to 17,790,000 new common shares, 2.50% of 712,702,365 existing shares.	Maximum dilution is limited, but actual issuance remains unknown.

The most important caveat: the F-6 registration of 1,780,000,000 ADSs is not the offering size. Form F-6 registers the ADR facility and the related depositary-fee calculation. The actual equity offering size depends on the final ADS count and offering price in the F-1/A or final prospectus.

The stock read-through is Watchlist. This is an access and ownership-base event, not a fresh discovery of the HBM thesis. The alpha question is whether U.S. long-only and AI-memory investors use SKHY as a cleaner vehicle for SK Hynix exposure.

1. Why The June 24 F-1 Is Not Enough

Filing date	Form	Document	What changed
2026-06-24	F-1	Registration No. 333-296987	Initial ADS offering and Nasdaq listing preparation; key offering terms blank
2026-06-30	F-1/A	Amendment No. 1	Offering structure, new-share ceiling, use of proceeds and capex tables updated; ADS count and price still blank
2026-07-01	F-6	Registration No. 333-297185	Citibank depositary and one ADS = one-tenth common share

The company is SK hynix Inc. Its common shares trade on KOSPI under 000660. The ADSs are expected to trade on the Nasdaq Global Select Market under SKHY, subject to effectiveness and listing approval.

2. Offering Mechanics

Item	Latest status
Issuer	SK hynix Inc.
ADS ticker	`SKHY`, Nasdaq Global Select Market application
Depositary	Citibank, N.A.
Global coordinators	BofA Securities, Citigroup, Goldman Sachs, J.P. Morgan
ADS ratio	F-6: 1 ADS = 0.1 common share
ADS count	Blank in F-1/A
Offering price	Blank in F-1/A
Net proceeds	Blank in F-1/A
Lock-up	90-day lock-up may apply to the company and certain affiliates, subject to exceptions

The board authorized up to 17,790,000 new common shares. Against 712,702,365 existing shares, that is 2.50%. If fully issued, new shares would be about 2.44% of the post-issue share count. This is a ceiling, not the actual offering size.

3. F-6 ADS Registration Is Not Equity Proceeds

The F-6 registers 1,780,000,000 ADSs, each representing one-tenth of one common share. It also uses a fee calculation of $5.00 per 100 ADSs, giving a proposed maximum aggregate fee-basis amount of $89,000,000 and a registration fee of $12,290.90.

That fee basis is not the gross equity offering. Investors should not read 1.78 billion ADSs as the number of ADSs that will be sold in this offering. The real offering size remains blocked until the final ADS count and offer price are filed.

4. Use Of Proceeds: Yongin, P&T7 And EUV

The F-1/A frames proceeds around general corporate purposes, including capital expenditures. The filing specifically discusses KRW 45.5 trillion of Korea production-facility capex and approximately KRW 11.9 trillion of EUV scanner cost, with EUV delivery expected by December 2027.

Project	Target completion of planned investments	Total expected cost	Invested by May 31, 2026	Additional planned investment
Yongin Fab 1	End of 2030	KRW 31.0T	KRW 4.4T	KRW 26.6T
Cheongju P&T7	End of 2030	KRW 19.0T	KRW 0.1T	KRW 18.9T
Total	End of 2030	KRW 50.0T	KRW 4.5T	KRW 45.5T

Project	2026	2027	2028	2029	2030	Total
Yongin Fab 1	KRW 7.4T	KRW 10.1T	KRW 6.6T	KRW 2.5T	KRW 0.0T	KRW 26.6T
Cheongju P&T7	KRW 0.5T	KRW 2.1T	KRW 2.7T	KRW 5.8T	KRW 7.8T	KRW 18.9T
Total	KRW 7.9T	KRW 12.2T	KRW 9.3T	KRW 8.3T	KRW 7.8T	KRW 45.5T

The business section and capex table use different milestones. The business section points to the first Yongin cleanroom opening in 1Q27 and P&T7 construction completion by end-2027. The capex table targets completion of planned investments by end-2030. That is not necessarily a contradiction; it separates phase-opening from full investment completion and equipment installation.

5. The Narrative: HBM, Custom HBM And eSSD

The F-1/A does not present SK Hynix as a plain DRAM-cycle company. It emphasizes HBM, server DRAM and eSSD as structural AI-infrastructure products.

Market	IDC-based position in 1Q26
DRAM including HBM	No. 2 by revenue, 29.1% share
HBM	No. 1 by revenue, 56.4% share
NAND flash	No. 2 by revenue, 18.5% share

Custom HBM is especially important. The filing defines it as HBM that integrates certain GPU or ASIC functions into the HBM base die according to customer configuration. That shifts the story from memory pricing alone toward memory becoming part of the accelerator design stack.

6. Recent Financials

Metric	1Q25	1Q26	Change
Revenue	KRW 17.639T	KRW 52.576T	+198.1%
Profit for the period	KRW 8.108T	KRW 40.346T	~5.0x
Total assets	KRW 176.108T at 2025-end	KRW 222.829T at Mar. 31, 2026	+26.5%
Total equity	KRW 120.667T at 2025-end	KRW 164.380T at Mar. 31, 2026	+36.2%

Product	1Q26 revenue	Share
DRAM	KRW 40.659T	77.3%
NAND	KRW 11.574T	22.0%
Other	KRW 0.343T	0.7%

The U.S. accounted for KRW 33.999T, or 64.7% of 1Q26 revenue. That makes the ADR event strategically coherent: the investor access point is moving closer to the customer-revenue geography.

7. Shareholders, Dividends And Governance

Shareholder	Shares	Ownership
SK Square	146,100,000	20.50%
National Pension Service	57,439,774	8.06%
BlackRock	36,407,157	5.11%
Capital Research and Management	25,149,374	3.53%
Others	445,979,195	62.58%
Treasury shares	1,626,865	0.23%

The filing says the maximum offering size reflects the Korean Monopoly Regulation and Fair Trade Act requirement that SK Square maintain at least 20% ownership.

The 2025-2027 base dividend policy is KRW 1,500 per share per year, paid quarterly at KRW 375. The 2025 total dividend was KRW 3,000 per share after additional annual distribution. This ADR story is not a dividend story; it is a capex and AI-memory access story.

8. Risks

The main risk factors remain memory cyclicality, HBM manufacturing complexity, AI infrastructure slowdown, customer concentration, U.S.-China restrictions, raw materials, water, power, equipment, foreign-exchange volatility and ADS-specific rights. ADS holders exercise voting and dividend rights through the depositary and can face conversion, cancellation or redeposit limitations.

9. Investment View

SKHY can make SK Hynix easier to own for U.S. AI-memory investors. It can also make the comparison with Micron’s U.S.-listed AI-memory premium more direct.

But this is not yet an immediate buy signal. The final ADS count, offering price, net proceeds, effective date and first trading date remain unavailable. The event belongs on the watchlist until those terms are filed.

Watch List

Watch item	Why it matters
Next F-1/A or final prospectus	ADS count, offering price and net proceeds
SEC effectiveness	Start of real offering timeline
First `SKHY` trading date	U.S. investor access becomes observable
KRX common vs ADS parity	Premium, discount or arbitrage pressure
Offer price relative to KRX	Demand strength and near-term flow risk
U.S. AI-memory ETF inclusion	Potential structural demand
2Q/3Q HBM ASP and shipments	Whether the filing narrative converts into earnings
Custom HBM and eSSD commentary	Whether SK Hynix becomes a memory platform, not only an HBM supplier
Yongin, P&T7 and Indiana milestones	Whether capex turns into supply capacity on schedule

Sources

FADU 2Q26 Earnings Preview: Likely Beat, Not a Mega Surprise

Sat, 04 Jul 2026 05:00:00 +0900

Context This is a follow-up to FADU as Korea’s Sandisk beta, FADU’s P, Q and new-segment test, Korean semiconductor Top 50 two-month alpha candidates and the HBF/HBC commercialization calendar. FADU should be analyzed as an eSSD controller and AI-storage bottleneck name, not as a confirmed HBF beneficiary.

TL;DR

My FADU 2Q26 estimate is KRW 69-72B revenue and KRW 11-12.5B operating profit. The midpoint is KRW 70.5B revenue, KRW 11.8B operating profit and 16.7% operating margin.

The local consensus snapshot as of July 3, 2026 shows 2Q26 revenue of KRW 60.5B and operating profit of KRW 9.3B. The midpoint estimate is 16.5% above revenue consensus and 26.9% above operating-profit consensus. A beat is likely, but this is not a mega-surprise bar.

The key is not the gross amount of 2Q contract disclosures. Major supply contracts disclosed during the quarter total about KRW 150.8B. A mechanical pro-rata allocation would imply roughly KRW 35.4B of 2Q exposure. But amended disclosures in June pushed some contract end dates and payment terms later. Treating the full pro-rata amount as incremental 2Q revenue would be too aggressive. A more reasonable 2Q incremental recognition range is KRW 9.5-12.5B.

The stock call is Wait. FADU closed at KRW 79,600 on July 3, 2026, implying roughly KRW 3.98T market cap. Against FY2026E revenue of KRW 301.5B and operating profit of KRW 44.8B, that is about 13.2x sales and 89x operating profit. At this valuation, a good 2Q print is not enough. Investors need evidence of second-half delivery normalization, backlog conversion and controller-mix durability.

0. Setup

Item	Detail
Company	FADU
Ticker	440110 / KOSDAQ
Analysis date	2026-07-04 KST
Price basis	July 3, 2026 close, KRW 79,600, local KR database
Consensus basis	July 3, 2026 local consensus snapshot, 2Q26 revenue KRW 60.5B, OP KRW 9.3B
Main question	How much of the 2Q contract disclosures can actually be recognized as 2Q revenue?
Stance	Wait; conditional entry after proof

This preview uses public filings, public media reports and local market data. It does not use private management color.

1. Baseline From 1Q26

Metric	1Q26
Revenue	KRW 59.54B
Operating profit	KRW 7.69B
Net income	KRW 10.21B
SSD controller revenue	KRW 47.9B
SSD controller mix	80.4%
SSD finished-product revenue	KRW 9.9B
SSD finished-product mix	16.6%

The important point is mix. FADU earns a higher-quality multiple only if it is seen as an eSSD controller and firmware bottleneck company, not as a finished-SSD assembler. The 1Q controller mix of 80.4% is therefore more important than the headline profit turn.

The reported 1Q gross profit was KRW 34.65B. Against revenue of KRW 59.54B, this implies a 58.2% gross margin. Subtracting KRW 7.69B operating profit gives implied SG&A of about KRW 26.96B.

Item	Formula	Value
1Q gross margin	34.65 / 59.54	58.2%
Implied 1Q SG&A	34.65 - 7.69	KRW 26.96B
1Q operating margin	7.69 / 59.54	12.9%

2. Backlog And 2026 Delivery Base

FADU’s 1Q-end backlog was reported at $116.208M, roughly KRW 175.9B. The 2026 delivery portion was $95.131M, roughly KRW 144.0B. Adding 1Q revenue gives a visible annual revenue base above KRW 200B, before the rest of the 2Q disclosures.

Item	Amount
1Q-end backlog	$116.208M
KRW equivalent	~KRW 175.9B
2026 delivery portion	$95.131M
KRW equivalent	~KRW 144.0B
1Q26 revenue	KRW 59.54B

The caveat is timing. A 2026 delivery commitment does not mean 2Q recognition.

3. 2Q Contract Disclosures

Disclosure timing	Product	Contract amount	Contract period	Pro-rata 2Q exposure
Apr 13	Enterprise SSD controller	KRW 15.196B	Apr 10-Oct 9	~KRW 6.8B
May 6	Enterprise SSD controller	KRW 50.012B	May 5-Dec 8 after amendment	~KRW 13.1B
May 22	Enterprise SSD controller	KRW 28.684B	May 21-Dec 11	~KRW 5.7B
May 28	eSSD controller	~KRW 46.5B	May 28-Jan 1, 2027	~KRW 7.2B
May 29	Finished SSD	KRW 10.373B	May 28-Oct 11	~KRW 2.6B
Total		~KRW 150.8B		~KRW 35.4B

The June amendments matter. The KRW 50B contract end date moved from Nov. 13 to Dec. 8, and payment terms moved from the next Friday after delivery to within 30 days after delivery. A separate KRW 20.3B controller contract was also extended from June 12 to July 17. These do not prove demand weakness, but they do weaken the 2Q cutoff.

4. Revenue Model

Component	Estimate
1Q revenue baseline	KRW 59.5B
Incremental 2Q recognition from new contracts	+KRW 9.5-12.5B
Finished products and other movement	+/- KRW 1.0B
2Q26 revenue estimate	KRW 69-72B
Midpoint	KRW 70.5B

This is more conservative than the earlier June framework that looked for KRW 90-105B of 2Q revenue. The reason is new information: amended disclosures increased the risk that part of the revenue recognition slips into 3Q.

5. Operating Profit Model

Segment	Revenue assumption	Gross margin assumption	Gross profit
Controllers	KRW 57.5B	61%	KRW 35.1B
Finished SSD / modules	KRW 11.0B	15%	KRW 1.7B
Other	KRW 2.0B	50%	KRW 1.0B
Total	KRW 70.5B	~53.5%	~KRW 37.7B

Using KRW 25.9B of SG&A, the midpoint operating profit is about KRW 11.8B, or a 16.7% margin. The implied 2Q gross margin is below 1Q’s 58.2% to reflect finished-product mix and cutoff risk.

6. Scenario Table

Scenario	Revenue	Operating profit	OPM	Interpretation
Bear	KRW 62-65B	KRW 9-10.5B	14.5-16.2%	In line to slight beat
Base	KRW 69-72B	KRW 11-12.5B	15.9-17.4%	Clear beat
Bull	KRW 78-83B	KRW 14.5-17B	18.6-20.5%	Major surprise

Base is the central case. Bull requires faster shipment, limited impact from amendments and controller mix near 80%.

7. Consensus Bar

Metric	Consensus	Midpoint estimate	Delta	Delta %
Revenue	KRW 60.5B	KRW 70.5B	+KRW 10.0B	+16.5%
Operating profit	KRW 9.3B	KRW 11.8B	+KRW 2.5B	+26.9%
OPM	15.4%	16.7%	+1.3ppt

A meaningful beat needs revenue above KRW 70B and OP above KRW 11.5B. A true estimate-reset print likely needs revenue above KRW 75B and OP above KRW 14B.

8. Price, Flow And Valuation

Item	Value
July 3 close	KRW 79,600
5D return	-4.1%
20D return	-27.5%
60D return	+57.9%
Estimated market cap	~KRW 3.98T
20D foreign net buying	~KRW 227.3B
20D institutional net selling	~KRW -117.6B
Foreign ownership	24.11% on Jun 8 to 28.25% on Jul 3

Foreign flow is the positive signal. Institutional confirmation is not there yet. On valuation, FY2026E revenue of KRW 301.5B and OP of KRW 44.8B imply about 13.2x sales and 89x operating profit. This is not cheap. It requires continued proof.

9. Entry, Catalysts And Falsifiers

Action rule	Requirement
Earnings-confirmation entry	2Q revenue above KRW 70B, OP above KRW 11.5B, controller mix near 80%
Strong-beat entry	Revenue above KRW 75B, OP above KRW 14B
Valuation entry	If the print is only moderate and the stock derates toward below KRW 3.0T market cap or below 10x FY2026E sales

Catalysts are the 2Q result, 2Q-end backlog, 2026 delivery update, controller mix, normalization of amended contracts and 3Q revenue acceleration.

Falsifiers are revenue below KRW 65B, OP below KRW 10B, controller mix below 75%, gross margin below 50%, further delays in the KRW 50B or KRW 20.3B contracts, no 3Q backlog-conversion guide, or continued dependence on two customers near 80%.

Final View

FADU probably beats consensus in 2Q26. My midpoint is KRW 70.5B revenue and KRW 11.8B operating profit. But this is a moderate beat, not a clean buy signal at any price.

FADU remains a valid AI-storage bottleneck candidate. The stock needs stricter evidence now: 2Q revenue above KRW 70B, controller mix near 80%, and proof that delayed contracts convert into 2H revenue.

Sources: Daum News, The Elec, DailyInvest, KIND filing, The Bell, Datatooza, local KR price, flow and consensus databases as of July 3, 2026.

H1 2026 Review: AI Infrastructure Bottlenecks Broadened, but the Market Did Not

Tue, 30 Jun 2026 22:35:00 +0900

Related context This is a half-year follow-up to Samsung 2Q26 preview, Micron FY3Q26 earnings, the NVIDIA elasticity framework for Samsung and SK Hynix, how hard it was to beat the KOSPI benchmark, the KOSPI 60-day disparity framework, the Real Money Flow framework, and the AI supercycle midgame note.

TL;DR

H1 2026 looked like a broad risk-on period from the index level. Under the surface, it was much narrower. Korea was a KOSPI megacap crowding market. The U.S. was an AI infrastructure market where leadership moved from megacap AI platforms into memory, storage, servers, semiconductor equipment and power. Neither market was a broad “everything went up” market.

The headline Korean index was extraordinary. KOSPI rose from 4,214.17 on December 30, 2025 to 8,476.48 on June 30, 2026, up 101.1%. KOSDAQ fell from 925.47 to 916.18, down 1.0%. Across roughly 2,700 Korean stocks, 856 rose and 1,788 fell. The median stock return was -12.6%. The index doubled, but the average stock was in a weak market.

The U.S. story was also more specific than the index suggests. The S&P 500 gained 9.2% and Nasdaq gained 11.1%, but the real story was AI infrastructure bottlenecks. SanDisk, Micron, Western Digital, Dell, Marvell, ARM, Applied Materials and Lam Research were much stronger than NVIDIA. The market was buying the places where the AI factory actually bottlenecks.

The H2 strategy is not simply “find cheap stocks.” The better rule is to buy proven bottlenecks with real revenue and contracts on pullbacks. Option-like equipment, gaming and biotech event names should be sized smaller. Continuous-revenue bottlenecks in memory, eSSD, MLCC, substrates, power, shipbuilding, defense and large biotech platforms deserve more attention.

One-Line Conclusion

H1 was risk-on, but it was not broad. Korea was dominated by Samsung Electronics and SK Hynix crowding, while the U.S. rewarded the physical bottlenecks beneath AI compute. H2 is about separating recurring-revenue bottlenecks from option-like stories and leveraged flow risk.

0. Data Basis and Reconciliation

The data basis is:

Area	Basis
Korea indices and stocks	June 30, 2026 close
U.S. indices and stocks	June 29, 2026 close
Korea flows	KR local DB, Kiwoom amount basis
U.S. stock returns	US local DB and yfinance adjusted close
Secondary inputs	Naver Finance, local screeners and session flow analysis

Some figures differ slightly by dataset and reference date. For example, the final KOSPI return is +101.1% using the June 30 close, while some local basket snapshots show +96.7%, and the June 23 interim review showed +90.4%. U.S. AI stock returns also differ depending on adjusted-close treatment and vendor handling.

This post reconciles those differences as follows.

Data type	How it is used
June 30 Korea close	Main H1 performance benchmark
June 29 U.S. close	Main U.S. H1 performance benchmark
June 23 interim snapshot	Supporting evidence for breadth and buyer structure
Adjusted-close versus local DB differences	Used directionally when conclusions are unchanged
Less-verified single figures	Treated as H2 watch items, not as core proof

The point is not false precision. The conclusion is stable: indices were strong, breadth was weak, and AI infrastructure returns moved from GPUs into memory, storage, servers, equipment and power.

1. Index Performance: KOSPI Doubled, KOSDAQ Fell

Market	Basis	H1 return
KOSPI	4,214.17 on 2025-12-30 to 8,476.48 on 2026-06-30	+101.1%
KOSDAQ	925.47 to 916.18	-1.0%
S&P 500	2025-12-31 to 2026-06-29	+9.2%
Nasdaq	2025-12-31 to 2026-06-29	+11.1%
Russell 2000	2025-12-31 to 2026-06-29	+21.3%
SMH semiconductor ETF	2025-12-31 to 2026-06-29	+75.5%
SOXX semiconductor ETF	2025-12-31 to 2026-06-29	+104.2%

At the index level, Korea was spectacular. KOSPI doubled in six months. But KOSDAQ fell. Within the same Korean market, large-cap AI infrastructure and growth stocks lived in different worlds.

The U.S. index story also understates the internal rotation. The S&P 500 and Nasdaq were up around low double digits, while semiconductor ETFs were up far more. SOXX gained 104.2%, close to the KOSPI’s gain, showing how concentrated the global AI hardware trade became.

2. Korea: The Index Doubled, the Median Stock Fell

Item	Value
Observed Korean stocks	roughly 2,700
H1 advancers	856
H1 decliners	1,788
Median stock return	-12.6%
KOSPI H1 return	+101.1%
KOSDAQ H1 return	-1.0%

This is an unusual combination. The index was in a historic bull market, while the median stock was weak. That is why the lived investor experience was so different from the index headline.

The main H1 leaders were:

Stock	H1 return
Samsung Electro-Mechanics	+756%
Jusung Engineering	+626%
SK Square	+361%
SK Hynix	+307%
Jeju Semiconductor	+294%
Daeduck Electronics	+226%
Simmtech	+201%
Samsung Electronics	+179%

The Korean market’s structure was:

Samsung Electronics and SK Hynix
→ Samsung Electro-Mechanics, MLCC and FC-BGA
→ substrates, PCB and semi equipment
→ power equipment, industrials and some financials

KOSDAQ, many biotech names, platforms and parts of secondary batteries lagged. Korea was strong, but most Korean stocks were not.

3. Korea Was Not a Foreign-Buying Market

YTD flow data on the local Kiwoom amount basis:

Investor group	YTD cumulative
Foreigners	-KRW 142T
Institutions	+KRW 45T
Retail	+KRW 77.7T
Financial investment	+KRW 75.4T
Investment trusts	-KRW 1.7T
Private funds	-KRW 3.8T
Pension funds and others	-KRW 7.0T

The key point is that Korea was not a simple foreign-buying market. Foreigners sold Samsung Electronics and SK Hynix heavily. Domestic retail, financial investment accounts and institutions absorbed the flow.

Stock	YTD return	Foreign cumulative flow
Samsung Electronics	around +160%	-KRW 72.6T
SK Hynix	around +291%	-KRW 57.1T
SK Square	around +333%	-KRW 7.6T
Samsung Electronics preferred	around +125%	-KRW 3.2T

Foreigners sold the core megacaps while those stocks rallied. That is not a normal foreign-led rally. It was a domestic absorption market.

Foreigners did not sell everything. They rotated into select bottlenecks and global-fund-friendly assets such as Samsung Electro-Mechanics, Doosan Enerbility, Samsung SDI, FADU, Hanwha Ocean and Sanil Electric. The right description is not “foreigners bought Korea.” It is “foreigners sold the largest memory megacaps, domestic liquidity absorbed them, and foreigners reallocated into select second-layer bottlenecks.”

4. The Hidden Buyer Was Financial Investment, Not Long-Only Real Money

Institutional buying was not the same as long-only accumulation. The strongest institutional subcategory was financial investment, not investment trusts, private funds or pension funds.

Flow category	Interpretation
Financial investment	ETF, derivatives, program and rebalance-linked flow
Investment trusts	More likely to represent long-only fund demand, but weak in H1
Private funds	Mixed thematic, long-short and event-driven flow, but weak cumulatively
Pension funds and others	Long-term real money, but net selling in H1
Retail	Key absorber of megacap AI infrastructure flow and leverage

This matters. Real money can sustain a trend. Financial-investment and program flow can be powerful, but it is more vulnerable to futures expiry, ETF caps, derivative positioning and leverage mechanics.

H1 Korea was both a fundamental rerating market and a derivatives/ETF market. Confusing the two leads to repeated mistakes.

5. KOSDAQ Lacked Buyers, Not Stories

KOSDAQ had plenty of stories: biotech, robotics, CPO, SOCAMM, semiconductor equipment, gaming and AI software. The index still lagged because the buyer base was weak.

The June 23 interim snapshot was already clear:

Index	YTD	From high
KOSPI	+90.4%	-10.0%
KOSDAQ	-5.7%	-27.3%

The final June 30 structure was the same: KOSPI +101.1%, KOSDAQ -1.0%.

KOSDAQ struggled because:

Retail capital was tied up absorbing Samsung Electronics, SK Hynix and large AI infrastructure names.
Institutional real money did not buy KOSDAQ broadly.
Foreign buying was narrow, focused on names such as FADU, Jeju Semiconductor and Simmtech.
Rates and FX pressure weighed on biotech and unprofitable growth multiples.

The condition for KOSDAQ recovery is:

KOSDAQ recovery =
retail capital returning
+ institutional real-money buying
+ Samsung Electronics and SK Hynix moving sideways rather than falling
+ KOSDAQ earnings revisions stabilizing

If Samsung and SK Hynix crash, money does not automatically rotate into KOSDAQ. Risk-off comes first. The best environment for KOSDAQ is not megacap collapse. It is megacap consolidation.

6. The U.S. Market Was Not Just NVIDIA

The real U.S. signal was not “AI went up.” It was:

The physical bottlenecks after the GPU outperformed the GPU leader.

Representative H1 return ranges:

Stock	H1 return range
SanDisk	+645-764%
Micron	+263-301%
Western Digital	+247-279%
Intel	+235-257%
Dell	+224-232%
Marvell	+211-227%
ARM	+200-214%
Applied Materials	+171%
AMD	+152%
Lam Research	+141%

Big Tech was mixed:

Stock	H1 return
Google	+13%
Amazon	+4%
Apple	+4%
NVIDIA	+3-5%
Broadcom	+7%
Meta	-15%
Microsoft	-22% to -23%
Oracle	-25%
Palantir	-31%

The market bought the points where the AI factory bottlenecks:

GPU
→ HBM and DRAM
→ eSSD and storage
→ servers and racks
→ optical and interconnect
→ semiconductor equipment
→ power and cooling

This is directly relevant to Korea. HBM maps to Samsung and SK Hynix. eSSD and NAND map to Samsung, FADU and Jeju Semiconductor. MLCC and FC-BGA map to Samsung Electro-Mechanics and substrate names. Power maps to Doosan Enerbility and electrical equipment. Equipment maps to HPSP, TES, VM and Jusung.

7. Macro Was Not a Pure Tailwind

Asset or indicator	H1 move
Dollar index	+3.1%
USD/KRW	+7.9%
U.S. 10-year yield	4.16% to 4.39%
WTI oil	+24.1%
Gold	-6.8%
Bitcoin	-33.2%

This was not a clean multiple-expansion macro environment. The dollar rose, the won weakened, U.S. yields rose and oil rose. Gold and Bitcoin fell.

Stocks rose anyway because the AI infrastructure earnings cycle overwhelmed macro pressure. That is important for H2. If rates, the dollar and oil all rise together again, the most crowded bottleneck stocks can suffer multiple compression even if demand remains strong.

The realistic risk is:

AI demand stays strong
+ earnings remain good
+ rates, dollar and oil remain firm
+ valuation and leveraged flows crack first

That is the same stock-elasticity problem discussed in the NVIDIA/Samsung/SK Hynix framework.

8. Non-Consensus Truth 1: Leadership Moved From Compute to Memory

The first non-consensus point is U.S. AI leadership. NVIDIA is still the center of AI. But incremental stock performance moved into Micron, SanDisk, Western Digital, Dell, Marvell and ARM.

That does not mean compute no longer matters. It means the marginal return moved down the stack.

Question	Meaning
Is this company a cost item in AI capex?	It can be pressured when customers optimize costs.
Is this company a bottleneck item in AI capex?	It can reprice if customers need supply.
Is this company only a narrative?	It is vulnerable during catalyst gaps.

Cost items get pressured. Bottlenecks rerate. Narratives require timing discipline.

9. Non-Consensus Truth 2: The Real Top Came Before the Index High

Breadth weakened before the index did. The June index high was driven by Samsung Electronics, SK Hynix and a small set of AI infrastructure names. KOSDAQ and the median stock had already weakened earlier.

Weak median stock
+ weak KOSDAQ
+ strong KOSPI memory megacaps
= index high and weak lived market at the same time

For H2, the key indicators are breadth, ADR, KOSDAQ/KOSPI relative strength, megacap memory trading-value share, and KOSDAQ retail/institutional joint buying.

10. Non-Consensus Truth 3: Multiples Moved Before Fundamentals Fully Arrived

Samsung, SK Hynix, Micron and Samsung Electro-Mechanics all had strong fundamentals. But stock returns moved faster than near-term earnings. The market changed the multiple.

Old frame:
memory = commodity cycle = peak-earnings discount

H1 frame:
memory = AI infrastructure bottleneck = supply shortage plus contract premium

The H2 risk is not just bad earnings. It is good earnings with lower multiples if the market starts treating memory as commodity again. Investors should watch EPS revision speed, beat magnitude and FY2027 visibility, not just the next quarterly beat.

11. Non-Consensus Truth 4: Leverage Plumbing Amplified Tail Risk

Korean memory megacaps were affected not only by fundamentals, but also by leverage plumbing: 2x and 3x Samsung/SK Hynix ETFs listed offshore, domestic leverage and inverse products, futures and program flow.

Simplified upside loop:

stock up
→ leveraged ETF rebalancing buy
→ futures and program support
→ more megacap upside
→ more momentum capital

Downside loop:

stock down
→ leveraged ETF rebalancing sell
→ futures and program pressure
→ retail margin and forced-selling risk
→ larger drawdown

Korea can therefore move more violently than Micron even when the fundamental direction is similar.

12. Non-Consensus Truth 5: Supply Moved From Strong Hands to Weaker Hands

The important question is not only who sold. It is who bought.

H1 saw foreigners and some institutional accounts reduce exposure in memory megacaps while retail and financial-investment accounts absorbed the flow.

Positive read	Negative read
Domestic liquidity is strong	Shares may have moved into weaker leveraged hands
KOSPI held despite foreign selling	Forced-selling risk rises if margin unwinds
Large-cap buyer base widened	It may also represent distribution from better-informed accounts

H2 monitoring should combine foreign cash flow, retail margin, deposits, ETF balances, and program flow.

13. Non-Consensus Truth 6: Foreigners Bought Globally Explainable Korean Assets

Foreigners did not buy every Korean theme. The stocks they bought were globally explainable.

Stock	YTD foreign net buying
Samsung Electro-Mechanics	+KRW 2.28T
Doosan Enerbility	+KRW 1.71T
Celltrion	+KRW 1.52T
Samsung SDI	+KRW 1.33T
Doosan	+KRW 1.11T
APR	+KRW 1.08T
Hyundai Rotem	+KRW 0.85T
LG Energy Solution	+KRW 0.85T
FADU	+KRW 0.80T
Hanwha Ocean	+KRW 0.76T

Common features:

power, nuclear, shipbuilding and defense
global quality industrials
large battery names
large biotech platforms
direct AI infrastructure bottlenecks

For H2, foreign long-only capital is more likely to buy assets it can explain in an investment committee than obscure small-cap stories.

14. Non-Consensus Truth 7: Revenue Pattern Beat Company Quality

In H1, the quality of the revenue pattern mattered more than abstract company quality.

Weak revenue-pattern names:

Type	Weakness
Equipment	Order gaps can create earnings gaps.
Gaming	Revenue visibility breaks around launch cycles.
Biotech event stocks	Trial and licensing gaps create narrative gaps.
Some robotics and AI software	Expectations move before revenue.

Strong revenue-pattern names:

Type	Strength
Memory	Pricing and volume improved together.
eSSD and storage	AI server storage bottleneck rerating.
MLCC and substrates	Server and package bottlenecks.
Power equipment	Data-center power demand plus policy capital.
Shipbuilding	Long backlog visibility.
Defense	Export backlog and government budgets.
Large biotech platforms	Contracts, production and platform value are explainable.

The H2 rule is simple:

Size option-like names smaller. Size recurring-revenue bottlenecks larger.

15. Bonus Hypothesis: Korea Became a Leveraged U.S. AI Proxy

Korea’s market identity changed in H1. Historically, Samsung and SK Hynix were proxies for EM, China and global IT inventory cycles. In H1 2026, the linkage to U.S. AI capex became much stronger.

Session analysis suggested a stronger relationship between U.S. prior-day AI/semi moves and Korea next-day memory megacaps, including an interim correlation estimate of 0.42. That figure needs formal revalidation, but the price action is consistent with the direction.

The issue is diversification. If Korea behaves like a leveraged U.S. AI capex proxy, domestic buffers weaken when the global AI factor cools.

Another watch item is capex. Session analysis classified large SK Hynix capex commitment as a peak-confidence checkpoint. The exact amount and scope should be rechecked against filings before it becomes a core argument. But historically, peak capex announcements often arrive when demand looks strongest.

Key H2 capex questions:

Question	Meaning
Are customer contracts supporting capex?	Lowers supply-glut risk.
Does capex impair FCF?	Earnings can look good while cash flow weakens.
Does supply arrive near demand peak?	Raises 2027 cycle risk.
Is capex HBM and AI-memory focused?	Must be separated from commodity DRAM oversupply risk.

16. H2 Checklist

Checkpoint	Why it matters
Samsung 2Q26 Core OP	Core profit power matters more than reported headline OP.
SK Hynix 2Q and 3Q HBM allocation	Determines whether the crowded winner can reaccelerate.
Micron FY4Q26 guide delivery	Tests whether U.S. AI memory premium can persist.
DRAM and NAND pricing	Confirms margin defense beyond HBM.
KOSDAQ retail and institutional buying	Needed for KOSDAQ recovery.
Financial-investment and program flow	Real plumbing behind megacap volatility.
USD/KRW, U.S. 10-year yields and oil	Put a ceiling on AI bottleneck multiples.
AI capex contracts and FCF	Shows whether demand is paid for by cash flow, not just narrative.

17. H2 Strategy

Area	Strategy
AI infrastructure core	Keep exposure, but buy pullbacks rather than chasing.
Memory megacaps	Samsung is the unreflected-alpha candidate. SK Hynix is the reacceleration candidate after pullbacks.
HBM lower-stack bottlenecks	Buy only where orders, margins and customer qualification are visible.
KOSDAQ	Do not buy the whole index first. Start with semiconductor equipment, test and profitable growth names.
Option-like equipment, gaming and biotech	Smaller sizing due to catalyst-gap risk.
Global quality cyclicals	Continue to monitor shipbuilding, defense, power, nuclear and large biotech platforms.

The H2 strategy in one sentence:

Keep AI infrastructure core exposure, reduce option-like positions, and shift toward recurring-revenue bottlenecks and Korean quality cyclicals that global capital can explain.

18. Final View

The crowd may remember H1 2026 as the best Korea/semiconductor supercycle year. That is not wrong. KOSPI doubled and memory, substrates, equipment and power leaders exploded.

But the more important story is:

leadership became extremely narrow
multiples moved before fundamentals fully arrived
foreigners sold the core megacaps while domestic liquidity absorbed them
financial-investment and ETF/program flows amplified the market
KOSDAQ lacked buyers, not stories
U.S. AI leadership moved from GPU into memory, storage, servers and power

The H2 question is not only “how much more upside is left?” The better questions are:

Does stock elasticity remain positive even after strong earnings?
Can domestic liquidity keep absorbing foreign selling?
Do KOSDAQ buyers return?
Do AI bottlenecks convert from narrative into revenue and contracts?

H1 was risk-on, but it was not broad. It was a compressed leadership market. H2 requires more precision: real bottlenecks over cheap stocks, recurring revenue over one-off catalysts, and Core OP plus FCF over headline earnings.

Data Note

Korea data: KR local DB, Kiwoom investor flow data and Naver Finance support as of the June 30, 2026 close.
U.S. data: US local DB and yfinance adjusted close as of the June 29, 2026 close.
Interim snapshots: June 23 KOSPI/KOSDAQ relative performance and session flow analysis.
U.S. June 30 regular-session results are not included.
Less-verified capex single figures are treated as H2 monitoring items rather than core proof.

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.

Samsung 2Q26 Preview: Micron’s Surprise Has Been Erased, Core OP Is the Real Test

Mon, 29 Jun 2026 14:35:00 +0900

Related context This is a follow-up to Micron FY3Q26 earnings, the NVIDIA elasticity framework for Samsung and SK Hynix, Samsung-Hynix-Micron parity, Samsung shareholder return and DS bonus buyback flow, and Samsung versus SK Hynix in HBM4E 12-high.

TL;DR

The number that matters in Samsung Electronics’ 2Q26 preview is not simply reported operating profit. The market needs to separate reported OP from Core OP, defined as reported OP plus the catch-up accrual for performance-based compensation.

At current prices, Micron’s FY3Q26 surprise has almost disappeared from Korean memory equities. Micron retained an 8.0% gain versus its pre-earnings price as of June 26. Samsung Electronics is down 5.9% versus the June 24 close based on June 29 intraday data, while SK Hynix is down 0.5%. The June 25 reaction was real, but most of it was erased by June 26 and June 29.

Samsung’s 2Q26 DRAM pricing is not the downgrade factor. Conventional DRAM contract prices are estimated to be up 58-63% QoQ, while mobile LPDDR is up 70-83% QoQ. A blended Samsung DRAM ASP assumption in the high-50% QoQ range is therefore not aggressive.

The reasonable range for 2Q26 reported OP is KRW 87-92 trillion, with a midpoint of KRW 89.5 trillion. Adding back KRW 15-17 trillion of catch-up bonus accrual implies Core OP of roughly KRW 104-106 trillion. In other words, the correct reading is not “earnings fell from the KRW 100 trillion expectation to KRW 89 trillion.” It is closer to “the memory core business reached more than KRW 100 trillion of profit power, but Samsung is recognizing excess-profit sharing costs in 2Q.”

The rating is Watchlist / Conditional Buy. Conditions are clear: reported OP should be at least KRW 87 trillion, Core OP should be at least KRW 102 trillion, the downgrade should be explained by accruals rather than DRAM/NAND price or shipment weakness, and 3Q26 OP durability around KRW 100 trillion should remain credible.

Main Point

Micron’s surprise was reflected in Samsung and SK Hynix for one day on June 25, but it has mostly been erased. Samsung is back to a price that barely embeds the Micron read-through. The real test is not headline 2Q OP, but Core OP above KRW 100 trillion and the durability of 3Q memory earnings.

1. Price reaction: is Micron’s surprise still in Korea memory?

Micron reported FY3Q26 results after the U.S. close on June 24, 2026. For Korea, the clean baseline is the June 24 close.

Stock	Jun 24 close	Jun 25 close	Jun 26 close	Jun 29 intraday	Current vs Jun 24
Samsung Electronics	KRW 340,500	KRW 358,500	KRW 339,500	KRW 320,500	-5.9%
SK Hynix	KRW 2,580,000	KRW 2,917,000	KRW 2,673,000	KRW 2,568,000	-0.5%
Micron	USD 1,048.51	USD 1,213.56	USD 1,132.33	latest DB value Jun 26	+8.0%

Source: KII local daily price database for Korea and U.S. stocks; June 29 intraday Korea prices from Naver Finance API at 13:19 KST.

SK Hynix reacted most directly on June 25, rising 13.1%. But it has since nearly returned to its pre-event level. Samsung reacted 5.3% on June 25, then fell below the pre-event price. Micron still held an 8.0% gain as of June 26.

The conclusion is simple: at current prices, Korean memory has not retained Micron’s earnings read-through. It was reflected for one day, then erased by macro pressure, KOSPI overheat adjustment and company-specific uncertainty. The unreflected alpha is more visible in Samsung. SK Hynix is more of a pullback-and-reacceleration candidate.

2. What Micron confirmed

Micron FY3Q26 was a strong benchmark for the memory cycle. Revenue was USD 41.456 billion, non-GAAP EPS was USD 25.11, non-GAAP gross margin was 84.9%, and FY4Q26 guidance called for USD 50.0 billion revenue and USD 31.00 non-GAAP EPS. These are official company numbers. Micron FY3Q26 earnings release

The key read-throughs are:

Micron signal	Meaning for Samsung and SK Hynix
USD 41.456B revenue and 84.9% non-GAAP GM	Memory pricing, mix and shortage-driven operating leverage are very strong.
FY4Q guidance of USD 50B revenue and USD 31 EPS	Pricing and shipments are unlikely to collapse immediately after 2Q.
Data center revenue above USD 25B	AI memory demand is broader than HBM alone.
HBM4 and HBM4E roadmap	2027 allocation and customer qualification remain the next battleground.
Strategic Customer Agreements	Long-term contracts may reduce the old memory cyclicality discount.

This matters because Samsung now has to prove the same point with its own numbers: whether core memory profitability is already in the KRW 100 trillion range.

3. Reported OP versus Core OP

The working formula is:

Reported OP = accounting operating profit
Core OP = Reported OP + catch-up accrual for performance-based compensation

Base case:

Item	Base case
Reported OP	KRW 89.5T
Catch-up bonus accrual	KRW 15-17T
Core OP	KRW 104-106T

The important point is that the expected OP cut is not driven by weaker DRAM or NAND assumptions. In the research input, Kiwoom’s 2Q26 OP estimate fell from KRW 100T in May to KRW 89T on June 29. But over the same period, DRAM ASP assumptions moved from +55% QoQ to +58% QoQ, and NAND from +72% QoQ to +75% QoQ. The downgrade is therefore better explained by bonus accruals, non-memory weakness, DX cost pressure, and continued Foundry/S.LSI losses.

Item	May estimate	Late-June estimate	Meaning
2Q26 OP	KRW 100T	KRW 89T	Headline lower
DRAM ASP	+55% QoQ	+58% QoQ	Price assumption stronger
NAND ASP	+72% QoQ	+75% QoQ	Price assumption stronger
Downgrade driver	not applicable	Bonus, non-memory, DX, Foundry/S.LSI	Not a memory pricing problem

4. Official 1Q26 base

Samsung’s 1Q26 official results provide the base:

Item	1Q26
Company revenue	KRW 133.9T
Company operating profit	KRW 57.2T
DS revenue	KRW 81.7T
DS operating profit	KRW 53.7T

Source: Samsung Electronics 1Q26 results.

DS operating profit of KRW 53.7T is the key base for the 2Q bonus accrual discussion. If 1Q did not fully accrue the new profit-sharing structure, 2Q can carry a catch-up cost.

5. DRAM pricing is supportive

Product group	Estimated 2Q26 price trend	Interpretation
Conventional DRAM	+58-63% QoQ	Kiwoom’s +58% assumption is near the lower end of the market range.
PC DDR5	+43-48% QoQ	PC demand is not strong, but pricing still rose.
PC DDR4	+35-40% QoQ	Lower than DDR5, but still strong.
Mobile LPDDR4X	+70-75% QoQ	Mobile DRAM helps pull blended ASP higher.
Mobile LPDDR5X	+78-83% QoQ	Positive for Samsung’s blended DRAM ASP.
Consumer DRAM	+45-50% QoQ	Customer price resistance exists, but pricing is still up.

The core conclusion is that a high-50% QoQ blended DRAM ASP assumption is reasonable. If reported OP comes in near KRW 89T, it should not be interpreted as DRAM price weakness unless Samsung’s own disclosures say otherwise.

6. Bonus accrual treatment

The existence of DS performance-based compensation is not new. The unsettled question is scale, timing and disclosure quality.

Analytical purpose	How to treat bonus accrual
2Q earnings quality	Add it back to estimate Core OP
3Q run-rate	Exclude 1Q catch-up, include normal ongoing bonus expense
FY26 EPS	Include it
FY27 valuation	Include it if recurring
Shareholder return and FCF	Include it
Short-term event trade	Watch both reported OP and Core OP

The two mistakes to avoid are: treating all bonus expense as a one-off, or treating all of it as structural deterioration. The catch-up portion matters for 2Q interpretation. The recurring profit-sharing structure matters for FY26/FY27 EPS and valuation.

7. Sell-side OP distribution

Source	2Q26 OP estimate	Interpretation
BNK	KRW 77.7T	Most conservative bonus retroactive accrual assumption
Samsung Securities	KRW 81.3T	DRAM ASP +46%, NAND ASP +65%, DS bonus reflected
Mirae Asset	KRW 84.1T	Strong pre-accrual cost scenario near KRW 35T
FnGuide consensus	KRW 86.2T	Early-June average, may not fully reflect latest prices and accruals
Kiwoom Jun 29	KRW 89.0T	Higher DRAM/NAND price assumptions, but bonus/non-memory costs reflected
KB	KRW 90.0-90.2T	Memory price up more than 50%, DS OP in the high-KRW 80T range
Daishin	KRW 91.0T	Memory price strength with partial cost reflection
Hana	KRW 92.0T	Stronger pricing and shipment assumptions

Investment framing:

Reported OP range	Interpretation
Below KRW 84T	Bear case. Costs and non-memory drag are very large.
KRW 87-92T	Base cluster. This is the realistic range.
Above KRW 94T	Bull case. Costs are limited or price/shipments are even stronger.

8. Scenario model

Item	Bear	Base	Bull
DRAM blended ASP	+50%	+56-58%	+62-63%
DRAM bit growth	+4%	+5%	+6%
FX effect	+1.0%	+1.5%	+2.0%
DRAM revenue	KRW 86-88T	KRW 91-93T	KRW 95-97T
DRAM OP margin	71-73%	73-75%	76-78%
DRAM OP	KRW 62-64T	KRW 67-70T	KRW 72-75T
NAND OP	KRW 18-20T	KRW 20-22T	KRW 22-24T
Foundry/S.LSI OP	-KRW 1.5T to -KRW 1.0T	-KRW 1.0T to 0	0 to KRW 0.5T
Display/DX/Harman combined	-KRW 0.5T to KRW 1.0T	0 to KRW 1.0T	KRW 1-2T
Reported OP	KRW 82-86T	KRW 87-92T	KRW 94-98T

Base midpoint:

Item	Midpoint
Revenue	about KRW 181T
Reported OP	about KRW 89.5T
OPM	about 49.4%
Core OP	KRW 104-106T

9. Sensitivities

Variable	OP sensitivity
DRAM ASP +1ppt	+KRW 0.4-0.5T
DRAM bit growth +1ppt	+KRW 0.4-0.5T
NAND ASP +1ppt	+KRW 0.1-0.2T
Bonus/accrual +KRW 1T	-KRW 1.0T
DX cost +KRW 1T	-KRW 1.0T
Foundry/S.LSI loss +KRW 1T	-KRW 1.0T

The OP downgrade from KRW 100T to KRW 89T despite stronger DRAM/NAND price assumptions implies roughly KRW 12-13T of cost, accrual and non-memory drag.

10. Reported OP and Core OP matrix

Reported OP	Bonus accrual	Core OP	Interpretation	Action
KRW 92T or higher	about KRW 15T	KRW 107T or higher	Clear beat	Buy possible
KRW 89-92T	about KRW 15T	KRW 104-107T	In-line positive	Buy weakness
KRW 87-89T	about KRW 15T	KRW 102-104T	Consensus area	Staged entry possible
KRW 84-87T	about KRW 15T	KRW 99-102T	Weak miss	Needs explanation
KRW 84-87T	KRW 25-35T	KRW 109-122T	Headline miss, but neutral to positive if full-year pre-accrual	Verify cost nature
KRW 80-84T	about KRW 15T	KRW 95-99T	Cannot be explained by bonus alone	Wait or reduce
Below KRW 80T	any	not meaningful	Thesis damage likely	Avoid

11. Catalysts

Event	Timing	Importance	Watch item
Samsung 2Q26 preliminary results	likely early July 2026	Very high	Reported OP at least KRW 87T
Samsung 2Q26 final results and call	likely late July 2026	Very high	Bonus accrual, DS core profit, 3Q outlook
3Q26 DRAM/NAND contract prices	July-August 2026	Very high	Flat to up QoQ
HBM4E, SOCAMM2, eSSD customer comments	2H26	High	Evidence of share gain
Bonus-share buyback disclosure	TBD	Medium-high	Distinguish compensation buyback from cancellation buyback
Micron FY4Q26 results	likely late September 2026	High	Whether USD 50B revenue and USD 31 EPS are delivered

12. Valuation

The Kiwoom June 29 estimates summarized in the research input are:

Item	2026F	2027F
Revenue	KRW 723.6T	KRW 820.1T
Operating profit	KRW 372.9T	KRW 421.3T
EPS	KRW 43,429	KRW 48,177
PER	7.8x	7.0x
PBR	3.5x	2.3x
ROE	53.9%	39.6%

A KRW 430,000 target price implies 9.9x 2026F EPS and 8.9x 2027F EPS. This is valid only if 3Q26 OP remains above KRW 100T, 2027 EPS stays near KRW 48,000, bonus and stock compensation are already embedded in EPS, HBM4/eSSD contributions are visible, and DRAM/NAND prices do not fall sharply.

13. Red Team and invalidation

Risk	What would invalidate the thesis
Macro failure	Rates and credit spreads slow AI capex, hyperscaler capex guidance falls, AI server orders are delayed.
Micro failure	Samsung does not gain HBM4/eSSD share, HBM4 qualification is delayed, Foundry/S.LSI losses do not shrink.
Accounting failure	Reported OP falls below KRW 84T without enough bonus/accrual explanation, or Core OP falls below KRW 100T.
Capital allocation failure	Compensation buybacks are mistaken for shareholder return, while recurring profit-sharing erodes FY27 EPS and FCF.

Hard invalidation conditions:

Condition	Meaning
Reported OP below KRW 84T and not explained by at least KRW 25T of pre-accrual	More than timing noise
Core OP below KRW 100T	Core memory profit power weaker than expected
3Q26 OP outlook below KRW 100T	Peak-out risk
3Q DRAM/NAND prices fall QoQ	Pricing thesis damaged
No HBM4/eSSD share gains	2027 EPS and multiple expansion limited
Foundry/S.LSI loss persists or widens	Samsung discount persists

Final view

At current prices, Micron’s surprise is almost gone from Samsung and SK Hynix. Samsung has returned to a price that barely embeds the Micron read-through. SK Hynix has almost returned to pre-event levels after a strong June 25 reaction.

Samsung’s opportunity is now specific: if 2Q preliminary results show that the headline OP decline is mostly bonus-accrual timing, while Core OP is above KRW 100T and 3Q durability remains intact, the stock can reprice.

The relative conclusion is straightforward:

Stock	Interpretation
Samsung Electronics	Unreflected alpha if Core OP and 3Q durability are confirmed
SK Hynix	Pullback and reacceleration alpha, not a “cheap” stock
Micron	Cleanest proof of the AI memory cycle, but already repriced

One sentence: Micron’s earnings were reflected in Korean memory names for one day, but most of that effect has been erased. For Samsung, the next battle is not headline 2Q OP but Core OP above KRW 100 trillion and 3Q durability.

HBF And HBC Commercialization Calendar: Milestones, Stock Triggers, And Misclassified Baskets

Sun, 28 Jun 2026 23:20:00 +0900

HBF And HBC Commercialization Calendar: Milestones, Stock Triggers, And Misclassified Baskets

This note follows the earlier HBM, HBF and HBC technology comparison. The goal here is not to restate the technology. The goal is to separate timing, evidence, equity exposure and misclassified stock baskets.

TL;DR

HBF and HBC are real directions, but both are early.
HBF should be read as second-half 2026 sample, early 2027 first inference-device sample, and 2028 or later revenue potential.
HBC starts with Qualcomm AI250. Qualcomm AI200 is the pre-HBC product. AI250 with HBC Gen 1 is targeted for commercial sampling in mid-2027. Revenue relevance is more likely 2028 or later.
The closest 2026 trigger is not revenue. It is physical evidence: HBF sample delivery and Qualcomm AI200 shipment.
HBF equity exposure is narrow. SanDisk is the most direct listed option, but the stock is still primarily a NAND and data-center SSD cycle stock. SK Hynix has HBF exposure, but its stock is still driven by HBM. FADU is not a confirmed HBF beneficiary.
HBC exposure is also narrow. Qualcomm is the home asset. In Korea, Samsung Electro-Mechanics is the most direct confirmed component exposure through FC-BGA substrates for Qualcomm AI200, based on media reports. Hanmi Semiconductor is an HBM and HBF equipment name, not an HBC name.

Why The Distinction Matters

HBF and HBC sound similar, but they are not the same layer.

Item	What it is	Current state	Key question
HBM	High-bandwidth DRAM memory	Mass production	Can 2027 pricing and allocation hold?
HBF	NAND-based high-bandwidth flash layer	Standardization and pre-sample	Can samples and customer adoption appear?
HBC	Qualcomm accelerator architecture	Roadmap and design targets	Can Qualcomm win data-center customers?

SanDisk and SK Hynix have announced HBF standardization work under OCP. SanDisk has targeted first HBF samples in the second half of 2026 and first AI inference devices with HBF in early 2027. Qualcomm, meanwhile, describes HBC as a near-memory compute architecture inside its Dragonfly data-center roadmap, with HBC Gen 1 commercial sampling targeted with AI250 in mid-2027.

Commercialization Clock

Milestone	HBF	HBC
Current state	Pre-sample. Most numbers are simulations or targets	No shipped HBC product. Public numbers are design targets
First physical evidence	HBF sample delivery in 2H26	AI200 shipment in 2H26, but AI200 is not HBC
Defining point	First HBF inference-device sample in early 2027	AI250 with HBC Gen 1 sample in mid-2027
Revenue timing	2028 or later, likely 2028-2030	2028 or later
Largest swing factor	NVIDIA or major AI-platform adoption	Qualcomm data-center customer orders and revenue
Direct listed exposure	SanDisk, with NAND caveat	Qualcomm; Samsung Electro-Mechanics as Korea component exposure

The practical conclusion is simple: 2026 is a sample and shipment watch, 2027 is a validation watch, and 2028 or later is the revenue watch.

HBF Stock Exposure

SanDisk is the cleanest HBF equity option because it is directly involved in HBF standardization and sample timing. But it is not a pure HBF stock. The core drivers remain NAND pricing, data-center SSD demand and AI storage backlog. HBF is an asymmetric option layered on top of that.

SK Hynix is involved in standardization, but HBF is not the main stock driver. HBM4 allocation, NVIDIA demand, DRAM pricing and memory operating margin remain far more important.

FADU should not be put into the HBF basket without evidence of HBF products, contracts or logic-die supply. Its real discussion belongs under eSSD controllers, AI storage, ICMS and KV-cache offload. The company also carries legal uncertainty from the IPO-related case reported in late 2025 and 2026.

HBC Stock Exposure

Qualcomm is the home asset for HBC. But its data-center business is still small relative to its smartphone, automotive and broader QCT base. The main triggers are AI200 shipment, AI250 sampling, customer purchase orders, data-center revenue and independent benchmark validation.

Samsung Electro-Mechanics is the most direct Korean component exposure. Media reports say it has begun producing FC-BGA substrates for Qualcomm AI200 at its Busan plant. This is exposure to Qualcomm’s data-center accelerator roadmap. It is not yet direct HBC Gen 1 exposure because AI200 is not HBC.

Samsung Electronics, SK Hynix and Micron may have LPDDR relevance, but HBC-linked LPDDR is not as economically powerful as HBM. For Korean memory majors, HBM remains the main thesis.

Hanmi Semiconductor is misclassified as an HBC beneficiary. Its TC bonder exposure belongs to HBM and potentially HBF, not Qualcomm HBC.

Priority Trigger Calendar

Priority	Theme	Trigger	Timing	Why it matters
1	HBF	First HBF sample delivery	2H26	First physical proof
2	HBC	AI200 shipment	2H26	First proof of Qualcomm data-center accelerator execution
3	HBC	Data-center revenue visibility	FY27	Most objective proof
4	HBF	NVIDIA stance	Unknown	Biggest two-way wildcard
5	HBC	AI250/HBC Gen 1 sample	Mid-2027	The defining HBC milestone
6	HBF	Anchor customer qualification	2027	From technology claim to customer adoption
7	Both	Independent benchmarks	2027+	Validation of design targets
8	HBC	Customer PO and deployment	2027-2028	Narrative turns into revenue

Final Take

HBF and HBC both address a real AI memory bottleneck. But the equity exposure is narrower than the theme suggests. HBF equals SanDisk option, with SK Hynix as a side branch. HBC equals Qualcomm, with Samsung Electro-Mechanics as the most direct Korean component exposure. FADU in HBF and Hanmi Semiconductor in HBC are weak classifications.

The right framework is to buy confirmation, not vocabulary. In 2026, watch samples and shipments. In 2027, watch qualification and AI250. In 2028 and beyond, watch revenue.

Sources include Qualcomm’s June 2026 data-center roadmap, SanDisk and SK Hynix HBF announcements, Samsung Electro-Mechanics product materials, and media reports on Samsung Electro-Mechanics and FADU. This is catalyst analysis, not investment advice.

What NVIDIA's Earnings Elasticity Says About Samsung And SK Hynix

Sun, 28 Jun 2026 22:00:00 +0900

What NVIDIA’s Earnings Elasticity Says About Samsung And SK Hynix

This is a follow-up to the Micron FY3Q26 review, the Samsung-Hynix-Micron parity series, the HBM4E allocation note, and the Samsung catch-up work. The question is simple: if Samsung Electronics and SK Hynix still have strong earnings, can their stocks keep rising with the same elasticity?

The NVIDIA case says no. Stocks do not react only to the level of earnings. They react to the rate of change, the second derivative of growth, and whether estimates are still being revised up faster than the stock has already moved.

TL;DR

NVIDIA’s first share-price elasticity slowdown started around calendar 3Q24. The more structural plateau began around calendar 4Q25 to 1Q26.
Revenue kept exploding, but the stock started responding less to absolute growth and more to the durability of surprise.
The Samsung and SK Hynix earnings cycle has not broken yet. Recent SK Hynix, Samsung, and Micron results still point to strong AI-memory demand.
The stock-cycle risk is different. Samsung and SK Hynix entered a first overheated zone around June 2026.
The main decision window is 4Q26 to 1Q27. Watch 2027 HBM contract pricing and volumes, HBM4 share, NVIDIA and hyperscaler AI capex, and memory operating-margin peak-out.
My equity call is conditional buy for Samsung Electronics and wait for SK Hynix. SK Hynix is the higher-quality HBM leader, but it is now a crowded winner. Samsung still has HBM4 and Vera Rubin catch-up optionality.

The NVIDIA Lesson

NVIDIA was the cleanest listed-equity case of the AI semiconductor cycle. The stock moved first, then the earnings caught up. But once the market fully believed the earnings path, excellent results no longer produced the same stock response.

NVIDIA reporting quarter	Revenue growth	Data-center growth	Calendar-quarter stock move	Read-through
FY23 Q4, reported Feb 2023	YoY -21%, QoQ +2%	n/a	2023 Q1 +90.1%	AI expectations moved before earnings
FY24 Q1, reported May 2023	YoY -13%, QoQ +19%	DC revenue $4.28bn	2023 Q2 +52.3%	Guidance shock
FY24 Q2, reported Aug 2023	YoY +101%, QoQ +88%	YoY +171%, QoQ +141%	2023 Q3 +2.8%	First digestion
FY24 Q4, reported Feb 2024	YoY +265%, QoQ +22%	YoY +409%, QoQ +27%	2024 Q1 +82.5%	Blackwell confidence
FY25 Q2, reported Aug 2024	YoY +122%, QoQ +15%	YoY +154%, QoQ +16%	2024 Q3 -1.7%	First inflection
FY26 Q3, reported Nov 2025	YoY +62%, QoQ +22%	YoY +66%, QoQ +25%	2025 Q4 about 0.0%	Structural plateau begins
FY26 Q4, reported Feb 2026	YoY +73%, QoQ +20%	YoY +75%, QoQ +22%	2026 Q1 -6.5%	Strong earnings, weak elasticity
FY27 Q1, reported May 2026	YoY +85%, QoQ +20%	YoY +92%, QoQ +21%	2026 Q2 to Jun 26 +10.4%	Earnings follow-through, limited multiple expansion

NVIDIA FY27 Q1 revenue was $81.6bn and data-center revenue was $75.2bn. The FY27 Q2 guide was $91.0bn, plus or minus 2%. Those are still exceptional numbers. Yet from September 2025 to June 2026, the stock barely moved compared with its earlier rise.

The lesson for Korean memory equities is clear: strong earnings are not enough. The stock has to be cheaper than the next estimate revision path.

Why Samsung And SK Hynix Have Not Broken Yet

The earnings cycle is still intact.

SK Hynix reported 1Q26 revenue of KRW 52.5763tn, operating profit of KRW 37.6103tn, and net income of KRW 40.3459tn. Samsung reported 1Q26 consolidated revenue of KRW 133.9tn and operating profit of KRW 57.2tn, with DS operating profit at KRW 53.7tn. Micron’s FY3Q26 was also very strong, with revenue of $41.456bn, non-GAAP EPS of $25.11, and non-GAAP gross margin of 84.9%.

That does not look like a demand collapse. It looks like a profitable shortage.

The problem is not the earnings cycle. The problem is the stock cycle. SK Hynix became South Korea’s most valuable company in June 2026, and Reuters reported that the stock was up more than 340% year-to-date. At that point, the market no longer needs proof that SK Hynix is an HBM winner. It needs proof that 2027 EPS estimates still have room to move up.

The Timing Framework

Stage	Expected timing	Interpretation	Key signal
First stock-elasticity slowdown	Already around June 2026	Overheated zone	SK Hynix market-cap leadership, retail FOMO, target-price chase
EPS-revision peak	4Q26 to 1Q27	Most important window	2027 HBM contract price and volume disappoint or confirm
Earnings peak-out	1H27 to 2H27	Base-case fundamental risk window	HBM ASP growth slows and memory margin rolls over
Full memory downcycle	2H27 to 2028	Not yet confirmed	Supply catches demand and inventories build

The stock can peak before earnings. That is the central point.

Five Break Conditions

Break condition	What it means
HBM ASP flattens or declines	The first sign is likely price momentum slowing, not demand disappearing
SK Hynix margin falls for two straight quarters	Peak earnings risk becomes observable
Samsung and Micron reduce SK Hynix’s HBM premium	The monopoly premium becomes a leadership premium
NVIDIA data-center QoQ growth falls into single digits for two quarters	AI capex is no longer accelerating fast enough
2027 supply catches demand	HBM4 and HBM4E ramp creates a new pricing regime

Stock Implications

Samsung Electronics is the cleaner conditional buy. The point is not that Samsung’s memory business is good. The point is that its HBM laggard discount can narrow if HBM4 and SOCAMM2 revenue show up in DS margin.

SK Hynix remains the quality leader, but new buying is less attractive. The stock now needs more than a strong 2026. It needs 2027 HBM contracts to prove that EPS can still be revised up.

Stock	View	Why
Samsung Electronics	Conditional buy	HBM4 and Vera Rubin catch-up can reduce the laggard discount
SK Hynix	Wait	Best HBM asset, but crowded and closer to peak-expectation risk
NVIDIA	Watchlist	AI leader, but earnings growth and share-price elasticity have separated
HBM equipment and substrates	Selective watch	Alpha can move down the bottleneck stack if megacap memory stalls

Final View

Samsung and SK Hynix are not in an earnings breakdown. They are in a stock-elasticity test. The right question is no longer whether HBM is good. The right question is whether 2027 EPS revisions can outrun the stock price.

That makes Samsung Electronics the more asymmetric Korean memory expression today. SK Hynix is higher quality, but more fully believed.

Sources: NVIDIA FY27 Q1, NVIDIA FY26 Q4, Samsung 1Q26, SK Hynix 1Q26, Micron FY3Q26, Reuters on SK Hynix.

Disclaimer: This is research and information only, not investment advice.

How To Express The Samsung Electronics Catch-Up Trade: Preferred Share Discount And Samsung C&T NAV Gap

Mon, 22 Jun 2026 23:10:00 +0900

Samsung Electronics common has already re-rated sharply on HBM, memory-cycle recovery, and free-cash-flow return expectations. The question is how to express the same thesis more efficiently.

TL;DR

Samsung Electronics preferred is the cleaner catch-up expression. It lagged the common by 37.8 percentage points YTD and traded at a 36.6% discount on 22 June 2026.
The discount is extreme: 36.6% versus a 19.4% three-year average, 18.1% median, +3.11 z-score, and 99.1st percentile.
Samsung C&T is not a one-for-one Samsung Electronics proxy. It is a NAV and affiliate-dividend story.
ETF cap spillover is secondary. A simple redistribution assumption implies only about KRW 24.6bn into Samsung Electronics preferred and KRW 46.0bn into Samsung C&T.

Key Data

Stock	2 Jan Close	22 Jun Close	YTD	Versus Samsung Electronics
Samsung Electronics	KRW 128,500	KRW 353,500	+175.1%	Baseline
Samsung Electronics preferred	KRW 94,400	KRW 224,000	+137.3%	-37.8pp
Samsung C&T	KRW 245,000	KRW 520,000	+112.2%	-62.9pp

Metric	Samsung Electronics	Samsung Electronics Preferred	Samsung C&T
20D return	+18.0%	+19.3%	+24.3%
60D return	+86.3%	+67.0%	+84.7%
Correlation with Samsung Electronics	1.000	0.949	0.779
Beta to Samsung Electronics	1.000	0.912	0.848

Preferred Discount

Metric	Value
Preferred/common ratio	63.37%
Current discount	36.63%
Three-year average discount	19.43%
Three-year median discount	18.14%
Current z-score	+3.11
Three-year percentile	99.1%

Preferred/Common Ratio	Implied Preferred Price	Upside
Current 63.4%	KRW 224,000	Baseline
70.0%	KRW 247,450	+10.5%
75.0%	KRW 265,125	+18.4%
80.0%	KRW 282,800	+26.3%
85.0%	KRW 300,475	+34.1%

Samsung C&T NAV

Samsung C&T is calculated to hold 298,818,100 Samsung Electronics shares. At KRW 353,500, that stake is worth about KRW 105.6tn. Samsung C&T’s estimated market cap is about KRW 84.3tn.

Metric	Value
Samsung Electronics stake value	About KRW 105.6tn
Samsung C&T estimated market cap	About KRW 84.3tn
YTD increase in Samsung Electronics stake value	About KRW 67.2tn
YTD increase in Samsung C&T market cap	About KRW 44.6tn
Reflection rate	About 66.3%

Reflection Rate	Implied Samsung C&T Price
50%	KRW 452,000
60%	KRW 494,000
70%	KRW 535,000
80%	KRW 577,000
100%	KRW 660,000

Flow Check

Stock	1D Foreign	1D Institution	1D Real Money	5D Foreign	5D Institution	5D Real Money
Samsung Electronics	-4,418	-1,164	+617	-11,499	+7,449	+2,568
Samsung Electronics preferred	+149	+389	+122	-270	-142	-512
Samsung C&T	-627	+401	+387	-936	+450	+597

Action Framework

Stock	Stance	Confirmation
Samsung Electronics preferred	Conditional Buy	Discount remains around 34-37%, Samsung Electronics trend intact, five-day foreign/institutional flow turns positive
Samsung C&T	Watchlist / Pullback Buy	Holds KRW 500,000, foreign/program selling slows, or breaks above KRW 535,000
Samsung Electronics common	Core Hold / cautious add	Foreign selling slows, program selling eases, 20-day moving average holds

The key conclusion is that the next Samsung Electronics catch-up alpha is more likely to come from preferred-share discount normalization than from ETF cap spillover.

Sources: Samsung Electronics shareholder return, Samsung C&T shareholder return, KCMI cap background, Korea Herald on KOSPI200 cap abolishment, KII internal snapshots as of 22 June 2026.

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

Samsung Electronics Is Not a Dividend Stock: FCF Return and the DS Bonus Buyback Flow

Mon, 22 Jun 2026 17:45:00 +0900

Related reads: Samsung DS bonus treasury-share math · Samsung HBM4E 12-high sample · Jensen Huang’s HBM4 three-vendor comment · Samsung-Hynix-Micron parity · AI HBM hub

TL;DR

Samsung Electronics’ official 2024-2026 shareholder-return policy is annual regular dividends of KRW 9.8 trillion plus 50% of free cash flow over the three-year period.
On KII’s local consensus database, FY2026E net income is about KRW 300.2 trillion. Even if only 50% of net income converts into FCF, the 50% FCF return pool is about KRW 75.1 trillion, or 3.34% of combined common and preferred market capitalization.
The DS performance-bonus treasury-share purchase should not be counted as pure shareholder return because the shares are not cancelled. But it can still create roughly KRW 20-22 trillion of company purchase demand, around 1% of common market cap.
Samsung is therefore not a high-dividend stock. The better description is an AI memory mega-cap with an FCF return option and a separate DS bonus-related buyback flow.
The key checkpoints are FY2026 FCF conversion, board-level additional return decisions, DS bonus buyback approval, and HBM4/HBM4E customer qualification and shipments.

1. Why This Matters

The common mistake is to analyze Samsung only through dividend yield.

That screen is not flattering. Samsung’s regular dividend commitment is KRW 9.8 trillion per year. Against roughly KRW 2,246.4 trillion of combined common and preferred market value as of June 22, 2026, that is only about 0.44%.

But the current policy is not just a fixed dividend. Samsung says it will return 50% of FCF for FY2024-2026. When AI memory earnings expand sharply, the absolute size of that FCF-linked return option changes.

The DS bonus treasury-share issue is a separate layer. It is not a cancellation buyback, so it should not be added to shareholder-return yield. Yet it can still matter as company purchase demand and possible free-float absorption.

Bucket	Shareholder return?	What investors should track
Regular dividend	Yes	KRW 9.8tn annually, stable but low headline yield
Special dividend	Yes	Possible if surplus FCF is meaningful
Buyback and cancellation	Yes	Direct EPS/BPS accretion
DS bonus treasury-share purchase	Conservatively no	Not cancelled, but creates purchase flow and partial lockup

2. Verified Baseline

Samsung’s official IR page states the FY2024-2026 return policy: annual regular dividend of KRW 9.8 trillion and total shareholder return of 50% of FCF. It also says the company may consider early capital return beyond regular dividends if meaningful surplus resources are expected at year-end. The 1Q26 dividend was KRW 372 per common and preferred share, with total payout of KRW 2.45 trillion. Source: Samsung IR.

Samsung’s 1Q26 earnings presentation shows consolidated revenue of KRW 133.9 trillion and operating profit of KRW 57.2 trillion. DS revenue was KRW 81.7 trillion and DS operating profit was KRW 53.7 trillion, about 93.9% of consolidated operating profit. Samsung also cited high-value-added AI product demand, higher ASPs due to supply shortages, and mass product sales of HBM4 and SOCAMM2 for NVIDIA’s Vera Rubin platform. Source: Samsung 1Q26 earnings presentation.

KII local consensus and Naver Finance data as of June 22, 2026:

Item	Value
Common share price	KRW 353,500
Common market cap	KRW 2,066.7tn
Preferred share price	KRW 224,000
Preferred market cap	KRW 179.7tn
Combined market cap	KRW 2,246.4tn
FY2026E revenue	KRW 700.5tn
FY2026E operating profit	KRW 363.2tn
FY2026E net income	KRW 300.2tn
FY2026E EPS	KRW 44,426
FY2026E PER	7.97x
FY2026E ROE	52.64%

The critical number is not just 7.97x PER. It is KRW 300.2 trillion of expected net income. At this profit scale, the same FCF-return policy becomes a much larger option.

3. FCF Return Scenarios

The formula is simple:

FCF = FY2026E net income × FCF / net income ratio
Total shareholder-return pool = FCF × 50%
Return yield = pool / combined common and preferred market cap

Using FY2026E net income of KRW 300.2 trillion and combined market cap of KRW 2,246.4 trillion:

FCF / net income	Estimated FCF	50% FCF return pool	% of combined market cap
50%	KRW 150.1tn	KRW 75.1tn	3.34%
60%	KRW 180.1tn	KRW 90.1tn	4.01%
75%	KRW 225.2tn	KRW 112.6tn	5.01%
100%	KRW 300.2tn	KRW 150.1tn	6.68%

Even the conservative 50% FCF conversion case produces a KRW 75.1 trillion return pool. That is very different from the 0.44% headline regular-dividend yield.

This does not mean the full amount becomes incremental dividend or cancellation. The regular dividend is already part of the return pool, and Samsung may prioritize capex, strategic investment, M&A, or cash buffers. Still, the absolute option value of the 50% FCF policy rises sharply when AI memory earnings scale up.

Yonhap reported that Samsung and labor representatives agreed to introduce a DS special management performance bonus funded by 10.5% of selected business performance. The bonus would be paid after tax entirely in treasury shares, with one-third immediately saleable, one-third locked for one year, and one-third locked for two years. Source: Yonhap.

MoneyToday estimated that Samsung could purchase about KRW 20.85 trillion of shares for this bonus, using FY2026 operating-profit consensus, a DS profit share of around 94%, a 10.5% bonus pool, and a 40% tax assumption. Source: MoneyToday.

Updating that math with KII’s June 22 consensus:

Estimated DS OP = FY2026E consolidated OP × 1Q26 DS OP share
DS bonus pool = estimated DS OP × 10.5%
After-tax share grant amount = bonus pool × (1 - tax assumption)

Item	Value
FY2026E consolidated OP	KRW 363.2tn
1Q26 DS OP share	93.9%
Estimated DS OP	KRW 341.0tn
Bonus pool rate	10.5%
Bonus pool	KRW 35.8tn
Tax assumption	40%
After-tax share purchase amount	KRW 21.5tn
% of common market cap	1.04%
% of combined market cap	0.96%

Scenario table:

DS OP assumption	10.5% pool	After-tax 60% grant	% of common market cap
KRW 300tn	KRW 31.5tn	KRW 18.9tn	0.91%
KRW 331tn	KRW 34.8tn	KRW 20.9tn	1.01%
KRW 341tn	KRW 35.8tn	KRW 21.5tn	1.04%
KRW 360tn	KRW 37.8tn	KRW 22.7tn	1.10%

For Samsung’s scale, 1% of market cap may look small. But a single-stock company purchase flow of KRW 20-22 trillion is difficult to ignore, especially if two-thirds of granted shares are locked for one to two years.

5. Do Not Double Count

The right way to combine the pieces:

Item	Size	% of combined market cap	Interpretation
Regular dividend	KRW 9.8tn annually	0.44%	Stable, low headline yield
FCF 50% return pool, assuming FCF/NI 50%	KRW 75.1tn	3.34%	Official return-policy potential, includes regular dividend
DS bonus after-tax treasury-share flow	KRW 20-22tn	about 0.9-1.0%	Not cancellation, but potential company purchase demand

The regular dividend is part of the FCF return pool. The DS bonus treasury-share purchase should be kept outside pure shareholder-return math. But it should still be tracked as a market-flow variable.

6. What the Market May Miss

The skeptical view is familiar: Samsung has lower HBM purity than SK Hynix, a more complex business mix, low headline dividend yield, and a foundry discount.

The variant view is also becoming clearer: if FY2026 net income is around KRW 300 trillion, the 50% FCF-return policy becomes much larger in absolute terms. HBM4E progress could narrow the HBM discount. The DS bonus stock flow can add company purchase demand. And stock-based compensation partially aligns DS employees with shareholders.

The Samsung thesis is not simply “cheaper than SK Hynix.” It is:

AI memory profit scale rises
 -> absolute value of the 50% FCF return policy rises
 -> special return / cancellation / buyback options improve
 -> DS bonus treasury-share purchase flow adds market demand
 -> HBM catch-up and shareholder-return optionality can be priced together

7. Samsung vs SK Hynix

SK Hynix has clearer pure HBM exposure and deserves a leadership premium. Samsung is different. It has lower HBM purity but much larger earnings scale, an official FCF-return policy, and a separate DS bonus-related purchase flow.

Item	Samsung Electronics	SK Hynix
Main thesis	HBM catch-up, DRAM/NAND leverage, FCF return, buyback flow	HBM leadership and high-margin AI memory exposure
Strength	Earnings scale, cash flow, return policy, broad optionality	Customer lock-in, HBM share, pure AI memory premium
Weakness	Complex mix, HBM catch-up still needs proof, foundry discount	High expectations, possible dilution of exclusivity premium
Rerating trigger	FCF conversion, additional returns, HBM4E qualification, DS buyback approval	HBM4 allocation defense, ASP discipline, long-term customer contracts

8. Decision Framework

Positive checkpoints:

Checkpoint	Meaning
FY2026E OP and net income revisions continue upward	Larger potential return pool
FCF conversion of 50% or more becomes visible	Confirms actual power of the return policy
Board announces additional buyback/cancellation or special return	Direct rerating trigger
Board approves DS bonus treasury-share purchase and schedule	Flow trigger
HBM4/HBM4E qualification and shipments expand	HBM discount narrows
DS margin stays high	Sustains both profit and bonus pool

Risks:

Risk	Meaning
HBM4/HBM4E qualification delay	Samsung’s HBM discount persists
FCF materially below net income	50% FCF return pool shrinks
Capex, M&A, strategic investment, or cash buffers take priority	Additional return delayed
DS bonus buyback is mistaken for cancellation	Market overcounts the effect
Employee selling after lockup	Future overhang
AI capex slowdown, higher rates, stronger dollar	Memory multiple pressure

Conclusion

Samsung Electronics is not a high-dividend stock. That remains true.

But the 2026 setup requires a different lens. AI memory earnings have lifted expected net income to roughly KRW 300 trillion. Samsung has an official policy to return 50% of FCF. Separately, the DS bonus structure may create KRW 20-22 trillion of treasury-share purchase flow.

The cleanest sentence is this:

The DS bonus treasury-share purchase is not cancellation-based shareholder return. But it can still operate as company purchase demand, while Samsung’s official 50% FCF return policy becomes more valuable as AI memory earnings scale.

The next checks are not headlines. They are 2Q26 FCF, DS operating profit, HBM4E qualification and shipments, board-level return decisions, and the actual DS bonus buyback schedule.

Sources And Coverage

Samsung shareholder-return policy: Samsung IR
Samsung 1Q26 results and DS data: Samsung 1Q26 earnings presentation
DS special bonus agreement reporting: Yonhap
DS bonus treasury-share purchase estimate: MoneyToday
Price and consensus: KII local consensus database and Naver Finance, collected June 22, 2026

HBM, HBF, and HBC: How to Tell AI Memory Technologies Apart

Mon, 22 Jun 2026 10:00:00 +0900

Context This post is the technical background companion to Who Foots the Bill for the 2027 Semiconductor Consensus?, CXMT IPO and Memory Price Risk, and Why the AI Supercycle Keeps Extending. Those posts covered demand, pricing, and valuation. This one examines what the three AI memory supply-side technologies actually are, how proven each one is, and how they relate to each other. Related hubs: AI HBM Hub and Exclusive Analysis Hub.

TL;DR

HBM, HBF, and HBC are not the same category of technology. HBM and HBF are memory components; HBC is the name of Qualcomm’s accelerator architecture. Lining all three up and comparing raw TB/s numbers in isolation will produce a misreading.
Each targets a different constraint. HBM attacks the bandwidth wall, HBF attacks the capacity wall, and HBC targets inference cost and power.
Maturity gaps are wide. HBM is in full production (5/5 stars); HBF is at simulation stage (2/5, first samples targeted for H2 2026); HBC is a blueprint (1/5, samples targeted for 2027).
The three are complements, not substitutes. Data “temperature” determines placement: hot data (KV cache) belongs in HBM, cold data (static model weights) belongs in HBF, and low-cost inference is where HBC aims.
No technology currently threatens HBM’s position as working memory for AI accelerators. Meaningful revenue contribution from HBF or HBC is unlikely before 2027-2028 at the earliest. [Inference]

Key Framing

In the AI memory race, only HBM is proven. HBF is a promise. HBC is a blueprint. The three are not competitors -- they fill different slots in the memory hierarchy.

1. The Real Bottleneck in AI Is Memory, Not Compute

GPU compute throughput has grown roughly 80x over the past decade, while memory bandwidth has grown only about 17x. [Fact: NVIDIA generation-by-generation spec comparison] That gap is the genuine bottleneck in today’s AI accelerators. Expensive AI chips spend a growing share of their time idle, waiting for data to arrive – and the speed at which an LLM generates a single token is almost entirely a function of how fast memory can ferry data to the processor.

That bottleneck breaks into two distinct walls.

The bandwidth wall: memory cannot deliver data fast enough. The decode phase of LLM inference must read the full set of model weights from memory for every token generated, which makes it permanently memory-bandwidth-bound.

The capacity wall: memory cannot hold enough data at once. The KV cache alone for a 128,000-token context window runs to roughly 343 GB – already exceeding the 192 GB of HBM capacity on the latest NVIDIA B200. [Fact: derived from public specifications]

HBM, HBF, and HBC each attack these two walls in a different way. And one of the three is not a memory component at all – it is a single company’s accelerator architecture. Missing that distinction collapses the entire comparison.

2. HBM: The Only Proven Technology, Solving the Bandwidth Wall

The Core Concept

The logic behind High Bandwidth Memory is elegant. Stack 8 to 16 DRAM dies vertically, place the stack directly beside the GPU on the same interposer, and connect them through thousands of fine electrodes called through-silicon vias (TSVs). The result is a data path 1,024 to 2,048 bits wide. Speed comes not from any individual pin running faster but from an extraordinarily wide channel – the same reason a 2,048-lane highway moves more total traffic than a narrow one-lane expressway, even if each lane is not the fastest.

Performance and Proven Maturity

A single HBM3E stack delivers roughly 1.2 TB/s of bandwidth; HBM4 targets approximately 2 TB/s. [Fact: JEDEC specifications] Both are 20 to 30 times the bandwidth of DDR5. Virtually every high-performance AI accelerator in production today – NVIDIA H100, H200, B200, AMD MI300 – ships with HBM.

Maturity rating: 5/5. HBM is the only one of the three technologies in full-volume production. Years of large-scale manufacturing have validated yield and reliability at scale. SK Hynix holds roughly 57 to 62 percent of the market and remains NVIDIA’s primary supplier. [Fact: company IR disclosures and market estimates]

Roadmap and Industry Structure

The roadmap runs from HBM4 (volume production 2026, introducing a logic-based base die) to HBM4E (custom configurations) to HBM5 (targeting 4 TB/s, 2029). Starting with HBM4, SK Hynix manufactures the base die on TSMC’s 12nm process, making HBM development a joint design effort between a memory company and a foundry. [Fact: company announcements] [Inference: the 4 TB/s target for HBM5 is a roadmap objective, not yet independently verified]

The supply side is a three-player oligopoly of SK Hynix, Samsung Electronics, and Micron, with TSMC anchoring the base die and interposer layer. The AI memory ecosystem is effectively a co-production across these five companies. [Inference: market size estimated at roughly $35 billion in 2025, growing to $45-58 billion in 2026, with significant variance across sources]

3. HBF: Targeting the Capacity Wall, Still at the Promise Stage

The Core Concept

The idea behind High Bandwidth Flash is clever: borrow HBM’s stacking geometry and interface approach, but swap the expensive DRAM inside for cheaper NAND flash. NAND can pack multiple bits per cell, delivering far greater capacity at a much lower cost per gigabyte. In August 2025, SanDisk and SK Hynix announced a joint standardization effort. SanDisk described the technology as designed to “augment, not replace, HBM.” [Fact: SanDisk/SK Hynix joint announcement]

SanDisk claims a single HBF stack can hold roughly 512 GB (8 to 16 times more than a comparable HBM stack) while delivering approximately 1.6 TB/s of bandwidth. [Fact: SanDisk fact sheet]

Maturity: An Honest Assessment

That is where the analysis needs to pause.

Maturity rating: 2/5. HBF has not entered mass production. There is no independent benchmark of a physical chip. The 512 GB and 1.6 TB/s figures are derived from SanDisk’s internal simulations; the fact sheet’s footnotes explicitly state “internal testing and simulations under specific model assumptions.” [Fact: SanDisk fact sheet footnotes] First silicon samples are targeted for H2 2026, with inference-device adoption expected no earlier than 2027. The manufacturing challenge is also non-trivial: stacking 238-layer NAND 16 dies high implies more than 5,000 total layers of accumulated process complexity.

NAND’s Weaknesses Define Its Role

Two physical constraints govern what HBF can and cannot do.

First, latency: NAND reads more slowly than DRAM. HBF’s read latency runs roughly 100 times longer than HBM’s – approximately 10 microseconds for HBF versus tens of nanoseconds for HBM. [Fact: inherent media characteristics]

Second, write endurance: NAND cells tolerate a finite number of write cycles, making the technology ill-suited to data that changes continuously.

These two constraints fix HBF’s role: a high-capacity store for model weights – data that is read repeatedly but almost never written after training completes. Within that role, both weaknesses become nearly irrelevant.

The Decisive Variable: NVIDIA’s Adoption Decision

NVIDIA has shown limited public interest in HBF. Signals from the company suggest it favors GPU-attached high-speed SSDs (eSSD) as the solution to the capacity problem. Among large hyperscalers, Google has shown the most interest in HBF. [Inference: based on industry observation and public reporting] Whether NVIDIA ultimately adopts HBF is the single biggest unknown determining whether the technology becomes a mainstream platform or a niche solution.

The industry structure is currently SanDisk-led with SK Hynix as the primary collaborator; Samsung has signaled participation. Kioxia has taken a different path, investing instead in eSSD. Sides have not yet hardened.

4. HBC: Not a Memory Standard – Qualcomm’s Accelerator Architecture

Getting the Terminology Right First

“HBC” is an acronym with a trap inside it. Placed alongside HBM and HBF, it looks like a third memory technology. It is not. HBC is the name Qualcomm gave to its own AI accelerator architecture – “High Bandwidth Compute” – announced at its investor day in June 2026. It is a single company’s brand, not an industry-agreed standard. [Fact: Qualcomm official announcement]

The “HBC” abbreviation also collides with AMD’s “High Bandwidth Cache Controller” (HBCC) and is easily confused with HBF. These are entirely unrelated concepts.

What this means for analysis: a “HBM vs HBF vs HBC” comparison is strictly two memory components versus one system architecture. Ignoring that asymmetry leads to conclusions like “HBC delivers 18x more bandwidth than HBM” – a comparison that mixes entirely different units and system-level reference points.

Technology and Approach

The memory medium inside HBC is not HBM – it is LPDDR, the low-power DRAM used in smartphones. Qualcomm’s approach is to stack this inexpensive LPDDR directly on top of a compute logic die in 3D, collapsing the distance between memory and processing to near zero. The goal is to run inference cheaply and efficiently without the cost of a silicon interposer or expensive HBM.

Qualcomm claims 6x bandwidth per watt compared to HBM, and 200x capacity per watt compared to SRAM. [Fact: Qualcomm announcement] Both figures are Qualcomm’s own claims, normalized at the card or rack level. They are not directly comparable to per-stack HBM specs; reading them alongside HBM figures without adjusting for the reference basis will produce misleading conclusions.

Maturity and Ground Truth

Maturity rating: 1/5. The Qualcomm AI250 accelerator that embeds HBC does not yet exist as silicon. Sampling is targeted for mid-2027. Every performance figure released is a design target. [Fact: Qualcomm announcement] If HBF is at the simulation stage, HBC is at the blueprint stage. Reports have circulated that Microsoft and Meta have placed orders for Qualcomm AI accelerators, but those reports remain unconfirmed. [Inference: based on secondary press coverage, not independently verified]

The correct frame for reading HBC is not “a competing memory standard to HBM” but rather “one company’s system-level strategy to run inference cheaply without using HBM." If it succeeds, it could capture a portion of the low-cost inference segment that currently uses HBM-based accelerators – but it does not present a scenario in which HBM as a category is displaced. The supply side is also effectively Qualcomm alone.

5. Complements, Not Substitutes: Data Temperature Determines Placement

Headlines declaring “HBF will kill HBM” or “HBC will replace HBM” surface regularly. The analytical conclusion from working through the technology is that these claims are overstated. The three technologies are fundamentally complementary.

That conclusion does not rest on abstract market logic. It follows from the physical nature of the data itself.

LLM inference puts two categories of data into memory.

Model weights (cold data): fixed after training, read continuously during inference, almost never written. A large, slow, inexpensive memory tier (HBF) is an entirely adequate home for this data. NAND’s write-endurance constraint is nearly irrelevant when the data is read-only.

KV cache (hot data): updated continuously as a conversation or generation sequence progresses. It must live in the fastest available memory (HBM). Moving it to a slower tier slows the entire inference pipeline.

With a roughly 100x latency gap between them, HBM and HBF cannot occupy the same working-memory role. HBF does not displace HBM; it receives what HBM cannot hold, functioning as a lower tier in the hierarchy. HBC operates on an entirely different track, targeting low-cost inference economics rather than serving as a memory upgrade.

The most likely direction for future AI accelerator platforms is not a binary choice between HBM and HBF but a three-tier stack of HBM plus HBF plus high-speed SSD, with each layer serving data at the appropriate temperature. [Inference: technology roadmap analysis]

Partial competition does exist. Whether capacity extensions are solved through HBF or through eSSD is still an open contest. In the low-cost inference segment, HBC-based and HBM-based accelerators will overlap to some degree. And if HBF or HBC succeeds in making “large capacity at low cost” a viable option, some customers will have a reason to buy less HBM at the margin. But all of these competitive dynamics play out in 2027 and beyond.

6. Milestones to Watch

All three technologies are at different points on the journey from announced to proven.

HBM is the ongoing story. The next inflection points are HBM4 production yield, the rollout of custom HBM configurations tailored to specific hyperscaler needs, and the transition to hybrid bonding. SK Hynix, Samsung, Micron, and TSMC will capture the large majority of AI memory value creation in the near term. [Inference: market share projections]

For HBF, two milestones dominate. First: the arrival of first physical samples and independent benchmarks in H2 2026. Every number published to date is a simulation result; the moment real silicon can be tested independently is when the technology’s claims can be evaluated against reality. Second: whether NVIDIA elects to adopt HBF. An NVIDIA design win would rapidly crystallize the HBF ecosystem; the absence of one leaves the technology in a limited-customer camp.

For HBC, the first gate is Qualcomm’s AI250 reaching silicon in 2027 and passing independent verification. Confirmed orders from Microsoft and Meta would signal whether an HBC-compatible ecosystem is forming or whether Qualcomm remains the sole actor.

Hype Check

Several claims circulate in coverage of this space that warrant caution.

“HBF and HBC will replace HBM”: The latency gap and role differentiation make a complementary structure far more plausible than a replacement narrative. The displacement story is overstated.

“16x capacity, 4 TB VRAM, 6x power efficiency”: The majority of these figures are internal simulations or design targets, not independently measured results.

“HBM faces serious competitive threat in 2026”: Meaningful revenue contribution from HBF or HBC is unlikely before 2027 to 2028 at the earliest.

“HBC is a next-generation memory technology”: This is a category error. HBC is Qualcomm’s accelerator architecture, not a memory standard.

The correct starting point for analysis is separating marketing headlines from independently verified facts – and recognizing where the evidence stops and the promise begins.

This post is a technology analysis based on publicly available primary, secondary, and tertiary sources and company announcements. It is not a recommendation to buy or sell any security or product. Performance figures cited for HBF and HBC are predominantly company claims, simulation results, or design targets that have not yet been independently verified. This field is evolving rapidly; readers are encouraged to cross-check the latest primary sources before drawing investment conclusions.

What To Actually Watch In Korea's MSCI Review On June 24: Watch List Probability And The KB Financial Trade

Mon, 22 Jun 2026 08:00:00 +0900

Connecting the dots This piece follows How rare is an account that beats the pure KOSPI benchmark, The KOSPI 60-day disparity +28.6% overheat framework, and Who pays for the 2027 semiconductor consensus. Where the earlier pieces dealt with a “Samsung Electronics and SK Hynix hyper-concentration tape,” this one separates, by probability and by trade, whether the June 24 MSCI announcement is the event that changes that concentration. It synthesizes two independent analyses (a probability frame and a Red Team check) without conflict.

TL;DR

Korea’s probability of an immediate Developed Markets promotion in 2026 is around 1%. The substance of the June 24 event is not a “Developed promotion” but whether Korea is re-added to the Developed Markets Watch List (the pool under review for reclassification).
Combining the two analyses, the probabilities are Watch List 30-35%, positive monitoring (base) about 53%, negative deferral 14%, immediate promotion 1%. The key drag is that in the MSCI 2026 accessibility review Korea still received “needs improvement (-)” on five items: FX, registration, information flow, clearing and settlement, and transferability.
The upside case rests on availability of investment instruments improving from “-” to “+", and on the government preparing a 24-hour FX market for July 2026 and an offshore won settlement network for 2027. That said, as of the assessment cut-off date (May 31), neither was yet in operation, so there is a shortage of “verified investor experience.”
The most actionable conclusion is not a chase of large-cap semiconductors but a conditional re-rating trade in KB Financial. At roughly 0.87x 2026F P/B on the current price, the market has over-loaded AI and KOSPI beta but under-priced the potential decline in the bank’s cost of capital.
Do not buy unconditionally ahead of the announcement. The higher-expected-value approach is to enter on price terms after confirming the “under review / Watch List / market adoption” wording.

1. What To Actually Watch On June 24

MSCI has announced that it will release the results of its 2026 annual market classification review shortly after 22:30 CEST on June 23, 2026, which is shortly after 05:30 KST on June 24, 2026 in Korean time.

The five key questions for Korea are these.

Will Korea be promoted directly from Emerging to Developed? → Very unlikely, since it is procedurally hard to skip the Watch List and consultation stages.
Will Korea be re-added to the Developed Markets Watch List (the pool under review for reclassification)? → The crux of this event.
Does MSCI view Korea’s FX reform as a “sufficiently implemented and verified change,” or as “planned but unverified”? → The decisive point.
Will MSCI recognize the 24-hour FX market for July 2026 and the offshore won settlement network for 2027 as irreversible reforms? → If it does, the Watch List probability rises sharply.
If added, is the timeline a 2027 decision and 2028 inclusion, or a 2028 decision and 2029 inclusion? → The latter is more conservative and realistic.

The MSCI classification framework uses three criteria: economic development, size and liquidity, and market accessibility. Developed markets require “very high” on accessibility and “unrestricted” availability of investment instruments.

2. Official Evidence Snapshot

2.1 The MSCI 2026 Accessibility Review’s Verdict On Korea

The MSCI 2026 Global Market Accessibility Review was published on June 18, with an analysis cut-off date of May 31, 2026. As a result, the 24-hour FX market scheduled to launch from July is not yet reflected in actual operating data.

Item	2026 Assessment	Read-through
Capital flow restriction	++	Restrictions on capital inflows and outflows are not a major issue
FX market liberalization	-	The core unresolved item
Investor registration & account setup	-	LEI/IRC coexist; omnibus-account friction
Information flow	-	English disclosure and dividend information still insufficient
Clearing & settlement	-	Settlement by investor ID; limited practical adoption
Transferability	-	In-kind and off-exchange allowed but not yet established
Short selling	+	Improved after the resumption, but regulatory complexity remains
Availability of investment instruments	+	Improved from “-” to “+” in 2026
Stability of institutional framework	+	Not yet at the developed-market “++” level

The most important point is that while Korea improved from six “-” marks in 2025 to five “-” in 2026, the gap to the developed-market standard is still wide.

2.2 The Bottlenecks MSCI Flagged Directly

MSCI assessed that Korea “does not yet have a fully deliverable offshore currency market, and onshore FX constraints persist.” The 24-hour onshore trading for July 2026 and the offshore won settlement for 2027 were described not as completed mechanisms but as future initiatives.

On registration and account setup, LEI has replaced the IRC, but the coexistence of both systems leaves friction; on information flow, English disclosure expanded in 2026 but further steps remain for 2027, and the overhaul of dividend procedures has only been partially adopted. On clearing and settlement, overdrafts and omnibus accounts have been allowed but actual adoption is limited, and short selling has been possible since the ban was lifted in early 2025, but MSCI stated it will continue to monitor operational friction and the compliance burden.

2.3 The Procedurally Important Wording

When considering a promotion, MSCI specifies that the change in classification “should be viewed as irreversible.” In 2025, it set as conditions for a Korea consultation “resolution of all issues, full implementation of the reforms, and sufficient time for market participants to assess the effects." Applied strictly, this wording lowers the probability of an addition in 2026, because as of May 31 the core FX reform had not yet been implemented and the offshore won settlement network is targeted for 2027.

3. Probability Scenarios (Synthesis Of Two Analyses)

The first analysis put the Watch List probability at 35%; the Red Team check put it at 30%. The difference comes from an assumption about “how preemptively MSCI trusts the policy roadmap.” Because the MSCI accessibility assessment is based on actual investor experience (feedback from asset managers, custodians, and brokers) rather than a roadmap alone, the more conservative 30% is the more defensible figure. The synthesis is as follows.

Scenario	First analysis	Red Team adjustment	Form of the 6/24 announcement wording
A. Added to the Watch List / pool under review for reclassification	35%	30%	“under review for potential reclassification to Developed Markets”
B. Addition deferred + positive monitoring (base)	50%	53%	“continue to monitor implementation and market adoption”
C. Addition deferred + negative wording	14%	14%	“fundamental accessibility issues remain unresolved”
D. Immediate Developed promotion	1%	1%	An immediate reclassification announcement (procedurally near-excluded)

Headline synthesis: Watch List 30-35%, base case is positive monitoring (about 53%), immediate promotion 1%.

Upside case (factors that raise the Watch List probability)

On size and liquidity, Korea is already a market that can be discussed as a developed candidate. The government, too, explains that the reason it remains in EM is market accessibility.
In the 2026 review, availability of investment instruments improved from “-” to “+". The listing of Korea-index-linked derivatives on international exchanges has broadened the range of accessible products.
The government roadmap is concrete. 24-hour dollar-won spot trading is scheduled to begin on July 6, 2026 (with pilot trading on June 29), and the offshore won settlement network is targeted for January 2027. This directly addresses the won accessibility issue MSCI has flagged the longest.
The consensus is also more favorable than in the past. NH Investment & Securities suggested that, if added to the Watch List, inclusion would follow after about 24 months of observation in 2028, with potential medium-to-long-term passive inflows of about $29.2bn (about KRW 44 trillion). Note that this is a brokerage estimate, not an official MSCI figure.

Downside case (why the probability cannot be lifted above 50%)

On the official accessibility table, Korea still has five “-” items. In particular, the “-” on FX market liberalization is critical. Developed-market FX hinges on onshore and offshore depth, real-time price transparency, hedging instruments, and best execution during large-scale rebalancing.
The core reforms are pre-implementation or in their early stages as of the announcement date. As of May 31, there is a shortage of actual investor-experience data.
Applied strictly, MSCI’s 2025 consultation conditions (“issues resolved, full implementation, sufficient time to assess”) make an addition in 2026 a decision pulled forward.
External surveys are not uniformly optimistic, either. In a report citing Bloomberg, a majority of 15 investors and strategists said more time is needed to prove the durability of the reforms, and expected Korea to remain in EM for the time being.

4. A Read-Through Guide By Announcement Wording

When the announcement lands at 05:30 on June 24, check the following phrases first.

Strongly positive: “under review for potential reclassification to Developed Markets,” “consultation on potential reclassification of Korea,” “Korea added to the Watch List.” In this case the posterior probability shifts quickly toward “actual inclusion in 2028-2029.”
Neutral / base: “continue to monitor the implementation and market adoption,” “assess effectiveness over time.” An addition in 2026 fails, but the 2027 track stays alive.
Negative: “fully operational offshore FX market is not available,” “operational friction persists,” “limited adoption.” Beyond non-addition, market expectations could break. Such phrasing already appears under the Korea item in the 2026 accessibility review.

5. Timeline Scenarios

If added to the Watch List in 2026: The fastest path is 2026 addition → 2027 decision → 2028 inclusion. But given that the offshore won settlement network only launches in January 2027 and that MSCI emphasizes actual investor experience, 2026 addition → 2028 decision → 2029 inclusion is more realistic. NH Investment & Securities likewise views about 24 months of observation followed by a 2028 announcement and a 2029 actual transition as the realistic scenario.
If not added in 2026: The focus shifts to June 2027. If the 24-hour FX market operates stably, the offshore won settlement network launches, and English disclosure, omnibus accounts, transferability, and settlement practice are felt in practice, the probability of a Watch List addition in 2027 rises. This path naturally runs 2027 addition → 2028/2029 decision → 2029/2030 inclusion.

6. Market Impact And Trades (Illustrations And Watch Points)

The names below are illustrations of the mechanism, not buy recommendations. The key point is that although this looks like an “event to buy all of Korea,” the actual benefit concentrates in financials, brokers, and large liquid names that can monetize improved foreign access and a lower cost of capital.

6.1 The Most Actionable Idea: KB Financial’s Conditional Re-rating

An improvement in MSCI accessibility translates into a lower cost of capital for Korean financials. KB Financial has been estimated at 2026F BPS of KRW 180,958, ROE of 11.3%, and P/B of 0.81x; on the June 22 price of KRW 157,000, the simple-conversion P/B is about 0.87x. A bank is not directly MSCI infrastructure, but it monetizes the narrowing of the risk premium from improved won and foreign access in the simplest way.

Call: Wait → conditional Buy.
Entry: (strong case) If the announcement explicitly states “under review” or “Watch List” and the share price is at or below 0.90x 2026F BPS, i.e. about KRW 163,000 or lower, enter. (base case) If there is no Watch List but a “market adoption” phrase appears and a disappointment sell-off brings it to KRW 154,000 or lower (about 0.85x BPS), enter.
Catalyst: The June 24 MSCI announcement, the June 29 24-hour FX pilot trading, the July 6 launch of 24-hour spot trading, and the 2Q26 results and shareholder-return update.
Invalidation: CET1 falling below 13%, a retreat in the payout ratio, a spike in credit cost, or “fundamental accessibility issues remain” being emphasized in the announcement.
Asymmetry: The downside is a Watch List miss and bank regulation; the upside is a re-rating from 0.87x to 1.0x P/B. On a simple calculation, that is about KRW 181,000.

6.2 Samsung Securities: Strong Event Beta, But Large Regulatory Risk

Samsung Securities at KRW 114,000 on June 22, on a P/E of 10.54x and P/B of 1.32x, is no longer a cheap price. 1Q26 net income was KRW 450.9 billion, up 81.5% year over year, beating consensus. If Korea is added to the Watch List, turnover and WM beta attach the fastest, but if leveraged-ETF and margin-loan regulation tightens, the Q turns down immediately.

Call: Watchlist. If, after a Watch List addition, the same-day gain is within 5% and there is no regulatory headline, a short-term trade is permitted. On a miss, approach only at KRW 108,000 or lower (about 1.25x P/B).
Key variable: Korea’s supervisory authorities have signaled that approval of Samsung Electronics- and SK Hynix-linked leveraged ETFs was hasty, and are considering stabilization measures. This is event beta, not a fundamental compounder.

6.3 Samsung Electronics / SK Hynix: Not MSCI Beneficiaries But A Crowded AI Trade

For the large-cap semiconductors, the valuation is dominated by the AI memory cycle rather than MSCI. SK Hynix is up more than 340% in 2026, on June 22 overtook Samsung Electronics to become Korea’s most valuable company, and the two names now make up more than 50% of the KOSPI. The reading that “MSCI expectations” are the cause of the semiconductor rally is weak. The actual driver is AI memory pricing and supply-demand.

Call: Avoid as an MSCI trade; Wait only on the memory cycle. Buying on MSCI alone is a thesis mismatch. The alpha source for these names is the hyperscaler CAPEX and HBM bottleneck seen in the 2027 semiconductor consensus piece, not MSCI.

6.4 Market Reaction By Scenario

Watch List addition: Direct passive rebalancing does not occur immediately, but the expectation of a “narrowing Korea discount” gets priced in first. Financial holding companies, brokers, exchange infrastructure, and improving-shareholder-return names move first. That said, in the actual inclusion phase, the EM exit and a low DM weight could produce some passive outflows (a brokerage estimate of about $5.2bn, KRW 8 trillion).
Positive monitoring (deferral): Some of the pre-priced expectation can unwind, but if the 2027 roadmap stays alive the drawdown is limited. The market again becomes sensitive to semiconductor earnings, the won-dollar rate, foreign flows, and the pace of dividend reform.
Negative deferral: If MSCI explicitly states that the FX reform “does not replicate developed-market practice,” the probability of a 2027 retry also falls and short-term expectation money exits quickly.

7. Final Call

The core sentence

The real question on June 24 is not "Has Korea become developed," but "Does MSCI trust Korea's institutional reform enough to put it in the pool under review for reclassification." Rather than buying ahead of the announcement, the higher-expected-value approach is to confirm the announcement wording and then enter a cost-of-capital beneficiary like KB Financial on price terms.

In the June 24, 2026 announcement, an “actual Developed promotion” for Korea is not an event to expect. The probability is around 1%. The key event is a Watch List re-addition, and combining the two analyses, that is a 30-35% positive surprise. The bulk of the official evidence still points to “continue to monitor” rather than an “early addition,” because as of May 31 five “-” marks remain and the core FX reform is only implemented after the announcement.

The base case is therefore “addition deferred + positive wording for a 2027 re-assessment.” Even so, factoring in the improvement in availability of investment instruments, the 24-hour FX and offshore won settlement roadmap, and the WGBI inclusion experience, the probability of entering the Watch List is clearly higher than in the past. Execution is led not by chasing semiconductors but by KB Financial’s conditional re-rating, and all you need to confirm is whether the announcement puts Korea among the “markets under review.”

The figures in this piece cite public materials noted in the text — the MSCI 2026 accessibility review and market classification framework, the Ministry of Economy and Finance, Reuters, Yonhap, Mirae Asset Securities, Samsung Securities, NH Investment & Securities, KCMI, and others — while the probabilities, passive-inflow estimates, and per-name re-rating ranges are estimates. The names are illustrations of the analytical flow, not investment recommendations; the actual investment decision and responsibility rest with the investor.

8. Evidence Classification Appendix

[Fact]

The MSCI 2026 Annual Market Classification Review is scheduled to be announced shortly after 22:30 CEST on June 23, 2026 (05:30 KST on June 24).
The data cut-off date for the MSCI 2026 accessibility review is May 31, 2026.
Korea received an assessment that it does not yet have a fully deliverable offshore currency market and that onshore FX constraints remain.
The availability of investment instruments item improved from “-” to “+” in 2026.
24-hour dollar-won spot trading is scheduled to begin on July 6, 2026, with pilot trading scheduled for June 29.
Samsung Securities’ 1Q26 net income was KRW 450.9 billion, up 81.5% year over year.
SK Hynix overtook Samsung Electronics on June 22 to become Korea’s most valuable company.

[Inference]

The 30-35% range is the defensible band for the Watch List probability, because the core FX reform has not been verified with actual operating data as of the announcement date.
KB Financial expresses the MSCI accessibility improvement more purely than the semiconductors. At about 0.87x 2026F P/B, if the cost-of-capital decline thesis holds, there is room to approach 1.0x.
Samsung Securities benefits quickly but carries large regulatory beta, making it a lower-quality trade than KB Financial.
The alpha source for Samsung Electronics and SK Hynix is the AI memory bottleneck, not MSCI.

[Speculation]

If the announcement is at the “positive monitoring” level, disappointment selling could emerge, but if the 2027 path stays alive the drawdown in financial holdings may be limited.
On a Watch List addition, the short-term gain may be larger for the brokers, but on a 3-6 month holding-quality basis KB Financial is better.
In the actual DM inclusion phase, the EM exit and a low DM weight could produce some passive outflows.

[Blocked]

The final 2026 market classification result has not yet been announced.
MSCI’s official passive inflow/outflow magnitudes are unconfirmed (the KRW 44 trillion inflow and KRW 8 trillion outflow are brokerage-industry estimates).
The real-time FY26 consensus multiples for Samsung Electronics and SK Hynix cannot be fixed from official disclosures alone.

Sources

Q2 Earnings: Exporters That Can Still Benefit From KRW Weakness

Sun, 21 Jun 2026 21:35:00 +0900

Context This post follows our work on Korea market liquidity and narrow leadership, quality names to revisit in Q3, Korean Semiconductor Top 50 two-month alpha candidates and target-price gap screens. It belongs in the Exclusive Analysis Hub.

TL;DR

KRW weakness is supportive for Korean exporters into Q2 earnings. But the trade is not “buy all exporters.” The better screen is narrower: high dollar revenue, meaningful won-denominated costs, improving Q2 volume or pricing, and estimates that have not fully caught up.

Using FRED DEXKOUS daily observations, the 2026 Q2-to-date USD/KRW average through June 12 is 1,493.7. That is 1.9% above the Q1 average of 1,465.6 and 6.7% above the 2025 Q2 average of 1,399.8. The June 1-12 average is even higher at 1,526.2.

The final ranking from our screen is: Kolmar Korea, PSK Holdings, Wonik QnC, FADU, LIG Nex1, HD Hyundai Heavy Industries and Pearl Abyss. The cleanest mix is Kolmar Korea. Within semis, PSK Holdings and Wonik QnC look more comfortable than the already-crowded mega-cap and substrate names.

Core View

FX is not the variant perception by itself. The variant wedge is FX plus volume or pricing plus under-reflected consensus. Only stocks where those three line up should be treated as Q2 FX-alpha candidates.

1. The FX Setup Is Real

We recalculated the USD/KRW averages directly from FRED DEXKOUS daily observations. FRED is business-day data, so the averages use available observations.

Period	Observations	Average USD/KRW	Window
2026 Q1	61	1,465.6	Jan 2-Mar 31, 2026
2026 Q2 to date	52	1,493.7	Apr 1-Jun 12, 2026
2025 Q2	63	1,399.8	Apr 1-Jun 30, 2025
Jun 1-12, 2026	10	1,526.2	Jun 1-Jun 12, 2026

The math is straightforward: Q2-to-date is 1.9% above Q1 and 6.7% above the prior-year Q2. For exporters with dollar revenue and won costs, this creates a translation and margin tailwind, although actual P&L sensitivity still depends on hedging, dollar cost, inventory, shipment timing and revenue recognition.

2. Export Data Also Supports The Setup

The FX story would not matter if volume were weak. But recent export data also points in the same direction. Korea’s May exports rose 53.2% year over year, with semiconductor exports up 169.4%. May ICT exports rose 128.9%, ICT semiconductor exports rose 169.2%, and computers and peripherals rose 259.6%. For June 1-10, total exports rose 85.9%, semiconductors 205.8%, and computer peripherals 259.4%.

Data point	Growth	Read-through
May total exports	+53.2% YoY	Export demand is not weak
May semiconductor exports	+169.4% YoY	Semis have price, volume and mix support
May ICT exports	+128.9% YoY	AI, semis and peripherals lead
May ICT semiconductors	+169.2% YoY	Memory and system demand are strong
May computers/peripherals	+259.6% YoY	Server and storage read-through to check
Jun 1-10 total exports	+85.9% YoY	June started strong
Jun 1-10 semiconductors	+205.8% YoY	Semi momentum persisted into late Q2
Jun 1-10 computer peripherals	+259.4% YoY	AI server/storage linkage remains relevant

The investor question is not who benefits from FX. It is who benefits from FX without the benefit already being fully reflected in price and consensus.

3. Where The Structure Is Best

Segment	FX effect	Earnings-surprise probability	What to check
Semis/HBM/memory	Dollar revenue, partial won costs	High, but expectations are high	Pricing, shipments, HBM mix, inventory marks
MLCC/substrates/AI parts	Dollar revenue and premium mix	High if price and volume improve together	AI mix, product mix, orders
Defense exports	Dollar/euro revenue, won costs	Medium-high	Delivery schedule, revenue recognition, project margin
Shipbuilding	Dollar backlog	Medium	Hedging, long-cycle recognition, steel cost, progress
Games/software	Overseas revenue, won labor cost	Potentially high	New title, traffic, launch timing, monetization
Cosmetics/food exports	Export revenue and won costs	Medium-high	US/Japan/China channels, raw materials, marketing

The best setup has three traits: dollar revenue and won costs; improving Q2 volume or price; and recent consensus upgrades that have not fully moved the share price.

4. Priority Candidates

Source: Korea Invest Insights internal screen, as of June 19, 2026 for price, flow and consensus. Flow is trailing 20 trading days in KRW 100mn units. OP is in KRW 100mn.

Rank	Name	Core logic	Q2 OP	FY26 P/E	20D return	20D flow	View
1	Kolmar Korea	Cosmetics exports, won cost base, lower valuation burden	923	12.2x	-4.4%	Foreign +482 / Inst. +133	Buy candidate
2	PSK Holdings	HBM/back-end equipment, high margin, institutional flow	183	20.3x	+3.4%	Inst. +346	Buy candidate
3	Wonik QnC	Semi consumables, utilization recovery, less crowded	280	15.5x	+4.9%	Inst. +240	Buy candidate
4	FADU	SSD/storage re-rating, strong foreign/program flow	93	113.4x	-3.6%	Foreign +4,120 / Inst. -2,866	High-beta candidate
5	LIG Nex1	Defense exports, strong institutional and real-money flow	1,052	53.3x	+2.6%	Foreign +340 / Inst. +1,172	Valuation gated
6	HD Hyundai Heavy Industries	Shipbuilding, defense and nuclear option, KRW tailwind	9,933	23.8x	+14.4%	Foreign -1,440 / Inst. +1,746	Pullback candidate
7	Pearl Abyss	Overseas revenue, won labor cost and FX leverage	1,450	6.8x	-17.0%	Foreign +79 / Inst. -173	Event candidate

5. Name-By-Name Judgment

Kolmar Korea is the cleanest mix. It has export-linked beauty manufacturing exposure, a won cost base, a reasonable FY26 P/E of 12.2x, and a -4.4% 20-day price move despite positive foreign and institutional flow. The main diligence items are whether overseas demand flows through to margin and whether raw materials or marketing costs offset the FX benefit.

PSK Holdings and Wonik QnC are the preferred semi names for this specific FX-alpha screen. They are not as crowded as Samsung Electronics, SK Hynix, Samsung Electro-Mechanics, Daeduck Electronics, Simmtech or Korea Circuit. PSK Holdings adds HBM/back-end equipment exposure and institutional buying. Wonik QnC adds semi consumables and utilization recovery with a lower valuation burden.

FADU has the biggest right-tail risk but also the largest risk. The foreign buying is powerful, but the institution selling is equally important. This is not a clean long-only accumulation signal. It is a high-beta storage re-rating option that needs SSD controller revenue and customer diversification to show up.

LIG Nex1 has a good defense-export and FX structure, but valuation is the gating item. At 53.3x FY26 P/E, the company needs more than FX. It needs export orders, project margin and evidence that the European/NATO air-defense channel can become a durable pipeline.

HD Hyundai Heavy Industries benefits from a dollar backlog, but shipbuilding FX effects are slower because of hedging, progress accounting, contract duration, steel cost and labor. Pearl Abyss has clear overseas-revenue and won-cost leverage, but the stock is really a new-title event story, not a pure FX story.

6. Lower-Priority And Excluded Names

Name	Reason
iFamilySC	Low P/E and export-consumer logic, but flow size is small
VT	Oversold export-consumer candidate, but flow confirmation is needed
Silicon2	Export platform structure is attractive, but recent flow is weak
ST Pharm	Possible dollar-revenue structure, but biotech flow needs to recover
TLB	SOCAMM/substrate candidate, but foreign flow is weak
HD Hyundai	Undervalued shipbuilding holding company, more NAV/beta than direct exporter

Name	Reason for exclusion or chase discipline
Samsung Electronics / SK Hynix	Q2 surprise probability is high, but expectations and price already reflect a lot
Samsung Electro-Mechanics	Strong thesis, but price and target upside are less comfortable
Daeduck Electronics	Good company, but 20D +24.8% and limited target upside
Simmtech	SOCAMM expectation reflected, 20D +29.2%
Korea Circuit	20D +68.3%, weaker new-entry expected value
PSK / Eugene Tech / TES	Fundamentals may be good, but short-term price moves are already hot

7. Final Ranking

Rank	Name	Action	Reason
1	Kolmar Korea	Buy candidate	Best balance of FX, exports, valuation, pullback and flow
2	PSK Holdings	Buy candidate	HBM back-end exposure, institutional flow and lower chase risk
3	Wonik QnC	Buy candidate	Semi consumables recovery, lower valuation and less crowded setup
4	FADU	High-beta candidate	Foreign buying and storage re-rating, but high P/E and institution selling
5	LIG Nex1	Valuation gated	Defense-export quality is strong, valuation is the hurdle
6	HD Hyundai Heavy Industries	Pullback candidate	Good strategic exposure, but FX P&L recognition is slower
7	Pearl Abyss	Event candidate	FX structure is good, but the real driver is the new-title event

The portfolio-manager summary: Q2 KRW weakness is supportive for Korean exporters, but generic exporter exposure is too obvious. The better candidates are companies where FX works with volume, price or margin and where consensus has not fully caught up. Kolmar Korea is the cleanest current setup. In semis, PSK Holdings and Wonik QnC are more comfortable than crowded mega-cap or substrate names. FADU has convexity but needs limited sizing. LIG Nex1 needs valuation discipline.

Missing Evidence

This is a screen, not a final trade ticket. The next layer should check company-level hedge ratios, exact dollar revenue and dollar cost exposure, the broker notes behind Q2 consensus, trailing five-day detailed investor flow, and actual entry/exit levels.

Sources And Data

FRED DEXKOUS daily observations through June 12, 2026. FRED DEXKOUS
Korea May export data: total exports +53.2%, semiconductors +169.4%. KDI Economic Information Center
Korea May ICT exports: ICT +128.9%, ICT semiconductors +169.2%, computers/peripherals +259.6%. MOTIR
Korea June 1-10 export data: total exports +85.9%, semiconductors +205.8%, computers/peripherals +259.4%. KDI Economic Information Center
Stock-level price, flow and consensus: Korea Invest Insights internal screen, as of June 19, 2026. No internal filesystem path is exposed in this public post.

Who Pays For The 2027 Semiconductor Consensus? Hyperscaler Payability For Samsung, Hynix, Micron, Nvidia

Sun, 21 Jun 2026 20:00:00 +0900

Connecting the dots This piece is the synthesis of Why The AI Supercycle Runs Longer: IPO Funding And Memory/Storage, The $5.3tn Era Of AI Data Center CapEx, Goldman’s Token Demand vs JPMorgan’s Memory ASP, and Sam-Hy-Mu Parity: When Korean Memory Got Cheap Again. Where those earlier pieces looked at demand, pricing, and valuation separately, this one ties them together under a single question: “Can demand actually afford to pay for the 2027 earnings consensus?"

TL;DR

The 2027 consensus does not hinge on consumer electronics demand, but on whether hyperscalers, AI cloud, and sovereign AI keep absorbing the prices of HBM, GPUs, and server DRAM.
The verdict splits. Hyperscalers can pay conditionally, governments and sovereign AI have the political will to pay but unclear ROI, and PC and smartphone OEMs are already in the unpayable zone.
In numbers, the combined 2027E CAPEX of the big four (Microsoft, Google, Meta, Amazon) alone is roughly ~$782.2bn, against combined FCF of roughly ~$119.9bn. They can pay on an accounting basis, but the residual cash cushion is thin.
Nvidia’s FY2028 revenue consensus of roughly ~$551.7bn equals about 70.5% of the big four’s CAPEX. For that number to hold, you need not just the four but the entire global AI CAPEX pool including Oracle, OpenAI, sovereign AI, China, and enterprise.
The 2027E earnings of the three memory makers are even more sensitive to sustained memory ASPs and the durability of the HBM supply shortage than to hyperscaler CAPEX. That is why the most interesting asymmetry is Samsung Electronics (a conditional candidate where the peak discount may be excessive).

1. The Core Question And The Fiscal-Year Adjustment

The purpose of this analysis is singular: to judge whether the 2027 earnings outlook for Samsung Electronics, SK hynix, Micron, and Nvidia is mere sell-side optimism, or a level verifiable against the payability of end demand.

Five core questions:

What are the 2027E revenue and earnings outlooks, and what does the picture look like once we adjust for fiscal-year differences?
Can you simply add up the outlooks for the three memory makers and Nvidia, or does double-counting creep in?
Can hyperscalers’ 2027E CAPEX, FCF, and external financing capacity cover this revenue?
Are governments and sovereign AI large enough to independently sustain 2027 earnings?
Can PC and smartphone OEMs pass rising memory prices through to end consumers?

A note on fiscal years: Samsung Electronics and SK hynix close in December, so they line up with CY2027. Micron’s FY2027 ends in August 2027. Nvidia’s FY2027 ends in January 2027, so it largely captures 2026 results; to see “2027,” you have to look at FY2028 (ending January 2028) alongside it.

2. The 2027E Earnings Consensus Snapshot

The figures below are not company guidance but external consensus and market data (per MarketScreener, closing prices of June 19, 2026). KRW figures are noted at roughly USD 1 = KRW 1,454 for ease of comparison.

Company	Comparison period	2027E revenue	2027E operating profit / EBIT	2027E net income	Valuation	First-pass verdict
Samsung Electronics	CY2027	KRW 856.8tn (~$589.0bn)	KRW 460.1tn	KRW 373.4tn	KRW 354,000, 2027E P/E 6.12x	Consensus is aggressive; the stock is discounted as a “temporary peak”
SK hynix	CY2027	KRW 483.8tn (~$333.0bn)	KRW 377.1tn	KRW 297.1tn	KRW 2,764,000, 2027E P/E 6.71x	HBM leadership is already largely priced in. A 6-7x peak P/E signals disbelief in durability
Micron	FY2027	~$190.98bn	~$157.20bn	~$122.92bn (EPS ~$111)	$1,133.99, FY2027E P/E ~10.2x	Maximum memory beta, but with 2028 slowdown coming into view, the lowest margin of safety for new buyers
Nvidia	FY2028	~$551.69bn	~$362.0bn	~$304.84bn (EPS ~$12.85)	$210.69, FY2028E P/E ~16.4x	Not expensive if the numbers are right. The issue is the scale of customer CAPEX those numbers require

The dispersion in forecasts is wide. For SK hynix in particular, KB’s estimated 2027E operating profit was about 71% higher than the FnGuide consensus at the time (KRW 209.3tn). When the market’s 2027 picture for the same company is this divided, it means the consensus is not a “settled future” but a “high-conviction assumption about price durability.”

And the 2027E operating margins for the three memory makers include figures in the 70% range. That is not a typical memory cycle, but a number that is only possible when a structural shortage plus long-term contracts hold.

3. “Don’t Add Them Up”: The Double-Counting Trap

Start with the most important interpretation. The HBM, DRAM, and NAND revenue of the three memory makers and Nvidia’s GPU revenue partially overlap from the standpoint of end-demand spending.

Nvidia’s revenue embeds the cost of systems, boards, and networking, including HBM. The memory makers’ HBM revenue enters as the upstream of that system cost, then gets reflected in the price of the AI servers hyperscalers buy. So if you simply add “Nvidia revenue + memory makers’ revenue” as a final spend figure, you count the same dollars twice.

From an investment standpoint, the revenue at each node can all be real. But the end buyer has to pay for all of it within the same AI data center budget: GPUs, HBM, commodity DRAM, SSDs, networking, servers, power, cooling, and construction. So rather than summing revenue, it is more rigorous to compare against the “total AI data center CAPEX pool."

4. Payability Test By Demand Source

4.1 Hyperscalers: Conditionally Payable

Summing the big four’s 2027E CAPEX and FCF gives the following (on a consensus basis).

Demand source	2027E CAPEX	2027E FCF	Interpretation
Microsoft	~$170.4bn	~$59.3bn	Ample investment capacity, but FCF dilution
Alphabet	~$231.1bn	~$25.2bn	CAPEX absorbs most of FCF
Meta	~$159.5bn	~$11.4bn	A structure where residual FCF gets very thin
Amazon	~$221.2bn	~$24.0bn	Simultaneous investment in AWS, logistics, and AI
Total	~$782.2bn	~$119.9bn	Payable on an accounting basis, but limited cushion

The arithmetic is simple. CFO proxy = CAPEX + FCF = $782.2bn + $119.9bn = roughly ~$902.1bn. In other words, the claim that “hyperscalers have no cash flow to pay 2027 AI costs” is wrong. The problem is the opposite. Too much cash flow is being reallocated into AI CAPEX. In this structure, buybacks, dividends, M&A, existing infrastructure replacement, power contracts, and data center leases all compete for the same cash.

And the four alone are not enough. Nvidia’s FY2028 revenue of ~$551.69bn equals about 70.5% of the four’s $782.2bn CAPEX (551.69 / 782.2 = 70.5%). For Nvidia’s revenue to materialize as the consensus expects, the four must pay not only for GPUs but also for non-GPU infrastructure (power, cooling, servers, networking, construction) at the same time, so you need the entire global AI CAPEX pool including Oracle, CoreWeave, OpenAI/Stargate, xAI, Chinese cloud, Middle Eastern sovereign AI, and enterprise.

Goldman Sachs estimates 2026-2031 AI CAPEX at roughly $7.6tn, and Morgan Stanley estimates 2027E global data center spending at roughly ~$812.7bn. The scale is within range to explain the consensus, but both also present a financing gap: hyperscaler cash flow alone is insufficient, requiring external financing such as corporate bonds, ABS, and private credit.

4.2 Economic Payability: Different From Cash

What you can pay for in cash and what you can afford economically are different. Goldman Sachs typically puts the useful life of AI chips at 4-6 years. If 2027 AI CAPEX is $0.9tn-$1.1tn, then depreciation alone implies an annual economic cost of roughly ~$150bn-$275bn per year ($0.9tn ÷ 6 years = $150bn / $1.1tn ÷ 4 years = $275bn). On top of that come power, cooling, leasing, networking, operations, model-training costs, and financing costs.

So hyperscalers’ payability ultimately depends on how fast AI inference revenue, enterprise AI seat expansion, ad/search/commerce improvements, and cloud GPU rental income absorb this depreciation. Through 2027 they can hold on with cash flow and the financing markets. But to keep investing at the same pace in 2028-2029, AI monetization has to be far clearer.

Verdict: Cash payability is conditionally TRUE; payability on an economic ROIC basis is MIXED.

4.3 Government And Sovereign AI: Supplementary Demand, Not A Standalone Payer

The demand is real. The EU announced an AI investment mobilization totaling EUR 200bn through InvestAI, and Saudi Arabia’s HUMAIN envisions an AI factory of up to 500MW and hundreds of thousands of GPUs with Nvidia (with a Phase 1 mention of 18,000 GB300 units). OpenAI’s Stargate is also a $500bn, 10GW program over four years (private-led, not a pure government budget).

[Inference] Government and sovereign AI have lower price elasticity than private cloud (driven by defense, data sovereignty, and industrial-policy goals). They therefore act as a demand buffer for Nvidia, HBM, and power infrastructure. But in terms of scale and pace of execution, they are closer to subsidies, long-term power contracts, credit enhancement, and anchor customers than substitutes for hyperscaler CAPEX. Announced amounts should not be read as revenue outright. China in particular faces limited direct benefit to Nvidia and Micron due to export controls and localization.

Verdict: The claim that government demand alone sustains the 2027 supercycle is FALSE. It matters as downside protection when combined with hyperscalers.

4.4 Consumer Electronics OEMs: Close To Unable To Pay

[Fact] Gartner forecast 2026 PC shipments at -10.4%, smartphones at -8.4%, and DRAM+SSD prices at +130%, which it expects to push PC prices +17% and smartphone prices +13%, with the memory share of a PC’s BOM rising from 16% in 2025 to 23% in 2026. IDC likewise sees smartphones and PCs responding with price hikes, spec cuts, and launch delays as AI data centers absorb wafers.

[Inference] Consumer electronics OEMs do not “pay” memory prices so much as respond with lower shipment volumes, reduced content per device, higher selling prices, and a downward model mix shift. Premium devices can absorb ASP increases, but the mid-to-low-end market has high price elasticity, so demand destruction appears. In other words, PCs and smartphones are not the party justifying the 2027 memory consensus. That consensus effectively depends on demand for AI servers, HBM, enterprise SSDs, and DDR5 RDIMM.

Verdict: Not the core payer, but the victim of rising prices. FALSE.

5. Supply Bottleneck: 2027 Is A P Problem More Than A Q Problem

[Fact] Gartner forecasts 2026 semiconductor revenue of $1.320tn and 2027 of $1.555tn, with memory revenue of ~$633.3bn in 2026 and ~$748.1bn in 2027. It sees 2026 DRAM prices +125% and NAND +234%, with meaningful price relief potentially limited until the end of 2027.

[Fact] TrendForce sees the HBM share of wafer input potentially rising to 18% at end-2025, 22% at end-2026, and 30% at end-2027, but the HBM bit-supply share may stay at only around 13% even in 2027. This means HBM eats up a lot of DRAM wafers, yet total bit supply grows only modestly.

So 2027 earnings are a question less of demand volume Q than of price P and product mix, and of who captures the incremental margin generated under constrained supply. That is why, even if commodity demand from PCs and smartphones slows, the HBM and server DDR5 shortage will not ease immediately, since wafer allocation toward AI servers has already moved.

6. Korea Read-Through And Stock Verdicts (Illustrative Watch Points)

The names below are illustrative basket examples and watch points. They are an organization of investment hypotheses, not buy recommendations.

Stock	Call	One-line thesis	Key risk
Samsung Electronics	Watchlist → conditional Buy	If HBM4/HBM4E catch-up and DS operating leverage are confirmed, an early-6x 2027E P/E may be an excessive “peak discount”	HBM4 supply delays, failure to secure key-customer allocation, sharp drop in DS operating margin
SK hynix	Wait	HBM leadership is the most certain, but the stock is already largely priced in. Earnings durability is attractive, but the margin of safety for new entry is low	Share/margin decline as rivals improve HBM4 yields, order adjustments
Nvidia	Watchlist	An FY2028E P/E in the 16x range is low if the numbers are right, but the burden of validating customer CAPEX and AI ROI has grown	In-house ASIC penetration, China regulation, falling GPU rental income, power bottlenecks
Micron	Avoid for new buys	Memory beta is large, but an FY2027E P/E in the 10x range plus 2028 slowdown risk make the asymmetry weaker than Korean memory	Early DRAM/NAND peak-out, inferior HBM share, FY2028E EPS downgrades

The best setup is a “bottleneck node with pricing power that is still discounted as a peak." By that criterion, the most interesting one right now is Samsung Electronics. The market is still discounting HBM4, packaging, and memory price durability as a “peak cycle,” so if HBM4 catch-up is confirmed, there is asymmetry for an early-6x P/E to re-rate. SK hynix is high quality but the entry price is the issue; Nvidia is a quality asset but customer ROI validation has to come first; and Micron has the weakest asymmetry for new buying.

7. Red Team: When This Conclusion Is Wrong

Rate and credit risk: If rates rise or credit spreads widen, hyperscaler CAPEX wobbles first, given its heavy reliance on external financing.
ASIC insourcing: Google’s TPU, Amazon’s Trainium, and Meta’s in-house chips reshape Nvidia’s ASP ceiling and the HBM demand structure beyond 2027.
Memory capacity additions: If Samsung and Micron expand HBM capacity aggressively and SK hynix’s spread narrows, 2027E margins plunge.
“Revenue holds, FCF and monetization disappoint”: The most dangerous signal is not CAPEX cuts but CAPEX held steady with weak AI monetization. Even if 2027 revenue holds, order intensity could break sharply in 2028.

8. Monitoring Checklist

Indicator	Why it matters
Big four CAPEX revisions	The top-level demand proxy for Samsung, Hynix, Micron, and Nvidia earnings
FCF after CAPEX	The key line between “able to pay” and “overspending”
Nvidia data center backlog / lead times	Visibility into FY2028 $550bn+ revenue
HBM LTAs, prepayments, capacity reservations	The most direct evidence of memory ASP durability
DRAM/NAND contract price vs spot	Gauging overheating and the top of the commodity memory supercycle
Cloud GPU rental income	A proxy for actual monetization of AI infrastructure
Inference cost and AI service margins	Hyperscalers’ economic payability
PC/smartphone unit elasticity	Whether consumer electronics demand destruction is occurring
Data center power permitting and PPAs	Risk of supply delays and CAPEX deferral

9. Final Conclusion

The core sentence

What pays for the 2027 semiconductor consensus is not the consumer, but hyperscaler CAPEX. So the core question is not "Is there AI demand?" but "Do AI revenue and GPU utilization rise fast enough to justify CAPEX even beyond 2027?"

The high-earnings outlook for the four companies in 2027 is not mathematically impossible. The CAPEX of hyperscalers and the broader AI infrastructure ecosystem has already entered the $0.8tn-$1.0tn range, and including external financing, it can pay for 2027 supply-chain revenue. But three conditions must hold simultaneously.

Hyperscaler CAPEX sustains at least $0.8tn-$1.1tn in 2027, and players outside the four (Oracle, OpenAI, sovereign AI, China, enterprise) fill in Nvidia revenue and the non-GPU costs together.
Memory prices are locked in not as a “temporary shortage premium” but through long-term contracts, prepayments, and capacity reservations.
AI infrastructure converts into actual revenue. To keep investing at the same pace in 2028-2029, inference revenue, AI SaaS, ad/search improvement, and GPU utilization must justify depreciation and power costs.

In short, the 2027 earnings outlook holds as a “hyperscaler-led conditional-payability scenario," but it does not hold on government demand and consumer electronics OEM demand alone. In particular, the earnings of the three memory makers are even more sensitive to sustained memory ASPs and the durability of the HBM shortage than to hyperscaler CAPEX. So what to watch right now is not an “AI semiconductor basket,” but the bottleneck node that has pricing power and is still discounted as a peak.

The figures in this piece cite the MarketScreener consensus, company official results, and Gartner, IDC, Goldman Sachs, Morgan Stanley, TrendForce, Reuters, Mirae Asset, and KB materials as noted in the text, and all are market expectations as of mid-2026, not actual results. Stock names are examples illustrating the flow of analysis, not investment recommendations; actual investment decisions and responsibility rest with the investor.

Sources

KOSPI 60-Day Disparity At +28.6%: Not A Top Call, But A Partial Risk-Reduction Signal

Sun, 21 Jun 2026 02:35:00 +0900

Context This is a follow-up to How Rare Is It To Beat The Pure KOSPI Benchmark?, ETF Flow Is Leading The Korean Market, Korea Has Liquidity, But Breadth Has Broken and Have Foreign Investors Returned?. Related hubs: Exclusive Analysis and Korea Daily Market Hub.

TL;DR

As of June 19, 2026, KOSPI closed at 9,052, its 60-day moving average was 7,039, and its 60-day disparity was +28.6%. That is clear medium-term overspeed.
But disparity overheat is not automatically a market top. Reproducing 355 trading days from January 2, 2025 to June 19, 2026 shows that 20/60/120-day overheat signals were followed by positive average 5-day and 20-day returns. In a strong trend, overheat can first mean acceleration.
However, when 60-day disparity rises above +20%, the probability of a -5% drawdown within the next 10 trading days rises from a 16% baseline to 63%.
That 63% must be handled carefully. The 41 signal days are not 41 independent observations. They compress into 8 episodes, and only 6 had completed forward windows.
The current read is therefore: partial risk reduction is justified, but this is not a top call. Disparity and volatility gates are on; the 20-day moving-average slope is still rising.

Core Point

KOSPI's +28.6% 60-day disparity does not say "this is the top". It says the trend has become fast enough that investors should reduce chase risk, harvest part of the risk budget, and prepare for a pullback window.

1. What was tested

The analysis independently reproduced a KOSPI disparity framework. The questions were:

Do the current KOSPI disparity numbers tie out?
Did disparity overheat historically raise correction probability?
Should the current reading be treated as a sell signal?

Item	Setting
Index	KOSPI, `^KS11`
Period	2025-01-02 to 2026-06-19
Sample	355 trading days
Price	Close
Disparity	`close / N-day moving average - 1`
Windows	20, 60, 120 days
Correction definition	A -5% or worse low within the next 10 trading days
Baseline	Same correction frequency across all valid days

This is a market-risk framework, not a trading instruction.

2. Current state: the core numbers tie out

Metric	Original	Reproduction	Read
KOSPI close	9,052	9,052	Matched
MA20 / disparity20	8,322 / +8.8%	8,322 / +8.77%	Matched
MA60 / disparity60	7,039 / +28.6%	7,039 / +28.61%	Matched
MA120 / disparity120	6,088 / +48.7%	6,088 / +48.69%	Matched
MA20 5-day slope	+4.25%	+4.25%	Matched
RSI(14)	57	SMA 57.3 / Wilder 65.9	Formula difference
ATR(20)%	3.61%	Wilder 4.15 / SMA 4.70	Formula difference

The key moving-average and disparity numbers tie out almost exactly. The RSI difference is explainable by formula choice: SMA RSI is close to 57, while Wilder RSI is 65.9. Both remain below 70. The conclusion is unchanged: this is not primarily an RSI overbought signal. It is a medium-term disparity signal.

3. Overheat did not predict tops well

Signal	Signal days	Avg +5D return	Avg +20D return	Negative +20D share
Disparity20 ≥ 7%	74	+2.68%	+9.07%	24%
Disparity60 ≥ 12%	120	+2.26%	+6.88%	22%
Disparity120 ≥ 20%	107	+2.55%	+9.11%	23%
Baseline	335	+1.86%	+7.83%	19%

This is the first uncomfortable point. In this 2025-2026 trend, overheat did not reliably identify the top. Average returns after overheat signals were still positive.

That makes sense in a strong trend. A large distance above moving averages often means money is still being forced into the leaders.

4. But 60-day disparity did raise pullback odds

Condition	Signal days	-5% correction within 10 trading days
Baseline	All valid days	16%
Disparity60 ≥ 15%	89	36%
Disparity60 ≥ 18%	54	48%
Disparity60 ≥ 20%	41	63%
Disparity60 ≥ 22%	33	67%
Disparity60 ≥ 25%	26	65%

The signal is useful, but not as a top-calling tool. It is a pullback-risk warning. Above +20% disparity60, the probability of a -5% drawdown within 10 trading days rose about fourfold versus the baseline.

5. The statistical caveat: 41 days are only 8 episodes

The 41 days of +20% disparity are not independent. Grouped into runs, they become 8 episodes:

#	Episode	Signal days	Peak disparity60	-5% correction within 10D
1	2025-11-03	1	+22.3%	Yes
2	2026-01-28 to 01-30	3	+21.6%	Yes
3	2026-02-03 to 02-04	2	+22.8%	Yes
4	2026-02-12 to 03-03	10	+33.6%	Yes
5	2026-05-04 to 05-18	10	+32.1%	Yes
6	2026-05-21 to 06-05	10	+36.6%	Yes
7	2026-06-09	1	+21.6%	Not complete
8	2026-06-15 to 06-18	4	+29.9%	Not complete

The six completed episodes all produced a -5% correction. That is strong, but the sample is tiny. The right lesson is not to hard-code exact thresholds. The right lesson is to treat extreme 60-day disparity as a risk-budget tightening signal.

6. Which indicators mattered?

Indicator	Pre-correction average	Full-sample average	Difference	Signal quality
Disparity60	+20.35	+11.33	+9.02	Strong
ATR%(Wilder)	+2.56	+2.03	+0.53	Medium to strong
MA20 slope	+2.74	+1.81	+0.93	Medium
Disparity20	+5.06	+3.50	+1.56	Weak
RSI(Wilder)	+64.53	+65.46	-0.93	Little signal

The ranking is intuitive. A 20-day average follows price quickly, so it gets digested fast. A 60-day average moves more slowly, so extreme 60-day disparity captures medium-term overspeed better. RSI was not the useful axis in this sample.

7. The current three-gate read

Gate	Rule	Current value	Status
Medium-term overspeed	Disparity60 ≥ 18-20%	+28.6%	On
Volatility expansion	ATR ≥ 2.2%	4% range	On
Trend deterioration	MA20 slope slowing or turning down	+4.25% rising	Not yet

Two gates are on. That argues against aggressive chasing.

But the third gate is not on. The 20-day moving-average slope is still rising. Therefore the current signal is not a broad-market exit signal. It is a partial rebalancing signal.

8. Use dynamic levels, not fixed index levels

One common mistake is to hard-code an index level. If the 60-day moving average is 7,039 and the re-entry zone is roughly 8% above it, investors may anchor on something like KOSPI 7,600.

But moving averages move. In a strong trend, the 60-day average will keep rising.

Static frame: KOSPI around 7,600
Dynamic frame: the then-current 60-day moving average plus roughly 8%

The dynamic frame is better because the pullback zone moves up as the trend matures.

9. Why this matters more in a narrow market

The current KOSPI overheat is not a broad-market overheat. It is heavily influenced by Samsung Electronics and SK Hynix.

That links directly to the pure KOSPI benchmark analysis. In 2026, KOSPI has behaved less like the average Korean stock and more like a concentrated AI-memory benchmark.

So the next useful test is not just KOSPI disparity. It is:

Question	Why it matters
Is KOSPI overheated?	Index-level risk budget
Is KOSPI ex Samsung/SK Hynix overheated?	Broad-market versus mega-cap effect
Are memory mega-caps driving the signal?	Whether the signal is narrow leadership
Are second-line stocks also stretched?	Whether this is a broad extension

10. Practical usage

Situation	Read	Action frame
Disparity60 12-18%	Strong trend	Hold and verify trend
Disparity60 18-20%	Medium-term overspeed	Reduce new chase risk
Disparity60 20%+ with high ATR	Pullback probability rising	Reduce part of risk budget
Disparity60 20%+ and MA20 slope slows	Distribution risk rising	Consider further reduction
Disparity60 cools to 8-12%	Overheat easing	Re-underwrite with flow and earnings
Break below the 60-day average	Different regime	Stop using simple pullback-buy rules

The one-line rule: 60-day disparity overheat is not a short signal. It is a rebalancing signal inside a strong trend.

Final view

KOSPI’s +28.6% 60-day disparity is real overheat. But the reproduced data does not support treating it as a confirmed top.

Overheat signals were followed by positive average returns in this trend. At the same time, +20% or greater 60-day disparity materially raised the odds of a -5% pullback within 10 trading days.

The right conclusion is balanced:

This is not the zone to abandon the bull market. It is the zone to stop chasing, harvest part of the risk budget, and wait for either trend deterioration or a cleaner pullback.

Coverage Health

As of: June 19, 2026 close.
Data: KOSPI ^KS11, 2025-01-02 to 2026-06-19, 355 trading days.
Reproduced: 20/60/120-day disparity, RSI 14, ATR 20%, 5-day slope of the 20-day moving average, and -5% forward-10D correction events.
Confirmed: current disparity values, post-overheat forward returns, disparity60 threshold correction rates, RSI formula difference.
Not yet confirmed: 2020-2022 cross-validation, KOSPI ex Samsung Electronics/SK Hynix disparity, and robustness of re-entry rules in a true bear-market transition.
This is a market-risk framework, not investment advice.

CXMT IPO And Memory Price Risk: HBM Is Not The First Place To Break

Sun, 21 Jun 2026 02:20:00 +0900

Context This is a follow-up to the HBM4E 12-high battle, Jensen Huang’s HBM4 three-vendor signal, Samsung-Hynix-Micron parity, and the longer AI supercycle framework. Related hubs: AI HBM hub and Exclusive Analysis hub.

TL;DR

CXMT’s STAR Market IPO is a real policy-capital event for China’s DRAM industry. China’s securities regulator approved the registration, and the Shanghai Stock Exchange prospectus points to wafer-line upgrades, DRAM technology upgrades and next-generation DRAM R&D.
But this is not an immediate HBM price-collapse signal. HBM pricing depends on qualification, TSV stacking, base die, thermals, packaging, long-term agreements and roadmap synchronization. It is not priced like client DDR5.
The 2026 risk is not outright memory price decline. It is the peak-out of the rate of price increases. HBM and high-capacity server DRAM should remain more resilient; PC DDR5, mobile LPDDR and consumer NAND are the first areas to watch.
The 2027 risk is product bifurcation: AI server memory may remain tight while client DRAM and NAND face more supply pressure.
SK Hynix remains the most defensive AI memory long. Samsung Electronics has the largest HBM rerating option if HBM4/HBM4E execution works. Micron remains the U.S. AI memory proxy and valuation benchmark.

Key Framing

The wrong question is "Will Chinese DRAM supply break DRAM prices?" The better question is "Which product, for which customer segment, at what point in time, and through which new supplier will see pricing pressure?"

1. Why the CXMT IPO matters

ChangXin Memory Technologies, or CXMT, is the core company behind China’s DRAM self-sufficiency push. In June 2026, the China Securities Regulatory Commission approved CXMT’s STAR Market IPO registration. The Shanghai Stock Exchange prospectus describes CXMT as China’s largest DRAM manufacturer and one of the global top-tier capacity players, while also noting that it still trails Samsung Electronics, SK Hynix and Micron in technology and scale. (CSRC approval, SSE prospectus)

The use of proceeds is the core point.

Use of proceeds	Amount	Investment meaning
12-inch wafer manufacturing-line upgrades	RMB 7.5bn	Better efficiency and lower unit cost
DRAM process technology upgrades	RMB 13.0bn	Process migration and product competitiveness
Next-generation DRAM R&D	RMB 9.0bn	Long-term high-value-product optionality
Total	RMB 29.5bn	Structural strengthening of China’s DRAM supply base

This is not just a listing. It is a policy-backed attempt to improve cost, process capability and product breadth in DRAM. That makes CXMT important. But it should be framed correctly: CXMT is not yet the company that breaks HBM pricing. It is the supplier that can cap pricing power first in client DDR5, LPDDR and selected commodity DRAM markets.

2. DRAM is no longer one product

The memory cycle should not be reduced to one “DRAM price” number.

Product	Main customers	Pricing mechanism	CXMT impact
HBM3E/HBM4/HBM4E	NVIDIA, AMD, hyperscaler ASICs	Qualification, stacking yield, LTAs, roadmap lock-in	Very low
High-capacity RDIMM/MRDIMM	AI servers, data centers	CSP contracts and tight server supply	Low
Standard server DDR5	Enterprise servers	Server demand and OEM qualification	Low to medium
PC DDR5	PC OEMs, module vendors	Spot/contract prices, inventory, China localization	Medium to high
Mobile LPDDR	Smartphones and tablets	Mobile OEM procurement and power efficiency	Medium
Enterprise SSD	Data centers	AI storage demand and NAND allocation	Medium
Consumer NAND	PC, mobile, retail	Consumer demand and inventory	High

HBM is a stack-and-package product tied to AI accelerator qualification. Customers care about power, thermals, speed, yield, packaging reliability and supply roadmap. Client DDR5 is much more price-sensitive and inventory-sensitive. That is where CXMT matters first.

3. DDR5 and LPDDR need to be separated

Item	DDR5	LPDDR
Main use	PCs, servers, workstations	Smartphones, tablets, low-power notebooks and edge AI devices
Key requirement	Bandwidth, capacity, platform compatibility	Low power, small footprint, mobile packaging
Price sensitivity	PC and server OEM inventory	Smartphone shipments and mobile AP cycles
CXMT impact	Can appear quickly in client DDR5	Can expand gradually through China’s mobile supply chain

If CXMT wins more client DDR5 share, it can reduce the pricing power of Samsung, SK Hynix and Micron in the China PC and module chain. That does not automatically translate into HBM4 pricing pressure.

4. Cost convergence is happening, but the gap remains

One public die-size comparison illustrates the direction. It is not a precise cost model, but it shows why CXMT can pressure commodity pricing while still being behind the leaders in leading-edge economics.

Item	CXMT G4 16Gb DDR5	Micron 1β 16Gb DDR5	Read-through
Die size	66.99mm²	35.83mm²	CXMT is about 1.87x larger
Gross die per wafer	Lower	Higher	Larger die reduces wafer output
Illustrative yield	70%	90%	Actual yields are not public
Relative good-die cost	About 2.4x	1.0x baseline	CXMT may still be cost-disadvantaged

The point is not the exact number. The point is that CXMT’s first pricing impact should appear where policy support and domestic demand can offset cost disadvantages: client DRAM, mobile DRAM and selected standard products.

5. 2026 risk: peak-out in price increases

Memory pricing is still strong. TrendForce expects conventional DRAM contract prices to rise 58-63% QoQ in 2Q26 and NAND Flash contract prices to rise 70-75% QoQ, driven by AI server demand, long-term supply agreements and suppliers reallocating capacity toward server applications. (TrendForce)

The investment question is not whether prices are rising now. The question is when the rate of increase starts to slow.

Product	2026 price-decline risk	Reason
HBM3E/HBM4/HBM4E	Low	Qualification, supply shortage and AI GPU roadmaps
High-capacity RDIMM/MRDIMM	Low	AI server demand and CSP priority
Standard server DDR5	Low to medium	Demand is good, but standardization raises sensitivity
PC DDR5	Low to medium	OEM inventory and China supply can matter
Mobile LPDDR	Medium	Smartphone demand and China supply-chain adoption
Consumer NAND	Medium	Post-rebound inventory risk

For 2026, the base case is not a memory price collapse. It is a strong price environment with a rising need to watch the peak in price momentum.

6. 2027 risk: bifurcation

By 2027, product separation may become much more visible.

Product	2027 price risk	Key variable
HBM4/HBM4E	Low to medium	Three-vendor multi-sourcing and HBM ASP discipline
High-capacity RDIMM/MRDIMM	Low to medium	AI server shipments and CSP LTAs
SOCAMM/server LPDDR	Medium	Inference-server adoption and supplier expansion
Standard server DDR5	Medium	General server demand and inventory
PC DDR5	Medium to high	China supply and OEM bargaining power
Mobile LPDDR	Medium to high	Localization in China’s smartphone chain
Consumer NAND	Medium to high	Capacity response and inventory

This means the equity question becomes: can HBM and AI server memory mix offset client DRAM and NAND pressure?

7. Company implications

SK Hynix

SK Hynix is the cleanest defensive AI memory exposure. CXMT is not the main risk for SK Hynix. The main risks are HBM ASP discipline, multi-sourcing by NVIDIA and hyperscalers, and any pause in AI capex.

Samsung Electronics

Samsung has the biggest rerating option if HBM4/HBM4E catch-up works. But Samsung also has more exposure to broad DRAM, mobile memory, NAND and consumer electronics. In other words, Samsung needs HBM rerating to offset commodity-memory sensitivity.

Micron

Micron remains the U.S.-listed AI memory proxy and a valuation benchmark for Korean memory stocks. If Micron keeps its premium while Samsung and SK Hynix EPS remain intact, Korea’s memory discount becomes a live opportunity. If Micron’s premium breaks, Korean memory will likely feel the same short-term pressure.

8. What to monitor

Watch item	Why it matters
DDR5 spot falling for four weeks or -10% monthly	Early client DRAM peak-out signal
DDR5 contract-price QoQ slowdown	Spot pressure migrating into contract pricing
PC and smartphone OEM inventory	Demand-side pressure on client DRAM/LPDDR
CXMT wins at China OEMs	Domestic price ceiling moving lower
CXMT qualification at global OEMs	China risk becoming global pricing pressure
China server RDIMM quality	Whether pressure extends beyond client DRAM
HBM4/HBM4E contract pricing	AI memory premium durability
HBM wafer input and capex	Whether shortage persists
CSP capex guidance	The top-down variable for AI memory
eSSD versus client NAND pricing	NAND bifurcation signal

Final view

CXMT’s IPO is important, but it is not an immediate HBM bear case. It is a signal that the memory cycle is splitting.

China supply can cap client DDR5, LPDDR and some commodity DRAM pricing. HBM and AI server DRAM remain in a separate pricing structure for now. The key equity question is whether HBM and AI server mix can absorb future weakness in client DRAM and NAND.

In portfolio language: SK Hynix is more about HBM price discipline and AI capex than about CXMT. Samsung Electronics must prove HBM rerating can offset commodity-memory exposure. Micron remains the U.S. AI memory proxy and the valuation yardstick.

Coverage Health

As of: June 21, 2026.
Verified sources: CSRC registration approval, SSE prospectus, TrendForce memory price outlook.
Supporting sources: China media reports on CXMT performance and share; reporting on Chinese DRAM adoption in the PC supply chain.
Not verified: CXMT actual yields and cost structure, HBM contract prices, exact global OEM qualification status, 2027-2028 product-level supply contracts.
This is a product-level price-risk framework, not a short-term price target note.

How Rare Is It To Beat The Pure KOSPI Benchmark In This Narrow Market?

Sat, 20 Jun 2026 10:30:00 +0900

Context
This is a follow-up to our narrow-market, foreign-flow and semiconductor relative-alpha work. Related hub: Exclusive Analysis Hub.

TL;DR

Korea looks like a roaring bull market at the index level. As of June 19, 2026, KOSPI traded around the 9,050 level and was up roughly 115% YTD.
Under the index, the market is extremely narrow. In our proprietary 2,730-stock Korea dataset, the median stock return was -7.6%, only 37% of names were up, and only about 30% traded above their 200-day moving average.
Monte Carlo simulations show how hard the pure KOSPI benchmark was to beat. A random equal-weight 20-stock portfolio beat KOSPI only 1.0% of the time. A 30-stock version beat it only 0.5% of the time.
Turnover-weighted portfolios did much better, with 10-stock and 30-stock versions beating KOSPI 60.1% and 70.8% of the time. But that was essentially a concentrated exposure to the most popular mega-cap AI-memory winners.
The conclusion: in 2026, pure KOSPI is not “the average Korean equity market.” It is closer to a concentrated Samsung Electronics / SK Hynix AI-memory benchmark.

Core View

This year's KOSPI was not the average path. It was the narrowest path. Beating it required being heavily and persistently exposed to the Samsung Electronics, SK Hynix and high-turnover AI-memory winner cohort.

The Index And The Median Stock Diverged

The KOSPI level around 9,050 on June 19 is visible on external index screens such as Investing.com and Trading Economics. KOSDAQ, by contrast, was near 966.6 and up roughly 11% YTD on public screens.

Our proprietary breadth dataset shows the internal divergence.

Metric	Value
KOSPI YTD	around +115%
KOSDAQ YTD	around +11%
Median Korean stock	-7.6%
Advancing-stock ratio	37%
Cumulative A-D line	-20,391
Stocks above MA200	30%
Top-decile return	+88% or more
Top-1% return	+537% or more

The gap between KOSPI and the median Korean stock is roughly 122 percentage points. That is not a normal broad bull market. It is a narrow leadership regime.

Monte Carlo Result

We simulated random Korean equity portfolios from a 2,730-stock universe, using returns from the start of 2026 to June 19. Each case ran 20,000 trials.

Portfolio type	Holdings	Beat KOSPI	Median portfolio return
Equal-weight random	5	5.3%	+2.6%
Equal-weight random	10	2.8%	+9.8%
Equal-weight random	20	1.0%	+14.1%
Equal-weight random	30	0.5%	+15.9%
Turnover-weighted	10	60.1%	+127%
Turnover-weighted	30	70.8%	+131%

Only 7.9% of individual stocks beat the KOSPI return threshold.

The interpretation is straightforward. Diversification reduced benchmark-relative performance because the benchmark itself was heavily concentrated in a tiny winner cohort. Turnover weighting did better because it leaned into the same crowded winners that drove the index.

What It Means

An active account lagging pure KOSPI in 2026 was not necessarily a failed account. It may have simply avoided the extreme concentration embedded in the benchmark.

The right evaluation has three layers:

Layer	Question
Absolute return	Did the account make money?
Median-stock comparison	Did it beat the typical Korean stock?
Pure KOSPI comparison	Did it beat the mega-cap AI-memory benchmark?

Those are different questions. A diversified account could do very well in absolute terms and still lag pure KOSPI.

Final View

In 2026, beating the pure KOSPI benchmark required more than good stock picking. It required concentration in the narrow mega-cap AI-memory path, early entry and long holding periods. Diversification was not wrong, but it carried a relative-performance cost.

Coverage: proprietary 2,730-stock Korea dataset, price returns from the start of 2026 to June 19, 2026, and 20,000 Monte Carlo trials per portfolio structure. External index levels checked against Investing.com, Trading Economics, Yahoo Finance KOSDAQ and MarketWatch KOSDAQ.

Exclusive Analysis on Korea Invest Insights

Are Samsung Electronics and SK Hynix Really Oversold Versus the 2027 Consensus? The Worst Case Is Already the Price

TL;DR

1. 2027 Profit Estimates: Same Companies, Same Year, Numbers That Diverge Nearly Fourfold

2. Should We Even Trust These Numbers? The Low-P/E Trap

3. Reverse-Engineering: What Scenario Is the Current Price Actually Pricing?

4. The Real Reason Views Diverge: The Cause Is 2028, the Symptom Is 2027

5. What the Overlay Reveals

6. The Practical Frame: What to Watch and When to Move

Closing: Two Facts Worth Acting On

Related Posts

Samsung and SK Hynix 2028E Profit Valuation: Cheap-Looking Numbers and Cycle Tests

TL;DR

1. Current Coordinates

2. The Point-Estimate Valuation

3. How to Estimate 2028E

4. Scenario Valuation

5. What the Current Price Already Embeds

6. Samsung Versus SK Hynix

7. Practical Stance

8. Invalidation

Final View

Evidence and Limits

Sources

AI Token Value Today and Tomorrow: Value Added for Memory Companies

TL;DR

1. The Real Question: Who Keeps the Dollar?

2. The Current Value Waterfall

3. The Inverted Pyramid

4. Three Convergence Paths

5. Long-Run Equilibrium: Workflows Capture the Dollar

6. Capture Tiers

7. Multiples: Cheap Looking Is Not the Same as Cheap

8. The Memory Question: Can Mid-Cycle Net Income Reach 100 to 140 Trillion Won?

9. Samsung, SK Hynix, and Micron Are Not the Same Trade

10. Cloud, Power, and Vertical AI

11. What to Monitor

Final View

Samsung Electronics recorded the world's largest quarterly operating profit. Signals and noise poured in today.

TL;DR

1. Why should we separate signal and noise?

2. Officially confirmed values: Sales 171 trillion won, operating profit 89.4 trillion won

3. Stock price and supply and demand: Foreigners sold after good numbers

4. Four axes of Samsung Electronics thesis

5. Signal 1: Legacy DRAM and NAND prices are stronger than expected

Legacy DRAM

Legacy NAND

Supply Structure

6. Signal 2: It is difficult to say that AI CapEx has decreased yet

7. Signal 3: AI server supply chain is not yet broken

8. 소음 1: TrendForce spot 데이터는 약하지만 붕괴는 아니다

9. Noise 2: Morgan Stanley peak logic is only partially correct

10. Noise 3: ASML and TSMC are high bar events

11. Noise 4: Press multiple for interest rates and exchange rates

12. Noise 5: Foreign outflow and leverage piping

13. Korean market performance is also not good

14. Then what should I do?

15. Conclusion

Basis classification

Evidence Ledger

Which NVIDIA Quarter Does Samsung's Earnings Selloff Resemble?

TL;DR

1. What was strange about Samsung’s earnings reaction

2. Why compare Samsung with NVIDIA

3. Timing adjustment: NVIDIA reports after the close

4. NVIDIA’s 17-quarter post-earnings reaction table

5. The closest analogue is NVIDIA Q4 FY26

6. Why the August 2024 NVIDIA sell-on is the wrong template

7. Did NVIDIA recover immediately after large post-print selloffs?

8. The real low and recovery took weeks

9. The rebound trigger had to kill the exact fear

9-1. August 2024: the Fed addressed macro fear

9-2. February 2025: tariff fear required a policy reversal

9-3. February 2026: GTC addressed AI CapEx peak fear

10. Applying the lesson to Samsung

11. Why the Fed put is weaker this time

12. Practical view: wait for proof, then scale in

13. Samsung versus SK hynix

14. Conclusion: Samsung needs better disproof, not just better earnings

Evidence classification