The consensus story about Samsung Electronics is that you buy it for the memory supercycle. That story is incomplete, and the part everyone is missing is not in the earnings headline — it is in the cash-flow appendix. Samsung’s FY2024–2026 shareholder return policy commits the company to returning 50% of free cash flow, and that three-year window closes with FY2026 — roughly four months from now. Run the subtraction the company itself has not run in public: first-half 2026 operating cash flow of ₩145.35tn less ₩31.24tn of capital expenditure gives free cash flow of ₩114.11tn. Half of that is ₩57.06tn owed to shareholders. Samsung has actually paid ₩19.46tn. That leaves roughly ₩37.6tn outstanding from the first half alone — about 2.5% of a ₩1,525.9tn market capitalisation, before you count the second half, and before you count 2024 and 2025. At a spot price of ₩261,000 (KOSPI close, 1 September 2026), this Samsung stock prediction sets a 12-month bull case of ₩380,000 and a bear case of ₩185,000.
That ₩37.6tn number is the single most underpriced fact about this stock, and the reason is mundane: it requires reading two Samsung documents together that are never published together. The quarterly deck carries the cash-flow statement; the investor-relations site carries the policy, and neither cross-references the other. Samsung also has an explicit, dated precedent for what happens when a three-year window closes with a surplus. On the FY2018–2020 programme, the company states plainly that the leftover portion of 50% of three-year free cash flow after regular dividends — ₩10.7tn — “was paid out as a special cash dividend on April 16, 2021.” That residual was struck against a far smaller earnings base; the 2026 equivalent is arithmetically several times larger, and it falls due inside the window this bull case runs over. Everyone is modelling DRAM prices. Almost nobody is modelling the calendar.
Key facts
- Spot ₩261,000, +0.38% on the day, KOSPI regular-session close 15:30 KST — Naver Finance / Daum Finance, 1 September 2026
- Q2 2026 revenue ₩171.5tn (+130% YoY); operating profit ₩89.5tn against ₩4.7tn a year earlier — Samsung Q2 2026 earnings deck, 30 July 2026
- Device Solutions operating profit ₩89.2tn on ₩127.5tn revenue — a 70% divisional margin, versus 1.43% in Q2 2025 — Samsung Q2 2026 deck
- Memory revenue ₩120.8tn, +471% YoY — Samsung Q2 2026 deck (Samsung publishes no ASP or bit-growth percentages; this revenue line is the closest available proxy)
- Device eXperience swung to a ₩0.8tn operating loss, with Mobile eXperience alone at ₩-0.7tn — Samsung Q2 2026 deck
- Net cash ₩167.59tn at 30 June 2026, up from ₩86.70tn a year earlier — Samsung Q2 2026 deck
- SK Hynix Q2 2026 revenue ₩79.3187tn, operating profit ₩60.5426tn, 76% margin — SK Hynix, 29 July 2026
- Consensus forward EPS ₩48,339 and BPS ₩86,052, putting spot on 5.40x forward earnings and 3.03x book — Naver Finance, data as of June 2026
What is actually happening, and why the numbers look fake
Samsung’s second quarter is the kind of print that makes analysts re-check the spreadsheet. Revenue of ₩171.5tn, up 130% year on year. Operating profit of ₩89.5tn against ₩4.7tn in the same quarter of 2025 — a 1,814% increase. Group operating margin of 52.2%. Earnings per common share of ₩10,849, against ₩737. Return on equity of 56%, up from 5%.
Practically all of it is memory. Device Solutions, the semiconductor division, produced ₩89.2tn of the ₩89.5tn group operating profit, on ₩127.5tn of revenue that grew 357% year on year. Memory revenue inside that was ₩120.8tn, a 471% increase. Samsung does not disclose average selling prices or bit-growth percentages, so that revenue line is the only public proxy for what has happened to memory pricing — and it should be read as a proxy, not as a price index, because it blends volume, mix and price together.
The mechanism is not mysterious. Think of memory as a toll road with three operators and no way to build a fourth lane inside two years: AI server demand arrived as a step-change in traffic, while capacity is set by fab schedules fixed years earlier. The deck’s own outlook says demand for server DRAM, enterprise SSD and HBM is “expected to keep the market undersupplied.” On product, Samsung reached industry-first mass production of HBM4 in February 2026 using 1c DRAM and a 4nm base die, and was first to ship HBM4E samples to global customers in May. Those two milestones ended the narrative, dominant through 2024 and 2025, that Samsung had structurally lost high-bandwidth memory. Our coverage of NAND pricing and the multi-year capacity lag set out why relief on the supply side is not a 2026 or 2027 event.
On the mid-cycle question, Ryu Hyung-keun, Research Analyst at Daishin Securities, put it this way in an interview published by SK Hynix on 14 August 2026: “Market expansion is likely to continue at least through to the end of 2027.” He was discussing SK Hynix and the memory market generally, not issuing a Samsung target — but the horizon he names is precisely the one this piece runs over.
The 40-point gap that became six — and the buyback Samsung has not announced
Here is the comparison that reframes the investment case. Twelve months ago Samsung’s semiconductor division was, in margin terms, barely a business. Today it is near parity with the company that spent two years being described as its conqueror.
| Operating margin, semiconductor/memory | Q2 2025 | Q2 2026 |
|---|---|---|
| Samsung Device Solutions | 1.43% (₩0.4tn on ₩27.9tn) | 69.96% (₩89.2tn on ₩127.5tn) |
| SK Hynix (company level) | 41.44% (₩9.2129tn on ₩22.232tn) | 76.33% (₩60.5426tn on ₩79.3187tn) |
| Gap in Samsung’s disfavour | 40.0 points | 6.4 points |
Samsung closed a forty-point margin gap to six points in four quarters. And on absolute scale it has already won: Samsung’s semiconductor division is 61% larger by revenue and earns ₩28.7tn more operating profit per quarter than SK Hynix earns in total. The remaining six points are a mix question — SK Hynix’s revenue is more concentrated in high-bandwidth memory — not a competence question. The 2024 thesis that Samsung cannot execute in AI memory is one the Q2 numbers have already retired.
The response side is where the two companies diverge sharply, and it is the strongest argument in the bull case. SK Hynix’s board resolved on 19 August 2026 to repurchase and fully cancel ₩40tn of treasury shares — roughly 24.07 million shares, about 3.3% of issued stock, executed over about three months from 20 August. It is the largest treasury-share cancellation in the history of Korean listed companies, and SK Hynix simultaneously widened its policy from “within 50%” to “over 50%” of cumulative 2025–2027 free cash flow, because it judges its intrinsic value “is not fully reflected in its current stock price.” We covered the market reaction to that programme when it followed a 9.8% single-day fall, and the divergence in shareholder-return posture between the two Korean memory makers in this comparison of the two payout policies.
Samsung, sitting on ₩167.59tn of net cash, has announced no equivalent. It bought back ₩5.63tn of stock in Q2 and ₩13.24tn across the first half, and paid ₩6.21tn of dividends in the quarter. Against a ₩57.06tn first-half obligation, that is a third of the run-rate. Samsung’s silence is not neutral information — under its own published policy it is a deferral, not a decision.
The arithmetic: ₩185,000 < ₩261,000 < ₩380,000
The constraint that keeps a price target honest: every level has to be somewhere the stock has actually traded, or a multiple the market has actually paid. Neither target is an extrapolation.
Spot is ₩261,000. The bear case of ₩185,000 is ₩76,000 below it, a fall of 29.12%. The bull case of ₩380,000 is ₩119,000 above it, a gain of 45.59%. The base case of ₩295,000 is ₩34,000 above spot, a gain of 13.03%. The horizon is twelve months, to end-September 2027.
Anchor those to fundamentals. Consensus forward EPS is ₩48,339 and book value per share is ₩86,052 (Naver Finance, June 2026 basis). Spot therefore trades on 5.40x forward earnings — Naver’s own estimated-PER field reads 5.40x, which ties out exactly — and 3.03x book.
| Case | Price | vs spot | Forward P/E | P/B | Reference point |
|---|---|---|---|---|---|
| Bull | ₩380,000 | +45.59% | 7.86x | 4.42x | ~1.5% above the 52-week intraday high of ₩374,500 (19 June 2026) |
| Base | ₩295,000 | +13.03% | 6.10x | 3.43x | Just above the August 2026 intraday high of ₩288,000 |
| Spot | ₩261,000 | — | 5.40x | 3.03x | 1 September 2026 KOSPI close |
| Bear | ₩185,000 | −29.12% | 3.83x | 2.15x | Just under the 50-day intraday low of ₩189,200 |
Read the right-hand column carefully, because it is the whole argument. The bull case is a level the stock stood on ten weeks ago. The base case reclaims a level it touched three weeks ago. The bear case sits fractionally below ground it held in February 2026, between the weekly closes of ₩181,200 and ₩190,100. A de-rate from 5.40x to 3.83x forward earnings is a savage repricing of the multiple — but it is not a 2023-style memory bust, and it does not require earnings to collapse.
The flow data adds a second layer. Foreign ownership stands at 46.72%, down from a 52.62% peak on 31 October 2025 and off a trough of 46.50% on 29 July 2026. Overseas investors sold nearly six percentage points of the register through the strongest earnings expansion in the company’s history — and are now, marginally, buying back. The stock is 28.00% below its record close of ₩362,500 set on 18 June 2026, while up 117.68% year to date and 286.09% over twelve months. That combination — a violent drawdown inside a violent uptrend — is what a market looks like when it cannot decide whether it is paying for a cycle or a re-rating. Our SK Hynix bull and bear analysis and our Micron price prediction frame the same dilemma at the other two suppliers, and the broader index question is set out in our look at whether Korea is cheap enough to buy.
The bear case, stated by an analyst and confirmed by Samsung’s own segments
The cleanest statement of the bear case does not come from a bear. In an interview published on SK Hynix’s newsroom on 13 May 2026, Rok-ho Kim, Research Analyst at Hana Securities, is recorded arguing that because the market has already reflected much of the expected expansion in AI demand and price increases, “the key variable going forward will be how long high profit margins can be maintained.” Again, he was analysing SK Hynix rather than Samsung. But at a 70% divisional margin, that sentence is the entire risk. No version of the bear case requires demand to fall. It only requires margins to normalise while the multiple compresses — the classic double-squeeze of a late-cycle semiconductor.
Samsung’s own segment reporting supplies the second bear argument, and it is one no competitor has. Samsung is the only memory company that is also a victim of the memory supercycle. Device eXperience — phones, TVs, appliances — swung to a ₩0.8tn operating loss in Q2, with Mobile eXperience alone at ₩-0.7tn, on revenue of ₩32.3tn. The deck’s stated reason is that MX earnings “declined due to elevated component cost pressures across the industry.” Samsung’s own memory prices made Samsung’s own phone business unprofitable. SK Hynix and Micron have no internal victim; they capture the price rise with nothing on the other side of the ledger. That is a genuine structural disadvantage during the upswing — and, symmetrically, a hedge on the way down.
The third bear signal is in the balance sheet, and it is a signal about management conviction. Financial investments rose ₩50.2tn quarter on quarter to ₩104.82tn. Property, plant and equipment rose only ₩6.56tn. Capital expenditure actually fell 17.6% quarter on quarter, to ₩14.11tn from ₩17.13tn. A company that believed the supercycle had years to run would be converting cash into capacity. Samsung is converting it into financial instruments. (Note that Samsung publishes no full-year 2026 capex target; the only public figure is the ₩31.24tn first-half actual. Any specific annual number attributed to the company is an estimate, not a disclosure.)
So the sentence that captures the position is this: Samsung is earning the most money in its history, is not spending it on capacity, is not yet returning it to shareholders, and is losing money on phones because of its own prices. Every one of those four clauses is a fact from the same deck. A bull reads the middle two as stored value; a bear reads them as a board that does not believe its own numbers. Chinese capacity matters too — as we examined in our coverage of CXMT’s 466% Shanghai debut, is a commodity-DRAM threat rather than an HBM one for now, but it shortens the runway on the low end.
What happens next
Three concrete predictions, each with its causal chain.
First: Samsung announces a capital-return decision before the end of Q1 2027. The FY2024–2026 window closes with this financial year, the company states it will “consider executing an early return of capital beyond regular dividends” where it forecasts a significant surplus, and the surplus is now too large to characterise as anything else. The FY2018–2020 residual was settled by special cash dividend in April 2021 — roughly four months after that window closed. The same clock, applied to a window closing 31 December 2026, points at the first four months of 2027. This is the single event most likely to move the stock from the base case toward the bull case, and it is a policy mechanism rather than a forecast.
Second: the margin gap to SK Hynix closes below four points or reopens above ten — it will not sit at six. The gap is a product mix effect. Samsung is scaling HBM4 into volume after February’s mass-production start and shipped HBM4E samples in May; if those convert to revenue share, mix converges. If the qualification cycle slips, Samsung’s mix stays weighted to conventional server DRAM and NAND, whose prices turn first. Watch Q3 and Q4 divisional margin, not group revenue.
Third: the phone business is the tell, not the noise. If Mobile eXperience returns to profit in the second half without memory prices falling, Samsung has proven it can price through its own input costs and the vertical-integration objection dies. If MX stays loss-making into 2027 while DS margins hold, the market will correctly start valuing Samsung as a memory pure-play with a structurally loss-making handset business attached — and that is a lower multiple than 5.40x, not a higher one.
The honest summary is that Samsung at ₩261,000 is not priced as a company earning ₩89.5tn a quarter. It is priced at 5.40x forward earnings and 3.03x book because the market assumes those earnings are borrowed from a cycle that ends. The bull case does not require the market to be wrong about the cycle. It requires Samsung to hand back money it has already promised, before the promise expires.
Frequently asked questions
What is the Samsung stock prediction for the next 12 months?
This analysis sets a bull case of ₩380,000 (+45.59%), a base case of ₩295,000 (+13.03%) and a bear case of ₩185,000 (−29.12%) against a spot price of ₩261,000, over a twelve-month horizon to end-September 2027. On consensus forward EPS of ₩48,339 those levels are 7.86x, 6.10x and 3.83x forward earnings respectively.
Why is Samsung trading at only 5.4x forward earnings?
Because the market treats memory profits as cyclical rather than durable. Samsung’s Q2 2026 operating profit of ₩89.5tn compares with ₩4.7tn a year earlier; investors are discounting the probability that a 70% divisional margin reverts. Trailing PER on the June 2026 basis is 11.71x, roughly double the forward figure — the arithmetic of an earnings base not yet fully in the trailing twelve months.
How much does Samsung still owe shareholders under its current policy?
Under the FY2024–2026 policy of returning 50% of free cash flow, first-half 2026 free cash flow of ₩114.11tn generates a ₩57.06tn obligation. Samsung has returned ₩19.46tn in buybacks and dividends over that period, leaving roughly ₩37.6tn outstanding from the first half alone, before the second half and before earlier years in the window.
Is Samsung behind SK Hynix in HBM?
Not on the milestones it publishes. Samsung reached industry-first mass production of HBM4 in February 2026 and shipped the first HBM4E samples to global customers in May 2026. SK Hynix retains a higher company-level margin at 76% versus Samsung’s 70% divisional margin, reflecting revenue mix rather than technology position.
Why did Samsung’s phone division lose money in a record quarter?
Mobile eXperience posted a ₩0.7tn operating loss on ₩32.3tn of revenue, which Samsung attributes to elevated component cost pressures across the industry. Memory is a primary smartphone input, so the price increases driving Device Solutions to a 70% margin also raise the bill of materials in Samsung’s own handsets. Samsung is the only major memory maker exposed to both sides of that trade.
What would invalidate the bear case?
A capital-return announcement of scale before the FY2024–2026 window closes on 31 December 2026, a return of Mobile eXperience to profitability without memory prices falling, or evidence that HBM4 and HBM4E are converting into revenue share fast enough to close the mix gap to SK Hynix. Any of the three removes a distinct pillar of the ₩185,000 case.
Analysis, not investment advice. Prices and multiples are as of 1 September 2026, sourced from Naver Finance, Daum Finance and Samsung’s own Q2 2026 investor materials. Samsung does not issue price targets; no forecast here is attributable to the company or any Samsung executive.