NextFin News - South Korea's two memory-chip giants are preparing the largest shareholder-return programs in the country's listed-company history, with SK Hynix moving first on a 40 trillion won buyback and Samsung Electronics expected to unveil a package worth more than 100 trillion won later this month. Combined, the two programs would return well over $130 billion to investors and mark the most forceful attempt yet to close the valuation discount that has long dogged Korean equities.
The Situation: Two Announcements, One Market-Wide Pivot
SK Hynix's board approved a resolution on August 19 to repurchase and fully cancel 40 trillion won of treasury shares — roughly $28.6 billion — in what the company said would be the largest share cancellation ever announced by a publicly listed South Korean company. The repurchase covers as many as 24 million shares, about 3.3% of total outstanding stock, based on a reference price of 1,662,000 won per share, and will run from August 20 to November 19. Alongside the buyback, SK Hynix raised its shareholder-return target from "within 50%" of cumulative free cash flow to "over 50%" for the 2025-to-2027 period, and said additional returns would be announced with its third-quarter results.
Samsung Electronics is set to follow. The company is preparing a new shareholder-return policy worth more than 100 trillion won — approximately $71.75 billion — to be announced after a board meeting later in August, according to industry officials cited in local reporting. The program would deploy 50% of free cash flow, consistent with the framework Samsung adopted for 2024-to-2026, under which it committed to returning half of FCF while maintaining an annual regular dividend of 9.8 trillion won.
The scale of the shift is stark. By one estimate, SK Hynix's total shareholder returns last year came to roughly 14.3 trillion won, including about 2.1 trillion won in cash dividends and 12.2 trillion won in share cancellations — meaning the new 40 trillion won repurchase alone is nearly three times that full-year total. JPMorgan analyst Jay Kwon estimates SK Hynix could return at least another $130 billion through 2027 beyond the current buyback, a figure equal to roughly 16% of the company's market value.
The backdrop is record profitability fuelled by the AI memory boom. SK Hynix reported second-quarter net profit up 1,242% year-on-year, capping a fiscal 2025 in which it earned 42.9 trillion won in net profit on revenue of 97.1 trillion won. Samsung posted record second-quarter revenue of 171.5 trillion won, up 28% quarter-on-quarter, driven by AI-related semiconductor demand. Yet both stocks have punished investors: SK Hynix closed at 1,500,000 won on August 19, down 9.75% on the day and still more than 40% below its June record high.
The question the market now faces is not whether the cash is real — it is. The question is whether these payouts represent a durable change in how Korea's corporate giants treat shareholders, or a one-time distribution of peak-cycle profits just before the memory cycle turns.
Why the Payouts Are Arriving Now — and Why the Timing Is Not Accidental
The convergence of three forces explains the timing: peak-cycle cash generation, a depressed share price that makes repurchases accretive, and intensifying political and activist pressure to share the wealth.
SK Hynix ended the second quarter with net cash of about 69 trillion won. A 40 trillion won buyback represents roughly 58% of that war chest committed in a single three-month window. At a share price that has fallen more than 40% from its peak — and trades at roughly half the average broker target price, according to Seoul-based research — each repurchased share retires claims on future earnings at a discount. Management is using the cycle's cash to defend a valuation the market has rejected.
"A commitment to its own shares on this scale over the next three months indicates SK Hynix does not think memory pricing is about to roll over," said Josh Gilbert, an analyst at trading platform eToro.
Pressure from shareholders has been building for months. In early August, retail-investor platform ACT launched a campaign to force an extraordinary shareholders' meeting at Samsung, urging the company to buy back about $32 billion worth of shares and to cap performance bonuses tied to operating profit. The campaign explicitly sought backing from the National Pension Service and domestic and overseas asset managers. Around the same time, Samsung's finance chief told investors on a July 30 earnings call that the company would "faithfully execute the shareholder return policy as promised" and would "update it soon."
There is also a competitive dynamic at work. U.S. rival Micron has pledged to return 100% of its excess cash to shareholders. For Korean chipmakers long criticized for hoarding cash and cross-shareholding within family-controlled conglomerates, the American benchmark has become a governance cudgel. Sanjeev Rana of investment firm CLSA noted that the 40 trillion won buyback should satisfy investor expectations, particularly because more buybacks and special dividends could be announced at a later stage.
The immediate market reaction captured the relief. SK Hynix shares jumped as much as 13% on the Korea Exchange after the plan was unveiled, and Samsung climbed as much as 10% on reports of its own coming program. The bounce in SK Hynix helped lift the benchmark KOSPI, which has leaned heavily on the two chipmakers: since late 2024, Samsung and SK Hynix together have accounted for roughly 51% of the index's gains, according to Seoul-based brokerage LS Securities.
Cyclical Cash, Structural Framework — Separating the Two
This is the crux of the investment question, and getting it wrong flips the conclusion. The cash is cyclical. The framework is structural. They must be judged separately.
Memory semiconductors are among the most cyclical products in the global economy. SK Hynix's fiscal 2025 operating margin reached 49%, and its second-quarter 2026 operating margin hit 58%. These are peak-cycle margins, the kind that appear at the top of a supercycle, not averages that persist through a decade. History is unforgiving here: the industry rode a boom in 2017-2018, crashed in 2019, recovered in 2021-2022, slumped through 2023, and only returned to profitability on the back of AI-driven demand for high-bandwidth memory. When AI capital expenditure slows, or when Chinese capacity from producers such as CXMT ramps into legacy nodes, memory pricing rolls over and free cash flow collapses with it.
That is precisely why the "over 50% of cumulative free cash flow" pledge is structurally meaningful. It is self-adjusting: in a downturn, the absolute payout shrinks automatically, so the company does not borrow to fund dividends at the bottom of the cycle. SK Hynix has also raised its fixed annual dividend 25% to 1,500 won per share, up from 1,200 won — a floor that persists regardless of FCF volatility. Samsung's 9.8 trillion won annual regular dividend, committed through 2026, plays the same role.
The structural case rests on more than financial engineering. Korea's "value-up" program, the rise of organized retail activism, and the precedent of multi-year board commitments have created political and competitive pressure that does not reverse when memory pricing does. A payout ratio anchored to FCF, combined with a fixed-dividend floor and a board-level commitment spanning multiple years, is a governance regime change — not a one-off gift.
The distinction matters because it dictates what investors should underwrite. Underwrite the framework surviving, and the valuation re-rating has room to run. Underwrite the absolute won amounts persisting at peak levels, and you are betting the AI supercycle has no down leg — a bet the last three memory cycles have punished.
The Second-Order Read: What the Market Is Not Pricing
The first-order effect of these buybacks is mechanical price support. JPMorgan's estimate that SK Hynix could return another $130 billion through 2027 — 16% of market capitalization — is a direct bid under the stock. That much is priced in, at least partially, after the double-digit rally.
The second-order effect is what the market has largely overlooked: a coordinated reduction in retained cash changes the supply-and-demand mechanics of Korea's benchmark index itself. Samsung and SK Hynix together drove approximately 98% of the operating-profit growth across KOSPI 200 companies over the past year, according to Seoul-based brokerage Heungkuk Securities. When the two companies that dominate index earnings simultaneously shrink their free float through cancellations, the domestic supply of benchmark shares tightens even as foreign investors have been net sellers of Korean chips. That is support for the KOSPI that is independent of the earnings cycle — a mechanical bid, not a sentiment bid.
There is a second transmission channel, this one running through commodity pricing. Returning cash to shareholders instead of flooding it into fabrication capacity is quietly bullish for memory pricing discipline. The 2023 downturn was caused by oversupply: fabs kept producing while demand contracted, and spot prices collapsed. If the Samsung–SK Hynix duopoly chooses shareholder returns over a capital-expenditure arms race, the next downcycle has a shallower floor. Governance reform, in other words, propagates into the very commodity prices that generate the cash to fund the reform.
The third-order expectation gap sits at the end of that chain. The market has priced "big buybacks." What it has not priced is whether these payout frameworks survive a genuine memory downturn. If SK Hynix and Samsung maintain their FCF-linked ratios and fixed dividends through a downcycle, the Korea discount closes structurally. If they suspend the variable portion at the first sign of weakness, the multiple snaps back to where it started. The durability of the framework — not the size of this year's check — is the variable that determines whether this rally is a regime change or a cyclical head-fake.
The Counter-Thesis: Peak-Cycle Cash, and a Promise the Cycle Can Break
The strongest case against the structural read is simple, and it is backed by the market's own positioning. Memory is cyclical. AI capital expenditure is front-loaded. And Chinese competition is arriving.
SK Hynix shares remain more than 40% below their June record high. That drawdown is not noise; it is the market discounting a slowdown in the durability of AI spending among the company's Big Tech customers. Brokerage target prices tell the same story from the other direction: the average broker target for SK Hynix stood near 3.37 million won in late July, roughly double the prevailing share price. Investors are being asked to believe that the cash generated at the top of the cycle will keep flowing — and history says it will not.
The counter-thesis is correct on the cash, but incomplete on the framework. The structural argument does not require profits to stay at record levels. It requires only that the payout framework survive the downturn. Samsung's current 50%-of-FCF policy was announced in January 2024, in the middle of the last memory downturn, and it maintained the fixed 9.8 trillion won dividend through 2024-to-2026. SK Hynix's 1,500 won per-share fixed dividend plays the same role. The variable "over 50% of FCF" portion will shrink when FCF shrinks — and that is a feature of a sustainable policy, not a bug.
Still, the falsifying signal is concrete and observable. If either SK Hynix or Samsung cuts its fixed dividend, or if memory spot prices fall more than 20% from current levels while capital-expenditure guidance stays flat, the structural-governance thesis is wrong: the payouts were cyclical after all, and the framework was cosmetic. Until that signal prints, the burden of proof sits with the skeptics.
What Comes Next: Beneficiaries, the Exposed, and the Scenarios
The mechanism, once cashed in, points to clear winners and losers. Beneficiaries include Korean equities broadly, as a tightening KOSPI float supports the index mechanically; retail and foreign investors holding Samsung and SK Hynix directly; and memory pricing itself, if capacity discipline holds. The exposed are the equipment and materials suppliers that counted on Korean chip capex continuing to expand — shareholder returns that substitute for investment redirect cash away from their order books — and, in a downside scenario, bondholders if either company were to fund payouts with debt rather than net cash. Given that both companies sit on large net-cash positions, that last risk remains remote.
The outlook splits cleanly by time horizon. In the short term, the buyback execution window — SK Hynix's runs from August 20 to November 19 — provides mechanical support and should ease volatility. In the medium term, third-quarter results will show whether free cash flow actually supports the "over 50%" pledge; SK Hynix has flagged that additional returns would be announced alongside those results. In the long term, the question is structural: if the framework survives a memory downturn, Korea's valuation discount closes permanently; if it does not, this rally is a cyclical head-fake.
Three scenarios frame the path. The base case is that Samsung announces a program exceeding 100 trillion won later this month, both companies maintain payout ratios above 50% of FCF through 2027, and the KOSPI retains its mechanical support. The upside case is that AI spending proves durable, memory pricing holds, and both firms add special dividends — in which case JPMorgan's $130 billion estimate for SK Hynix materializes and the Korea discount compresses decisively. The downside case is that AI capex slows, memory prices fall more than 20%, and the variable portion of the pledge is quietly suspended — sending shares back toward their June lows.
What to watch is specific: the date and size of Samsung's board announcement; SK Hynix's third-quarter free cash flow and any special-dividend declaration; the trajectory of memory spot prices; and the fixed-dividend decisions for fiscal 2026. Each of these either confirms the framework's durability or triggers the falsifying signal.
Korea's chip giants are not just returning cash. They are betting that the governance premium they were never paid is finally worth more than the cycle they have spent decades surviving — and this time, the market has a concrete way to call their bluff.
Explore more exclusive insights at nextfin.ai.
