SK hynix’s October 28 results strengthen the operating case, while replacing dividend speculation with an approved quarterly payment and a broader—but partly forward-looking—return plan. The unresolved investment question has changed: it is now how much of exceptional memory profitability survives new capacity and distributions, not whether the previously reported special dividend has already been approved. We preserve our dated model rather than add a dividend bonus or reduce shares without the corresponding cash cost.
Record operating profit and falling net profit describe different things
The October 28 preliminary-results Form 6-K reports consolidated K-IFRS figures for July–September. Revenue rose 31.9% from Q2 and operating profit 37.6%, while profit for the period fell 57.7%. These are preliminary figures, subject to completion of the external review; they are not an audited annual result.
| KRW trillion | Q2 2026 | Q3 2026 |
|---|---|---|
| Revenue | 79.318746 | 104.615483 |
| Operating profit | 60.542608 | 83.333818 |
| Operating margin, calculated | 76.33% | 79.66% |
| Profit before income tax | 122.708355 | 50.708500 |
| Profit for the period | 93.922593 | 39.775593 |
The arithmetic is informative without pretending to be a complete earnings bridge. Pre-tax profit minus operating profit was positive ₩62.166tn in Q2 but negative ₩32.625tn in Q3. That residual includes the aggregate net effect of non-operating items; it is not a separately disclosed Kioxia gain or loss. The preliminary statement does not reconcile every component. A falling bottom line therefore cannot, on its own, establish deteriorating memory operations; equally, record operating profit cannot be treated as cash shareholders have already received.
HBM4 progress and longer contracts improve visibility, not certainty
In its July release, SK hynix said HBM4 mass shipments had begun and long-term agreements involved around ten customers. The new issuer update reports 14 customers, with contracted volume covering more than half of 2027–2028 production capacity. The company expects HBM4 to exceed HBM3E in both shipments and revenue in Q4, and investment to increase significantly next year. Those are company expectations, not completed production or guaranteed profit.
The contract structure matters more than the customer count alone. SK hynix describes market-linked pricing within agreed ranges. Such a structure can improve planning and reduce some demand uncertainty without fixing every future margin. Capacity covered by agreements is also not revenue already recognized. Delivery, qualification, product mix and enforceable commercial terms determine what ultimately reaches earnings and cash.
Our interpretation is stronger operating visibility alongside a continuing reinvestment test. More advanced memory can support richer product economics, but manufacturing ramps require spending before all their output is sold. Expanding capacity can eventually loosen the scarcity that supports current pricing. A company can correctly identify strong demand while an investor still overestimates how long peak margins persist.
What is approved, what is completed and what remains planned
The shareholder-return Form 6-K sets a Q3 dividend of ₩20,000 per share: ₩10,000 from earnings and ₩10,000 special. It plans most Kioxia investment gains to be returned through Q3 and year-end special dividends; the annual dividend still requires the relevant procedures. The ₩100tn annual return estimate includes August’s ₩40tn buyback, excludes employee-compensation shares and extends through the FY2026 annual-dividend payment date.
| Capital action | October 28 status | Important boundary |
|---|---|---|
| Q3 dividend | ₩20,000 per share; issuer describes roughly ₩14tn total | Half the per-share payment is special, not all recurring earnings |
| New ₩20tn repurchase resolution | Approximately ₩10tn for shareholder returns and ₩10tn for employee compensation | Only the return portion is expected to be cancelled |
| Approximately ₩100tn FY2026 returns | Expected total includes earlier ₩40tn repurchase | Not ₩100tn of wholly new October cash payments |
This resolves the question posed in our October 26 dividend-report analysis, but it does not confirm the rumor in its original form. The approved quarterly payment combines an earnings component and a special component. It is not a ₩16tn special dividend. Nor should the annual plan be added on top of the earlier buyback when that program is already included.
The newsroom says the August repurchases were completed last week and cancellation would follow. Purchase and cancellation remain distinct steps. Buying shares for employee compensation also differs from permanently retiring them: later transfer to employees can restore economic share claims. A valuation should follow actual shares outstanding, treasury treatment and awards rather than assume every repurchased share disappears forever.
Liquidity supports choices; it does not make distributions free
The issuer reports quarter-end cash and cash equivalents of ₩114.5tn and borrowings of ₩15.8tn, giving ₩98.7tn net cash. Those are September 30 balances, not post-distribution October liquidity. They do not establish free cash flow after all planned capital expenditure, taxes and shareholder returns. The company also confirms that it discontinued the Solidigm IPO review, preserving the funding issue explored in our Solidigm financing analysis.
A special dividend distributes existing corporate resources; it is not another operating gain on top of the investment realization that funded it. A buyback uses cash to reduce some share claims. The benefit depends on price paid and alternative investment returns. Both actions belong in a single bridge from opening cash through operating generation, investment proceeds, capital spending, taxes, financing and distributions to closing liquidity.
Valuation: preserve the historical model, review the new bridge
Our October 15 FY2027 valuation published Bear ₩713,000, Base ₩1,857,000 and Bull ₩3,018,000, with an approximately October 2027 horizon and dividends excluded. We preserve these dated references, but the current capital-allocation bridge is under review. This article does not establish revised targets or mechanically add the new dividend to them.
The historical Base is [(₩480tn revenue × 55% operating margin + ₩4tn recurring net non-operating income) × 75% after tax × 99% owners’ allocation] ÷ 750m modeled economic shares × 7 = ₩1,857,240 before rounding. The 750m is an analytical share budget, not the legal dividend-entitlement denominator. Extraordinary investment gains are excluded, and there is no separately added cash-per-share asset.
Q3’s 79.66% operating margin is well above our modeled 55%, but a quarterly peak cannot be inserted as the next full year’s margin without a durability case. Equally, completed repurchases and approved distributions require an actual cash-and-share reconciliation before changing the denominator or recurring financing income. This is why stronger operating evidence can raise confidence while leaving the numerical revision unfinished.
For a dated comparison, the Stock Analysis/S&P Global page accessed October 28 shows 40 analysts, a one-year mean of ₩3,096,288, median ₩3,050,000 and range ₩1,535,000–4,700,000. The preserved Base is 40.025% below that mean; Bear and Bull are 76.972% and 2.528% below. These are broadly comparable one-year horizons, not identically dated models or reconciled dividend conventions. The vendor does not establish that every analyst has incorporated today’s results.
The greater-than-20% difference primarily reflects our more cautious recurring earnings path and model conventions. Our next revision requires evidence about sustainable product margins, the scale and return of new investment, actual repurchases and remaining cash. A higher payout alone does not justify raising value; nor would a lower retained cash balance justify ignoring profitable growth. We have not verified a complete old-to-new broker target change attributable to these results, so none is presented as a confirmed post-results action.
Next evidence: HBM4 mix and delivery execution; the economics and timing of contracted capacity; cash conversion after taxes and investment; and the completed share and cancellation record. Follow the dated history in our SK hynix research and news hub. This is educational investment information, not personalized advice; high margins, cycle exposure and forecast uncertainty can still produce substantial losses.
Sources and original documents
SK hynix — preliminary consolidated results, Form 6-K, October 28, 2026; quarter ended September 30, original units KRW million.
SK hynix — expected FY2026 shareholder returns and implementation plan, Form 6-K, October 28, 2026.
SK hynix — Q3 results, product plans, liquidity and capital actions, October 28, 2026.
SK hynix — Q2 results and earlier HBM4/LTA position, July 29, 2026.
Stock Analysis/S&P Global — KRX:000660 forecasts, accessed October 28, 2026; per-broker dates vary.
AHNIM STUDIO — original model, October 15, 2026; historical assumptions preserved.
Leave a Reply