SK Hynix Reports Explosive Earnings: Profit Up Sevenfold, 100 Trillion Won Shareholder Return, Shortage Through 2030 — So Why Was the Market Reaction Muted?

On October 28, SK Hynix released its fiscal 2026 third-quarter results. Operating profit was KRW 83.33 trillion, up 632% year over year and about 9% above market expectations. Revenue was KRW 104.62 trillion, up 328% year over year, the first time quarterly revenue has crossed the KRW 100 trillion mark, and a single quarter’s revenue now exceeds that of all of last year. The operating margin reached 80%, a record for a single quarter.

Alongside the results, SK Hynix announced a shareholder return plan of roughly KRW 100 trillion (about USD 75 billion) for fiscal 2026, and offered a clear medium- to long-term judgment: the memory chip supply-demand imbalance will persist through 2030.

However, the report also showed net profit of KRW 39.78 trillion, far below the market expectation of about KRW 60 trillion. The 216% year-over-year increase looks substantial, but it is down 58% quarter over quarter. There is a gap between operating profit and net profit that needs explaining.

Key Takeaways from the Earnings Report

Source of the sevenfold profit: the main driver is not HBM, but conventional DRAM and NAND

The market is used to tying SK Hynix’s performance directly to HBM. But breaking down the Q3 profit structure, the main driver of this earnings surge was not HBM. The company disclosed that average DRAM prices rose about 20% quarter over quarter in Q3, and average NAND prices rose in the mid-20% range. Counterpoint Research director Hwang MS put it bluntly: “HBM is indeed seeing price increases, but this quarter’s sharp jump in earnings is mainly attributable to rapid price hikes in conventional DRAM and NAND.”

Behind this is a structural shift on the supply side. The three major manufacturers, Samsung, SK Hynix and Micron, have moved large amounts of advanced-node capacity toward HBM, systematically squeezing the supply of conventional DRAM and NAND. The stronger AI data center demand for HBM, the tighter the supply of general-purpose memory. As a core HBM supplier, SK Hynix benefits on two fronts: rising HBM volumes and prices, and the price elasticity of conventional memory as supply is crowded out.

HBM4 will become the main product from Q4, and server DDR5 and enterprise SSDs are also continuing to ramp up. But these are “incremental gains to come.” The foundation of Q3 profit is that the conventional memory price upcycle is now fully underway.

The KRW 100 trillion shareholder return plan

The plan has two components: at least KRW 30 trillion in cash dividends and at least KRW 60 trillion in share buybacks and cancellations, together exceeding 50% of free cash flow. Of this, the Q3 diidend is KRW 20,000 per share, including a special dividend of KRW 10,000 per share derived from investment gains on Kioxia.

Previously, the company announced in August and completed a KRW 40 trillion buyback program. The additional KRW 60 trillion in returns is a further step up on top of that.

The scale of this return needs to be understood in context. SK Hynix’s market capitalization at one point exceeded USD 1 trillion in the first half of 2026, but its share price has since fallen sharply from the June high, and around October 21 the Korean market as a whole suffered concentrated selling by foreign investors and leveraged funds. Announcing a KRW 100 trillion return plan at this moment does more than “give money back to shareholders.” It also signals management’s view that the current share price does not fully reflect the company’s cash flow generation capacity.

The company’s remarks during its earnings communications reinforced this signal: quarterly earnings have not yet hit their peak, and the upward momentum will not reverse before 2027 at the earliest.

“Shortage through 2030”: the supply constraint is structural, not cyclical

The judgment that “demand will exceed supply through 2030” is the sentence in this report with the greatest medium- to long-term pricing significance. SK Hynix’s reasoning is structural constraints on the supply side: the technical difficulty of advanced process nodes keeps rising, building new semiconductor fabs takes considerable time, and the pace of capacity release cannot keep up with the pace of demand expansion.

The demand-side story is also changing. The company believes short-term demand is driven by enterprise and consumer AI applications, while medium- to long-term demand will come from “physical AI” scenarios such as autonomous driving and robotics. The base of AI investment has expanded from a handful of hyperscale cloud providers to ordinary enterprises and even nations. This diffusion means the durability of demand no longer depends on the capital spending rhythm of a single customer group.

A harder piece of evidence: DRAM and HBM capacity through 2029 is already fully sold out, and the company has signed long-term agreements with 14 major customers, locking in capacity equal to more than 50% of total 2027–2028 capacity, with some customers still negotiating to extend the agreement terms further.

“Shortage through 2030” is not a vague optimistic outlook; it is backed by sold-out capacity and signed contracts. This is also why the company dared to announce a return plan on the scale of KRW 100 trillion on the same day net profit missed expectations: management places more weight on its judgment of cash flow sustainability than on accounting fluctuations in a single quarter’s net profit.

The net profit gap

Operating profit was KRW 83.33 trillion and net profit KRW 39.78 trillion; the difference between the two needs explaining. The company disclosed non-operating losses of KRW 32.63 trillion for the quarter. Analysts point to several overlapping factors: tax treatment related to the Kioxia investment, foreign exchange gains and losses, one-off non-operating items, and the accounting impact of the large shareholder return plan.

The nature of this gap matters. It does not come from deterioration at the operating level; the 80% operating margin is itself proof of operating efficiency. It comes from non-operating line items and accounting treatment, which do not affect the company’s ability to generate cash and do not change the pricing trend of DRAM and NAND. But for investors who use net profit as their core valuation anchor, this shortfall is a variable that needs ongoing tracking: if the non-operating losses are one-off, Q3 net profit is the low point; if they contain structural factors, the sustainability of earnings quality needs to be reassessed.

Market Reaction

After the earnings and the return plan were announced, SK Hynix’s share price rose as much as nearly 8% in the Seoul market, and U.S. overnight trading gains at one point widened to about 7%. But the pre-market quote on the Nextrade platform was up only about 3%, and the gain narrowed to about 6% at the final close.

The muted reaction has a specific backdrop. On the same day as the earnings release, the Federal Reserve’s rate decision and the earnings of Alphabet and Microsoft were also due, and investors chose to stay cautious before placing big bets. More broadly, the recent earnings season for the AI supply chain has shown a pattern: beating expectations is no longer enough to reignite the earlier market euphoria, because “beating expectations” is becoming the baseline scenario rather than incremental information.

But SK Hynix stands in sharp contrast with Seagate, which reported on the same day. Seagate’s results and guidance also beat expectations across the board, yet its shares once plunged 8% after hours, because of excessive gains for the year and crowded positioning, which made record profits merely “what was expected.” That SK Hynix managed to hold onto a positive return in this environment owes much to the “incremental information” quality of its KRW 100 trillion return plan, which is the key differentiating factor.

The market’s response of a moderate rise rather than a frenzied rally is itself reasonable. In an earnings season where “beating expectations” is no longer scarce, what SK Hynix delivered was not just another quarterly figure but a judgment that cash flow will not reverse before 2027. Whether that judgment holds depends on whether the DRAM and NAND price upcycle can withstand the pace of Chinese capacity coming online, and whether HBM4 customer qualification and pricing can carry on the profit momentum from conventional memory.