SK hynix Posts Record AI-Memory Profit — But Can an 80% Margin Last?

SEOUL, October 28, 2026 — The global artificial-intelligence boom has produced a striking set of numbers at South Korean memory-chip maker SK hynix: third-quarter revenue of 104.62 trillion won and operating profit of 83.33 trillion won. But the bigger story is not simply the size of one quarter’s earnings. It is how AI infrastructure demand is reshaping the supply, pricing and investment decisions behind the chips that make powerful processors useful.

The figures — and what they do not mean

In its October 28 earnings announcement, SK hynix reported revenue up 32% and operating profit up 38% from the previous quarter. Its operating margin was approximately 80%, while net income was 39.78 trillion won — materially below operating profit. The distinction matters: operating profit, net income and cash available for dividends are different measures, and a record quarter does not guarantee that such margins will persist.

The company attributed its performance to stronger prices for key memory products and higher sales of advanced server memory, including DDR5 and high-bandwidth memory (HBM). HBM sits close to AI accelerators, moving data quickly enough to keep expensive processors working. Without adequate memory capacity and bandwidth, additional computing chips alone cannot deliver their theoretical performance.

Why contracts matter as much as record profits

SK hynix said it had long-term agreements with 14 major customers covering volumes equivalent to more than half of its total production capacity in both 2027 and 2028. That offers visibility into future demand, but it should not be mistaken for a guarantee of fixed prices or permanent profitability: the company said its contracts allow prices to respond to market conditions within agreed ranges.

Reuters separately reported that SK hynix expects demand for HBM to exceed the amount it can supply and that the company views memory shortages as a potential problem through 2030. These are company assessments and forecasts, not independently established facts about future market conditions. A slowdown in AI spending, faster competitor expansion or changes in chip architecture could alter the outlook.

HBM4 is the next competitive test

SK hynix plans to increase output from its M15X facility and expects its newer HBM4 products to overtake HBM3E in both shipment volume and revenue during the fourth quarter. That is a meaningful technology transition: customers want not just more memory but better energy efficiency, speed and packaging integration.

Competition remains a key uncertainty. Other memory suppliers are expanding advanced products, while Reuters reported on October 28 that China’s CXMT is preparing limited HBM3E production despite substantial yield challenges. The Chinese company’s plans are reported, not proof that it can immediately match the output, cost or performance of established suppliers.

A huge shareholder return — alongside a costly expansion

SK hynix outlined approximately 100 trillion won of shareholder returns for 2026, including dividends and share repurchases or cancellations. It also intends to increase investment significantly next year to expand manufacturing capacity. Those two commitments create an important question for investors: can the current AI-memory profit cycle fund both substantial payouts and expensive new factories if demand or prices weaken?

For buyers of servers, computers and other electronics, the implications are broader than one stock. Sustained demand for advanced memory may raise the cost of building AI systems and compete for manufacturing resources used by more conventional devices. Yet forecasts of shortages are inherently uncertain, and consumers should not interpret a chipmaker’s earnings release as a reliable prediction of retail prices.

The takeaway: SK hynix’s record quarter provides unusually strong evidence of today’s AI-memory demand. Its multi-year contracts and HBM4 ramp offer a clearer picture of where the industry is heading — but whether the current margins can survive the next expansion cycle remains the critical unanswered question.

Illustrative image: a silicon wafer photographed by Laura Ockel via Unsplash; not an image of an SK hynix production line.