SK Hynix posted its strongest quarterly performance on record, as demand for memory chips used in artificial-intelligence infrastructure continued to exceed available supply. The South Korean chipmaker said the imbalance could persist through 2030, while much of its future production is already reserved by major customers.
Record sales and profit
Revenue reached 104.6 trillion won in the third quarter, up 328% from a year earlier, according to the company’s earnings release and reporting by Decrypt. It was the first quarter in which SK Hynix’s sales exceeded 100 trillion won. Operating profit climbed 632% to 83.3 trillion won, producing an operating margin of 80%.
An operating margin measures how much profit a company retains from its core business after operating costs. In SK Hynix’s case, the result means it generated about 80 won in operating profit for every 100 won of revenue during the quarter. Analysts surveyed by Bloomberg had expected operating profit of 77.5 trillion won.
Net income was 39.8 trillion won, 216% higher than a year earlier but 58% lower than in the second quarter. Chief financial officer Kim Woo-hyun said the difference reflected 32.6 trillion won in net non-operating expenses, which are costs outside the company’s core chip-sales operations. Those expenses included an 11.6 trillion-won currency loss and a 22.9 trillion-won loss from investment valuations.
AI memory backlog extends into future years
SK Hynix told investors that memory demand is likely to remain greater than supply through 2030. It also said its expected DRAM and high-bandwidth memory, or HBM, volumes through 2029 are effectively sold out.
DRAM is widely used in servers, personal computers and smartphones. HBM is a stacked form of memory positioned alongside AI processors, where it helps supply data at the high speeds required by demanding computing workloads. The technology has become an important component of AI accelerators and the data-center systems that support them.
Kim said the supply constraint reflected the complexity of producing advanced chips and the long lead times required to build new manufacturing facilities. SK Hynix said some customers had recently changed their HBM specifications, which it viewed as evidence that demand was overwhelming supply. Customers are building more AI accelerators while working with limited chip availability, the company said.
The company added that customers’ requested volumes are currently more than twice what SK Hynix can supply. That gap helps explain why the manufacturer is providing production visibility several years ahead, rather than relying only on short-term orders.
Long-term contracts and questions over AI spending
SK Hynix has signed long-term agreements with 14 key customers, including major global cloud providers. The contracts cover more than half of the company’s production capacity for 2027 and 2028. They generally run for five years, allow prices to move within an agreed range and include protections such as customer deposits.
The company expects investment to rise significantly in 2027, when the first phase of its Yongin chip complex is scheduled to enter full-scale operation. The planned expansion comes as investors debate whether the rapid growth in AI infrastructure spending can continue, particularly when some AI businesses are finding financing more difficult to secure.
During the earnings question-and-answer session, Citigroup analyst Peter Lee asked what supported confidence in AI investment despite concerns that monetization was taking time. Those concerns had also been raised by Citadel founder Ken Griffin, who said eight days earlier that investors were “almost certainly” in an AI bubble.
Chief executive Kwak Noh-Jung pointed to reported year-over-year cloud revenue growth of 30% to 80% at major cloud customers and cloud order backlogs that had increased twofold to fivefold. He also said AI spending was spreading beyond a small group of large technology companies to non-technology businesses, government-backed sovereign AI projects and on-premises systems operated on companies’ own servers.
SK Hynix expects higher memory prices and limited supply to weaken demand from PC and smartphone manufacturers. It said demand from AI infrastructure should offset that decline. At the same time, Elon Musk wrote on October 25 that debt markets had almost entirely shut off access to money for firms without substantial non-AI revenue, highlighting a financing risk for parts of the broader AI infrastructure market.
Moneyiar market picture
Moneyiar market data at the given time, October 25, 2026, showed the following prices and performance:
- Solana (SOL): $101.5978856709, up 0.03% over 24 hours and down 1.52% over seven days.
- AKEDO (AKE): $0.031015752, up 0.8% over 24 hours and 2.5% over seven days.
- USD.AI (CHIP): $0.0443846724, down 0.29% over 24 hours and up 9.36% over seven days.
- Loaded Lions (LION): $0.0011817703, down 2.39% over 24 hours and down 3.06% over seven days.
These figures are Moneyiar data recorded at approximately 22:38-22:39 on October 25, 2026, and are separate from SK Hynix’s reported quarterly results.
