SK Hynix Tops 100 Trillion Won Quarterly Sales Milestone as AI Demand Drives Record Profit

South Korean memory chip manufacturer SK Hynix reported record-breaking third-quarter financial results for 2026, exceeding 100 trillion won in quarterly sales for the first time as AI-driven component demand accelerated.

The company outperformed consensus estimates across top-line revenue and profitability while unveiling an expansive shareholder return plan totaling approximately 100 trillion won ($25 billion) for the fiscal year.

Investor optimism followed board approval of a 14 trillion won cash dividend for the third quarter alongside a 20 trillion won share repurchase program. The quarterly cash payout equals 20,000 won per share, combining a 10,000-won base dividend tied to operational results with a 10,000-won special dividend funded by gains realized from its strategic investment in Kioxia. Of the approved share buyback, 10 trillion won will be retired, while the remaining portion is allocated for employee equity compensation.

These capital allocations reflect the chipmaker’s goal of returning over 50% of cumulative free cash flow to shareholders. The full fiscal year 2026 return framework commits to at least 30 trillion won in cash dividends and over 60 trillion won in equity repurchases and cancellations. This total includes a previously announced 40 trillion won share buyback completed in late October.

For the quarter ending September 2026, net sales reached 104.62 trillion won ($78.86 billion), representing a 32% quarter-on-quarter increase and a 328% jump year-on-year. Operating profit expanded to 83.33 trillion won—up 38% from the second quarter and 632% from the prior year—generating an operating margin of 80%. Net income reached 39.78 trillion won. The results surpassed forecasts compiled by LSEG SmartEstimates, which anticipated revenue of 98.34 trillion won and operating profit of 76.72 trillion won.

Earnings momentum was driven by pricing gains and expanding shipments of high-value memory architectures. DRAM performance was underpinned by rising demand for server DDR5 modules and initial commercial shipments of next-generation HBM4 chips. Management projects HBM4 to eclipse HBM3E in both shipment volume and revenue during the fourth quarter, establishing itself as the primary revenue generator. In NAND flash memory, sales were boosted by high-capacity 321-layer triple-level cell eSSDs for enterprise data centers, with 321-layer capacity targeted to exceed half of domestic production by year-end.

Robust operational cash generation fortified the company’s financial structure, boosting cash and cash equivalents to 114.5 trillion won and expanding net cash to 98.7 trillion won following debt repayments.

Alongside its financial results, SK Hynix confirmed the termination of its review regarding an initial public offering or external capital raise for its U.S. storage subsidiary, Solidigm. Company leadership cited macro financial market volatility and valuation corrections across technology stocks, concluding that an equity market listing would not achieve its capital goals.

To secure mid-to-long-term demand, the manufacturer has finalized long-term supply agreements with 14 major global technology customers, including prominent cloud service providers. These agreements lock in over 50% of planned production capacity for 2027 and 2028 through flexible pricing frameworks designed to adjust within agreed parameters based on market conditions. Capital expenditures are set to increase significantly next year to support domestic and international infrastructure build-outs, including full-scale operations at the Yongin Phase 1 site.

Looking forward, SK Hynix projects that structural supply deficits across the memory sector will endure through 2030. Manufacturing complexities inherent in advanced process nodes and extended fabrication facility lead times are expected to keep production lagging behind market demand. While short-term demand remains anchored by enterprise and consumer AI applications, mid-to-long-term adoption is expected to broaden into physical AI domains, including robotics and autonomous navigation platforms.