SK hynix, Inc. Q3 FY2026 earnings call

• Record Financial Results: Achieved all-time quarterly records for operating profit (KRW 83.3 trillion) and margin (80%), driven by sustained AI memory demand and price increases.
• HBM4 Leadership: Full-scale mass production of HBM4 began in Q3. It is expected to surpass HBM3E in both shipment volume and revenue share in Q4, becoming the flagship product. Yield rates are stable and exceeding targets.
• Supply-Demand Imbalance: Structural supply constraints persist through 2030 due to complex leading-edge processes and fab construction lead times. Demand significantly outpaces supply.
• Long-Term Agreements (LTAs): Signed LTAs with 14 key customers, securing more than 50% of total capacity for 2027 and 2028. This improves demand visibility and stabilizes profitability.
• Shareholder Returns: Announced a total shareholder return plan of approximately KRW 100 trillion for FY2026. Includes KRW 30 trillion+ in cash dividends and KRW 60 trillion+ in share acquisitions/cancellations. A special dividend of KRW 10,000 per share reflects gains from the Kioxia investment.
• Solidigm IPO Decision: Discontinued plans for Solidigm’s IPO due to market volatility, opting instead to retain strategic control while continuing to invest in AI data center storage competitiveness.

Total revenue reached KRW 104.6 trillion, a record high, up 32% sequentially and 328% year-on-year. Operating profit was KRW 83.3 trillion (up 38% QoQ, 632% YoY) with an operating margin of 80%. Net income was KRW 39.8 trillion.

DRAM: Revenue growth driven by high-ASP products. HBM4 mass production commenced, and server DDR5 shipments rose over 40% sequentially. DRAM ASP increased in the high 20s percentage points due to strong prices and expanded HBM4 sales mix.

NAND: Bit shipments decreased slightly as the company shifted focus to high-value products like 321-layer TLC enterprise SSDs. However, ASP rose in the mid-20s percentage points due to price strength and an expanded SSD mix.

• 2027 Growth Expectations: Anticipates moderate but robust growth for 2027, with DRAM bit shipment growth in the mid-20% range and NAND in the high-teens percentage range. This reflects achievable demand given current supply constraints.
• Q4 Shipment Plans: Expects meaningful sequential growth in DRAM shipments (mid-20%) driven by 1c-nm process expansion and M15X ramp-up. NAND bit shipments are expected to increase by mid-single digits sequentially.
• CapEx Outlook: Capital expenditures in 2027 are expected to increase significantly compared to 2026 to support leading-edge process expansion and the Yongin fab ramp-up.
• Profitability Trend: Expects HBM profitability to remain high, far above past cycle peaks, with significant year-over-year growth in HBM revenue and profit driven by HBM4's higher price and improved yields.

• Exchange Rate Volatility: The strong Korean won caused a foreign exchange-related net loss of KRW 11.6 trillion, impacting reported earnings despite operational strength.
• Investment Asset Valuation: Valuation losses on investment assets totaled KRW 22.9 trillion, contributing to non-operating expenses.
• Supply Chain Constraints: Difficulty in fully meeting customer volume requirements in the near term due to the complexities of implementing leading-edge processes and long lead times for new production facilities.
• Demand Uncertainty: While AI infrastructure investment is projected to grow, there is potential for adjustment in spending pace as customers evaluate project feasibility and profitability.

Q: Citigroup asked about the sustainability of AI investment momentum given slow monetization. Management cited two reasons for continued robustness: 1) Hyperscalers report 30-80% cloud revenue growth driven by AI, with expanding backlogs and paying users, proving business viability. 2) The investor base has structurally diversified beyond big tech to include sovereign AI and enterprises, reducing reliance on single entities and offsetting potential slowdowns in one sector with demand from others.

A: Goldman Sachs asked about the implications of LTAs covering 50% of 2027/2028 capacity. Management explained that this secures medium-to-long-term demand visibility, effectively selling DRAM/HBM volumes through 2029. This creates a reliable earnings base while allowing non-LTA capacity to flexibly capture opportunities from next-generation products and market improvements.

Q: Bank of America raised concerns about overproduction risks amid aggressive fab investments (e.g., Yongin). Management clarified that investment is staged; shell structures are built early, but clean rooms and equipment move-in are phased based on demand. They assess not just CapEx forecasts but also customers' financial health and supply chain readiness. Production ramps only align with verified demand and profitability, ensuring flexibility if conditions change.

Q: SK Securities asked if lowering HBM stack counts would hurt shipments or profitability. Management stated that lower stacks (e.g., 8-stack vs 12-stack) do not proportionally reduce capacity because die density increases. Since demand exceeds supply double-fold, volume requirements remain unchanged. Profitability will improve YoY due to higher HBM4 pricing, better yields, and cost competitiveness, remaining well above historical peaks.

Q: Mirae Asset asked for updates on 4F square technology preparation. Management outlined a strategy to maximize the proven 6F square platform via EUV scaling before transitioning to 4F square or 3D DRAM. They prioritize yield, reliability, and cost competitiveness, applying new cell structures only when technologically and economically ready, ensuring commercialization timing aligns with product maturity.