SK Hynix posted its first-ever quarter above 100 trillion KRW in revenue, reporting 104.6155 trillion KRW in Q3 FY2026 with an 80% operating margin [1][1]. HBM4 shipments fully ramped in the quarter, and the company projects HBM4 will surpass HBM3E in both volume and revenue starting Q4 FY2026 [1]. With long-term agreements already covering more than 50% of 2027-2028 capacity [1], the AI memory investment cycle has moved decisively from speculative to contracted.
What is Covered in this Article
- Record Q3 FY2026 financials: first 100T KRW quarter [1][1]
- HBM4 ramp and DDR5 growth as primary AI revenue engines [1]
- Structural supply-demand imbalance forecast through 2030 [1]
- Long-term agreements covering 50%+ of 2027-2028 capacity [1]
- Roadmap: M15X expansion, 321-layer NAND, and High Bandwidth Flash [1][1]
The News: SK Hynix released its Q3 FY2026 financial results on October 28, 2026, marking the first time in the company's history that quarterly revenue exceeded 100 trillion KRW [1]. Operating revenue reached 104.6155 trillion KRW, operating profit hit 83.3338 trillion KRW at an 80% margin, and net profit came in at 39.7756 trillion KRW, representing a 38% net profit margin [1][1]. Revenue and operating profit grew 32% and 38% QoQ respectively [1], and surged 328% and 632% year-over-year [1]. HBM4 shipments fully ramped in Q3, with 321-layer TLC enterprise SSD sales also expanding [1][1]. During the financial results announcement, CEO Kwak Noh-jung disclosed the suspension of the Solidigm IPO review, citing unfavorable market conditions [1], while unveiling a total shareholder return program of approximately 100 trillion KRW for FY2026 [1].
SK Hynix Breaks 100T KRW: AI Memory Demand Is Now Structural
Analyst Take: These results are not a cyclical spike. SK Hynix's 80% operating margin and 632% year-over-year operating profit growth [1] reflect a fundamental repricing of AI-grade memory, driven by the same enterprise AI adoption wave that Futurum's survey data shows is now mainstream: 45.9% of data decision-makers report already using AI tools for data management [2]. The company's financials are, in effect, a real-time ledger of that adoption.
HBM4 and DDR5: The AI Infrastructure Revenue Engine
HBM4 shipments fully ramped in Q3 FY2026, with server-grade DDR5 sales expanding in parallel to drive DRAM segment growth [1]. SK Hynix projects HBM4 will surpass HBM3E in both shipment volume and revenue starting Q4 FY2026, establishing it as the company's flagship product [1]. This trajectory maps directly to enterprise demand: 52.3% of data decision-makers identify generative and agentic AI tools or platforms as a top investment priority [2], and hyperscalers are translating that enterprise pull into accelerating data center buildouts. The company plans to expand M15X factory capacity to maximum levels by year-end specifically to meet HBM4 demand [1]. On the NAND side, 321-layer TLC eSSD sales expanded in Q3, with plans to raise 321-layer process share to over 50% of domestic Korean capacity by Q4 FY2026 [1], reinforcing SK Hynix's position across the full AI storage stack.
Supply Constraints Are Structural, Not Cyclical
SK Hynix forecasts the memory market supply-demand imbalance will persist through 2030, driven by increasing process complexity and long facility construction timelines [1]. This is not a temporary shortage. Advanced node transitions require more process steps, longer yield ramp times, and capital-intensive tooling that cannot be accelerated by demand alone. For enterprise data infrastructure buyers, this has a direct operational implication: growth in data capacity and/or complexity is already cited as a top pressure by 42.7% of data decision-makers (n=839) [2], and that pressure will compound against a constrained supply environment for years. Demand and supply are moving in opposite directions, and SK Hynix is positioned at that intersection.
Contracted Demand: LTAs Signal a Maturing Investment Cycle
Long-term supply agreements with 14 major customers, including large CSPs, already cover more than 50% of 2027 and 2028 total production capacity [1]. These agreements use a pricing structure that reflects market conditions within a customer-agreed range, providing both revenue stability and flexibility. CEO Kwak Noh-jung stated that the company's technology competitiveness is converting customer trust into long-term agreements, shifting the business toward durable demand visibility and earnings stability. This is a meaningful structural shift: AI infrastructure spend has moved from discretionary to committed. Enterprise data leaders should note that AI-augmented and agentic automated analytics ranks as a top trend for 38.6% of data decision-makers (n=839) [2]; these LTAs confirm that the underlying infrastructure layer is now being locked in at scale. The company also reported a net cash position of 98.7 trillion KRW as of Q3 end [1], giving it the financial capacity to fund the next generation of capacity expansion without use risk.
Roadmap: HBF and Next-Generation NAND as AI Throughput Solutions
Beyond current-generation products, SK Hynix is developing High Bandwidth Flash as a core AI throughput and cost-efficiency solution for NAND flash in the medium-to-long term [1]. HBF targets the gap between DRAM bandwidth and NAND density, offering a potential path to higher-capacity AI inference and training configurations at lower cost per bit. Combined with the planned Yongin Phase 1 fab coming online next year and continued 1c-node DRAM process ramp, the company's roadmap is designed to sustain its lead through the next product generation. For enterprise data platform architects, this signals that the memory tier of the AI stack will continue to evolve rapidly, with new form factors and performance profiles requiring ongoing infrastructure reassessment.
What to Watch
- HBM4 volume crossover: whether HBM4 shipments and revenue definitively surpass HBM3E in Q4 FY2026 results, confirming the product transition timeline [1]
- LTA pricing dynamics: how the market-reflective pricing structure within long-term agreements holds as additional HBM4 capacity comes online from M15X expansion in Q1 2027 [1][1]
- Supply imbalance indicators: whether competitor capacity additions in early 2027 begin to close the structural gap SK Hynix forecasts persisting through 2030 [1]
- HBF commercialization timeline: when High Bandwidth Flash moves from roadmap to customer qualification, signaling the next inflection in AI storage architecture [1]
- Enterprise AI adoption acceleration: whether the share of data decision-makers already using AI tools for data management grows beyond 45.9% in the next survey cycle, sustaining downstream memory demand [2]
Sources
1. SK海力士发布2026财年第三季度财务报告, Skhynix
2. 1H 2026 Data Intelligence, Analytics, and Infrastructure Decision Maker Survey Report, Futurum Research, March 2026
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Read the full Futurum Group Disclosure.
Other Insights from Futurum:
AI Platform Investment: SK Hynix 54T Won Deal
SK hynix's HBF Technology: A Major shift for AI Infrastructure?
SK Hynix ADR Issuance Strategy
Author Information
This content is written by a commercial general-purpose language model (LLM) along with the Futurum Intelligence Platform, and has not been curated or reviewed by editors. Due to the inherent limitations in using AI tools, please consider the probability of error. The accuracy, completeness, or timeliness of this content cannot be guaranteed. It is generated on the date indicated at the top of the page, based on the content available, and it may be automatically updated as new content becomes available. The content does not consider any other information or perform any independent analysis.
