During its Q2 2026 earnings call, Samsung Electronics stated that the AI computing boom is driving a rapid surge in server memory demand, and the supply-demand gap in the memory market is expected to widen further next year. The company is continuously upgrading its product mix, with a rapid increase in the proportion of high-value-added products such as HBM4 and server SSDs. Meanwhile, it is locking in the majority of its capacity through long-term supply agreements (LTAs) and accelerating the expansion of its Taylor fab in the US and advanced process deployment to continuously consolidate its leading position in the AI memory market.
AI Drives Memory Demand Higher; Supply-Demand Imbalance to Intensify Next Year
In the Q2 earnings call, Samsung Electronics noted that although demand in the mobile phone and PC sectors is currently weakening due to memory price hikes, the growth rate of new demand for server DRAM, SSDs, and HBM far exceeds this. Hyperscale data centers are stepping up infrastructure investments to maintain their AI leadership and are continuously making additional supply requests. Memory demand will remain strong in the second half of the year.
With the widespread adoption of agentic AI, the demand growth trend for both general-purpose servers and AI servers is accelerating. Despite Samsung's efforts to expand production, demand is growing even faster, and the supply-demand gap is expected to widen further next year.
Q3 Bit Shipment Growth Expected; Product Mix Optimized to Capitalize on Market Trends
In Q2 2026, Samsung Electronics' DRAM bit shipments grew by 10%-13% quarter-on-quarter, exceeding expectations, while NAND bit shipments grew by 1%-3% QoQ, in line with expectations. The average selling price (ASP) of DRAM increased by 44%-46% QoQ, and the NAND ASP rose by 67%-69% QoQ. Benefiting from AI computing demand, Samsung's DRAM and NAND bit shipments both hit record highs, with server products accounting for their highest-ever proportion.
To address this challenge, Samsung Electronics plans to optimize its product mix based on customer demand and inventory levels. For the third quarter, DRAM bit shipments are expected to grow by 4%-6% QoQ, and NAND bit shipments are projected to increase by 7%-9% QoQ. Samsung stated that it will rely on its industry-leading technological strength and diversified product lines to actively respond to user demand and maintain its leading position in the AI memory market. In an environment where supply is expected to remain tight next year, Samsung is committed to achieving a balanced operation between HBM and general-purpose DRAM, aiming to bring their market shares to a comparable level.
HBM4 to Account for Over 60% of Sales in H2; SSD Sales to Exceed 60% of Total NAND This Year
Samsung Electronics announced that customer evaluations for HBM4 have been successfully completed. As customers expand their production scale in the second half of the year, market demand is growing rapidly. The company is steadily increasing capacity as planned to expand HBM4 supply. During the earnings call, Samsung revealed that HBM4 sales in Q3 are expected to more than triple QoQ, and H2 sales are projected to account for over 60% of the company's total HBM sales. Regarding next year's HBM business, Samsung stated that it has already finalized HBM supply agreements for 2027. With its industry-first capability to sample HBM4E, the company is highly confident in commercializing with key customers as scheduled.
Demand for server SSDs has surged across AI servers, general-purpose computing servers, and servers dedicated to key-value and caching. The NAND business significantly drove Samsung's overall memory business performance in Q2. Samsung noted that its NAND business is being restructured into a high-value-added industry centered on server SSDs. Server SSD sales this year are expected to grow by over 20% YoY, accounting for more than 60% of total NAND sales. Samsung also disclosed that its PCIe 6.0 SSDs, developed to meet high-performance TLC storage demands, have received positive feedback from major customers. The company aims to be the first to capture the initial market for next-generation AI platform PCIe 6.0 SSDs.
LTAs Lock in 70% of Capacity; Taylor Phase 2 Fab Targeted for Mass Production in 2030
To cope with the persistent supply shortage of AI memory, Samsung has launched a new contract framework, planning to allocate 60%-70% of its total capacity through LTAs. Samsung is signing long-term supply agreements on a "5-year rolling" basis, with a base contract term of five years and an option to extend the contract by one year through annual negotiations. The company has already completed signings with five global data center customers, and negotiations with five other major AI-related customers are in their final stages. The number of customers seeking long-term contracts is expected to continue growing. To strengthen contract enforcement, the new agreements include advance payment clauses (equivalent to multi-year deposits). Approximately one-quarter of the total advance payments has already been received, with the scale expected to expand as new contracts are signed. In terms of pricing, Samsung has shifted away from previous models to adopt differentiated pricing based on customer and product categories, while setting price caps in general memory contracts to mitigate investment risks caused by severe market price fluctuations.
Additionally, to promptly respond to continuously growing customer demand, Samsung Electronics is preparing to begin construction of its Taylor Phase 2 fab by the end of this year, with a target of achieving mass production in 2030. The existing Taylor production facilities are progressing with commissioning preparations this year as planned, after which 2nm process capacity will be gradually increased. Samsung previously announced that it has secured 2nm process orders from major CSPs as well as AI and HPC customers, and expects the number of 2nm process orders this year to more than double compared to last year.