SK hynix Demonstrates Competitiveness as a Comprehensive "AI Memory Solution Provider" at FMS in the US

Showcasing a Full Range from HBM to DRAM, NAND, and Next-Generation Storage
Highlighting Global Big Tech Partnerships with NVIDIA and Dell
Unveiling the "Tiered Memory" Vision and HBF Standard Specifications

SK hynix demonstrated its technological leadership as a comprehensive "AI memory solution provider" in the global market by unveiling a wide array of next-generation artificial intelligence (AI) memory products at "FMS (Future of Memory and Storage) 2026," the world's largest semiconductor exhibition.

From the 4th to the 6th (local time), the SK hynix exhibition booth at the Santa Clara Convention Center in California, USA, where FMS 2026 was held. SK hynix

On August 7, the company announced that it participated in FMS 2026, held at the Santa Clara Convention Center in California, USA, from August 4 to 6 (local time). SK hynix operated an exhibition booth and showcased a key memory product portfolio covering the entirety of AI systems.

The exhibition hall was organized into specialized zones: the "Partnership Zone," highlighting collaboration with major big tech (large IT companies); "The Smart Edge," which featured cutting-edge NAND flash technologies; and "Next Gen. Tech," which introduced next-generation connectivity technologies.

In the "Partnership Zone," a mock-up of NVIDIA's next-generation AI accelerator "Vera Rubin" was displayed alongside a model of SK hynix's HBM4 high-bandwidth memory, providing a direct demonstration of the partnership with NVIDIA. Additionally, advanced DRAM products such as HBM4E (12-stack), HBM4 (16-stack), and the server-specialized SOCAMM2 module (based on low-power DRAM) were unveiled. Also on display were the "64GB DDR5 Server Memory Module (RDIMM)" included in Dell's latest AI server system "R7625" and the high-performance enterprise solid-state drive (eSSD) "PS1110 E3.S."

On both sides, visitors could compare the specifications and performance of major server DRAM products suited for AI workloads—such as the "256GB DDR5 RDIMM" (optimized for high capacity) and the "128GB DDR5 MRDIMM" (featuring high bandwidth)—along with SK hynix's flagship eSSD lineup, including the ▲PEB210 E1.S 9.5mm ▲PEB210 M.2 2280 ▲PS1101 E3.S.

At the "Smart Edge" zone, the "375-layer 4D NAND wafer" drew the attention of attendees. This product delivers 2.5 times higher performance per watt compared to previous generations and is optimized for AI infrastructure environments such as data centers, where both power efficiency and performance are critical. SK hynix plans to start mass production of high-performance eSSD based on this technology in the first half of next year. They also showcased high-performance mobile and PC storage devices (such as LPDDR6, LPDDR5/5X, UFS 5.0), presenting solutions to implement "physical AI" technologies used in autonomous vehicles and robotics.

In the "Next Gen. Tech" zone, SK hynix demonstrated its next-generation compute express link (CXL)-based memory technology. The exhibition included a live demonstration of "pooled memory" technology, which aggregates multiple memories to vastly enhance the data processing speed of AI agent services, earning an enthusiastic response from visitors.

Seongseong Kim, Head of Solution Development Division at SK hynix, delivered a keynote speech on June 4th (local time) at FMS 2026 held at the Santa Clara Convention Center in California, USA, under the theme "Efficient Implementation of AI Infrastructure Based on Hierarchical Memory in the Era of AI Agents." SK hynix

Together with the popularity of the exhibition booth, the keynote address delivered on the first day attracted intense interest from global semiconductor professionals. Sungseong Kim, Head of Solution Development at SK hynix, and Wookseong Kang, Head of Next-Generation Product Planning, presented on the topic "Efficient Implementation of AI Infrastructure Based on Hierarchical Memory in the Era of AI Agents." They analyzed that, as AI agents quickly move toward commercialization, the volume of data to process is soaring, and it is difficult to overcome the limits of performance and efficiency with a single semiconductor product. They advocated for a next-generation architecture based on "tiered memory"—which effectively integrates various memory types with different speeds and capacities to minimize data movement.

Kim emphasized, "As AI expands from learning-focused models to inference and agentic AI, the competitiveness of infrastructure depends not only on the individual performance of memory modules like HBM, DRAM, NAND, and SSD, but also on how optimally they are arranged and connected to minimize data movement."

Kang added, "SK hynix possesses a broad portfolio spanning all layers of AI infrastructure—from HBM to CXL-based memory and eSSD. Our role going forward will be to collaborate with customers from the design phase and deliver optimal solutions tailored to their specific needs."

On the final day of the event, Eechul Lim, Head of Solution AT at SK hynix, joined global companies such as Google and SanDisk in discussions on high-bandwidth flash (HBF), a next-generation storage technology. HBF is a new concept memory that combines the high speeds of HBM with the high capacity advantages of SSD. Lim explained, "HBF is a key solution that increases the scalability of AI systems and reduces operating costs. To achieve this, a co-design approach—which involves designing hardware and software together—is crucial."

This content was produced with the assistance of AI translation services.