SK Hynix, Nvidia discuss AI memory alliance at GTC 2026 in San Jose

SK Group Chairman Chey Tae-won and SK Hynix President Kwak Nojeong attended Nvidia’s GTC 2026 in San Jose, California, on the 16th local time, to view exhibits together and discuss a semiconductor alliance with Nvidia.

GTC is described as a global AI conference where major companies and developers share the latest in artificial intelligence and accelerated computing, highlighting trends in AI hardware and software ecosystems.

At the event, SK Hynix outlined plans to align with global AI industry trends and to discuss mid- to long-term collaboration with Nvidia, noting meetings with other major tech players to exchange insights on AI development and infrastructure changes.

SK Hynix said its memory products aim to minimize bottlenecks in AI training and inference while maximizing performance, stating that these components are being integrated into Nvidia’s AI infrastructure as part of the partnership.

The company is presenting under the theme “AI Memory Spotlight,” with an exhibition space designed to showcase AI memory technology and solutions. The layout includes a Nvidia collaboration zone, a product portfolio zone, and an events zone, using interactive content to illustrate concepts.

In the Nvidia collaboration zone, SK Hynix highlights how its memory products—such as HBM4, HBM3E, and SOCAMM2—are employed in Nvidia AI platforms, featuring memory configurations shown through models and physical displays that resemble GPU-based accelerator setups.

Displays include a liquid-cooled eSSD developed in collaboration with Nvidia, along with SK Hynix LPDDR5X modules showcased inside Nvidia’s AI supercomputer, DGX Spark.

The product portfolio zone presents core AI-infra memory products, including HBM4 and HBM3E, high-density server DRAM modules, LPDDR6, GDDR7, eSSD, and automotive memory solutions, giving visitors a comprehensive view of memory offerings aimed at AI workloads.

Visitors can use a joystick to select products and learn about their features and applications on nearby screens, while a dedicated “HBM 16-layer stacking” game forms part of the exhibition experience.

For the United States and global markets, the gathering signals continued collaboration between Nvidia’s U.S.-based AI platform ecosystem and Korean memory suppliers. The partnership underscores the importance of memory technology in AI infrastructure, with potential implications for supply chains, pricing, and the pace of AI deployment in data centers and cloud services.