SK Hynix executives will strengthen global AI partnerships at Nvidia GTC 2026. SK Group Chairman Choi Tae-won (최태원) and SK Hynix CEO Kwak No-jung (곽노정), among other senior executives, will attend the event in person. During the event, they plan to meet global big tech companies to share insights on AI technology development and changes in infrastructure architecture and to discuss medium- and long-term cooperation plans.

SK Hynix is participating in GTC (GPU Technology Conference) 2026, held in San Jose, California, from March 16 to 19 local time, and will showcase the competitiveness of its AI memory products including HBM4, it said on March 17. Nvidia GTC is a global AI conference where major companies and developers worldwide share the latest technologies and industry direction in AI and accelerated computing.

SK Hynix will set up an exhibition space under the theme "Spotlight on AI Memory" and unveil its AI memory product lineup. "We have installed memory products in Nvidia AI infrastructure that can minimize data bottlenecks and maximize performance in AI training and inference," the company said. "Based on our partnership with Nvidia, we will demonstrate the competitiveness of memory technology, a key infrastructure in the AI era," it added.

The exhibition hall consists of three areas: an Nvidia collaboration zone, a product portfolio zone and an event zone. At the entrance, the Nvidia collaboration zone presents, through models and actual products, cases where HBM4, HBM3E and SOCAMM2 are applied to Nvidia's AI platform. It will also display Nvidia's AI supercomputer DGX Spark, which includes a liquid-cooled eSSD and LPDDR5X developed in collaboration with Nvidia.

The product portfolio zone introduces the full lineup of AI memory products, including HBM4 and HBM3E, high-capacity DRAM modules for servers, LPDDR6, GDDR7, eSSD and automotive memory solutions. Visitors can use a joystick to select products of interest and check their features and application cases.

In the event zone, SK Hynix will run an "HBM 16-layer stacking game" inspired by the HBM stacking structure. The hands-on experience of stacking virtual memory chips is designed to help visitors understand TSV processes and high-stack packaging technology.

Technical sessions will be held alongside the exhibition. Do Seung-yong (도승용), a vice president in charge of DT, will appear as a co-presenter on March 17 in the "Building the Future of Manufacturing" session and propose cases for using AI technology in process innovation and AI infrastructure plans. On March 19, Moon Dong-wook (문동욱), head of AI Platform SW, will in the "How HBM4 Unlocks Efficient LLM Serving at Scale" session cite memory bandwidth and capacity as key constraints for LLM inference and introduce HBM4-based solutions.

A SK Hynix official said, "As AI technology advances, memory is becoming a key factor that goes beyond a simple component and determines the structure and performance of overall AI infrastructure." "Based on memory technology capabilities that cover all areas of AI, from data centers to on-device, we will work with global partners to shape the future of AI," the official added.