Nvidia CEO Jensen Huang confirmed on June 5, 2026, that Samsung Electronics, SK Hynix, and Micron Technology have all been certified to supply HBM4 memory chips for Nvidia’s Vera Rubin AI accelerator platform during his visit to Seoul [1, 2, 3].
HBM4 is a next-generation high-bandwidth memory product essential for Nvidia’s latest AI platform, Vera Rubin, which is already in volume production [4, 5, 1, 2, 3]. Huang said, “三家供應商都已取得認證資格。三家供應商都已投入生產，並且都在全力競爭支援Vera Rubin” (All three suppliers are certified, have begun production, and are competing fully to support Vera Rubin) [1].
The three suppliers dominate the global computing storage semiconductor market and compete fiercely for Nvidia’s HBM4 business [4, 5, 2, 3]. Each Nvidia Vera Rubin AI server system is equipped with several terabytes of HBM4 memory to handle high-demand AI workloads [2, 3].
All three companies have begun mass production of HBM4 to support Nvidia’s platform [5, 1, 2, 3]. The Vera Rubin chip is Nvidia’s latest generation AI accelerator and Nvidia plans to start shipments in autumn 2026 [1, 2, 3].
During his trip, Huang also confirmed Nvidia’s intention to establish a new AI and robotics-focused R&D center in South Korea. The company is actively recruiting to staff the facility [2, 3].
Nvidia’s move to certify multiple major suppliers aims to secure stable access to cutting-edge memory technology critical for the Vera Rubin platform’s performance and release schedule. The shipments planned for later this year are a key milestone in Nvidia’s AI hardware rollout [1, 2].
