NVIDIA partners with SK Hynix on next-gen AI memory; codeveloping for Vera Rubin and autonomous fabs

NVIDIA and SK Hynix announced a multiyear technology partnership on June 7 to codevelop next-generation memory for the global AI factory buildout, with SK Hynix committing supply aligned to NVIDIA's infrastructure roadmap spanning AI supercomputers, personal AI PCs, and robotic platforms. The deal builds on years of deep co-engineering collaboration and addresses the extended development cycles of advanced memory as AI factories scale globally. SK Hynix will codevelop memory for NVIDIA Vera Rubin AI supercomputers, Vera CPUs, RTX Spark-powered PCs and Jetson Thor robotic computing platforms.

Beyond memory codevelopment, the two companies will apply NVIDIA's CUDA-X libraries and PhysicsNeMo framework to accelerate SK Hynix's semiconductor simulation and design workflows. SK Hynix is also developing fab digital twins using NVIDIA Omniverse libraries and OpenUSD pipelines to enable autonomous manufacturing operations using NVIDIA's cuOpt decision optimization engine. The partnership does not disclose financial terms, duration, or specific volume commitments, but signals deep alignment on AI memory supply as a strategic advantage.

For architects planning multi-year AI infrastructure deployments, this partnership tightens memory supply visibility and links SK Hynix's next-generation capacity directly to NVIDIA's hardware roadmap. However, memory prices will remain elevated; the deal codifies supply prioritization for NVIDIA platforms over commodity DRAM markets. SK Hynix' dual commitment to capacity expansion and AI-accelerated fab automation suggests sustained investment in manufacturing efficiency, though relief on DRAM/HBM costs remains a multi-year story.