NVIDIA and SK hynix have unveiled a multiyear technology partnership to codevelop next-generation memory for the global AI factory buildout, extending years of deep co-engineering into a formal roadmap that spans NVIDIA's Vera Rubin AI supercomputers, Vera CPUs, RTX Spark-powered PCs and Jetson Thor robotic computing platforms. The June 7 announcement lands as memory supply chains struggle to keep pace with hyperscaler and physical AI demand.
A Memory Roadmap Locked To NVIDIA Silicon
The multiyear agreement supports supply to address the extended development cycles of advanced HBM and DRAM as AI factories scale worldwide. SK hynix will diversify into new markets NVIDIA is creating — spanning AI infrastructure, personal AI and physical AI — codeveloping bespoke memory for Vera Rubin supercomputers, Vera CPUs, RTX Spark PCs and the Jetson Thor T2000/T3000 robotics compute stack.
AI Applied To Semiconductor Design And Manufacturing
The pact goes beyond capacity. SK hynix is already deploying NVIDIA CUDA-X libraries and the PhysicsNeMo framework to accelerate semiconductor simulation, TCAD workflows, computational lithography and in-house engineering codes. By extending those tools into the wider EDA ecosystem, the collaboration paves the way for three-way partnerships among chipmakers, NVIDIA and EDA software vendors. "AI factories are the engines of the next industrial revolution, and advanced memory is essential to their performance," NVIDIA CEO Jensen Huang said in a joint statement.
Autonomous Fabs Get An Omniverse Digital Twin
SK hynix will also advance factory digital twins for autonomous fab operations. Using NVIDIA Omniverse libraries and OpenUSD pipelines, SK hynix teams can build 3D scenes of complex semiconductor manufacturing environments, then combine NVIDIA cuOpt decision optimization and the NVIDIA Metropolis platform to orchestrate autonomous mobile robots and fab assets. The companies are exploring how to connect those digital twins with legacy MES software and agentic AI so systems can reason over fab data and automate manufacturing decisions.
Layered On A Wider Korea AI Push
The deal deepens NVIDIA's Korean supply chain at the same moment Seoul is doubling down on chips. It follows South Korea's $880B AI and chip drive and SK Hynix's $28B Nasdaq uplisting, and dovetails with SK hynix's recent analog in-memory AI chip collaboration with TetraMem. "Together, we are codeveloping the next generation of memory for AI factories and applying AI to how we design and manufacture semiconductors," said SK Group Chairman Chey Tae-won.
Reporting based on coverage from NVIDIA Newsroom, SK hynix and TechPowerUp.