NVIDIA Partners With SK Hynix for Next-Gen AI Memory Development

NVIDIA and SK Hynix signed a multi-year technology partnership to co-develop next-generation memory for AI infrastructure. The agreement targets supply constraints in advanced memory fabrication and capital investments. Both companies will apply artificial intelligence tools to semiconductor design using NVIDIA CUDA-X libraries and PhysicsNeMo. SK Hynix plans to advance factory digital twins through NVIDIA Omniverse, OpenUSD, and cuOpt.

Multi-year agreement targets supply constraints and advanced memory fabrication

The collaboration focuses on developing memory for Vera Rubin AI supercomputers and Vera CPUs. The partnership also covers RTX Spark-powered PCs and Jetson Thor robotic computing platforms. Jensen Huang described advanced memory as essential to the performance of AI factories. He noted that SK Hynix plays a central role in delivering memory technologies for NVIDIA AI computing platforms.

SK Hynix will diversify its manufacturing into new markets including personal AI and physical AI. The companies aim to support the global expansion of AI infrastructure from frontier model training to agentic systems. Both firms will integrate AI into their semiconductor design workflows. SK Hynix will use NVIDIA's software stack to optimize factory operations.

The partnership addresses supply constraints by supporting advanced memory development and fabrication. NVIDIA and SK Hynix will co-develop the next generation of memory for AI factories. The agreement covers multiple product lines including Vera Rubin supercomputers and Jetson Thor robotic platforms. Jensen Huang emphasized that the collaboration supports the accelerating expansion of global AI infrastructure.

Jensen Huang called SK Hynix an extraordinary partner in delivering advanced memory technologies. He stated that both companies will codevelop next-generation memory for AI factories. The partnership aims to support frontier model training and physical AI systems. SK Hynix will apply NVIDIA's CUDA-X libraries and PhysicsNeMo to semiconductor design.

The multi-year agreement targets supply constraints in advanced memory fabrication. Both companies will use NVIDIA's Omniverse, OpenUSD, and cuOpt to advance factory digital twins. The collaboration covers Vera Rubin AI supercomputers, Vera CPUs, RTX Spark PCs, and Jetson Thor robotic platforms. Jensen Huang highlighted that advanced memory is essential for the performance of AI factories.