Nvidia, SK Hynix to co-develop AI memory for data centers

Nvidia and SK Hynix signed a multiyear deal to develop next-generation memory chips for AI data centers. The partnership uses Nvidia's CUDA-X library and PhysicsNeMo framework to speed up semiconductor simulations. SK Hynix will expand into personal AI and physical AI applications like autonomous robots and vehicles. Nvidia CEO Jensen Huang emphasized stable memory supply as essential for AI factory development. The collaboration supports AI training, agentic AI, and chip manufacturing simulations.