As demand for AI infrastructure accelerates worldwide, how will memory technology keep pace? NVIDIA and SK hynix are expanding their collaboration across development.
NVIDIA and SK hynix have entered into a multiyear technology partnership to develop next-generation memory technologies for AI infrastructure while expanding collaboration in semiconductor design and manufacturing.
The agreement is intended to support growing demand for advanced memory as AI computing capacity expands globally. Both companies said the partnership aligns memory development with NVIDIA’s future AI infrastructure roadmap, helping ensure supply can meet the requirements of increasingly large-scale AI systems.
As part of the collaboration, SK hynix will work with NVIDIA on memory technologies for a range of future computing platforms, including the NVIDIA Vera Rubin AI supercomputer architecture, Vera central processing units (CPUs), RTX Spark-powered personal computers and Jetson Thor robotic computing systems.
The partnership also reflects NVIDIA’s expansion into markets spanning AI infrastructure, personal AI and physical AI applications.
Beyond hardware development, the companies will jointly explore the use of AI to improve semiconductor engineering and production processes. SK hynix is already employing NVIDIA’s CUDA-X software libraries and PhysicsNeMo framework to accelerate semiconductor simulations, including technology computer-aided design (TCAD), computational lithography and other internal engineering workflows.
The companies said the initiative could support broader collaboration between chip manufacturers, electronic design automation software providers and NVIDIA to improve semiconductor development tools and simulation capabilities.
Manufacturing innovation forms another key part of the agreement. SK hynix is developing digital twin models of semiconductor fabrication facilities to support more autonomous factory operations.
These virtual environments will use NVIDIA Omniverse technologies, OpenUSD-based workflows and AI-driven optimisation tools to model, simulate and improve manufacturing processes.
The digital twin systems are also expected to assist in managing autonomous mobile robots and factory assets through NVIDIA’s cuOpt optimisation engine and Metropolis platform. The companies are also exploring ways to integrate digital twins with existing factory software and agentic AI systems to automate tasks and support operational decision-making.
