NVIDIA and SK hynix have entered a multiyear technology partnership to co-develop next-generation memory essential for the global AI factory buildout. This collaboration aligns SK hynix’s memory expertise with NVIDIA’s AI infrastructure roadmap, covering innovations for Vera Rubin AI supercomputers, Vera CPUs, RTX Spark-powered PCs, and Jetson Thor robotic platforms.
How is this partnership shaping the future of AI infrastructure?
The agreement addresses the extended development cycles and capital-intensive nature of advanced memory production. By working together, the companies aim to ensure that memory supply keeps pace with the rapid expansion of AI factories worldwide. According to Jensen Huang, founder and CEO of NVIDIA, advanced memory is the engine of the next industrial revolution, and this partnership secures the necessary technology for everything from frontier model training to agentic and physical AI.
The partnership extends beyond hardware supply. The companies are integrating AI into the semiconductor design process itself, using NVIDIA CUDA-X libraries and PhysicsNeMo to accelerate simulations and engineering workflows.
What role does simulation play in modern semiconductor manufacturing?
SK hynix is leveraging NVIDIA’s software ecosystem to modernize its design and manufacturing. By using CUDA-X libraries, the company is speeding up technology computer-aided design (TCAD) and computational lithography workflows. The integration of the PhysicsNeMo framework allows for core-workload acceleration, which helps engineers run complex simulations more efficiently. This initiative creates a new model for three-way collaboration between chipmakers, software vendors, and NVIDIA.
How are digital twins transforming factory operations?
To achieve fully autonomous fab operations, SK hynix is building digital twins of its manufacturing facilities. These virtual replicas use NVIDIA Omniverse and OpenUSD to visualize and optimize 3D factory environments. According to the companies, this setup allows for the simulation of complex manufacturing flows before they are implemented on the physical floor.
Look for the use of NVIDIA cuOpt and the Metropolis platform in these digital twins. These tools enable the optimization of autonomous mobile robots and other assets, helping manufacturers reason over fab data to automate decision-making.
Frequently Asked Questions
What specific NVIDIA platforms will use this new memory?
SK hynix is co-developing memory for NVIDIA Vera Rubin AI supercomputers, NVIDIA Vera CPUs, NVIDIA RTX Spark-powered PCs, and NVIDIA Jetson Thor robotic computing platforms.
Why are digital twins important for semiconductor fabs?
Digital twins allow manufacturers to simulate factory environments in 3D. This helps optimize the movement of assets and robots, ultimately driving toward fully autonomous fab operations.
How does this partnership impact semiconductor design?
The companies are applying AI to the design process. By using NVIDIA CUDA-X and PhysicsNeMo, SK hynix can accelerate semiconductor simulations and TCAD workflows, shortening the time required to bring new designs to market.
Interested in the future of AI hardware? Subscribe to our newsletter for the latest updates on infrastructure and semiconductor innovation, or join the conversation in the comments below.
