Nvidia and Korea's MSIT widen sovereign AI pact to chips, data centers and talent

NVIDIA and South Korea's Ministry of Science and ICT signed off on a wider sovereign AI pact, covering everything from AI chips and data centers to talent, startups, foundation models, agents and security. The deal gives Korean NPU and XPU makers a shot at plugging into NVIDIA's NVLink Fusion ecosystem, the interconnect tech built to move data fast between GPUs and other processors. For Seoul, it's another push toward tech independence in the AI race.

NVIDIA is putting training on the table too: certification programs, hackathons and technical seminars aimed at squeezing more out of its GPUs. Those line up with Korea's existing 'AI for All' program, which is meant to spread AI access across industries and institutions rather than keep it locked inside a few labs.

On the model side, NVIDIA will team up with three selected consortia and four of Korea's leading foundation model builders. They'll work on NVIDIA Dynamo, the company's software for optimizing and scaling AI inference, to make Korean-specific models faster and easier to run at scale.

The headline collaboration is with SK Telecom and Rebellions. Backed by MSIT, the trio will use NVLink Fusion to tune AI models and workloads for Korean use cases as part of the sovereign AI infrastructure build-out.

It's the latest brick in a wall NVIDIA has been laying in Korea for months. In July 2026, Seoul National University and NVIDIA launched a multi-year NVIDIA AI Technology Center Joint Lab for research, talent and infrastructure. Back in June, NVIDIA linked arms with LG Group, SK Group and Hyundai Motor Group on AI infrastructure and agentic and physical AI applications, while NAVER is building an NVIDIA-powered AI factory of its own.