NVIDIA adds inference and skills programs to South Korea’s sovereign AI push

NVIDIA and South Korea’s Ministry of Science and ICT are broadening their sovereign AI collaboration, pairing support for Korean model builders with workforce training and a potential opening for domestic chip companies. NVIDIA’s November 8 announcement spans both the infrastructure used to run AI and the skills needed to use it.

The ministry-level effort reaches beyond computing hardware. NVIDIA lists data centers, talent, startups, models, agents and security among its areas of cooperation. Its concrete commitments describe three different kinds of work: helping run models, developing technical skills and exploring chip-company participation.

Helping Korean models run at scale

Under Korea’s AI for All initiative, NVIDIA will work with three selected consortia and four Korean foundation model builders. They will use NVIDIA Dynamo software to optimize and scale inference—the process of running a model to produce answers or other outputs. The stated focus is serving AI, rather than a commitment to train a new model.

A potential route for Korean chip companies

For Korean companies developing specialized processors, described as NPUs and XPUs, the collaboration opens an exploratory path toward NVIDIA’s NVLink Fusion ecosystem. NVIDIA says the agreement lays the groundwork for those companies to consider participation. That is narrower than announcing that particular Korean chips have joined the ecosystem or are ready for deployment: participation remains a possibility, not a completed integration.

Training targets better use of GPUs

The workforce component includes NVIDIA training and certification programs, hackathons and technical seminars. NVIDIA ties those activities to stronger GPU utilization: the goal is to help participants make better use of the computing hardware. This part of the collaboration addresses operating skills, rather than promising additional chips or data-center capacity through the training programs themselves.