NVIDIA and South Korea’s KAIST (Korea Advanced Institute of Science and Technology) announced a joint AI research laboratory on July 23, based at KAIST’s Kim Jaechul Graduate School of AI in Seoul. The lab will focus on building agentic AI models and multi-step reasoning systems tailored to the Korean language and Korean industry, according to NVIDIA’s announcement.
A five-year, $300 million commitment
NVIDIA describes the partnership as a $300 million collaboration running an initial five years, including roughly $50 million a year in compute contributions. The companies say the funding will support at least 10 KAIST researchers annually, give them NVIDIA internships, and create full-time NVIDIA roles for the program’s top graduates.
The lab pairs NVIDIA’s Nemotron open models and its Cloud Partner computing infrastructure with KAIST’s academic research, aiming to build a pipeline from lab research to enterprise and national AI deployments. Hyunwoo Kim, an incoming KAIST faculty member who will head the lab, said AI research “is entering a new era — one that requires frontier talent, large-scale infrastructure and deep collaboration across academia and industry.” NVIDIA chief scientist Bill Dally called Korea “home to leading AI researchers and one of the world’s most advanced technology ecosystems.”
Part of a broader sovereign AI push
The Seoul lab extends NVIDIA’s campaign to help individual countries build their own AI capacity instead of depending entirely on foreign providers — the same argument Jensen Huang made in a July letter signed by dozens of companies backing open-weight models. It also lands weeks after Seoul unveiled its own $880 billion push into AI and semiconductors, underscoring how central government-backed AI infrastructure has become to South Korea’s industrial strategy. NVIDIA and KAIST frame the lab as a template that could be extended to other countries seeking to combine Nemotron and local compute into sovereign AI programs of their own.