NAVER and NVIDIA to Build Gigawatt-Scale Sovereign AI Factory in South Korea
In brief: NAVER will deploy a full-stack NVIDIA DSX AI factory at its GAK Sejong hyperscale data center — 55 MW by H1 2027, scaling to 100 MW, 200 MW, and eventually 1 GW — to build sovereign AI infrastructure for Korean government, enterprise, and cloud customers.
Key facts
- Deployment roadmap: 55 MW (H1 2027) → 100 MW (H2 2027) → 200 MW (2028) → 1 GW long-term
- Built on NVIDIA DSX, providing high-density accelerated compute with energy efficiency and disaster-resilience
- NAVER becomes the first Korean company in NVIDIA's Nemotron Coalition, contributing to open model pre-training, post-training, and RL
- Compute will power next-generation HyperCLOVA X, the Seoul World Model, and NAVER's agentic AI services
- Jensen Huang's Korea visit also produced AI factory agreements with SK hynix, SK Telecom, LG Group, and Hyundai Motor Group
Why it matters
South Korea is building one of the world's first nationally-scaled sovereign AI compute bases outside the US, EU, and China. For NAVER, this transforms GAK Sejong from a cloud provider into an AI factory capable of training frontier-class models domestically — reducing dependence on foreign compute and positioning HyperCLOVA X as a credible global multilingual model challenger.
Read more
- NVIDIA Newsroom — NVIDIA
- Korea Herald — Korea Herald