NVIDIA and NAVER plan gigawatt-scale sovereign AI factory in South Korea

NVIDIA and NAVER said on June 7, 2026 that the South Korean internet company will build out sovereign AI infrastructure starting at 55 megawatts and scaling toward gigawatt capacity, running on the NVIDIA DSX platform at NAVER's GAK Sejong hyperscale data center in Sejong, South Korea. The project is part of a wider push by governments and national champions to keep frontier AI compute and data within their own borders.

NAVER will use NVIDIA's full-stack platforms, models and software to advance its HyperCLOVA X models, which it fine-tunes from NVIDIA Nemotron 3 Ultra, and its Seoul World Model, built with NVIDIA Cosmos world foundation models, NVIDIA said. The company is also the first Korean firm to join the NVIDIA Nemotron Coalition, a group of developers building on NVIDIA's open model family.

NAVER plans to launch an AI Agent Platform in Korea in the second half of 2026, powered by NVIDIA NemoClaw blueprints. "Useful AI has arrived, and demand for AI factories is extraordinary," NVIDIA chief executive Jensen Huang said. NAVER founder and chairman Haejin Lee said the company is building sovereign AI infrastructure to serve Korea's industries and global customers.

The announcement frames the buildout in capacity, not cost: NVIDIA and NAVER did not disclose the investment behind the project, the timeline for reaching gigawatt scale, or how much capacity the initial 55 megawatts represents in delivered systems.

The deal lands the same week NVIDIA detailed a separate memory partnership with SK hynix, underscoring how much of NVIDIA's South Korea strategy now runs through national infrastructure rather than one-off chip sales.

For NAVER, the bet is that owning sovereign capacity — rather than renting it from foreign clouds — becomes a competitive edge as Korean firms and the public sector adopt AI.