South Korea's NAVER and NVIDIA on June 7 announced a sweeping partnership to expand NAVER's sovereign AI infrastructure on the NVIDIA DSX platform, starting at 55 megawatts at the GAK Sejong data center and scaling toward gigawatt-class AI factories serving enterprises, industries and government.
The Build-Out
The first wave of NVIDIA DSX-powered capacity will sit inside NAVER's hyperscale GAK Sejong campus in Sejong, South Korea. NVIDIA DSX is a codesigned stack spanning chips, systems, software, facilities and partner technologies, paired with DSX MaxLPS software to maximize token throughput per megawatt and DSX OS to orchestrate multi-tenant AI factory operations. NAVER plans to pair the platform with its own large-scale GPU operations expertise to drive down token cost as it ramps toward gigawatt scale.
What Runs On It
The new infrastructure will fuel NAVER's HyperCLOVA X family of models, which the company is fine-tuning on top of NVIDIA's Nemotron 3 Ultra open model to deliver Korean and global enterprise customers what NAVER calls a more culturally fluent foundation model. NAVER joined the NVIDIA Nemotron Coalition and will launch an AI Agent Platform in Korea in the second half of 2026 powered by NVIDIA NemoClaw blueprints, alongside a Seoul World Model built on NVIDIA Cosmos world foundation models using NAVER's proprietary urban street-view data.
Why It Matters
The deal stands out as one of the most concrete sovereign AI infrastructure commitments in Asia, with NAVER positioning itself as a Korean alternative to U.S. hyperscalers for government and regulated enterprise workloads. "Useful AI has arrived, and demand for AI factories is extraordinary," NVIDIA CEO Jensen Huang said, while NAVER founder and chairman Haejin Lee framed the partnership as a way to help customers "move from AI experimentation to production-scale AI factories." It joins NVIDIA's gathering Korean push that already includes recent agreements with SK hynix on memory and Hyundai on robotics, as well as LG Group's AI factory for robotics and mobility.
What Comes Next
NAVER said the new DSX-powered cloud will also serve customers in Europe and the Middle East, where it is competing to host sovereign AI workloads under local data-residency rules. The company has not disclosed a final gigawatt target or capex figure, but reaching gigawatt scale would put GAK Sejong in the same conversation as the largest AI factories now being commissioned by U.S. hyperscalers and frontier model labs.
Reporting based on coverage from NVIDIA, GlobeNewswire and Engineering.com.