- The GAK Sejong buildout would more than triple the 55-megawatt plan the companies floated in June
- NVIDIA intends to put $1 billion into NAVER, with Brookfield eyeing up to $9 billion under a nonbinding term sheet
- The added capacity is meant to give developers, businesses and government users production-scale compute for models and agents
NAVER, NVIDIA and Brookfield have proposed expanding a sovereign AI factory in South Korea to 200 megawatts by 2028. The buildout at NAVER’s GAK Sejong data center would more than triple the 55-megawatt deployment the first two companies announced in June.
NVIDIA said Friday it plans to invest $1 billion in NAVER. Brookfield has signed a nonbinding term sheet for up to $9 billion, with NAVER covering the balance. NVIDIA’s investment depends on customary closing conditions and on NAVER locking in at least $9 billion in committed financing.
What Will the Expanded South Korea AI Factory Support?
NVIDIA said the added capacity will give AI developers in South Korea and the United States production-scale compute for building models, agents and services. Enterprises, industry customers and government organizations are the intended users.
The 200-megawatt factory is expected to run on NVIDIA’s Vera Rubin and Blackwell platforms. NAVER, an NVIDIA Cloud Partner, said it intends to grow its NVIDIA infrastructure deployment to 1 gigawatt over time.
How Does the Expansion Build on a Previous NAVER-NVIDIA Agreement?
NAVER committed in June to extending GAK Sejong with the NVIDIA DSX platform, starting at 55 megawatts and pointing toward gigawatt-scale capacity. The DSX platform packages chips, systems, software and facility design into a single stack, with separate software layers handling throughput optimization and multi-tenant lifecycle management.
Friday’s announcement came during South Korean President Lee Jae-myung’s AI summit visit to San Francisco.
Brookfield counts NVIDIA as a founding partner in its Global AI Infrastructure Program. The firm has invested in South Korea since 2014.
What AI Models and Agents Is NAVER Developing With NVIDIA?
NAVER is building its HyperCLOVA X models on NVIDIA’s Nemotron 3 Ultra open models, applying its own data and training methods. It is the first Korean company to join the NVIDIA Nemotron Coalition.
An AI agent platform for the Korean market is slated for the second half of the year, running on NVIDIA Agent Toolkit software. NAVER is separately developing a Seoul World Model from proprietary street-view and spatial data, built on NVIDIA Cosmos world foundation models.
