NAVER, NVIDIA, and Brookfield partnership commits to tripling Korea AI data center capacity to 200 MW by 2028, marking concrete expansion timeline
NAVER, NVIDIA, and Brookfield have unveiled an ambitious plan to expand Korea's sovereign AI factory infrastructure, boosting capacity at the GAK Sejong data center from 55 megawatts (MW) to 200 MW by 2028. Announced on July 25 during Korea's AI Summit in San Francisco, this project underscores the nation's drive to position itself as a global AI powerhouse.
The expansion builds on NAVER's June 2026 commitment to develop gigawatt-scale AI infrastructure using NVIDIA's DSX platform. The initial 55 MW deployment, slated to go live in 2027, will scale to 200 MW within a year of operation, with long-term plans to reach 1 gigawatt (GW).
To fund this growth, NVIDIA is committing $1 billion, while Brookfield plans to provide up to $9 billion under a nonbinding term sheet. NAVER will secure the remaining financing. This collaboration integrates NAVER's expertise in hyperscale data centers, NVIDIA's full-stack AI platform, and Brookfield's investment capabilities to create scalable, multi-tenant AI cloud infrastructure capable of serving enterprises, industries, and government organizations.
The expanded AI factory will provide critical compute power for developing advanced AI models and applications, including NAVER's proprietary HyperCLOVA X foundation models. NAVER aims to enhance its AI services with NVIDIA's cutting-edge platforms, such as the NVIDIA Vera Rubin and Blackwell architectures, which optimize token throughput while lowering costs. This infrastructure will also support open model initiatives like NVIDIA's Nemotron 3 Ultra.
"AI factories are the infrastructure nations need to compete in the intelligence era," said NVIDIA CEO Jensen Huang. "This partnership will drive Korea's AI competitiveness by providing the scale and sovereignty necessary to fuel innovation."
NAVER plans to launch an AI agent platform in late 2026, integrating NVIDIA's Agent Toolkit software. The company is also developing a Seoul World Model using NVIDIA Cosmos technology, which could have implications for urban planning and autonomous systems.
NAVER's vision extends beyond domestic infrastructure to include a global AI footprint. Earlier this year, NAVER Cloud CEO Kim Yu-won outlined plans for a "Global AI Factory" spanning Asia, the Middle East, and Europe. By leveraging partnerships with NVIDIA and Brookfield, NAVER aims to evolve into a leading operator of global AI infrastructure, aligning with Korea's broader digital transformation strategy.
Since January 2026, NAVER has operated Korea's largest AI cluster, featuring 4,000 NVIDIA B200 GPUs, which have cut training times for large-scale models by over 90%. The expanded GAK Sejong facility will build on this foundation, offering production-scale AI compute to both domestic and U.S.-based innovators.
With AI adoption accelerating across industries, sovereign infrastructure projects like NAVER's are increasingly viewed as strategic assets. For Korea, this expansion represents an investment in its technological future and a bid to reduce dependence on foreign AI infrastructure. The collaboration positions Brookfield as a key player in the global AI infrastructure sector, building on its $100 billion portfolio of data centers, semiconductors, and power generation assets.
As the infrastructure comes online starting in 2027, it will likely unlock new opportunities for AI startups and established players alike, both in Korea and abroad. This could have ripple effects across industries, from autonomous vehicles to advanced healthcare solutions, as businesses leverage the expanded compute capacity to build next-generation AI applications.