NVIDIA, Naver, and Brookfield forge $10 billion AI infrastructure partnership to expand South Korea's sovereign computing capacity

NVIDIA, Naver, and Brookfield have announced a planned $10 billion investment to expand South Korea's sovereign AI factory infrastructure to 200MW by 2028, with a long-term goal of reaching 1GW. The partnership aims to bolster national AI competitiveness and provide scalable compute for innovators in both Korea and the US.

In a significant move to bolster national artificial intelligence capabilities, NVIDIA Corporation, Naver Corp., and Brookfield Asset Management have unveiled a strategic partnership to substantially expand South Korea's sovereign AI infrastructure. The collaboration, announced during South Korean President Jae Myung Lee's AI Summit visit to San Francisco, outlines a planned investment of up to $10 billion to scale the existing NVIDIA DSX AI factory at Naver's GAK Sejong hyperscale data center.

The proposed expansion will increase the facility's capacity from the 55 megawatts announced in June 2026 to 200 megawatts by 2028, effectively more than tripling its initial size. Beyond this immediate target, the consortium has articulated a long-term vision to scale the deployment of NVIDIA AI infrastructure to 1 gigawatt, positioning the facility as a cornerstone of Korea's sovereign AI ambitions. The GAK Sejong data center, located in Sejong, South Korea, will serve as the physical hub for this expanded infrastructure, which is designed to provide production-scale AI compute for developing next-generation models, agents, and AI-powered services.

According to official disclosures, NVIDIA plans to invest $1 billion into Naver Corp. as part of the agreement. Concurrently, Brookfield has entered into a nonbinding term sheet to fund up to $9 billion, with Naver covering the remaining project costs. The completion of NVIDIA's investment is contingent upon standard closing conditions and Naver finalizing at least $9 billion in committed financing from other sources. This financial structuring underscores the scale and complexity of the initiative, leveraging the combined strengths of a technology leader, a local internet giant, and a global infrastructure investor.

The partnership represents a convergence of specialized expertise. Brookfield brings its substantial AI infrastructure investment platform, which manages approximately $100 billion in assets across the value chain, including data centers and power generation. Naver contributes its full-stack AI and data center operating experience, gained from running Korea's largest internet platform. NVIDIA provides its accelerated computing platform and the NVIDIA DSX platform, an end-to-end, co-designed stack for AI factories that includes chips, systems, and software such as DSX MaxLPS for optimizing token throughput and DSX OS for lifecycle management.

This development is poised to have a profound impact on multiple stakeholders. For South Korea, it represents a strategic leap forward in securing sovereign AI capabilities, reducing reliance on foreign infrastructure and fostering a domestic AI ecosystem. The expanded compute capacity is expected to fuel innovation among Korean startups, enterprises, and government organizations, enabling them to develop and deploy competitive AI models at scale. For the broader Asia-Pacific region, the project establishes a significant, trusted AI hub that could attract international AI cloud customers and serve as a model for other nations seeking to build their own AI infrastructure.

The implications for global businesses and investors are multifaceted. The partnership highlights the growing strategic importance of sovereign AI infrastructure as a competitive differentiator for nations. It signals a significant capital deployment into physical AI assets, creating opportunities and benchmarks for similar large-scale projects worldwide. Furthermore, the collaboration between a leading chipmaker and a major investment firm validates the economic viability and long-term demand for dedicated AI compute, potentially influencing investment strategies in the technology and infrastructure sectors. The use of NVIDIA's Vera Rubin and Blackwell platforms within the factory also points to the continuous demand for cutting-edge hardware to support increasingly complex AI workloads.

The partnership builds on a pre-existing relationship between Naver and NVIDIA, which includes collaboration on open model development. Naver is advancing its HyperCLOVA X models based on NVIDIA Nemotron 3 Ultra open models and is the first Korean company to join the NVIDIA Nemotron Coalition. Furthermore, Naver plans to launch an AI agent platform in Korea later this year, powered by NVIDIA's Agent Toolkit, and is developing a Seoul World Model using its proprietary urban data and NVIDIA Cosmos world foundation models. These initiatives suggest a deep integration that extends beyond pure infrastructure provision into the application layer of AI development.