Naver is joining forces with NVIDIA to construct a colossal global artificial intelligence (AI) factory with gigawatt (GW) capacity.
This partnership is built on NVIDIA’s next-generation infrastructure platform, DSX. It’s more than just a tech alliance; it’s a comprehensive collaboration where both companies will jointly design and build the entire AI factory infrastructure, from identifying global demand to capital cooperation.
Notably, Naver is breaking new ground as the first South Korean company to build an AI factory of this magnitude, sharing both profits and risks with NVIDIA as a key global partner.
Naver disclosed these business plans on Monday. Lee Hae-jin, Naver’s Board Chairman, and Jensen Huang, NVIDIA’s Chief Executive Officer (CEO), met that afternoon at Naver’s headquarters in Bundang, Seongnam, to hash out the specifics of their joint ventures and strategies for global market expansion.
The two tech giants have agreed to establish leadership in the AI infrastructure ecosystem, setting their sights beyond Asia-Pacific to include European and Middle Eastern markets.
Naver will use its hyper-scale data center, Gak Sejong, as its primary hub. The company plans to kick off operations with 55 megawatts (MW) in the first half of 2027, ramping up to 100 MW later that year, and reaching 200 MW by 2028 to meet global demand.
This initiative is part of a staged roadmap aimed at ultimately creating GW-level infrastructure. To put it in perspective, one gigawatt is about four times the maximum capacity of Naver’s Gak Sejong, capable of housing hundreds of thousands of NVIDIA’s cutting-edge GPUs simultaneously.
The partnership will also intensify technical collaboration across the board. Naver will integrate its unique expertise in large-scale graphic processing unit (GPU) cluster operations and hyper-scale data center management with NVIDIA’s next-gen high-performance infrastructure platform, DSX. This fusion aims to boost data center efficiency and significantly enhance business viability.
In the realm of spatial intelligence, the collaboration will deepen with the creation of a Seoul World Model. This project will leverage Naver’s proprietary spatial modeling and street view data, integrating it with NVIDIA’s world foundation model, Cosmos.
Recently, Naver made history as the first South Korean company to join the NVIDIA NEMOTRON Alliance, a coalition of 12 global AI powerhouses including Cursor, MistralAI, and Perplexity. NEMOTRON is NVIDIA’s open-source AI foundation model.
Building on this, Naver is fast-tracking the enhancement of its own large language model (LLM), HyperCLOVA X. The company is combining NEMOTRON’s collaborative tech developments with its own data and accumulated learning expertise to boost performance and ensure global applicability.
Lee Hae-jin, Naver’s Board Chairman, expressed enthusiasm about the alliance, stating that this partnership enables us to offer tangible solutions for countries worldwide to develop their own sovereign AI capabilities. It’s a significant milestone, potentially catapulting Naver’s technological infrastructure to the global stage.