Nvidia and Naver are expanding their artificial intelligence infrastructure partnership and sovereign AI initiative, using the Korean internet giant’s existing data center in Sejong as the starting point to scale AI cloud capacity toward gigawatt-level AI factories.
Naver said Monday the two sides agreed to expand the capacity of its data center Gak Sejong, starting with 55 megawatts in the first half of 2027, before scaling overseas infrastructure to 100 MW later that year and 200 MW by 2028 to meet growing global demand. The phased rollout is part of a longer-term roadmap to build gigawatt-scale AI infrastructure.
One gigawatt is roughly four times the maximum capacity of Naver’s Gak Sejong data center and is large enough to accommodate hundreds of thousands of Nvidia’s latest GPUs.
Nvidia CEO Jensen Huang and Naver founder Lee Hae-jin were set to finalize their agreement at the Korean IT giant's second headquarters, Naver 1784 in Seongnam, later that day. The headquarters is considered a showcase for the company’s future technologies, including robotics, cloud computing, digital twins and private 5G networks.
"We are very encouraged that this alliance will enable us to present a concrete option for regions and countries around the world to build independent sovereign AI capabilities,” Lee said.
“The partnership is also meaningful in that it provides an opportunity for the competitiveness of Naver’s technological infrastructure to make a further leap into the global market.”
In a media prebriefing earlier Monday, Nvidia said it will work with Naver across its full AI stack, from DSX-based AI factories to Nemotron-powered sovereign AI models and Cosmos-enabled physical AI, as the US chip giant seeks to deepen its role in Korea’s AI infrastructure ecosystem.
“Korea’s ecosystem is very important for us. We want to make sure that we build our platform and support our community of AI startups and developers working with Naver," said Raj Mirpuri, vice president of global AI clouds and infrastructure ecosystem at Nvidia.
At the center of the gigawatt-scale expansion is Nvidia DSX, the company's next-generation high-performance infrastructure platform designed to help cloud providers deploy AI factories faster, maximize performance per watt and reduce token production costs.
The two companies will also expand cooperation in spatial intelligence. Naver plans to use its spatial modeling capabilities and street-view data with Nvidia Cosmos, the US chip giant’s world foundation model platform, to develop a Seoul World Model, the company said.
Naver also recently became the first Korean company to join Nvidia’s Nemotron Coalition, contributing to open model development across pretraining, post-training and reinforcement learning, according to Mirpuri. The group includes 12 global AI firms such as Cursor, Mistral AI and Perplexity.
Building on the coalition’s joint technology development, Naver said it plans to combine Nemotron’s advances with its proprietary data and accumulated training expertise to accelerate upgrades to HyperCLOVA X and improve its global applicability.
herim@heraldcorp.com
