Naver says it is partnering with Nvidia to expand its artificial intelligence infrastructure and advance its AI model efforts. The company plans to start with an expansion of its Gak Sejong data center in Sejong, central South Korea, using Nvidia’s DSX platform. Naver expects initial capacity of 55 megawatts by the first half of 2027, with a phased buildout that increases capacity over time, including plans to scale internationally toward a gigawatt level. Naver describes a one-gigawatt target as roughly four times the maximum capacity of the current Gak Sejong site and one that could support hundreds of thousands of Nvidia GPUs.
Beyond infrastructure, Naver also plans to participate in Nvidia’s Nemotron Coalition, a collaboration among open model builders and AI developers, to support development of its proprietary large language model, HyperCLOVA X. The companies also plan collaboration on “physical AI,” with Naver developing a “Seoul World Model” that combines its street-view and spatial modeling technologies with Nvidia’s Cosmos platform.
