Nvidia, SK Group Launch $500 Billion AI Infrastructure Plan
Nvidia and South Korea’s SK Group are pursuing an artificial-intelligence initiative valued at more than $500 billion, linking advanced processors, data-center capacity and high-bandwidth memory production. The plan is significant because competition in AI infrastructure increasingly depends not only on access to accelerators, but also on power availability and a reliable supply of HBM4, a critical component for training and running increasingly complex models.
The companies recently announced plans to build AI data centers in South Korea with 2 gigawatts of installed capacity, targeting operations in 2027. The facilities are expected to use Nvidia’s next-generation Vera Rubin chips alongside HBM4 memory from SK Group affiliate SK hynix. The partnership is designed to secure large-scale computing capacity while strengthening access to next-generation memory as demand from AI developers and cloud operators expands.
All Coverage
2 original reportsThe Backstory
The history behind this eventSK Group Partners With Nvidia on Gigawatt-Scale AI Data Center in Japan
South Korea’s SK Group is expanding its AI infrastructure footprint overseas with plans to build its first international “AI factory” in Japan. The facility will combine SK’s advanced memory with Nvidia GPUs and is being developed for gigawatt-scale computing with a low-power design, reflecting accelerating investment by Japanese and South Korean companies in computing capacity for generative AI.
SK Group plans to complete the next-generation AI data center in Japan between 2028 and 2029, with Nvidia contributing to the system design. The facility will provide high-performance computing capacity for AI training and inference. SK has not disclosed the exact location, investment amount or installed capacity, but has positioned the project as the group’s first overseas AI factory.
Subscribe to Mark Radar Weekly
Every Friday, the week's strongest signals in your inbox. Unsubscribe anytime.