Nvidia and South Korea’s SK Group have teamed up to invest over $500 billion in a massive push to build AI-focused data centers and cutting-edge memory technology. The announcement on July 24 reveals one of the largest infrastructure commitments ever seen in the AI sector, signaling a major shift in how semiconductor supply chains might evolve.
The centerpiece of this collaboration involves SK hynix, the memory chip division within SK Group, which will develop high-bandwidth memory (HBM) tailored specifically for Nvidia’s AI platforms. Meanwhile, SK Telecom plans a 2-gigawatt AI data center that will utilize Nvidia’s Vera Rubin chips and SK hynix’s upcoming HBM4 memory, with operations expected to start as early as 2027.
Building a Vertically Integrated AI Hub in South Korea
South Korea already houses two of the world’s top memory manufacturers, Samsung and SK hynix. Adding vast AI data center capacity to this mix creates a uniquely integrated AI ecosystem that few other countries can rival. This project builds upon previous deals made this year that laid the foundation for extensive AI hardware expansion across the region.
Besides SK Group’s efforts, Nvidia is collaborating with Naver, South Korea’s leading search engine and tech platform, along with Canadian infrastructure firm Brookfield, to boost an existing AI data center in the country. This demonstrates a broad strategy to solidify South Korea’s role as a global AI compute hub.
For Nvidia, such partnerships in South Korea complement earlier agreements in Japan, the UAE, and Saudi Arabia, as the company expands its global network of AI factories tied closely to its chips and software ecosystem. This long-term play aims to secure Nvidia’s dominance in both AI hardware and the supply chains supporting fast-growing AI applications.
