Nvidia has entered a massive partnership with South Korea's SK Group, valued at over $500 billion, to enhance AI infrastructure and lock down memory supply chains essential for AI computing.
At an AI conference in San Francisco, where South Korean President Lee Jae Myung was present, the companies revealed plans for SK Telecom to build a 2-gigawatt AI data center using Nvidia’s Vera Rubin hardware and SK Hynix’s latest HBM4 memory technology. This facility is expected to go live in 2027 and will serve global AI computing demands.
SK Hynix, a leading manufacturer of high-bandwidth memory, will collaborate closely with Nvidia to develop next-generation AI memory chips beyond HBM4, ensuring a steady supply amid worldwide shortages driven by AI's rapid growth. Nvidia’s enterprise VP Raj Mirpuri emphasized that this co-development will help stabilize their access to critical components.
Alongside this, Nvidia has committed $1 billion to South Korean cloud provider Naver to expand data center capabilities, highlighting the company’s broader push into the Asia-Pacific region’s AI market.
This deal contrasts with Samsung’s separate $200 billion memorandum with Broadcom focusing on memory chips and foundry services, underscoring intense competition among tech giants for AI supply chain dominance.
Shares of Nvidia closed slightly lower on the announcement day, but the strategic move positions Nvidia to meet growing AI hardware demands as the sector intensifies. CEO Jensen Huang has pointed to South Korea's rising role in the AI ecosystem as a key driver behind these investments.
