SK Group, Nvidia ink $500bn AI infrastructure deal

The SK Group–Nvidia partnership reflects a strategic shift in how South Korea approaches AI competitiveness. Rather than positioning itself merely as an adopter of AI technology, SK's chairman explicitly framed the country's role as a producer of "intelligence"—a signal that domestic AI infrastructure capacity is now viewed as critical to national economic advantage. The timing follows Nvidia's multi-year agreements announced last month with SK hynix, SK Telecom, Doosan Group, Naver, and LG Group, indicating a broader Nvidia commitment to South Korean manufacturing and deployment.

The deal is structured around two complementary moves: SK Telecom's cloud-infrastructure build (2GW by 2027) provides the computational backbone, while SK hynix's long-term partnership on AI memory supply and co-development addresses a bottleneck in large-scale training. Nvidia's DSX full-stack architecture—which bundles accelerated computing, integrated systems, and software—suggests the companies intend to optimize both energy efficiency and cost at scale, a necessity as demand for AI compute grows, especially across Asia-Pacific markets.

The parallel expansion involving Naver signals that South Korea is attracting multiple infrastructure investments concurrently. Naver's target of 1GW total deployment, combined with the 200MW Naver–Nvidia–Brookfield expansion (more than triple the prior 55MW plan), indicates accelerating capital commitments to the region. Brookfield's $9bn nonbinding term sheet underscores the scale of financing now flowing into AI infrastructure.