NAVER-NVIDIA Korea AI factory expands to 200MW; SK Group $500B partnership announced

NAVER, NVIDIA and Brookfield announced a major expansion of South Korea's sovereign AI factory infrastructure on July 24-25, scaling NAVER's NVIDIA DSX deployment at GAK Sejong data center from 55 megawatts to 200 megawatts—roughly 100,000 GPUs on the NVIDIA Vera Rubin platform. The $10 billion project is financed by Brookfield ($9 billion, nonbinding term sheet), NVIDIA ($1 billion direct investment), and NAVER (remaining). The expansion was announced at an AI Summit in San Francisco as South Korean President Jae Myung Lee laid out the nation's vision to become a pivot in the global AI supply chain. NAVER targets 1 gigawatt of NVIDIA AI infrastructure deployment over time.

In parallel, SK Group and NVIDIA announced a $500 billion-plus comprehensive partnership spanning AI infrastructure deployment, with SK Telecom building 2 gigawatts of Vera Rubin capacity, powered by SK hynix HBM4 memory. Hyundai Motor Group outlined a physical AI strategy with 50,000 NVIDIA Blackwell GPUs and integration of NVIDIA DRIVE Hyperion autonomous vehicle platforms. NVIDIA and KAIST (Korea Advanced Institute of Science and Technology) launched the first joint AI research lab between a Korean university and global tech company, focused on agentic AI and physical AI research.

For practitioners, Korea's announced buildout signals three infrastructure shifts: (1) sovereign AI factories now exist in Asia at scale, (2) memory supply (SK hynix HBM4) is being locked in via equity partnerships, and (3) physical AI and robotics are co-investments alongside foundation-model training capacity. The scale— ~2 GW from SK Telecom alone, 200 MW from NAVER— indicates production-grade demand for Vera Rubin and expected AI inference volume that will reshape global compute capacity distribution and margins.