Nvidia and the South Korean government said today that the country will deploy more than 260,000 of the company’s latest GPUs, split between sovereign AI infrastructure and private AI factories operated by the country’s largest industrial groups. The announcement came at the APEC summit.

The Ministry of Science and ICT is deploying up to 50,000 GPUs across the National AI Computing Center and sovereign clouds run by NHN Cloud, Kakao Corp. and NAVER Cloud, beginning with 13,000 Blackwell-class chips. Samsung Electronics is building a semiconductor AI factory with more than 50,000 GPUs, pairing them with Omniverse, cuLitho and Cosmos for digital twins, yield work and home robotics. SK Group is designing an AI factory of over 50,000 GPUs and what it calls Asia’s first industrial AI cloud, built on RTX PRO 6000 Blackwell Server Edition cards. Hyundai Motor Group gets 50,000 Blackwell GPUs for training and deploying models across manufacturing and autonomous driving, with roughly $3 billion attached. NAVER Cloud is expanding to more than 60,000 GPUs.

The two sides also framed the buildout as a platform play: Korean foundation models, AI-RAN and 6G work with Samsung, SK Telecom, KT, LG Uplus, ETRI and Yonsei, quantum-GPU research at KISTI, and a startup alliance under Nvidia Inception.