The South Korean government plans to distribute approximately 10,000 advanced graphics processing units (GPUs) to industry, academia, and research institutions starting in February next year, to accelerate national AI innovation. This initiative was announced by the Ministry of Science and ICT at the 2nd Meeting of Science and Technology Ministers on the 18th.
The government aims to secure over 52,000 GPUs by 2028 as part of its ‘AI expressway’ initiative. The first batch consists of 13,000 GPUs, including 10,080 NVIDIA B200 and 3,056 H200 units. Of these, 8,160 B200 and 2,296 H200 GPUs will be used directly by the government, with the rest allocated to private cloud companies such as Naver Cloud, Kakao, and NHN Cloud.
Each participating company will complete GPU deployment through cooling system installation, large-scale clustering, optimization, and beta testing. A pre-beta test is currently underway for 512 B200 GPUs to verify procedures and identify potential user inconveniences before full deployment.
To ensure fair and efficient allocation, the government has established a distribution framework and will select recipients through inter-agency demand surveys, expert reviews, and deliberations at the Science and Technology Ministers’ Meeting. Additional procurements include the construction of Supercomputer 6 (9,000 GPUs) and further GPU purchases (15,000 units) next year, with the goal of launching the National AI Computing Center by 2028.