3,000 GPUs to Support 52 National AI Projects
The government has selected 52 national AI projects and will provide 3,000 GPUs, essential high-performance computer components for AI learning, starting this month. The Ministry of Science and ICT (MSIT) announced on the 2nd that it held the 6th Ministerial Council for Science and Technology (in writing) to review and approve the 'Plan for Selecting National AI Projects for Nationwide AI Innovation'. National AI projects are a government-wide initiative aimed at fostering the national AI ecosystem and implementing national agenda items by supplying GPU resources in conjunction with government ministry projects.
This selection process involved receiving a total of 121 project proposals from 28 ministries, based on criteria such as strategic national importance, technological and social ripple effects, and the need for government-led execution. It was carried out after expert evaluations and preliminary reviews by a screening committee. Subsequently, through expert interviews for each project and a coordination committee, the final GPU allocation was determined.
Key projects include the Ministry of Trade, Industry and Energy's advancement of AI autonomous driving models for the transition to Software Defined Vehicles (SDVs), the MSIT's industry-specific foundation model projects, the Korea Meteorological Administration's development of a Korean-style AI weather and climate foundation model, the Ministry of Health and Welfare's AI fundamental healthcare initiatives, the Ministry of Culture, Sports and Tourism's revitalization of AI convergence content creation, and the Ministry of Oceans and Fisheries' development of integrated prediction technology based on Arctic route measurements.
Any unused GPU resources due to delayed project starts will be managed by accepting short-term demands from industry, academia, and research institutions through the National AI Computing Resource Support Portal. Deputy Prime Minister and MSIT Minister Bae Kyong-hoon emphasized, "We will strive to continuously identify AI innovation needs from both public and private sectors, and allocate the GPU resources secured by the government to where they are most needed."