South Korea Plans $3.5 Billion Frontier AI Program to Start in March 2027

South Korea plans to launch a 4.7 trillion won, or roughly $3.5 billion, program in March 2027 to build a homegrown frontier AI model, positioning the effort as a national-scale bid to compete with the most advanced systems from abroad. The Ministry of Science and ICT plans to choose a lead developer through a competitive tender after parliament approves the 2027 budget in December, with a winner possible as early as February 2027, according to Reuters. The funding is part of a proposed 9.4 trillion won AI budget request for 2027, Yonhap News Agency reported.\n\nThe plan combines state equity investment with private funding and would concentrate computing chips, data and talent on a single flagship project. Participation is expected to be open to single companies, consortiums or special-purpose vehicles, with startups, companies, universities and researchers allowed to take leading roles, according to Reuters. A ministry official told Yonhap News Agency that exact project details had not yet been decided.\n\nThe initiative is separate from South Korea's existing homegrown foundation model program, which will continue and select two final teams in a third-stage review in February 2027, according to Reuters. The Korea Times reported that the current national AI foundation model project will shift toward industry-specific models and real-world applications, while the new frontier project aims at a higher-performance model capable of competing with leading Chinese open-source models The Korea Times.\n\nOfficials have framed the effort as a sovereignty issue as global AI development becomes more capital-intensive. Science Minister Bae Kyung-hoon has stressed the need for a sovereign frontier model to avoid dependence on other countries amid intensifying competition between the United States and China, Yonhap News Agency reported. The investment scale also reflects a broader industry concern: recent research on training economics estimated that the largest AI training runs could cost more than $1 billion by 2027 if cost trends continue, limiting frontier development to the best-funded organizations arXiv.\n\nDevelopers and domestic AI companies are expected to view the plan as a major chance to secure compute, capital and national backing, especially because the government is considering equity investment rather than ordinary subsidy support. At the same time, the plan leaves unresolved questions about governance, model ownership, private-sector participation and how public returns would be shared if a special-purpose company is used. The tight sequence also adds execution risk: budget approval is expected in December, the lead developer could be selected in February, and the project is slated to begin in March 2027.