South Korea plans to launch a 4.7 trillion won ($3.5 billion) program to develop a frontier AI model starting in March 2027, Reuters reported, citing the Ministry of Science and ICT, as Seoul tries to compete in the race to build cutting-edge AI.
The ministry plans to select a lead developer through a competitive tender after parliament approves the 2027 budget in December, with a winner potentially chosen as early as February, according to Reuters. The government intends to combine state equity investment with private funding and concentrate computing chips, data and talent on the project, which aims to produce a model with leading performance.
Participation will be open to single companies, consortiums or special-purpose vehicles, the ministry said, and it wants startups, firms, universities and researchers to be able to play leading roles.
The Korea Times reported that the money is expected to include 3.9 trillion won for 10,000 Nvidia Vera Rubin GPUs and 800 billion won for training data, and that participating companies will have to match the government’s investment with private funding. The ministry plans to publish an outline of application criteria later this month, the paper said. The budget proposal still needs National Assembly approval, the Korea Herald noted.
Second Vice Minister of Science and ICT Ryu Je-myung told reporters that competing head-on with the largest US AI companies would be unrealistic, but that Korea could aim for a model competitive with leading Chinese open models if the planned public and private investment materializes, the Korea Herald reported. “Investment on the scale of 1,000 GPUs is not enough to catch up with frontier-class models,” Ryu said. “The scale has to be completely different.”
The new program sits alongside the country’s existing home-grown foundation model competition, which will continue and select two final teams in a third-stage review in February 2027, Reuters reported. Teams led by LG AI Research, SK Telecom and Upstage remain in that contest, according to the Korea Herald. “The goal of securing our own foundation model technology has not changed, and it will not be discontinued,” Ryu said.
The plan marks a shift from spreading compute across rival teams to a single pooled bet. Ryu said at the briefing that the government had weighed whether to keep giving companies more GPUs to compete “or whether we need to pool all available resources and concentrate them to close the gap with frontier models,” according to the Korea Times.