Backlash after 'Team Korea AI' loses in first round: "The risks of trying again are too high"

[The Financial News] After NAVER Cloud and NC AI were eliminated in the first-stage evaluation of the Proprietary AI Foundation Model Project, the government’s so-called “second-chance round” is facing headwinds as major companies have indicated they will not participate. In the industry, many believe the practical benefits of trying again are limited compared with the risks, raising doubts over whether the project itself is losing momentum.
According to the information technology sector on the 18th, NAVER Cloud, Kakao, and NC AI have all stated that they are “not considering” joining the government’s renewed recruitment of elite teams for the Proprietary AI Foundation Model Project. On the 15th, the Ministry of Science and ICT changed its original plan to eliminate only one team in the first-stage evaluation and instead dropped NAVER Cloud and NC AI. At the same time, it announced a plan to hold an additional call for one more team within the first half of this year, and to proceed with the second-stage evaluation with a total of four elite teams. This would give another chance not only to NAVER Cloud and NC AI, but also to Kakao, KT, Motif Technologies, Konan Technology, and the Korea Advanced Institute of Science and Technology (KAIST) consortium, which had failed to make the cut when five elite teams were initially selected. Other capable and willing companies would also be allowed to compete in the new elite-team selection.
However, the industry sees the likelihood of companies making another attempt as low. The project’s reputation has already been dented by the controversy over “originality” that led to the elimination of the NAVER Cloud consortium in the first-stage evaluation, and companies would face significant disadvantages competing against LG AI Research, SK Telecom, and Upstage, which have already advanced to the next stage. Participation in the project inevitably diverts corporate talent and resources, and given the rapidly changing AI landscape and intense competition, many companies believe it is more advantageous to focus on advancing their own technologies.
An AI industry insider said, “If the government suddenly says it will select one more company, there is bound to be pressure about having to compete against firms that have been working on the project since last year,” adding, “For companies that have already been eliminated, the incentives are weak, and because it requires substantial resources, it will be difficult for academia or startups to readily declare their intention to participate.” Another industry official commented, “Companies that were already dropped from the elite-team selection have now absorbed the shock and set their own business directions, so this ‘second-chance round’ can only come as a surprise,” and added, “Given the rapid global AI technology trends and fierce competition, companies are likely to judge that pouring energy into the project would be inefficient.”
There are also questions about the government’s evaluation criteria and about how useful the final models produced through the project will actually be. Critics point out that the criteria used in the initial elite-team recruitment and the first-stage evaluation have not been consistent, and that even if one or two final models are developed, it remains unclear how much they will contribute to strengthening the nation’s industrial competitiveness.
Still, some argue that there are incentives for small and mid-sized companies, given the strong government support on offer, including access to high-performance Graphics Processing Units (GPU). Developing AI models requires training based on GPUs, such as through reinforcement learning. For AI startups and small or mid-sized firms, being selected as an elite team—even if they are later eliminated—could still provide a valuable opportunity to receive GPU support for a certain period. One industry official noted, “Opportunities to use high-performance GPUs with government support are extremely rare in our country,” and added, “Any company with demand for Large Language Model (LLM) development could very well decide to participate.” The government has stated that if it fails to select an additional elite team, it will redistribute the remaining resources to the three companies that have already advanced to the second stage.
wongood@fnnews.com Joo Won-gyu Reporter