The South Korean government, led by the Ministry of Science and ICT (MSIT), is currently considering the abandonment of its "National Representative AI" project in favor of establishing a Special Purpose Company (SPC) to develop "Frontier AI" models. The "National Representative AI" initiative, which aimed to support domestic AI firms in developing proprietary foundation models, has faced significant criticism regarding its evaluation fairness. Specifically, the exclusion of startups—such as the case where Motif Technology was eliminated despite achieving the highest performance on the Artificial Analysis benchmark—has fueled concerns regarding the transparency of the evaluation metrics used to determine the top two remaining entities among finalists like SK Telecom, Upstage, and LG AI Research.

The proposed transition involves a shift toward a massive, capital-intensive public-private partnership model. The government intends to secure a 30% equity stake in the newly formed SPC, with a projected capital injection of 5 trillion KRW, while mandating that the remaining 70% be funded by private sector participants. This pivot raises substantial concerns regarding the marginalization of smaller AI startups. Given the massive financial threshold required for participation, industry analysts argue that the project will inevitably gravitate toward large conglomerates, effectively relegating startups to sub-contractor roles rather than primary developers of sovereign AI architectures.

The core methodology for the transition involves moving away from competitive, grant-based support for disparate model-building efforts toward a centralized development hub. This approach aims to unify resources to build a "Frontier AI," defined as the most advanced, general-purpose intelligence currently achievable. However, this strategic shift invites questions regarding the efficiency of public resource allocation. Given that over 200 billion KRW has already been spent on the existing evaluation process, the lack of a clear bridge or technological continuity between the defunct "National Representative AI" projects and the new SPC initiatives risks the total wastage of previous investments.

Technical and governance concerns persist, as stakeholders await the formal announcement from the MSIT. The central tension lies in whether this SPC model can effectively foster genuine innovation or if it will simply serve as a subsidy structure for dominant incumbents, potentially stifling the competitive diversity that the original project was intended to nurture. The success of this transition remains contingent on whether the government can establish a transparent governance structure for the SPC that avoids the pitfalls of the previous selection criteria, which were characterized by the opaque weighting of technical benchmarks versus subjective evaluation criteria.