The government has accelerated its "AI for All" project, which aims to provide free AI services to all citizens by the end of the year, but the domestic AI industry has expressed deep concerns about practical challenges despite opportunities such as Nvidia GPU support and large-scale user acquisition. These challenges include an ultra-tight development schedule, uncertain operating costs, and evaluation criteria that could be biased toward large platforms.

The Ministry of Science and ICT and the National IT Industry Promotion Agency (NIPA) announced today (22nd) the full launch of the "AI for All" project. The government has set ambitious goals to reduce dependence on foreign AI models, bridge the AI usage gap, and provide all citizens with free access to general-purpose AI chatbots and public AI agents based on domestic AI models. Furthermore, the government plans to establish a "one AI agent per citizen" system after 2027 to achieve the democratization of AI services.

The project timeline is extremely tight. The government plans to complete the selection of contractors by August, begin beta service in September, and launch officially by December. To this end, it will distribute up to 512 Nvidia GPU B200 units to 2-3 companies this year. Additionally, the government announced plans to support service operation costs with government budgets starting in 2027. Park Jin-ho, head of the AI Policy Planning Division at the Ministry of Science and ICT, led the project briefing, demonstrating the government's strong commitment.

In response to the government's policy direction, approximately 10 major domestic AI companies are expected to apply for the project. Large platform companies like Naver and Kakao, three mobile carriers, and promising AI startups are weighing their participation. The industry expects this project to secure valuable GPU resources, capture a massive user base nationwide, and build references needed for future government projects.

[Photo=Yonhapnews]

However, there are growing concerns about practical obstacles lurking behind the project. The first issue raised is the ultra-tight development schedule for "official launch by year-end." Concerns have been expressed that it is realistically difficult to establish and refine a stable service within such a short timeframe. Additionally, despite the planned government budget support for operating costs from 2027 onward, the uncertain scale and method of such support is fueling corporate anxiety.

In particular, there is strong dissatisfaction with the plan to use "number of users" as a key evaluation metric. The startup sector views this evaluation standard as being overwhelmingly favorable to large platform companies like Naver and Kakao, which already have substantial user bases. This could contradict the project's purpose of "securing diversity in the domestic AI ecosystem" and could result in an "unfair game" for startups. Consequently, some startups are reportedly considering participation in consortium formats rather than independent participation.

The government explained that it formulated the project plan by gathering industry opinions and will announce specific support measures once the budget is finalized. However, companies point out that relying solely on government support makes it difficult to ensure long-term "sustainability" of the project. There is concern that without developing their own innovative "revenue models," the project may become merely a short-term policy.

For the government's "AI for All" project to avoid being a temporary policy and instead foster a sustainable national AI ecosystem while achieving true AI democratization, it is urgent to realistically adjust the tight development schedule, provide substantial and specific operating cost support, establish fair evaluation criteria that can accommodate both large enterprises and startups, and provide support that encourages companies to independently secure revenue models.