Major economies around the world are waging a battle for dominance in artificial intelligence, with their national futures at stake. The contest has moved beyond the technological competitiveness of individual companies into a struggle between nations. In the United States, private companies pour in enormous capital to lead the AI industry, while China stakes everything on state-level support for its firms in response to U.S. controls on semiconductor and AI exports. Korea's government has also designated AI data centers as a national strategic industry, following semiconductors, and unveiled a blueprint for becoming one of the world's top three AI powers, calling for more than 1,000 trillion won ($720 billion) in investment through 2035.
In an interview with the Seoul Economic Daily on the 17th, Kim Kee-eung, the inaugural director of the National AI Research Hub, said, "The AI industry requires setting and executing budgets and investment plans with an eye on the nation's future over more than a decade, not one or two years." Kim added, "To build an AI ecosystem, nothing is more important than creating 'patient capital' of a long-term investment nature."
Kim, a professor at the Kim Jaechul Graduate School of AI at the Korea Advanced Institute of Science and Technology (KAIST), stressed, "From 2022 to 2025, Korea was a net exporter of AI talent, with people leaving for abroad." Kim said, "We need to offer extraordinary terms and lower the barriers to attracting overseas talent, so that we can create an environment where top-tier talent will return."
What is the role of the National AI Research Hub?
It is an organization launched in October 2024 with support from the Ministry of Science and ICT, an AI research consortium spanning academia, industry, research institutes and government. KAIST, Korea University, Yonsei University and POSTECH take part, together with 45 university faculty members, some 200 student researchers, 12 domestic partner companies and 14 overseas joint research institutions. Leading AI nations have government-level research anchors — the U.S. National AI Research Institutes, the United Kingdom's Alan Turing Institute and Canada's Vector Institute — and the hub was organized in line with this global trend. It can be described as a leading AI research hub that discovers and nurtures emerging AI researchers.
Why is the U.S. AI industry so strong?
In the United States, the government sets the stage through regulatory innovation while AI companies take the lead, creating "frontier models." Companies such as OpenAI, Google and Anthropic develop frontier models, and Big Tech invests hundreds of billions of dollars a year to build ecosystems, including data centers. The strengths also include a robust ecosystem of universities and research institutes that draw the world's best talent, and the enormous pool of investment funds that serves as "patient capital." The government's role is turning aggressive as well. In July last year, it drew up an "AI Action Plan" to ease regulation, expand infrastructure and support full-stack AI exports, and in November it launched the "Genesis Mission," presenting a blueprint to double the productivity of scientific research within a decade. This can be seen as a national grand plan comparable to the Manhattan Project, which developed the atomic bomb, and the Apollo program, which explored the moon.
Do you see it as a game changer for the industry landscape?
Yes. That is precisely why the United States is blocking computing access altogether through export controls on advanced AI chips to China. What is especially noteworthy is that the United States has defined AI not as a mere industrial technology but as an infrastructure of national competitiveness that runs through science, energy and defense, and is pouring astronomical sums into it. Rather than stopping at technological competition among individual models, it is expanding the field into a fight for control over data, computing, research infrastructure and the like.
China's "AI rise" is remarkable.
Its pace of growth is frighteningly fast. In its "Next-Generation AI Development Plan" in 2017, it set a goal of leading the world by 2030, and in August last year it drew up an "AI Plus Action Plan," moving past the stage of securing technology to integrate and apply AI across six core industries, including manufacturing and healthcare. The most striking part is its "open model strategy." After DeepSeek, Chinese companies such as Alibaba's Qwen and Moonshot AI's Kimi have adopted the release of the weights of high-performance models as a long-term competitive strategy. The performance gap with America's closed, top-tier models is narrowing steadily, and in the open-source ecosystem the presence of Chinese models is instead growing. Ultimately the aim is to make Chinese models the global standard, and for us that is a major threat.
What lies behind the strength of China's AI industry?
The internalization of technology stands out. With Huawei designing its own AI chip, "Ascend," companies are pushing full-stack self-reliance from domestic chips to models and services. A head start in expanding power infrastructure and the world's largest pool of AI research personnel are also advantages. If the United States concentrates on "closed frontier plus computing superiority," China is fighting back with a strategy of "open diffusion plus full-stack self-reliance." According to the Stanford AI Index, as of March this year the performance gap between U.S. and Chinese models was just 2.7 percentage points. China already leads the world in the number of papers and patents and in the use of industrial robots.
Where does Korea's AI level stand?
Korea ranks around sixth in the global AI index. But if the United States is set at 100 points, China scores 53.9 and Korea 27.3, so the gap with the leaders is indeed large. Korea's strengths are clear. It has outstanding competitiveness in semiconductors such as high-bandwidth memory (HBM) and in advanced industries such as robots and automobiles, and its digital-related infrastructure is world-class. It also ranks first in AI patents per capita. But there is an urgent need to shore up weaknesses, including a fragile base for private investment, delays in demonstrating strategic industries and a lack of global platforms. In particular, an environment is desperately needed in which capital is channeled into startups and venture firms along a long-term road map, and where patient capital waits over a long period.
What are the ways to become one of the top three AI powers?
A strategy of selection and concentration is required. It is hard to go head-to-head with the United States and China on sheer scale in areas such as computing, capital and talent. Rather than aiming for third place in every field, we need a strategy that intensively fosters strengths — such as AI infrastructure combined with semiconductors, physical AI linked to manufacturing sites, and original technology — and offsets weaknesses through cooperation with global networks. Consistency and speed in policy are also important. Since AI is a national task, rather than drawing up one- or two-year budgets, long-term investment plans that look a decade ahead, like America's Genesis Mission, must follow.
Isn't securing talent the key?
Yes. Computing and data can be bought with money, but top-tier talent gathers only when an ecosystem is built. When there is fair reward for taking on challenges, a culture that tolerates failure and a world-class research environment, talent from home and abroad naturally comes. But Korea's net AI talent migration index has been in net outflow for four straight years: minus 0.03 in 2022, minus 0.44 in 2023, minus 0.15 in 2024 and minus 0.35 last year. AI patents and papers per capita rank among the highest in the world, yet the reality is that people are leaving. For top-tier talent, we must guarantee extraordinary funding and research activity, and sharply lower the barriers to attracting talent, such as visas and settlement conditions. We must also overhaul the return system so that talent that has gone abroad can come back to work, bringing the experience gained at Big Tech.
It is encouraging that, with the AI Framework Act taking effect, Korea has come to have a comprehensive AI legal framework that is, along with the European Union (EU), among the most advanced in the world. Promotional mechanisms have also been put in place, such as a new system to verify AI products and services and priority purchasing of AI products in public procurement. I think the direction of striking a balance between regulation and fostering is right. But there is no shortage of areas to supplement. First, the criteria for judging "high-impact AI" must be predictable. If companies cannot even determine whether their own AI business is subject to regulation, it is hard for them to undertake aggressive investment. The transparency of standards and procedures must be raised, and the issue of training data must be solved. Laws must be refined so that the use of copyrighted works and personal information needed to train AI can be carried out without legal uncertainty. This is also a precondition for building homegrown models. In addition, given that laws cannot keep pace with the speed of AI technology, presidential decrees and guidelines
The AI Framework Act took effect in January this year.
