South Korea’s government unveiled a plan last week to invest more than 1 trillion won ($1.3 billion) this year to secure 10,000 graphic processing units (GPUs) as part of a national effort to strengthen artificial intelligence (AI) capabilities. But industry experts dismissed the move as a belated, symbolic gesture that does little to narrow the widening technology gap with the United States and China.
The plan, approved at a Cabinet meeting on April 18, allocates over 80% of a 1.8 trillion won supplementary AI budget to GPU purchases, according to government and industry sources. About 1.46 trillion won will be spent to secure 10,000 high-performance GPUs this year, which will be distributed to domestic firms, universities, and research institutes developing AI foundation models.
Other allocations include 193.6 billion won for Korea’s “World Best LLM (large language model) Project,” 75.2 billion won for AI semiconductor commercialization initiatives, and 55 billion won to establish an AI startup fund.
However, critics argue that the scale of investment falls far short of major players. In January, the U.S. government announced a $500 billion public-private partnership to build AI data centers over the next four years under its Stargate program. China has pledged 10 trillion yuan by 2030 to establish AI leadership.
“South Korea is chasing a market already dominated by the U.S. and China, and simply acquiring 10,000 GPUs won’t offer a meaningful solution,” said Kim Kyung-won, a professor of business at Sejong University. “This is a superficial, last-minute move with little practical effect.”
The lack of investment in cultivating AI talent is another point of concern. Just 45 billion won — or 2.4% of the supplementary budget — has been earmarked for workforce development. The Ministry of Science and ICT allocated 10 billion won to host global AI competitions, 5 billion won to attract foreign AI researchers to Korea, and 30 billion won to improve compensation and research opportunities for 400 postdoctoral researchers.
“Even with the hardware, it’s meaningless without skilled people to use it,” said Kim Yong-seok, head of Gachon University’s Semiconductor Education Institute. “Developing talent in areas like AI foundation model training and compiler development is far more urgent.”
The Paulson Institute at the University of Chicago estimates that 47% of the world’s top AI researchers are from China, followed by 18% from the United States and 17% from Europe. South Korea accounts for just 2%.
