South Korea to secure 10,000 GPUs as AI competition heats up

Government plans public-private cooperation to boost AI computing amid global chip rivalry.

South Korean Finance Minister Choi Sang-mok speaks during a trilateral meeting with U.S. Treasury Secretary Janet Yellen and Japanese Finance Minister Shunichi Suzuki on the sidelines of the IMF/G20 meetings at the U.S. Treasury in Washington, U.S., on April 17, 2024. Photo by Kevin Lamarque/Reuters

By Alana Salsabila and Anna Fadiah

South Korea has announced plans to secure 10,000 high-performance graphics processing units (GPUs) this year as it seeks to strengthen its artificial intelligence (AI) capabilities. The move comes amid growing international competition for advanced AI infrastructure, with countries racing to develop national AI ecosystems.

"As competition for dominance in the AI industry intensifies, the competitive landscape is shifting from battles between companies to a full-scale rivalry between national innovation ecosystems," said South Korea's acting President Choi Sang-mok in a statement on Monday.

The South Korean government plans to acquire the GPUs through a public-private partnership to accelerate the launch of its national AI computing center. This initiative is expected to bolster the country’s AI capabilities and help local developers train and deploy AI models more efficiently.

U.S. regulations and global AI chip restrictions

The announcement follows a recent move by the U.S. government to tighten regulations on AI chip exports. The new rules, introduced last month, restrict the export of high-performance GPUs and specialized processors used in AI development.

Under these regulations, the world is divided into tiers. South Korea is among 18 countries that are largely exempt from restrictions, while over 120 nations face export limitations. Countries like China, Russia, and Iran are barred entirely from accessing U.S.-designed AI chips.

The demand for GPUs has surged globally as AI development requires vast computing power. The number of GPUs needed for AI models depends on factors such as processing power, data size, and training duration.

South Korea's AI investment strategy

A Ministry of Science and ICT official told Reuters that South Korea has not yet finalized details on which GPU models to purchase. The government is expected to determine the budget, GPU specifications, and participating private companies by September.

The global GPU market is dominated by U.S. chipmaker Nvidia, which controls about 80% of the industry. Nvidia’s GPUs are widely used in AI computing, far surpassing competitors like Intel and AMD.

Meanwhile, OpenAI, backed by Microsoft, is looking to reduce its reliance on Nvidia chips. According to Reuters, OpenAI is developing its own AI silicon and is expected to finalize the design in the coming months before sending it for fabrication at Taiwan Semiconductor Manufacturing Co.

The rise of alternative AI chipmakers

The AI race is also being reshaped by new entrants in the semiconductor industry. Chinese startup DeepSeek has emerged as a competitor, focusing on AI models that optimize computational efficiency rather than sheer processing power. These advancements could help narrow the gap between Chinese-designed AI processors and the more powerful GPUs produced by U.S. firms.

As AI competition continues to escalate, South Korea’s investment in GPUs is seen as a strategic move to maintain its edge in AI research and development. The country’s AI computing center is expected to play a crucial role in fostering innovation and ensuring South Korea remains competitive in the evolving global AI landscape.