The technological capabilities and prospects of artificial intelligence (AI) are driving many countries to increase their investment in related hardware to keep pace with global competition. South Korea's acting president, Choi Sang-mok, stated on February 17 that the country will expand the supply of 18,000 high-performance GPUs (graphics processing units) by the first half of next year. Just last month, India also announced it would provide over 18,000 GPUs to build AI computing facilities to support local startups and developers. Experts interviewed by the Global Times indicated that current competition requires countries to achieve self-sufficiency and diversified supply in hardware, while also enhancing the level of collaboration between hardware and software. Ensure 10,000 GPUs by the End of the Year According to the Hankook Ilbo on February 17, Choi Sang-mok chaired a meeting of the "Special Committee on AI Computing Infrastructure" to gain a detailed understanding of the ongoing "National AI Computing Center" construction project. He stated that to enhance South Korea's AI competitiveness, the country will expand the processing capacity of 18,000 high-performance GPUs by the first half of next year to support businesses and research institutions. Choi Sang-mok also emphasized that the government will fully support efforts to become a true "AI powerhouse" (previously, South Korean media cited statistical agencies ranking the United States and China as the top two—editor's note). The government will ensure the supply of 10,000 high-performance GPUs through public-private partnerships within this year and expedite the opening of services for the "National AI Computing Center." By the first half of next year, the sixth supercomputer, equivalent to 8,000 GPU processing capabilities, will be built, focusing on supporting research institutions. The South Korean government will ensure the supply of 10,000 high-performance GPUs through public-private partnerships within this year. The image shows employees at Samsung's headquarters operating in a data center. (Visual China) Reuters quoted a South Korean Ministry of Science and Technology official as saying that the government has not yet decided which GPU products to purchase, but details such as budget, GPU models, and participating private companies will be finalized by September this year. The Asia Daily reported on the 18th that currently, South Korea ranks sixth in the global AI rankings, performing well in development and infrastructure, but lagging behind in talent cultivation, research investment, and venture capital. The Korea Herald reported that this year, the South Korean government increased its annual AI technology budget by 25% year-on-year, allocating 1.8 trillion won (approximately 91.1 billion RMB) for AI programs. Additionally, the government plans to launch a cross-departmental initiative aimed at enhancing the country's AI capabilities. Key Hardware in Multinational Competition According to Reuters, the number of GPUs required for AI large models depends on the advancement of the GPUs, the amount of data used to train large models, the size of the large models themselves, and how much time developers wish to spend training the large models. Currently, American chipmaker NVIDIA holds about 80% of the global GPU market share, leading competitors Intel and AMD Given the importance of GPUs for the development of artificial intelligence, countries such as India and Saudi Arabia, in addition to South Korea, are also purchasing this key hardware in large quantities. According to The Hindu, Ashwini Vaishnaw, India's Minister of Railways, Communications, and Electronics and Information Technology, stated at the end of last month that India aims to develop a domestic AI large model and will build a large AI computing facility using over 18,000 GPUs. Last March, the country announced an investment of $1.2 billion for the "IndiaAI Mission" project, with nearly half of the funds allocated for building computing infrastructure. Additionally, as early as 2023, Saudi Arabia planned to purchase thousands of high-performance NVIDIA chips. According to public reports, at that time, Saudi Arabia had already acquired at least 3,000 NVIDIA H100 chips. Some analysts believe that countries looking to develop AI are "stockpiling" GPUs and other high-end chips out of concern that it may become difficult to obtain these key hardware in the future. In January of this year, the Biden administration introduced new regulations limiting the export of American AI chips and technology before leaving office. France 24 reported that the U.S. classified countries into three tiers, keeping advanced computing capabilities within the U.S. and its allies while seeking ways to prevent exports to certain countries. The Hindu reported that the U.S. has restricted the quantity of GPUs that India can import. Saudi officials stated in an interview with U.S. CNBC last April that they hope to obtain NVIDIA's high-performance chips "within the next year." Competition Among National Innovation Ecosystems Choi Sang-mook stated in his speech that the competition for dominance in the AI industry has become increasingly fierce, shifting from competition between companies to competition among national "innovation ecosystems." However, "the case of DeepSeek tells us that even with insufficient capital and infrastructure, there are opportunities to surpass in performance." Huang Zhengliang, director of the New Industrialization Research Office at the Industrial Economic Research Institute of the Ministry of Industry and Information Technology, told Global Times reporters on the 18th that the competition in the artificial intelligence innovation ecosystem encompasses multiple fields, including hardware such as chips, precision instruments, and high-performance equipment; software aspects, including infrastructure supply, talent cultivation, and technology conversion efficiency; and environmental factors, including institutional construction, financing channels, and the alignment of supply and demand. He stated that the competition in innovation ecosystems poses new requirements for the construction and development of hardware industry chains in various countries, emphasizing that "key hardware must be self-controllable, non-key hardware should have diversified supply, and the level of collaboration between hardware and software must be improved." Liu Shaoshan, director of the Embodied Intelligence Center at the Shenzhen Institute of Artificial Intelligence and Robotics, told Global Times reporters that countries are no longer pursuing the highest cost-performance solutions for AI supply chain security; instead, they need to build their own solutions, creating more market opportunities. He further explained that for a country to gain an advantage in innovation ecosystem competition, it must have a sufficiently large market to absorb innovation and strong innovation capabilities to achieve supply Author of this article: Chen Zishuai, Mang Jiuchen, Source: Global Times, Original title: "AI Competition Drives Up Demand for Key Hardware, Countries like South Korea, India, and Saudi Arabia 'Stockpile' GPUs" Risk Warning and Disclaimer The market carries risks, and investment should be cautious. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article align with their specific circumstances. Investment based on this is at one's own risk