As the global race for artificial intelligence reaches new heights, the United States is projected to maintain its lead in AI supercomputing power by 2025, boasting a total AI computing capacity equivalent to 39.7 million H100 chips. This figure underscores the U.S.”s status as the benchmark in high-performance AI processing.
A recent analysis conducted by TRG Datacenters highlights that worldwide investment in AI infrastructure has surged to an unprecedented $200 billion. Countries are actively striving to enhance their computational capabilities to stay competitive in the rapidly evolving AI landscape. The study not only assessed AI supercomputing power but also evaluated the overall engagement of the workforce in AI-related activities, the presence of AI companies, and the readiness of government policies regarding AI.
The primary data source for this analysis was the Epoch AI Supercomputers dataset, which is recognized for its comprehensive insights into global AI infrastructure. The research also included metrics such as the share of the AI workforce, the number of AI companies within each nation, and an index measuring government readiness for AI implementation.
According to the findings, the U.S. not only leads in total AI computing power but also has the highest total power capacity at 19.8 kilowatts (MW), indicating a robust energy supply necessary for supporting AI advancements. Following the U.S. is the United Arab Emirates, with an AI computing power of 23.1 million H100 equivalents, significantly trailing the U.S. The UAE has over 188,000 AI chips and a total power capacity of 6.4 kilowatts.
Saudi Arabia ranks third, showcasing an AI computing power of 7.2 million H100 equivalents. The nation boasts more data center clusters than the UAE, along with a growing workforce engagement rate of approximately 2.2 percent in AI usage. However, it has a lower total power capacity of 2.4 kilowatts and around 54,000 AI chips.
In fourth place is South Korea, with an AI computing capacity of 5.1 million H100 equivalents. South Korea not only has a higher total power capacity of 3 kilowatts but also leads in workforce engagement, with around 50 percent of its workforce utilizing AI technologies.
France occupies the fifth position, with an AI computing power of 2.4 million H100 equivalents, which is less than half that of South Korea. France has 18 data center clusters and ranks second globally for the number of AI chips, just behind the U.S.
India is in sixth place with an AI computing power of 1.2 million H100 equivalents. The nation has a notable number of AI chips, totaling 493,000, and eight data center clusters, alongside a total power capacity of 1.1 kilowatts.
In seventh place, China shows an AI computing power of 400,000 H100 equivalents, which is markedly lower than that of India. Despite having the highest number of data center clusters at 230, China”s overall power capacity stands at 289 megawatts.
The United Kingdom and Northern Ireland follow in eighth place with an AI computing power of 120,000 H100 equivalents and a total power capacity of 99 megawatts, which is the second-lowest in the top ten. However, the UK has the second-highest number of active AI start-ups, totaling 4,700.
Finland comes in ninth, with an AI computing power of 72,000 H100 equivalents and a total power capacity of 110 megawatts, while having fewer data center clusters than the UK.
Lastly, Germany rounds out the top ten with an AI computing power of 51,000 H100 equivalents and a limited capacity of 25 megawatts across 12 data center clusters.
