AI supercomputers are growing faster than power infrastructure can handle. By 2030, a single system may need as much power as 9 nuclear plants.
The next generation of AI supercomputers is shaping up to be massive in size, cost, and power demand. A new study by researchers from Georgetown, Epoch AI, and Rand reveals just how immense the infrastructure required to train and run AI systems may become by the end of the decade.
Between 2019 and 2025, leading AI data centers’ hardware costs and power consumption have doubled annually. And if this exponential growth continues, we may see a single AI supercomputer in 2030 housing two million chips, costing $200 billion, and consuming 9 gigawatts of power, roughly what nine nuclear reactors produce.Power demand is outpacing efficiency gainsWhile data centers are becoming more efficient, it’s not enough to curb overall energy needs. The study found that computational performance per watt has improved 1.34x annually between 2019 and 2025. Yet, over the same period, power requirements have climbed 2x each year.“AI supercomputers are improving in energy efficiency, but the shift is not quickly enough to offset overall power growth,” notes Epoch AI. Their estimate shows that 9GW of power could keep the lights on in 7 to 9 million homes — a scale far beyond today’s capabilities.Currently, the most advanced AI supercomputer is xAI’s Colossus, which was built in 214 days and cost $7 billion. It runs on 200,000 AI chips and consumes 300 megawatts of electricity, equal to the usage of 250,000 households.How quickly are AI supercomputers scaling, where are they, and who owns them?Our new dataset covers 500+ of the largest AI supercomputers over the last six years.Here is what we found🧵 pic.twitter.com/X5ZOqXvavv— Epoch AI April 23, 2025Industry now dominates the AI supercomputer landscapeBack in 2019, private companies owned 40% of AI supercomputing power. By 2025, that figure has surged to 80%. The shift reflects how AI has moved from academic exploration to a commercial goldmine.“Where once supercomputers were used just as research tools, they’re now being used as industrial machines delivering economic value,” explains the Epoch AI paper.As a result, the size and scope of private sector systems have grown rapidly, at 2.7x annually compared to 1.9x for public sector machines. OpenAI’s $500 billion Stargate project and NVIDIA’s recent $500 billion commitment are just two examples of the massive capital flowing into AI infrastructure.The U.S. leads global AI computingAccording to the study, the United States controls about 75% of all AI supercomputing power today. China follows with 15%, while traditional supercomputing nations like Japan and Germany play smaller roles. However, the physical location of these data centers doesn’t necessarily reflect who uses them, as many offer remote access through the cloud.Still, hosting these facilities comes with major trade-offs. A report from Good Jobs First estimates that at least 10 U.S. states lose over $100 million annually in tax revenue due to generous incentives for data centers. Moreover, these centers consume vast amounts of water and land, straining local ecosystems.A possible cooling periodDespite the trajectory, not all signs point to unchecked growth. In early 2025, analysts at Cowen noted a “cooling” in the market, with hyperscalers like AWS and Microsoft pulling back some projects. Whether this signals a permanent slowdown or a strategic pause remains to be seen.But one thing is clear — the rise of AI is synonymous with the race to build bigger and more power-hungry machines.
AI Supercomputers Data Centers Epoch AI Nuclear Power Nvidia Investment Openai Stargate Power Consumption Xai Colossus
United States Latest News, United States Headlines
Similar News:You can also read news stories similar to this one that we have collected from other news sources.
HIV-funding cuts could lead to nearly 3 million extra deaths by 2030, study suggestsNicoletta Lanese is the health channel editor at Live Science and was previously a news editor and staff writer at the site. She holds a graduate certificate in science communication from UC Santa Cruz and degrees in neuroscience and dance from the University of Florida.
Read more »
World’s first fusion-fission plant aims to generate 100MW nuclear power by 2030China to build world's first fusion-fission hybrid power plant with 100 MW output by 2030.
Read more »
China Plans Ambitious Nuclear Fusion Target For 2030A multibillion-dollar project is set to be constructed in China, the country's latest move in a push for nuclear energy.
Read more »
Selena Quintanilla's killer, Yolanda Saldívar, denied parole, next eligibility in 2030Yolanda Saldívar was denied parole, the Texas Board of Pardons and Paroles confirmed in a statement posted to its webpage.
Read more »
602,000 metric ton Coca-Cola plastic packaging can enter world’s waterways by 2030Nearly 1.3 billion pounds of plastic waste is set to enter the world’s oceans and waterways after Coca-Cola scrapped 25% package reuse goal.
Read more »
Monster AI belly to devour 945 TWh power by 2030—US, China to fuel the feastSmall modular reactors (SMRs) are poised to transform data center power, as nuclear energy becomes essential for AI's energy needs.
Read more »