Electricity demand from data centers surged 17 percent in 2025, more than five times faster than overall global demand growth of 3 percent, as tech companies spent over US$ 400 billion on AI infrastructure, a figure that could increase by another 75 percent in 2026, according to a new report by the International Energy Agency (IEA). The agency projects that electricity demand from data centers will double by 2030, with demand linked specifically to AI expected to triple over the same period.
The Key Questions on Energy and AI report, details the mounting concerns that rapid global expansion of artificial intelligence (AI) is placing significant strain on energy generation and supply systems. While efficiency gains are reducing the energy required per AI task, total consumption continues to rise. This is driven by the scale of deployment and the growing use of more energy-intensive applications, including AI agents.
According to a press release, the pace of growth is already running into constraints due to shortages of critical infrastructure components including gas turbines, transformers and advanced semiconductors have tightened over the past year. Regulatory delays are also slowing grid connections and project approvals, further complicating expansion plans.
Fatih Birol, Executive Director, IEA, said, “The IEA was early in recognising that there is no AI without energy and that countries that provide secure, affordable and rapid access to electricity will be one step ahead. Now, we see that while AI is still an energy taker, it is also becoming an energy maker driving forward innovative solutions like next-generation nuclear reactors, flexible data centers and long-duration energy storage.”
The renewable energy sector accounted for 40 percent of corporate energy purchase agreements in 2025, while also accelerating investment in nuclear and geothermal energy. Agreements tied to small modular reactor projects have nearly doubled, rising from 25 GWs at the end of 2024 to 45 GWs.
Data center developers are using on-site generation and gas-powered systems in the United States, though many of these projects remain in early stages. The intermittent and fluctuating power demands of AI data centers are also increasing interest in battery storage as a supply stabilisation method.
The report also highlights potential economic gains from AI adoption. Existing applications could cut energy costs in heavy industries by 3 percent to 10 percent, though uptake remains uneven due to gaps in digital skills and limited access to high-quality data.