An International Energy Association (IEA) report expects demand from AI data centres to triple by 2030.

AI data centres are energy guzzlers, requiring a significant amount of electricity for computation. Just in the UK, it is predicted that the 140 proposed data centre schemes could collectively require 50GW of power – 5GW more than the country’s current peak electricity demand. 

The new IEA report – Key questions on energy and AI – shows how AI is rapidly transforming global electricity use.

Its analysis reveals that electricity demand from data centres rose by 17% in 2025, far outpacing the 3% growth in overall global electricity demand. AI-focused data centres are growing even faster as people make use of energy-intensive applications such as AI agents.

By 2030, the IEA projects that electricity consumption from data centres will double, while AI-specific demand is expected to triple.

At the same time, AI data centre development is increasingly coming up against a range of physical and regulatory bottlenecks. For one, the swelling pipeline of data centre projects is putting pressure on electricity grids, leading to delays in grid connection requests and slowing down planning and approval processes.

In response, the IEA report highlights that tech companies were responsible for signing around 40% of all corporate renewable power purchase agreements in 2025. These firms are investing in nuclear energy and geothermal as a way to help power their AI efforts. 

Additionally, the pipeline of conditional offtake agreements between data centre operators and small modular reactor nuclear projects has grown from 25GW at the end of 2024 to 45GW today. As an example, last year Amazon said it would build a small modular reactor facility in Washington state to provide carbon-free energy for its data centres.

Constrained by slow grid connections, some US-based data centre developers are also investing in on-site energy generation, particularly natural gas. However, IEA data from satellite-based tracking shows that many of these projects remain in their early stages, highlighting the technical and financial hurdles that need to be overcome. 

One of the key challenges is that AI data centres have rapid and large swings in demand, and meeting their power needs reliably can stretch the technical capabilities of on-site gas plants. For this reason, on-site battery storage is becoming a critical technology for the next generation of AI data centres, which could make them an asset to grids with the right incentives.

Fatih Birol, executive director at 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 centres and long-duration energy storage.”

Last month, major tech firms including Google, Amazon and OpenAI signed an agreement with the White House to bear all the costs associated with the new energy infrastructure needed to power their ever-expanding data centres.

An AI data centre integrated into a floating offshore wind platform is being developed by US technology start-up Aikido Technologies for commercial deployment in 2028.