The report, Key Questions on Energy and AI, explored AI’s growing energy footprint and options for meeting data centre power demand and found that the sector is continuing to grow at pace.
Driven by data centre investments, it revealed that the capital expenditure of five large technology companies soared to more than US$400 billion in 2025 and is expected to increase by an additional 75% in 2026.
It also found that, as more people use AI and energy-intensive use cases like AI agents, electricity consumption is poised to double by 2030, with power use from AI-enabled data centres expected to triple.
AI as ‘energy taker’ and ‘energy maker’
Electricity demand from data centres soared by 17% in 2025. AI-focused data centres climbed even faster, with the IEA finding they outpaced growth in global electricity demand by 3%. Fatih Birol, IEA executive director, recommends that technology and energy sectors should collaborate with policymakers to confront concerns around the rapid growth of AI.
“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,” said IEA executive director Fatih Birol.
The IEA also found that, although AI is contributing to massive global energy demand, it’s also powering innovation into clean power technologies like nuclear and renewables. It said its research confirms power consumption per AI task is declining, with efficiency improving at an unprecedented rate.
“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,” Birol added.
Paving the way for sustainable growth
To solve energy-related challenges, the wider technology sector has adopted new approaches, having accounted for roughly 40% of all corporate power purchase agreements (PPAs) for renewables signed in 2025. It is now helping to drive momentum in nuclear industries, with major deals signed in countries like the UK to drive growth.
According to the IEA’s research, the small modular reactor (SMR) pipeline with data centre operators has grown from 25 gigawatts (GW) at the end of 2024 to 45GW today, which it said indicates the momentum behind AI could accelerate the commercialisation of new energy technologies.
Robert Dunn, CEO at Start Campus, explained how sustainable data centre development in 2026 has already evolved beyond marketing narratives towards engineering-led solutions.
“The most significant advances are driven by the deployment of liquid cooling, waste heat recovery and reuse, zero-water cooling technologies and grid-interactive facility design,” he explained. “The paradigm has shifted. Sustainability and performance are no longer trade-offs. For AI infrastructure, they have become inseparable.”
He added: “High-density AI deployments demand the thermal management and power efficiency that sustainable engineering approaches inherently provide.”
How data centre operators can confront energy challenges
The IEA’s findings have come at the same time as AI deployment is coming up against a series of bottlenecks, which are limiting how much data centres can expand. It cites how supply chains for energy technologies like gas turbines, transformers and components like advanced chips are strained.
“The swelling pipeline of data centre projects is straining planning and regulatory systems [is] holding up grid connections and other necessary approvals,” the IEA said. “Constrained by slow grid connections, data centre developers are also advancing a large number of projects with onsite natural gas-based power generation, largely in the US.”
First-of-its-kind IEA data from satellite-based tracking has revealed that many of these projects remain in their early stages, the report said, highlighting technical and financial hurdles. The organisation also that, to meet AI data centre demand challenges, onsite battery storage is becoming a critical technology to make data centres grid assets.
“The energy sector is not yet taking full advantage of AI’s potential,” the IEA added, citing a lack of sufficient digital skills and data availability as key barriers to adoption.
As social concerns around data centres persist, with data centres viewed by the public as an energy concern, the IEA said policymakers can help facilitate smart integration of data centres into grids and incentivise them to operate more flexibly.
“The report finds that if the right mix of policies and infrastructure investment are in place, increases in electricity demand do not necessarily raise prices,” the report said.
Related stories
Is Muse Spark Meta’s bid to justify AI costs amid ‘superintelligence’ race?
Power, planning and regulation – confronting the data centre energy challenge
