A new IEA report examines AI’s increasing energy consumption, options for meeting data center power demands, and its impact on energy affordability, security, and the broader economy.
A new IEA report assesses how the relationship between energy and artificial intelligence (AI) is rapidly evolving, based on the latest data and analysis and close tracking of technological and economic developments in the field.
Built on IEA landmarks Energy and AI New analysis published today, based on the April 2025 report, finds that the sector continues to develop rapidly. Data center investments are expected to cause capital expenditures by the five largest technology companies to soar to more than $400 billion in 2025, with an additional 75% increase in 2026. Electricity demand from data centers will surge 17% in 2025, and electricity demand from AI-focused data centers will grow even faster, significantly outpacing the 3% growth in global electricity demand.
According to the report – Key questions about energy and AI – Power consumption per AI task is rapidly decreasing, increasing efficiency at a rate unprecedented in energy history. But more people are using AI, and energy-intensive applications such as AI agents are on the rise. As a result, power consumption by data centers is expected to double by 2030, and power use by AI-focused data centers is expected to triple.
At the same time, AI deployments increasingly face a variety of physical bottlenecks, limiting the speed of data center expansion in the short term. Supply chains for energy technologies such as gas turbines and transformers, as well as advanced chips and IT components, have tightened over the past year, with a bulging pipeline of data center projects straining planning and regulatory systems and holding up grid connections and other required approvals.
To solve the energy challenges at hand, the technology sector is taking new approaches. This will account for around 40% of corporate power purchase agreements for renewable energy signed in 2025 and is now a major source of impetus for the nuclear and advanced geothermal industries. The pipeline of conditional offtake agreements between data center operators and small modular reactor (SMR) nuclear projects has increased from 25 GW at the end of 2024 to 45 GW now, showing that the momentum behind AI could accelerate the commercialization of new energy technologies.
Data center developers are also pursuing a number of projects involving on-site natural gas-based power generation, primarily in the United States, constrained by slow grid connections. The first IEA data from satellite-based tracking shows that many of these projects are still in their early stages and highlights the technical and financial hurdles that need to be overcome. One of the key challenges is that demand for AI data centers can fluctuate rapidly and widely, potentially straining the technical capacity of on-site gas plants to reliably meet power demand. On-site battery storage is therefore becoming a key technology for the next generation of AI data centers and, with the right incentives, could become an asset to the grid.
“The IEA recognized early on that without energy there is no AI, and that countries that provide safe, affordable and rapid access to electricity will be one step ahead.” IEA Director General Fatih Birol. “Today, we see that while AI remains an energy stealer, it is also becoming an energy producer, driving innovative solutions such as next-generation nuclear reactors, flexible data centers, and long-term energy storage. Collaboration between policymakers and the energy and technology sectors remains critical to seize this opportunity to help countries modernize their energy systems and address bottlenecks and other concerns associated with the rapid growth of AI.”
He continued: “To foster greater dialogue, I am pleased to announce that we will soon be launching a new platform for governments and industry to regularly discuss energy and AI issues. In addition to providing the latest data, the IEA will continue to foster the collaboration needed to maximize the energy benefits of AI and overcome key challenges.”
This report shows that AI could be critical to global industrial innovation and competitiveness. We found that proven applications of AI can help companies in energy-intensive industries reduce energy costs by 3 to 10 percentage points. However, the report says the energy sector as a whole is yet to exploit the full potential of AI, with a lack of sufficient digital skills and data availability being key barriers to adoption.
Over the past year, public concerns surrounding AI have also increased, with data centers becoming a highly visible flashpoint for concerns about energy prices and the environment. The report finds that with the right mix of policy and infrastructure investment, increased demand for electricity does not necessarily lead to higher prices. However, because data centers have highly concentrated power loads and scale rapidly, they can pose special challenges of power affordability, often requiring investments in new generation assets and grids. Still, the report found that policymakers have tools in place to manage affordability issues, including tools that promote smarter integration of data centers into the grid and encourage more flexible data center operations.
