A recent analysis by the International Energy Agency (IEA) examines the rapidly evolving relationship between energy and artificial intelligence (AI), drawing on updated data and technological insights. The report notes that data centre investments by five major technology firms exceeded $400 billion in 2025 and are projected to increase by 75 % in 2026. Overall electricity use from data centres rose by 17 % last year, with AI-focused facilities growing faster than the 3 % increase in global power demand.
Despite significant improvements in energy efficiency—power consumption per AI task is declining at an unprecedented rate—the expansion of AI usage and energy-intensive applications means data centre electricity demand could double by 2030, and AI-centric power use may triple. However, expansion faces near-term constraints. Supply chains for gas turbines, transformers, advanced chips and IT components have tightened, while grid connection approvals and planning processes are under strain due to a surge in proposed projects.
To address these challenges, the technology sector is pursuing diverse strategies. In 2025, tech companies accounted for roughly 40 % of corporate renewable energy purchase agreements, and they are increasingly driving interest in nuclear and advanced geothermal projects. Conditional offtake agreements between data centre operators and small modular reactor projects have grown from 25 GW at the end of 2024 to 45 GW today.
Faced with slow grid connections, many data centre developers—particularly in the United States—are advancing onsite natural gas power generation. Early satellite-based data indicate that most of these projects are still in early stages, highlighting technical and financial hurdles. AI data centres’ rapid demand fluctuations pose challenges for onsite gas plants, making battery storage a key enabler for reliable operation and potential grid support.
IEA Executive Director Fatih Birol emphasized that while AI currently draws energy, it is also fostering innovation in next-generation nuclear reactors, flexible data centres and long-duration storage. He announced an upcoming platform to facilitate dialogue between governments and industry on energy and AI issues. The report also identifies AI’s potential to reduce energy costs by 3–10 percentage points in energy-intensive sectors, but notes that digital skills gaps and limited data availability remain barriers. Policymakers can manage affordability concerns by promoting grid-integrated, flexible data centre operations and supportive infrastructure investment.
Source: IEA News
