A new International Energy Agency (IEA) report, Key Questions on Energy and AI, shows that artificial intelligence is rapidly transforming global electricity use. In 2025, electricity demand from data centres rose by 17%, far outpacing the 3% growth in overall global power demand. AI-focused facilities are growing even faster. The IEA expects data centre electricity use to double by 2030, with AI-specific demand tripling as more users and energy-intensive applications, such as AI agents, increase consumption.
Although each AI task is becoming more energy-efficient, overall demand is still rising due to rapid expansion. The five largest tech companies made substantial capital investments in 2025, with spending expected to rise by a further 75% in 2026, largely driven by hyperscale data centre development.
At the same time, AI expansion is facing physical and regulatory constraints. Supply chain shortages for gas turbines, transformers and advanced chips, along with grid connection delays and complex permitting, are slowing data centre development. In response, tech companies signed around 40% of all corporate renewable PPAs in 2025 and are increasingly driving interest in nuclear and advanced geothermal. Conditional offtake agreements for small modular reactors (SMRs) have grown from 25 GW at the end of 2024 to 45 GW today.
With grid upgrades lagging, many US developers are turning to onsite natural gas generation, often paired with battery storage, to handle sharp swings in AI-driven demand. The report also notes that AI could reduce energy costs in heavy industry by 3–10 percentage points, but uptake remains limited due to gaps in digital skills and data access.
IEA Executive Director Fatih Birol said closer cooperation between governments, grid operators and tech companies is needed to modernise infrastructure, manage costs and make data centre demand more flexible for the grid. The IEA plans to launch a new government industry platform to support ongoing dialogue on energy and AI, as countries work to balance AI’s economic benefits with its rising electricity demand.
