All resources
Trending7 October 2026·5 min read

The IEA on AI and electricity: power is becoming the binding constraint

Data-centre electricity demand grew 17% in 2025, against 3% for global demand overall. Why that matters for hosting costs and model choices.

A Power Technology article dated 21 April 2026 summarises the International Energy Agency's report 'Key Questions on Energy and AI'. We have not read the IEA report itself; verify the figures at iea.org before quoting them.

Figures as summarised

  • Data-centre electricity demand rose 17% in 2025, compared with 3% growth in overall global power demand.
  • Data-centre electricity use is expected to double by 2030, with AI-specific demand tripling.
  • Technology companies signed about 40% of all corporate renewable power purchase agreements in 2025.
  • Conditional offtake agreements for small modular reactors grew from 25 GW at the end of 2024 to 45 GW.
  • Many US developers are turning to on-site natural gas generation.

Why it matters to a business buyer

Capacity limits can show up as higher prices, waiting lists for hosting in some regions, or rate limits at the provider. Efficiency is also a design choice: smaller models, caching and batching reduce both cost and energy use.

Include capacity and regional availability in vendor reviews, particularly if you need to keep data in a specific region.

Related capability

This is what our AI-Optimized Cloud & CI/CD practice is built around.

See how it's scoped