ConfX Global
InsightXOverviewReportsSector BriefingsLeadership PulseKnowledge SessionsInterviewsArticlesTools
Sector briefing · Sustainability

Data centres and the power demand of AI

How utilities and technology companies are planning for new demand.

ConfX editorial31 August 20263 min readEnergy and utilities
Download PDF

Why it matters

The growth of AI is increasing data-centre electricity demand and putting pressure on grid connections in several regions. Operators, utilities and governments are rethinking where and how data centres are powered.

Power availability is becoming a deciding factor in where data centres are built. Operators are working more closely with utilities and governments on grid capacity, flexible operation and low-carbon supply.

By the numbers

415 TWhused by data centres in 2024, about 1.5% of global electricity 1
945 TWhexpected by 2030, roughly double 1
Global data centre electricity use
2024
415 TWh
2030 (projection)
945 TWh
Source: IEA, Energy and AI, April 2025

What's changing

  • Grid connection timelines constrain new capacity in some markets.
  • Operators are signing long-term renewable power agreements.
  • Efficiency reporting and waste-heat reuse are receiving more policy attention, including under the EU Energy Efficiency Directive.

How leaders are responding

  • Coordinating early with network operators on capacity planning.
  • Exploring on-site generation and storage.
  • Improving cooling and workload efficiency.

What good looks like

  • Data centre siting decided with grid operators early
  • Power purchase agreements matched to new demand
  • Efficiency metrics reported for every facility

What to watch

  • Grid connection reforms in major markets.
  • Efficiency and reporting requirements for data centres.
  • On-site generation and storage.

Where to start

Include power availability, cost and carbon intensity in decisions on where AI workloads run.

Questions for your leadership team

  1. What grid capacity will our AI plans require?
  2. How will we source low-carbon power?
  3. How efficiently do we use the capacity we already have?

Sources

  1. IEA, Energy and AI, April 2025