AI Data-Center Power Crisis 2026

AI is turning data centres into a power-system issue. The practical question for industry is no longer whether AI workloads grow; it is where electricity, grid capacity, heat rejection and on-site generation come from, and what that means for industrial energy buyers.

Data-centre electricity demand is on track to roughly double by 2030

2025 data centres485 TWh2030 projection950 TWh
IEA central projection: data-centre electricity consumption roughly doubles from 2025 to 2030.

Source: IEA — Key Questions on Energy and AI (2026)

The IEA projects global data-centre electricity consumption to reach about 945 TWh by 2030, just under 3% of global electricity consumption. AI-focused data centres grow faster than the total, which is why the issue is moving from IT procurement into grid planning, power-plant permitting and industrial energy strategy.

The power constraint is local before it is global

Source: Inzonex PowerAtlas — Open power-plant emissions and capacity atlas (2026)

Global TWh numbers hide the operational problem: data-centre loads concentrate in particular grid nodes, substations and permitting zones. That makes local power plants, gas-turbine fleets, interconnection queues and heat rejection constraints more important than generic AI market forecasts.

What industrial buyers should watch

Source: Inzonex — Industrial heat-loss calculator (2026)

For industrial sites, the AI power boom can show up as higher local electricity prices, slower grid connections, competition for firm capacity and pressure to justify energy-efficiency projects. The near-term response is not abstract: measure standing losses, reduce waste heat, insulate exposed hot assets and make every kilowatt-hour of existing capacity work harder.

FAQ

How much electricity could data centres use by 2030?

The IEA central projection puts data-centre electricity consumption at roughly 950 TWh by 2030, close to double the 2025 level.

Why does this matter to factories?

Factories buy electricity from the same constrained grids. Data-centre growth can affect local prices, grid-connection timing and the business case for efficiency.

Sources

Related

All data reports →