AI data centre power demand set to double by 2030 as spending rises
The International Energy Agency says AI data centres could add power demand equal to another Great Britain by 2030. Tom's Hardware reports that grid limits, electricity costs and emissions are central concerns.

Key takeaways
- AI data centre power demand could double by 2030.
- The IEA tracks 93 large AI factories with 50GW of potential capacity.
- Grid upgrades may affect electricity bills, but no clear price link is established.
- Coal and gas emissions from data centres could rise by 70% by 2030.
In this article
Demand doubles by 2030
AI data centre power demand is expected to double by 2030, according to Tom's Hardware. The report draws on the International Energy Agency's World Energy Outlook 2026. It links the growth to $4 trillion in spending.
The IEA says about 120 gigawatts (GW) of data centre capacity is available worldwide today. It expects another 120GW to be completed by the end of the decade. About 75% of that new capacity would be used for artificial intelligence. The 120GW figure covers all types of data centres, not only AI sites.
The forecast shows how quickly AI infrastructure is becoming a major part of global electricity demand. It also explains why grid operators and communities are paying closer attention to data centre projects.
AI factories and grid pressure
The agency is tracking 93 large AI factories, or major AI data centres, that are being built or expanded. It used satellite imagery to estimate that their capacity grew from less than two gigawatts to 17GW in two years.
When finished, these sites could provide 50GW. The IEA says that would be like adding another Great Britain to power demand. The build-out is already straining electricity systems. Some regions face requests that exceed the peak load their networks can support.
This pressure is not only about the amount of power. It also concerns where the power comes from and how quickly new supply can be connected.
Electricity bills remain uncertain
Communities near data centres have raised concerns about higher utility prices. The IEA says its analysis has not found a clear link between data centre growth and rising electricity prices. It says the current evidence is not enough to draw a firm conclusion.
The outcome may depend on how the grid is used. If a data centre relies on spare capacity and covers its costs, bills could fall for all customers. If network upgrades are required, the costs may be spread across users and monthly bills could rise.
The IEA's comments suggest that local effects will vary. A project that uses spare capacity may have a different impact from one that forces expensive grid work.
Renewables and emissions

Nearly 50% of the electricity contracts signed by data centres are for renewable energy. But the IEA says actual generation is closer to 60% from coal and natural gas. It says reliability needs can keep fossil fuel plants running when renewable output is low.
The agency expects emissions from coal and natural gas used by data centres to rise by 70% by 2030. Some projects are building their own power supplies. The IEA names Elon Musk's Colossus 2 in Southaven, Mississippi. It says about 180GW is planned or being built in the US.
The gap between clean energy contracts and actual generation is important. It shows that adding renewable capacity alone may not remove the need for fossil fuel backup. The IEA says renewables are growing, but fossil fuels still play a major role.
Infrastructure bottleneck
Investment in AI data centres could reach $32 trillion by 2050, according to the IEA. The IEA says this would exceed the combined infrastructure spending on railways, electrification and the internet.
The grid remains a major bottleneck. New renewable plants can take one to five years to finish. Transmission projects may need three to five times longer. Large data centre projects can take a similar time if they are not delayed by long pauses. The IEA says transmission work can be slower than building new generation.
Some AI companies are installing jet engines at their sites, but turbine blades are in short supply. Tom's Hardware also notes that demand for data centre processors has surged.
The report shows that AI expansion is not limited to software. It is also shaping electricity demand, construction and the supply of computing hardware.
Sources
Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.



