Skip to main content

Data Explained

Add as Preferred Source
Get in touch
Data Shows Us How Much Energy AI Actually Uses

Table of Contents

The energy use of artificial intelligence (AI) is one of the most talked-about and least understood numbers in technology. 

 

The reality is that it is large, growing fast, highly concentrated, and still a modest share of the global total.

 

TL;DR

 

  • Data centers used about 415 terawatt-hours of electricity in 2024, roughly 1.5% of world consumption.
  • That is set to more than double to around 945 terawatt-hours by 2030, slightly more than Japan’s entire electricity use today.
  • The U.S. accounted for 45% of data center electricity in 2024, followed by China at 25% and Europe at 15%.

 

The energy figures here come from the IEA’s Energy and AI report, the most comprehensive global analysis of the subject.

 

ALSO READ: How AI Adoption Is Actually Measured, Explained

 

Where the power goes

 

 

Electricity use by AI-focused servers is projected to grow about 30% per year, far faster than conventional servers, which is why AI is the single biggest driver of the increase.

 

The U.S. story, and the honest caveat

 

Data centers are on course to account for almost half of U.S. electricity demand growth to 2030, and by the end of the decade the U.S. will consume more electricity for data centers than for producing aluminum, steel, cement, chemicals, and all other energy-intensive goods combined. 

 

Ireland is an even sharper case, where around 21% of national electricity already goes to data centers, potentially rising to 32% by 2026.

 

Yet the global figure deserves perspective. 

 

The rise in data center demand by 2030 is only about 8% of the total projected increase in electricity demand, less than the growth from electric vehicles or air conditioning. 

 

AI’s footprint is real and locally intense, but at the global level, it is not the dominant driver of energy growth.

 

ALSO READ: Does AI Adoption Actually Track With Productivity?

 

ELI5

 

The computers that run AI use a lot of electricity, and they will use about twice as much by 2030. Most of them are in the United States. 

 

It is a big number, but worldwide it is still a small slice of all the electricity people use.

 

Sources

 

IEA | IEA (2) |  Carbon Brief