IEA warns datacentre emissions to rise sharply as AI drives power demand

IEA warns datacentre emissions to rise sharply as AI drives power demand
AI fuels emissions surge

Rising electricity demand from artificial intelligence is reshaping the energy outlook for datacentres, with fossil fuels set to play a larger role in key markets through 2030. The International Energy Agency says the sector's carbon emissions are on track to increase by 70% by the end of the decade as the U.S. and China expand capacity to support AI growth.

Highlights

  • IEA projects datacentre electricity consumption will double by 2030, with AI driving about 75% of the growth to an extra 350TWh annually within four years.
  • AI factory capacity reached 17GW after a tenfold increase in 24 months, expected to hit 50GW soon, with global datacentre capacity headed to 240GW by 2030.
  • By 2030, 50% of Chinese AI datacentre power will come from coal and 60% of U.S. from gas, while the EU expects 70% of datacentre energy from renewables.

Power demand outlook for AI datacentres

As reported by The Guardian, citing the International Energy Agency, datacentres are expected to consume twice as much electricity by 2030 as they do today, with artificial intelligence accounting for about three-quarters of that growth. The agency said AI datacentres alone will require an additional 350 terawatt-hours of electricity a year within four years, more than Great Britain's annual power consumption.

The IEA based its forecast on data showing annual global shipments of AI servers have increased eighteen-fold since 2020. It also used space-based imagery to monitor the construction of 93 large AI factories worldwide, concluding that datacentre capacity has increased tenfold over the last 24 months to 17GW.

According to the agency's analysis, those AI factories are likely to reach a combined capacity of about 50GW once completed. Total existing global datacentre capacity, including sites with AI-optimised servers, now stands at 120GW, with 80% located in China, the U.S. and the EU, and another 120GW is due to be built by 2030 under its central scenario.

Fuel mix shifts and regional industry impact

The IEA said the sector's rapid expansion is increasing demand for fossil fuel generation faster than renewables in the two largest markets. By 2030, about half of the electricity used by China's AI datacentres is expected to come from coal, while in the U.S. gas is projected to supply about 60% of datacentre power demand.

The agency said plans for about 180GW of new gas-fired power plants linked to U.S. datacentres were on the table in August this year, close to the total gas capacity currently installed across the European Union. It added that U.S. datacentre developers' appetite for gas power is more than 16 times higher than in Canada, where plans cover 11GW, while proposed new gas capacity for the rest of the world totals just 4GW.

Outside the U.S. and China, renewables are expected to play a larger role in supplying datacentres. In the EU, the IEA expects renewables to provide about 70% of datacentre electricity by 2030, followed by nuclear and gas, with only a very small contribution from coal.

The report also said the broader global energy outlook has become more uncertain after disruption linked to the Middle East war. IEA executive director Fatih Birol said he does not see a credible path for the energy system to return to its previous state, as governments balance energy security, population growth, cooling demand and the expansion of the AI economy.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.