Policy
AI Datacenters to Boost Carbon Emissions 70% by 2030, IEA Warns
Fossil fuel reliance in the US and China will drive the surge as AI electricity demand doubles within four years.
Key takeaways
- Datacenter carbon emissions will rise 70% by 2030 as electricity consumption doubles, with AI driving three-quarters of the growth.
- The US plans 180GW of new gas-fired generation for datacenters—nearly equal to the EU's total gas capacity—while China will rely heavily on coal.
- AI datacenters will consume an additional 350 terawatt-hours annually by 2030, exceeding Great Britain's total yearly electricity use.
- Europe is charting a different course with 70% renewable energy expected to power its datacenters by decade's end.
- Global datacenter capacity has grown tenfold in 24 months to 17GW, with 93 large AI facilities under construction worldwide.
Fossil fuels will power AI's explosive growth
The world's datacenters are on track to generate 70% more carbon emissions by 2030, driven primarily by the United States and China's reliance on fossil fuels to meet surging artificial intelligence demand, according to a new forecast from the International Energy Agency.
The Paris-based watchdog projects that datacenters will consume twice as much electricity by the end of the decade compared to current levels, with AI accounting for three-quarters of that growth. The expansion translates to an additional 350 terawatt-hours of annual electricity consumption—exceeding the entire yearly power use of Great Britain.
Why it matters
While tech companies have positioned themselves as leaders in renewable energy procurement, the IEA's analysis reveals a stark gap between corporate sustainability pledges and infrastructure reality. The findings underscore how the AI race is reshaping global energy markets and threatening climate goals, particularly as the US and China pursue different fossil fuel strategies to maintain competitive advantage in AI development.
Divergent energy strategies emerge
The IEA's report, detailed by The Guardian, shows dramatically different approaches between the world's two AI superpowers. By 2030, approximately half of China's datacenter electricity will come from coal-fired plants, with the remainder supplied by renewables, nuclear power, and minimal natural gas.
The United States is charting a gas-heavy path. The agency expects natural gas to provide roughly 60% of American datacenter electricity by decade's end. Plans submitted as of August outlined approximately 180 gigawatts of new gas-fired generation linked to US datacenters—nearly matching the European Union's entire current gas capacity.
Canada presents a notable contrast within North America, with datacenter gas plans totaling just 11 gigawatts—more than 16 times less than US proposals. Outside the US and China, developers have proposed only 4 gigawatts of gas generation globally.
Europe bets on renewables
The European Union is pursuing a cleaner trajectory. The IEA forecasts that EU datacenters will draw approximately 70% of their power from renewable sources by 2030, supplemented by nuclear energy, gas, and minimal coal.
Despite datacenter operators becoming the world's largest corporate purchasers of renewable energy, the analysis shows fossil fuel appetite growing faster in absolute terms.
Tracking an opaque industry
The IEA based its projections on AI server shipment data—which has increased eighteenfold since 2020—and satellite imagery monitoring 93 large "AI factories" under construction worldwide.
Global datacenter capacity has expanded tenfold over the past two years to 17 gigawatts. The agency estimates completed AI facilities will reach combined capacity of about 50 gigawatts.
Total existing datacenter stock, including AI-optimized servers, now stands at 120 gigawatts, with 80% concentrated in China, the US, and the EU. Under the IEA's central scenario, an additional 120 gigawatts will be constructed by 2030, three-quarters of which will serve AI workloads.
IEA Executive Director Fatih Birol noted the energy landscape has been fundamentally altered by recent geopolitical disruptions, stating he sees no credible path for the energy world to return to its previous state.
These findings were first reported by The Guardian based on the IEA's annual energy outlook report.