A recent study published in the journal Nature Sustainability indicates that 403 hyperscale data centers in the United States are projected to consume between 68 and 99 terawatt-hours (TWh) of electricity from May 2024 to April 2025. This consumption accounts for 1.5% to 2.2% of the total US electricity usage during that period.
This level of electricity use is comparable to the annual consumption of a midsized country or US states such as Wisconsin or Indiana. The study focuses on hyperscale facilities, which are large structures being built to support technologies like generative AI.
The research highlights that fossil fuels remain the largest source of electricity for these hyperscale data centers. This reliance on fossil fuels contributes significantly to greenhouse gas emissions, raising concerns about the environmental footprint of the tech industry.
The findings suggest that considerable effort is needed to reduce pollution associated with the energy demands of AI and other data services. The study emphasizes that the environmental impact of computing infrastructure varies depending on the supporting electricity system.
The study provides a clear point of reference for tracking the energy consumption and estimated emissions of hyperscale data centers moving forward. Researchers were unable to differentiate between energy used for AI compute and more traditional data services within these facilities.
The team compiled a list of 403 data centers from an initial 675 facilities, verifying those with reported electrical capacity and suitable infrastructure. The average electrical capacity of these verified centers is approximately 40 megawatts.
The rapid expansion of hyperscale data centers, driven in part by the demand for large language models, has led to challenges in planning for clean and minimally impactful electricity supply. The study underscores the importance of understanding and addressing the energy demands of this growing sector.
The findings serve as a foundational estimate for the energy use and carbon pollution generated by these critical infrastructure components, informing future discussions on sustainability within the tech industry.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's briefNew every morning, and the back catalogue is archived by date.
The International Energy Agency (IEA) projects global data center power capacity will double to 240GW by 2030, with 75% of new capacity dedicated to AI. This expansion, including 93 large AI factories, will add significant strain to electricity grids, equivalent to the power demand of an entire Great Britain.
The International Energy Agency (IEA) projects that global electricity consumption by data centers could double by 2030, reaching approximately 1,000 terawatt-hours annually. This increase, combined with tight supply chains and limited grid capacity, could strain power grids and delay new data center projects. The IEA's Stated Policies Scenario indicates that significant grid upgrades and expansions are necessary to prevent congestion and ensure reliability.
The International Energy Agency (IEA) reports that global datacenters will increase carbon emissions by 70% by 2030, driven by a doubling of electricity consumption, with AI accounting for three-quarters of this growth. This surge is attributed to increased reliance on fossil fuels, particularly coal in China and gas in the US, to power new AI datacenters.
A Harvard-led study estimates 403 hyperscale U.S. data centers consumed 82 terawatt-hours (TWh) of electricity in one year, representing 1.8% of the national total. This consumption generated an estimated 25.7 million metric tons of CO2, with over half the power derived from gas and coal sources. The findings highlight the significant energy footprint and carbon emissions of large-scale data infrastructure.
A new study estimates that 403 hyperscale data centers in the US will consume between 68 and 99 terawatt-hours of electricity from May 2024 to April 2025. This consumption represents 1.5% to 2.2% of total US electricity use, equivalent to the usage of states like Wisconsin or Indiana. The study provides a baseline for tracking energy consumption by these facilities, including those supporting AI compute.
A new study in Nature Sustainability reports that 403 US hyperscale data centers consumed an estimated 82 terawatt-hours of electricity, or 1.8 percent of total US consumption, between May 2024 and April 2025. This consumption level exceeds the annual electricity use of 148 individual nations, with fossil fuels remaining the primary power source, contributing significantly to greenhouse gas emissions.