environmental impact of AI
In plain wordsThe effects of AI on the planet: the electricity, water, materials and land used to make chips, build data centers and train and run models, and the pollution that results.
The resource use and environmental effects attributable to AI across its lifecycle: manufacturing chips and servers (embodied emissions, mineral extraction, water for fabrication); building and powering data centersAI data center, n.: A building or campus packed with AI chips, power equipment and cooling, built to train and run AI models. The biggest use as much electricity as a city. (electricity, still largely fossil-generated in many grids, and water for cooling); trainingtraining, n.: The process by which an AI model learns: it is shown huge numbers of examples and its internal settings are adjusted, bit by bit, until its answers get better. models; serving them at inferenceinference, n.: Drawing a conclusion from evidence. In today's AI it most often means running an already-trained model to get an answer, as opposed to training it. time, which for widely used systems usually outweighs training; and disposing of hardware. The International Energy Agency estimated that data centers used about 415 TWh of electricity in 2024, around 1.5% of global consumption, and projected about 945 TWh by 2030, with AI the main driver of growth. Per-query footprints are small and falling (providers report fractions of a watt-hour for a median text prompt), but aggregate demand is rising fast and is concentrated on particular grids and in some water-stressed regions. Indirect effects, harmful (such as aiding fossil-fuel extraction) and beneficial (such as grid optimization and materials discovery), may outweigh direct ones but are hard to measure.
The sustainability report counted training runs but left out inference, where most of the energy actually went.
Origin
Emma Strubell, Ananya Ganesh and Andrew McCallum's 'Energy and Policy Considerations for Deep Learning in NLP' (2019) made AI's carbon footprint a research topic; Roy Schwartz and colleagues' Green AIGreen AI, n.: A movement to make AI research and products use less energy and computing power, and to report those costs openly alongside accuracy. (2019) and Emily M. Bender and colleagues' 'stochastic parrots' paper (2021) carried it into wider debate. The data-center build-out from 2023 made energy and water central policy issues.
Usage note
Estimates vary widely with assumptions about hardware, utilization, grid mix and system boundaries, and many developers disclose little. The EU AI ActEU AI Act, prop. n.: The European Union's law on artificial intelligence, the first broad AI law anywhere: it bans some uses, sets strict checks for risky ones and puts duties on the makers of the biggest AI models. requires providers of general-purpose AI models to document known or estimated energy consumption. Efficiency gains need not reduce total use (see Jevons paradoxJevons paradox, n.: The observation that making something more efficient can make people use more of it overall because it gets cheaper — as with coal in the 1800s, and with AI computing today.).
Related
- AI data centerAI data center, n.: A building or campus packed with AI chips, power equipment and cooling, built to train and run AI models. The biggest use as much electricity as a city.
- Green AIGreen AI, n.: A movement to make AI research and products use less energy and computing power, and to report those costs openly alongside accuracy.
- Jevons paradoxJevons paradox, n.: The observation that making something more efficient can make people use more of it overall because it gets cheaper — as with coal in the 1800s, and with AI computing today.
- power usage effectivenesspower usage effectiveness, n.: A score showing how much of a data center's electricity actually reaches its computers: total power drawn divided by the computers' share. A perfect 1.0 would mean no waste; the extra goes mainly to cooling.
- computecompute, n.: The raw computing power used to train and run AI, usually measured by how many specialized chips are used and for how long. With data and algorithms, it is a main driver of AI progress.
- inferenceinference, n.: Drawing a conclusion from evidence. In today's AI it most often means running an already-trained model to get an answer, as opposed to training it.
Sources
- Strubell, E., Ganesh, A. & McCallum, A. (2019). Energy and Policy Considerations for Deep Learning in NLP. Proc. ACL 2019.
- International Energy Agency (2025). Energy and AI. IEA, Paris. iea.org/reports/energy-and-ai/executive-summary
- Luccioni, A. S., Viguier, S. & Ligozat, A.-L. (2023). Estimating the Carbon Footprint of BLOOM, a 176B Parameter Language Model. Journal of Machine Learning Research 24.
Mentioned in
AI data centerAI data center, n.: A building or campus packed with AI chips, power equipment and cooling, built to train and run AI models. The biggest use as much electricity as a city., inference costinference cost, n.: What it costs to use an AI model after it has been built: the computing power, electricity and hardware needed every time it answers, usually billed per chunk of text., Jevons paradoxJevons paradox, n.: The observation that making something more efficient can make people use more of it overall because it gets cheaper — as with coal in the 1800s, and with AI computing today., liquid coolingliquid cooling, n.: Cooling computer chips by running liquid past them instead of blowing air. AI chips now run so hot and are packed so tightly that air alone can no longer keep them cool., Green AIGreen AI, n.: A movement to make AI research and products use less energy and computing power, and to report those costs openly alongside accuracy., Kate CrawfordKate Crawford, prop. n.: An Australian scholar who studies the social, political and environmental costs of AI, co-founded the AI Now Institute, and wrote 'Atlas of AI'., Magnifica HumanitasMagnifica Humanitas, prop. n.: Pope Leo XIV's 2026 letter to the Catholic Church and the world about artificial intelligence, arguing that AI must serve human dignity and the common good., orbital data centerorbital data center, n.: A proposed computer center circling the Earth, powered by almost constant sunlight, that would run AI without using land, water or power grids on the ground., power usage effectivenesspower usage effectiveness, n.: A score showing how much of a data center's electricity actually reaches its computers: total power drawn divided by the computers' share. A perfect 1.0 would mean no waste; the extra goes mainly to cooling.