# AI energy consumption

> Source: https://aiwiki.ai/wiki/ai_energy_consumption
> Updated: 2026-06-23
> Categories: AI Energy, AI Infrastructure
> From AI Wiki (https://aiwiki.ai), a free encyclopedia of artificial intelligence. Quote with attribution.

**AI energy consumption** is the electricity, and the associated water, land, and emissions, required to train and operate artificial intelligence systems, principally [generative AI](/wiki/generative_ai) and [large language models](/wiki/large_language_model) running in [data centers](/wiki/data_center). The single most-cited figure comes from the International Energy Agency, which projected in April 2025 that global data-center electricity demand would more than double to about 945 terawatt-hours (TWh) by 2030, with AI the leading driver, roughly as much electricity as Japan uses today.[1][2] After roughly fifteen years of nearly flat data-center electricity use, the build-out of AI accelerators since about 2022 has reversed that trend, turning compute infrastructure into one of the fastest-growing sources of new electricity demand in the United States and several other advanced economies, and a central question for [AI infrastructure](/wiki/ai_infrastructure) planning. The topic is genuinely contested: headline figures vary by an order of magnitude depending on what is counted, the scenarios are forward projections rather than measurements, and the companies that hold the most accurate data disclose the least. This article surveys the demand side (how much, and the split between training and operation), the supply side (the grid, water, carbon, and the scramble for new generation), and the principal disputes.

"Global electricity demand from data centres is set to more than double over the next five years, consuming as much electricity by 2030 as the whole of Japan does today," said IEA Executive Director Fatih Birol on releasing the report.[2]

## How much electricity does AI use?

Two reference points anchor most discussion. The International Energy Agency, in its April 2025 *Energy and AI* report, estimated that data centers consumed about 415 terawatt-hours (TWh) in 2024, roughly 1.5 percent of world electricity, and projected this would more than double to around 945 TWh by 2030 in its Base Case, just under 3 percent of global electricity, reaching about 1,200 TWh by 2035.[1] The IEA attributes data centers to roughly one-tenth of global electricity demand growth to 2030, but more than 20 percent in advanced economies.[2] For the United States specifically, the Lawrence Berkeley National Laboratory (LBNL) and US Department of Energy *2024 United States Data Center Energy Usage Report* (Shehabi et al., December 2024) found US data centers used about 176 TWh in 2023, or 4.4 percent of national electricity, up from 58 TWh in 2014, and projected a range of 325 to 580 TWh by 2028, equivalent to 6.7 to 12 percent of US electricity.[3][4] The report noted that US data-center power demand more than doubled between 2017 and 2023, largely because of AI servers.[3]

Other bodies project higher. The Electric Power Research Institute (EPRI), in updated 2025 scenarios, estimated US data centers could consume 9 to 17 percent of national electricity generation by 2030, an upward revision of roughly 60 percent from its 2024 work.[6] Goldman Sachs Research forecast in 2025 that global data-center power demand would rise about 50 percent by 2027 and as much as 165 percent by 2030 relative to 2023.[7] The wide spread between these figures reflects genuine uncertainty about chip shipments, utilization, and how fast AI adoption translates into sustained load.

| Source (publication date) | Geography | Baseline | Projection | Key assumptions |
|---|---|---|---|---|
| IEA, *Energy and AI* (Apr 2025) | Global | 415 TWh / ~1.5% (2024) | ~945 TWh / ~3% by 2030; ~1,200 TWh by 2035 | Base Case; AI the leading driver; announced policies and projects |
| LBNL / DOE (Dec 2024) | United States | 176 TWh / 4.4% (2023) | 325 to 580 TWh / 6.7 to 12% by 2028 | Range spans low to high AI accelerator growth and utilization |
| EPRI, *Powering Intelligence* (updated 2025) | United States | ~4% (2023) | 9 to 17% of generation by 2030 | Scenario-based; ~60% above EPRI's 2024 estimate |
| Goldman Sachs Research (2025) | Global / US | 2023 levels | +50% by 2027; +165% by 2030 | US data-center share of power demand roughly doubling from ~4% |
| MIT Technology Review (May 2025) | United States | n/a | AI to use >50% of data-center electricity by 2028 | Drawing on LBNL data; AI alone ~equivalent to 22% of US households |

A recurring caveat applies to all of these: they are projections built on assumptions, not observed outcomes, and several pre-date the largest 2025 and 2026 capacity announcements. They should be read as ranges, not point estimates.

## Does training or inference use more energy?

AI energy splits into training (the one-time cost of building a model) and inference (the recurring cost of running it for users). Training a frontier model is a large, concentrated draw. The most widely cited measured figure is for [GPT-3](/wiki/gpt-3): Patterson and colleagues estimated in 2021 that training the 175-billion-parameter model consumed about 1,287 megawatt-hours (MWh) of electricity and emitted roughly 502 metric tons of CO2-equivalent, comparable to the annual emissions of about 112 gasoline cars.[8][21] For later models the numbers are press estimates rather than disclosures: training [GPT-4](/wiki/gpt-4) has been reported at roughly 50 gigawatt-hours, described as enough to power San Francisco for about three days, though leading labs such as [OpenAI](/wiki/openai) do not publish exact figures.[8]

The more important point for the grid is that, in aggregate, inference now dominates. MIT Technology Review's May 2025 analysis concluded that inference accounts for roughly 80 to 90 percent of AI computing power, a share expected to grow as products embed AI into search, productivity software, and agents.[8] Other analysts using a different accounting put inference nearer 60 percent of the AI energy footprint, with training near 30 percent and fine-tuning the remainder; either way, serving models, not building them, is the larger and faster-growing load.

## How much energy does one ChatGPT query use?

Per-query figures became a flashpoint in 2025. [Epoch AI](/wiki/epoch_ai) reanalyzed the question in February 2025 and concluded that "a GPT-4o query consumes around 0.3 watt-hours for a typical text-based question," derived from assuming roughly one second of Nvidia H100 time per query, about 1,500 watts per GPU, and a 70 percent power-utilization factor.[22] That is about ten times lower than the widely circulated earlier estimate of roughly 2.9 to 3 watt-hours per ChatGPT query, which the researcher Alex de Vries had popularized in 2023 from a SemiAnalysis figure assuming far longer prompts.[22] Companies later published their own numbers in the same range: in August 2025 [Google](/wiki/google) released a methodology stating that the median Gemini text prompt uses about 0.24 watt-hours of energy, emits 0.03 grams of CO2-equivalent, and consumes about 0.26 milliliters of water,[9] and OpenAI's Sam Altman wrote in June 2025 that an average ChatGPT query uses about 0.34 watt-hours.[11]

The gap is largely about scope: company and Epoch figures tend to report a marginal or median text request and may exclude idle capacity, networking, and the embodied energy of hardware, while measured open-model results vary enormously by model size and modality. Epoch itself cautioned that long inputs and reasoning models cost far more, with a 100,000-token prompt approaching 40 watt-hours.[22]

| Task (source) | Energy estimate | Notes |
|---|---|---|
| Typical GPT-4o text query (Epoch AI, Feb 2025) | ~0.3 Wh | 1 s H100-time, 1,500 W, 70% utilization |
| Median Gemini text prompt (Google, Aug 2025) | 0.24 Wh | Plus 0.03 gCO2e and 0.26 mL water |
| Average ChatGPT query (Altman, Jun 2025) | 0.34 Wh | Company-stated average |
| Earlier ChatGPT estimate (de Vries, 2023) | ~2.9 to 3 Wh | Often cited; assumed long prompts; methodology disputed |
| Llama 3.1 8B response (MIT TR, measured) | ~0.03 Wh | Small open model, GPU plus overhead |
| Llama 3.1 405B response (MIT TR, measured) | ~1.9 Wh | Large open model |
| 5-second AI video (MIT TR, measured) | ~940 Wh | Hundreds of times an image; modality matters most |

## Why is the power grid the bottleneck?

The binding constraint is increasingly not chips but interconnection: the ability to connect new load and new generation to the grid. US interconnection queues held on the order of 2,600 gigawatts of proposed generation and storage in early 2026, far more than will be built, with multi-year waits. In Texas, the ERCOT large-load queue ran to roughly 410 GW, the majority of it data centers. Long-lead equipment compounds the delay: lead times for large power transformers stretched from roughly two years before 2020 to about five years, and combined-cycle gas turbine deliveries from the major OEMs, including [GE Vernova](/wiki/ge_vernova), pushed out to five to seven years against multi-year order backlogs.[19] These bottlenecks, more than electricity prices, are what slow projects and have driven the [AI data center moratorium](/wiki/ai_data_center_moratorium) debates in several US localities.

Siting matters as much as scale. Because operators cluster facilities where land, fiber, and permits align, the load lands unevenly on regional grids, and MIT Technology Review reported, drawing on academic work, that data-center electricity carries about 48 percent higher carbon intensity than the US grid average, partly because much of it sits on gas-heavy or coal-heavy systems.[8] The result is local: rising wholesale prices, deferred plant retirements, and ratepayer disputes over who pays for new transmission.

## How much water does AI use?

Cooling high-density AI racks consumes water, both directly through evaporative cooling at the facility and indirectly through thermoelectric power generation upstream. Direct figures are sparse because disclosure is voluntary. [Google](/wiki/google) reported using more than 5 billion gallons of water across its data centers in 2023, with a meaningful share drawn from water-stressed watersheds. Academic work by Shaolei Ren and colleagues at UC Riverside estimated that a short ChatGPT conversation (on the order of 20 to 50 exchanges, GPT-3 era) could correspond to roughly 500 milliliters of water once upstream generation is included, and projected global AI-related water withdrawal of 4.2 to 6.6 billion cubic meters by 2027.[12] Industry analysts have put 2025 AI data-center water use near 1 trillion liters, but such top-down estimates carry wide error bars.

The engineering response is to take water out of the loop. [Microsoft](/wiki/microsoft) has deployed closed-loop, zero-water-evaporation cooling at its [Fairwater](/wiki/microsoft_fairwater) campus in Mount Pleasant, Wisconsin, which the company says avoids more than 125 million liters of evaporative water per facility each year, and is researching in-chip [microfluidic cooling](/wiki/microsoft_microfluidic_cooling) to remove heat closer to the silicon. Direct-to-chip liquid cooling is becoming standard for the densest [Nvidia](/wiki/nvidia) GPU racks, where air cooling no longer suffices. As with PUE for energy, the industry tracks water usage effectiveness (WUE), though closed-loop designs can shift consumption from on-site water to additional electricity.

## How is AI affecting carbon emissions?

AI growth has visibly strained hyperscaler climate commitments. Google reported that its greenhouse-gas emissions rose about 48 percent between 2019 and its 2024 reporting, driven by data-center energy and supply-chain emissions, and in that report it stopped describing itself as maintaining operational carbon neutrality and reframed its targets as ambition-based.[20] Microsoft reported fiscal-2024 emissions roughly 23 percent above its 2020 baseline, again attributing the rise to AI and cloud expansion, while characterizing its carbon-negative-by-2030 goal as a marathon.[20] Both companies continue to sign large clean-energy contracts; Microsoft said it procured about 19 GW of new renewable capacity in 2024. The tension is structural: emissions are rising in absolute terms even as the companies buy record volumes of carbon-free power, because demand is growing faster than clean supply can be added.

## Is AI getting more efficient?

Efficiency has improved on two fronts. Facility overhead, measured by power usage effectiveness ([PUE](/wiki/pue), the ratio of total facility energy to IT energy), has fallen sharply at the leaders: Google reported a fleet-wide PUE of about 1.09 in 2024 against an industry average near 1.56 from the Uptime Institute, and other hyperscalers report similar figures, though the global fleet remains far less efficient.[10] Chip efficiency has also risen rapidly, with each GPU generation delivering more computation per watt. These gains are real, but they reduce energy per unit of work, not total energy, which is the crux of the rebound debate below.

## How is the AI power supply being built (nuclear and gas)?

The supply scramble has been the most visible part of the story since late 2024, and [nuclear power for AI](/wiki/nuclear_power_ai) has become its emblem. In September 2024 [Microsoft](/wiki/microsoft) signed a 20-year power-purchase agreement with [Constellation Energy](/wiki/constellation_energy) to restart Unit 1 at Three Mile Island, rebranded the Crane Clean Energy Center, providing about 835 MW; in November 2025 the US Department of Energy advanced a roughly $1 billion loan to support the restart, targeted for around 2027.[13][14] [Google](/wiki/google) agreed in October 2024 to buy power from small modular reactors (SMRs) built by Kairos Power, up to about 500 MW across six to seven units with a first deployment near 2030.[15] Amazon led investments of more than $500 million in X-energy and signed SMR agreements in Washington and Virginia, alongside a data-center campus next to the Susquehanna nuclear plant.[16] Meta issued a request for proposals for 1 to 4 GW of new nuclear, signed a 20-year agreement with Constellation to extend the 1.1 GW Clinton plant in June 2025, and contracted with [TerraPower](/wiki/terrapower), [Oklo](/wiki/oklo), and Vistra around its Ohio build-out.[17]

Nuclear, however, mostly arrives late in the decade. The near-term workhorse is natural gas, despite turbine backlogs, supplemented by renewables, batteries, and on-site generation.[19] Fuel-cell maker [Bloom Energy](/wiki/bloom_energy) has won data-center deals for behind-the-meter power that sidesteps the interconnection queue, and power-first developers such as [Crusoe Energy](/wiki/crusoe_energy) build generation and compute together. The practical outcome through about 2028 is a mix that leans on gas while clean firm capacity is contracted for the 2030s.

## What are the main debates about AI energy use?

Three disputes run through the subject. The first is measurement: per-query and per-facility figures depend heavily on system boundaries, so a 0.24 watt-hour median prompt and a multi-watt-hour large-model query can both be true for different questions, and critics argue that selective company disclosures risk understating the full footprint. The second is the rebound effect, or [Jevons paradox](/wiki/jevons_paradox): efficiency that makes inference cheaper tends to expand usage, so falling energy per query can coincide with rising total energy, an effect documented for AI in 2025 academic work.[18] The third is the projections themselves, which assume continued exponential adoption; a slower-than-expected return on AI investment, or faster algorithmic efficiency, could leave forecasts high, while a faster build-out could leave them low. What is not seriously disputed is the direction: after a long plateau, AI has made data centers a material and growing claim on electricity, water, and the grid, and the size of that claim through 2030 remains an open, actively contested question.

## References

1. International Energy Agency, *Energy and AI* (Executive summary), April 2025. https://www.iea.org/reports/energy-and-ai/executive-summary
2. International Energy Agency, "AI is set to drive surging electricity demand from data centres," 10 April 2025. https://www.iea.org/news/ai-is-set-to-drive-surging-electricity-demand-from-data-centres-while-offering-the-potential-to-transform-how-the-energy-sector-works
3. Shehabi, A., et al., *2024 United States Data Center Energy Usage Report*, Lawrence Berkeley National Laboratory / US DOE, December 2024. https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf
4. Berkeley Lab News Center, "Berkeley Lab Report Evaluates Increase in Electricity Demand from Data Centers," December 2024. https://newscenter.lbl.gov/2025/01/15/berkeley-lab-report-evaluates-increase-in-electricity-demand-from-data-centers/
5. US Department of Energy, "DOE Releases New Report Evaluating Increase in Electricity Demand from Data Centers." https://www.energy.gov/articles/doe-releases-new-report-evaluating-increase-electricity-demand-data-centers
6. EPRI, "Data Centers Could Consume up to 9% (updated 9 to 17%) of US Electricity by 2030," 2024 and updated 2025 *Powering Intelligence* scenarios. https://www.epri.com/about/media-resources/press-release/q5vu86fr8tkxatfx8ihf1u48vw4r1dzf
7. Goldman Sachs Research, "AI to drive 165% increase in data center power demand by 2030," 2025. https://www.goldmansachs.com/insights/articles/ai-to-drive-165-increase-in-data-center-power-demand-by-2030
8. MIT Technology Review, "We did the math on AI's energy footprint. Here's the story you haven't heard," 20 May 2025. https://www.technologyreview.com/2025/05/20/1116327/ai-energy-usage-climate-footprint-big-tech/
9. Google Cloud, "Measuring the environmental impact of AI inference," August 2025. https://cloud.google.com/blog/products/infrastructure/measuring-the-environmental-impact-of-ai-inference/
10. Google, "Efficiency" (data-center PUE). https://datacenters.google/efficiency/
11. Sam Altman, "The Gentle Singularity," 10 June 2025. https://blog.samaltman.com/the-gentle-singularity
12. UC Riverside News, "AI programs consume large volumes of scarce water" (Ren et al., *Making AI Less Thirsty*). https://news.ucr.edu/articles/2023/04/28/ai-programs-consume-large-volumes-scarce-water
13. Utility Dive, "Constellation plans 2028 restart of Three Mile Island unit 1, spurred by Microsoft PPA." https://www.utilitydive.com/news/constellation-three-mile-island-nuclear-power-plant-microsoft-data-center-ppa/727652/
14. CNBC, "Trump administration backs Three Mile Island nuclear restart with $1 billion loan to Constellation," 18 November 2025. https://www.cnbc.com/2025/11/18/trump-nuclear-three-mile-island-crane-loan-constellation-ceg.html
15. DatacenterDynamics, "Google signs nuclear SMR deal with Kairos for data center power," October 2024. https://www.datacenterdynamics.com/en/news/google-signs-nuclear-smr-deal-with-kairos-for-data-center-power/
16. Utility Dive, "Amazon announces small modular reactor deals with Dominion, X-energy, Energy Northwest." https://www.utilitydive.com/news/amazon-small-modular-reactor-deals-nuclear-dominion-x-energy-energy-northwest/730022/
17. Meta Sustainability, "Accelerating the Next Wave of Nuclear to Power AI Innovation," December 2024. https://sustainability.atmeta.com/blog/2024/12/03/accelerating-the-next-wave-of-nuclear-to-power-ai-innovation/
18. "From Efficiency Gains to Rebound Effects: The Problem of Jevons' Paradox in AI's Polarized Environmental Debate," ACM FAccT 2025 (arXiv:2501.16548). https://arxiv.org/abs/2501.16548
19. Turbomachinery Magazine, "Surging Gas Turbine Demand Fueled by Data Center, AI Growth," 2025. https://www.turbomachinerymag.com/view/surging-gas-turbine-demand-fueled-by-data-center-ai-growth
20. Fortune, "Big tech was embracing clean energy and turning a corner on climate change. Then AI data centers arrived," 2026. https://fortune.com/2026/03/29/big-tech-climate-change-goals-data-centers-ai-fossil-fuels/
21. Patterson, D., et al., "Carbon Emissions and Large Neural Network Training," 2021 (arXiv:2104.10350). https://arxiv.org/abs/2104.10350
22. Epoch AI, "How much energy does ChatGPT use?", 11 February 2025. https://epoch.ai/gradient-updates/how-much-energy-does-chatgpt-use