AI’s Climate and Energy FootprintAI’s Climate and Energy FootprintCoverage from MIT Sloan School of Management, Grist, and others
00/00/0000
DailyWeekly
Expanding AI requires data centers, chips, storage, and cooling, adding electricity demand whose climate impact depends on efficiency and the power grid.
The International Energy Agency estimated data centers used about 415 terawatt-hours of electricity worldwide in 2024. Research also examines emissions enabled when AI boosts oil and gas production, alongside potential benefits from clean-energy applications; the net effect is uncertain and depends on how these technologies are deployed. Choices about new generation, grid investment, corporate emissions accounting, and climate policy help determine both the environmental costs and who bears them.
Its modeling explains why potential AI efficiency gains do not automatically reduce warming without policies that limit rebound effects.
MIT Sloan School of Management / Dylan Walsh
Key Issues
01
Data-center growth is adding power demand to fossil-heavy, constrained grids
Expanding AI workloads are driving substantial data-center electricity use, while clean-power expansion and grid connections struggle to keep pace. Natural gas is a major source for U.S. data centers, linking near-term growth to emissions even as firms pursue clean-energy procurement.
Drawn from 6 articles
02
Rising technology emissions are straining corporate climate targets
Several large technology firms report rising electricity use and emissions amid AI infrastructure expansion, complicating progress toward climate commitments. The footprint also extends beyond operations to construction and other supply-chain emissions, which clean-energy purchases do not necessarily resolve.
Drawn from 6 articles
03
AI-enabled fossil-fuel production adds emissions outside conventional tech accounting
Oil and gas companies are applying AI to exploration and production, potentially increasing fossil-fuel output and emissions beyond the power used to run data centers. Reporting also highlights an accounting gap: these enabled emissions can remain outside technology companies’ reported climate footprints.
Drawn from 6 articles
04
AI’s climate benefits remain conditional and do not automatically offset its costs
Potential gains from AI applications in areas such as clean energy are presented as dependent on responsible deployment and policy choices, not as an automatic counterweight to AI’s energy use. Modeling and reporting also warn that efficiency gains can be offset by increased demand, while some claimed emissions reductions lack independent verification.
Drawn from 5 articles
Key Numbers
1.5% of global electricity consumption
data-centre share of global electricity use
2024
“The International Energy Agency estimated that data centres consumed about 415 terawatt-hours of electricity in 2024, approximately 1.5% of global electricity consumption. A typical text-based GPT-4o query has been estimated at around 0.3 watt-hours, but billions of queries, combined with training, image and video generation, and AI agents, create much larger aggregate demand.”
Gulf News
415 terawatt-hours
data-centre electricity consumption
2024
“The International Energy Agency estimated that data centres consumed about 415 terawatt-hours of electricity in 2024, approximately 1.5% of global electricity consumption. A typical text-based GPT-4o query has been estimated at around 0.3 watt-hours, but billions of queries, combined with training, image and video generation, and AI agents, create much larger aggregate demand.”
Gulf News
more than $420 billion USD
Data-center infrastructure spending by Amazon, Google, Meta and Microsoft
Four companies · 2025
“Amazon, Google, Meta and Microsoft spent more than $420 billion on data center infrastructure in 2025 and are expected to spend $745 billion in 2026. McKinsey estimates global data center construction through 2030 could reach $7 trillion, with 15% to 20% directed to power systems. By 2030, data centers could consume 3% to 5% of global electricity.”
Heatmap
32 million possible battery chemistries
Battery chemistries screened
“Microsoft’s Azure Quantum Elements screened 32 million possible battery chemistries in 80 hours and identified compounds that could reduce lithium requirements by as much as 70%.”
Heatmap
80 hours
Time taken to screen the battery chemistries
“Microsoft’s Azure Quantum Elements screened 32 million possible battery chemistries in 80 hours and identified compounds that could reduce lithium requirements by as much as 70%.”
Heatmap
Looking Back
269 Day Timeline
Articles published over time. Hover any bar for the period and its article count.
Jan 21
Mar 4
Apr 22
Jun 3
Jul 15
Sep 2
Oct 14
The Story So Far
No material change
No new-member articles were supplied, so there is no new evidence of a material change in the Topic.
Previously
Expanding AI use is increasing electricity demand from data centers and adding emissions across technology supply chains, complicating major firms’ climate commitments. AI tools are also being applied to fossil-fuel exploration and production, extending the potential climate impact beyond the power needed to run data centers. Clean-energy procurement and efficiency improvements are growing, but their effects must be considered alongside rising overall demand and emissions; the material also includes more limited examples of AI used for climate planning and adaptation.
History
10/14/2026
The story now gives greater weight to data centers’ water and local infrastructure impacts, alongside electricity use. Investor attention to firm power and infrastructure, and proposals for disclosure of AI-related resource use and enabled emissions, add market and accountability dimensions.
10/05/2026
The current version more explicitly says Google, Microsoft, Amazon, and Meta have reported emissions increases despite growing clean-energy purchases. This strengthens the corporate-emissions framing but does not materially change the story’s overall direction.
In a forthcoming climate-resilience framework, John Sterman and Jennifer Turliuk used En-ROADS to model AI's climate effects, finding economic rebound could outweigh efficiency gains and increase warming by 2100.
Former Microsoft sustainability workers Will Alpine and Holly Alpine used a macroeconomic model in new npj Climate Action research to estimate global emissions increases from AI-enabled oil and gas productivity.
AI developers and data-centre operators are increasing electricity and cooling demand worldwide as adoption expands, while cleaner grids and efficiency measures could limit climate impacts.
Researchers from Enabled Emissions Campaign and Purdue University reported that artificial intelligence could raise global emissions by increasing oil and gas productivity across modeled scenarios.
8/11/2026 • Markets & Industry & Innovation • General
A Claude Code user found during an eight-week analysis that agentic AI workloads consumed far more electricity and emissions on U.S. data-center grids than simple chatbot prompts.
In recent years, Aramco, Chevron, Equinor, ExxonMobil, Shell, and TotalEnergies have used AI to improve fossil fuel production across global operations, potentially increasing emissions as data-center electricity demand supports natural-gas use.
the article states that data center electricity demand in the united states could rise 60 to 80 percent by 2050, and policy choices could influence emissions and electricity costs
In 2026, climate-focused startups and investors pursued AI applications for emissions reduction and scientific discovery globally, while funding remained far below general AI investment.
9/22/2026 • Markets & Industry & Innovation • General
A recent peer-reviewed study found that oil companies using custom AI tools could create substantially more climate pollution than AI data centers, prompting calls for enabled-emissions disclosure across global fossil-fuel markets.
9/9/2026 • Markets & Industry & Innovation • General
Google reported a 37% 2025 electricity increase from AI data centers while operational emissions fell 2% due to clean-energy procurement and 24/7 carbon-free accounting.
During Climate Week NYC in New York, an August Nature-journal study found AI data-center power demand and fossil-fuel optimization produced more emissions than clean-energy AI applications saved.
Watershed proposed AI emissions estimation methods for token- and spend-based corporate reporting as California and the European Union develop AI energy disclosure requirements.
Yassine Charabi models AI data-center growth and chip manufacturing emissions, projecting delayed net climate benefits until late 2031 due to cumulative carbon debt.
Google reported 2025 electricity growth and rising Scope 3 emissions in 2026, citing faster AI infrastructure expansion than grid decarbonization in Asia.
7/4/2026 • Markets & Industry & Innovation • General
António Guterres proposed an AI Environmental Transparency Initiative in London, urging AI companies to disclose data-center emissions and switch to renewable electricity by 2030.
6/23/2026 • Policy, Politics & Governance • General
UK government revised Compute Roadmap figures in 2026, raising estimated AI data center greenhouse gas emissions to 123 MtCO2 and triggering parliamentary and campaign scrutiny over 2030 clean power and 2050 net zero.
4/24/2026 • Policy, Politics & Governance • General
United Nations University researchers estimate that expanding artificial intelligence will double data-center energy use, water use, and pollution by 2030.
6/3/2026 • Markets & Industry & Innovation • General
During a recent episode of The Ezra Klein Show, Nvidia CEO Jensen Huang argued that growing AI data-center demand requires more US electricity generation, including near-term fossil-fuel power.
Ketan Joshi and a nonprofit-commissioned analysis released during the AI Impact Summit in New Delhi this week found tech industry claims that AI reduces emissions conflate ML and generative AI and lack verification.
2/17/2026 • Markets & Industry & Innovation • General
In 2025, analysts examined how artificial-intelligence infrastructure, critical minerals, trade restrictions, and debt are affecting energy-transition prospects in the United States, China, Japan, and global markets.
Will Alpine, Nathan Geldner, Holly Alpine and Maksym G. Chepeliev found in a global energy-economy model that simultaneous AI adoption could increase emissions by improving fossil extraction faster than renewable generation.
Clara Vondrich links US hyperscale data-center expansion for AI to fossil-fueled electricity, methane warming, and mounting state regulation after New Yorks one-year ban.
At Climate Week NYC on September 21, researchers and energy companies discussed AI's climate effects, including data-center electricity demand, fossil-fuel use, and possible utility costs.
Analysts report that AI-driven electricity demand is making it difficult for major tech firms to meet 2030 greenhouse-gas goals, increasing reliance on natural gas for data centers.
Microsoft, Google, and Amazon face questions in August about whether custom AI tools supporting global oil and gas production conflict with climate commitments after research linked enabled emissions to substantially higher pollution.
8/19/2026 • Markets & Industry & Innovation • General
Holly Alpine, Will Alpine, Nathan Geldner, and Maksym Chepeliev reported in npj Climate Action that AI-enabled fossil fuel production could exceed data-center emissions by up to 13.3 times.
8/12/2026 • Markets & Industry & Innovation • General
Microsoft, Amazon, and Google reported higher emissions in new energy and emissions data as AI-driven electricity demand expanded in the United States.
7/10/2026 • Markets & Industry & Innovation • General
In 2025 greenhouse gas reports, Amazon and Google tied rising emissions to expanded AI data centers and electricity demand, with impacts across US power and supply chains.
7/1/2026 • Markets & Industry & Innovation • General
Microsoft may revise clean energy targets in 2025 after AI-driven data center expansion strains renewable sourcing, with local impacts raised in Vineland, New Jersey.
5/8/2026 • Markets & Industry & Innovation • General
Technology companies building AI data centers face electricity constraints as reported emissions rise, with natural gas supply and policy uncertainty complicating 2030 climate goals in the U.S. during 2024-2025.
4/5/2026 • Markets & Industry & Innovation • General
Google and Amazon reported higher sustainability emissions during ongoing AI data-center expansion, with Scope 3 increases driven by purchased hardware and capital goods.
7/2/2026 • Markets & Industry & Innovation • General
Google, Microsoft, Amazon, and other AI companies reported differing 2025 data-center emissions and resource impacts as electricity demand expanded globally.
Climate-tech investors at Climate Week in New York City discussed how AI-driven data-center demand is shifting financing toward firm power and infrastructure.
9/24/2026 • Markets & Industry & Innovation • General
Levine argues that AI data centers relying on fossil-fuel power can compound climate-related health harms in the United States, citing a Loudoun County facility's estimated annual air-pollution damages of $53 million to $99 million.
António Guterres urged data centers worldwide to run on renewables by 2030 as electricity demand from AI is projected to surge and grid emissions accounting remains contested.
7/28/2026 • Clean Energy & Emissions • General
LocalSYR / Seth Borenstein and Jennifer McDermott53
World and business leaders in New York addressed rising climate damages and energy costs as renewable additions reached records but fossil fuel growth persisted.
9/21/2026 • Clean Energy & Emissions • General
Mercury News / Seth Borenstein and Jennifer McDermott53
World and business leaders meeting in New York during Climate Week in 2025 assessed rising climate damages and energy demand while urging faster renewable deployment and warning about fossil-fuel dependence.
Global leaders opened Climate Week in New York during the United Nations General Assembly, urging faster climate action while addressing renewable growth and AI-related environmental pressures.
9/21/2026 • Clean Energy & Emissions • General
WHEC.com / Seth Borenstein and Jennifer McDermott53
During New York Climate Week, global leaders and energy organizations discussed faster renewable deployment and AI data-center impacts amid breached planetary boundaries and a 2025 fossil-fuel rebound.
9/21/2026 • Clean Energy & Emissions • General
AP News / Seth Borenstein and Jennifer McDermott53
Global leaders opened a week of climate talks on Monday in New York, addressing planetary risks, renewable deployment, fossil fuel use, and AI data center impacts.
Global leaders and business executives began New York Climate Week on Monday in New York, debating renewable deployment, fossil fuel dependence, and AI data-center impacts as climate risks intensified.
Microsoft, Amazon, and Google reported higher emissions as U.S. AI data centers expanded power demand, with Scope 3 capital goods driving large carbon totals in the latest year.
The International Energy Agency reports that global data-center and network electricity demand could nearly double by 2030, increasing digital carbon emissions worldwide.
In 2026, climate analysts and UK government advisers warned that AI data centres consume substantial energy and water, while food and travel generally create larger personal carbon footprints.
Google and Meta reported rising emissions during AI-driven data-center expansion in 2025 and 2024, challenging net-zero commitments despite increased clean-energy procurement.
7/31/2026 • Markets & Industry & Innovation • General
Microsoft, Amazon, and Google reported rising emissions linked to AI data-center growth in the United States, with scope 3 capital goods and supply chains driving most increases.
7/10/2026 • Markets & Industry & Innovation • General
António Guterres urged major AI companies at London Climate Action Week to disclose data center emissions, water, and land impacts and use renewables by 2030.
6/23/2026 • Markets & Industry & Innovation • General
United Nations University Institute for Water, Environment and Health quantified global AI data center carbon, water, and land footprints, projecting rising electricity and resource demands through 2030.
6/5/2026 • Markets & Industry & Innovation • General
Microsoft, Amazon, and Google reported higher CO2e emissions in the financial year ending March 2026, driven mainly by AI datacenter construction and supply-chain activity.
7/11/2026 • Markets & Industry & Innovation • General
Microsoft reported a 25% rise in greenhouse gas emissions in 2024-2025 as AI and data-center electricity demand expanded, alongside major growth in purchased electricity emissions.
7/9/2026 • Markets & Industry & Innovation • General
Cleanview, using planning documents and satellite surveillance, identified a methane gas power plant tied to Google data-center energy in Texas, prompting scrutiny over emissions goals.
Microsoft and Google sustainability reporting in fiscal year 2025 shows rising greenhouse gas emissions linked to AI data center electricity demand, using U.S. EIA 2024 state data comparisons.
7/14/2026 • Markets & Industry & Innovation • General
Microsoft, Amazon, Meta, and Alphabet increased 2026 AI infrastructure budgets, and the analysis says data-center power demand will accelerate firm low-carbon electricity build-out.
5/2/2026 • Markets & Industry & Innovation • General
Microsoft reported FY2025 emissions rose 25% year over year in 2026 sustainability reporting, citing AI data center expansion and reduced use of unbundled renewable energy certificates.
7/13/2026 • Markets & Industry & Innovation • General
Amazon reported a 16% greenhouse gas emissions increase in 2025 to about 81 million metric tons CO2e, citing data center construction and delivery fuel.
7/1/2026 • Markets & Industry & Innovation • General
António Guterres urged AI companies at London Climate Action Week to disclose data center carbon pollution and resource use and move to renewable electricity by 2030.
6/23/2026 • Policy, Politics & Governance • General
California will require large companies, including potentially Anthropic and OpenAI, to report operational emissions as AI data-center energy and environmental risks expand.
8/12/2026 • Economy, Business & Innovation • General
Anthropic and OpenAI are developing climate-reporting systems before potential public offerings as California's SB253 emissions requirements take effect in November.
8/12/2026 • Economy, Business & Innovation • General
Al Gore said AI data center electricity demand is a climate risk and urged clean power co-location rather than new data center restrictions, referencing Project Marvel in Bessemer, Alabama.
5/13/2026 • Policy, Politics & Governance • General
Bezos Earth Fund backs AI tools for climate and nature outcomes, as scrutiny links Amazon emissions growth in 2024 to AI data center expansion and transport.
3/24/2026 • Markets & Industry & Innovation • General
During New York Climate Week, Bill McKibben argued in New York that scrutiny of AI data centers should not displace emissions reduction and wider climate action.
9/22/2026 • Policy, Politics & Governance • General
AI Governance
Data Centers
Drones
Climate
Privacy
2026 Maine Senate
Bob Dylan
