The Green AI Paradox
As artificial intelligence reshapes our world, its environmental toll grows. Yet it may also be our best hope for fighting climate change.
The servers humming in Microsoft’s and Google data centers consume as much electricity as some small nations. Yet inside these energy-hungry machines, artificial intelligence is also busy fighting for the planet: optimizing wind farms, predicting the risk of forest fires and monitoring methane leaks. AI as both environmental villain and potential savior is the green paradox at the heart the technology.
AI could reduce global greenhouse emissions by 5–10% by 2030 — equivalent to eliminating the European Union’s entire carbon output. But the technology’s own carbon footprint is rising at an alarming rate — its power demand is projected to rise by 160% by 2030. The tension between technological progress and environmental preservation is being exacerbated with the exponential offer and demand for all things AI.
Megawatt Appetite
The engine rooms of the AI revolution are data centres housed in bland warehouses; they power the servers that run every single ChatGPT query and cool them down using water. The numbers are sobering: a simple query to ChatGPT requires ten times the electricity of a Google search, while training a single large language model produces as much carbon as five cars across their entire lifespans. America’s data centres have ballooned from 500,000 in 2012 to over 8 million today. They gobble up 4.59% of the country’s electricity — twice what they consumed in 2018 and potentially 12% of America’s power supply by 2028. Accordingly, their carbon footprint has risen to 105 million metric tonnes of CO2, or 2.18% of America’s emissions by August 2024.
Location makes things worse. Fully 95% of American data centres are located in areas where the electricity grid runs particularly dirty — typically 48% dirtier than the national average, given their proximity to coal country. “Multimodal” AI models, which process video and images alongside text, are accelerating this trend, and “reasoning” AI models will precipitate it. Global AI energy demand could surge tenfold by 2026, surpassing Belgium’s annual electricity consumption.
The Hidden Costs of Intelligence
The environmental impact of AI goes far beyond beyond energy bills. By 2030, AI infrastructure could use six times more water than Denmark. The mining of rare earth elements for AI microchips often involves environmentally destructive practices. Electronic waste from hardware that rapidly becomes obsolete adds to the challenge.
Sadly, the technology is increasingly being deployed in ways that will accelerate the rate of environmental degradation: the fossil fuel industry harnesses AI to find new deposits, fast fashion companies deploy it to expand niche markets for short-lived apparel, and fishing fleets use it to optimize their operations, accelerating ocean depletion. More disturbingly, AI has already been weaponized to produce climate change misinformation which has eroded public support for climate action — causing major American banks and asset managers to retreat from global climate action networks as a new climate skeptic Trump presidency begins.
A Real Tool for Green Innovation
While AI is emerging as a significant contributor to carbon emissions, it’s almost impossible to ignore its potential role in countering climate change. The technology shows promise across three fronts: slashing emissions, boosting resilience to climate change and turbocharging research into green technologies. AI holds the immense potential to bring considerable environmental and economic value to various industries.
Consider buildings, traditionally among the world’s most profligate energy consumers and responsible for 11.7% of global emissions, AI-powered management systems are delivering electricity savings of up to 29%. Agriculture, another major carbon culprit accounting for 11.7% of global emissions, employs now clever algorithms that lift efficiency by up to 40% through precision farming. Supply chains are growing smarter, thanks to AI crunching vast quantities of data in order to slash inventory waste and optimize logistics through better forecasting. The impact ripples through to transport networks, where intelligent traffic management is cutting fuel waste from idling vehicles. Even the electricity grid itself is getting more intelligent, as smart grids are designed to manage the intermittent nature of solar and wind power.
Meanwhile, AI is transforming the way scientists understand climate change by processing massive datasets that improve climate models and predict extreme weather events — boosting the resilience of the affected communities. Interestingly, AI is accelerating the search for breakthrough green technologies. By quickly prototyping new materials, it is already making carbon capture more economical. In manufacturing, generative AI is inventing alternatives to synthetic chemicals and plastics, nudging industries towards a more circular economy.
The Policy Puzzle
Policymakers are finally waking up to this challenge. Last year, the EU’s passsed the AI Act, which focusses on AI governance and includes provisions to encourage sustainable practices — however these provisions are largely voluntary. “General purpose” AI models like ChatGPT and Gemini will be soon required to track and report the energy used during both AI training and operations. Ironically, AI has emerged as a critical tool for processing precisely that same complex environmental data, feeding a virtuous circle of data-driven environmental improvements made by firms on the road to decarbonisation.
In America, the landscape looks bleaker. The Biden Administration issued a last minute Executive Order focused on advancing “frontier AI infrastructure” while promoting clean energy solutions. However, the new Trump Administration has signalled a rollback of environmental policies — impacting future efforts to regulate AI’s environmental footprint. A consortium of tech giants has unveiled Stargate, a $100bn infrastructure initiative that could expand to $500bn by 2028. The venture aims to address America’s AI bottleneck, with environmental concerns relegated to the footnotes. The race for AI leadership against China will durably keep them in the footnotes.
The United Nations Environment Programme (UNEP) has highlighted concerns about AI’s environmental costs, but the goals remain aspirational and the recommendations non-binding.
Big Tech’s Small Steps
Tech giants are already feeling the squeeze: Microsoft’s emissions have swelled by 29% since 2020, while Google’s have jumped 48% since 2019, both driven by AI-related expansion. At the same time, they are discovering their green conscience beyond carbon credit shopping. Google’s DeepMind unit has cut data centre cooling costs by 40% using smart algorithms. IBM hails new chips that are 14 times more energy efficient than their predecessors. Nvidia’s emphasis on accelerated computing for energy efficiency sparks hope for sustainable AI.
Yet these efforts illustrate a deeper irony. Many business leaders express the intent to invest in AI to advance sustainability in their operations, yet their actions frequently fall short of these goals: while 90% of business leaders expect AI to boost sustainability, just 44% have acted on their words. This gap underscores the need for a more robust approach to align intentions with tangible outcomes.
The Greener It Gets
The path of AI follows a familiar paradox of progress: as it grows more efficient, it threatens to increase its environmental toll. Like coal and electricity before it, AI’s success in becoming cheaper may turn it into a commodity consumed at the click of a button and everywhere, undoing the very environmental gains it promises. Silicon Valley’s response is telling. Tech giants’ quiet pivot to nuclear power, after years of championing solar and wind, suggests they grasp the scale of AI’s energy appetite, if not its implications.
Meanwhile, as AI’s carbon footprint grows, sustainability has emerged as a recognised source of enterprise value creation. But without a fundamental rethink of how AI is developed and deployed, the industry risks winning a bitter victory: achieving artificial intelligence while accelerating ecological collapse.
References
MIT Technology Review (2023), “Four ways AI is making the power grid faster and more resilient” — https://www.technologyreview.com/2023/11/22/1083792/ai-power-grid-improvement/
UNEP (2023), “An Eye on Methane: International Methane Emissions Observatory 2023 Report” — https://www.unep.org/resources/report/eye-methane-international-methane-emissions-observatory-2023-report
Google (2023), “Accelerating climate action with AI”, https://blog.google/outreach-initiatives/sustainability/report-ai-sustainability-google-cop28/
Goldman Sachs (2024), “AI is poised to drive 160% increase in data center power demand” — https://www.goldmansachs.com/insights/articles/AI-poised-to-drive-160-increase-in-power-demand
MIT Technology Review (2019), “Training a single AI model can emit as much carbon as five cars in their lifetimes” — https://www.technologyreview.com/2019/06/06/239031/training-a-single-ai-model-can-emit-as-much-carbon-as-five-cars-in-their-lifetimes/
MIT Technology Review (2024), “AI’s emissions are about to skyrocket even further” — https://www.technologyreview.com/2024/12/13/1108719/ais-emissions-are-about-to-skyrocket-even-further/
Scientific American (2025), “Biden Opens Publicly Owned Land to Data Centers Run on Clean Energy” — https://www.scientificamerican.com/article/data-centers-run-on-clean-energy-now-welcome-on-publicly-owned-land/
Harvard Business Review (2024), “The Uneven Distribution of AI’s Environmental Impacts” — https://hbr.org/2024/07/the-uneven-distribution-of-ais-environmental-impacts
UNEP (2024), “AI has an environmental problem. Here’s what the world can do about that” — https://www.unep.org/news-and-stories/story/ai-has-environmental-problem-heres-what-world-can-do-about
Stockholm Resilience Centre (2024), “AI could create a perfect storm of climate misinformation” — https://www.stockholmresilience.org/publications/publications/2024-10-21-ai-could-create-a-perfect-storm-of-climate-misinformation.html
New York Times (2025), “Big Banks Quit Climate Change Groups Ahead of Trump’s Term” — https://www.nytimes.com/2025/01/20/business/trump-climate-action-banks.html
Harvard Business Review (2023),“The Opportunities at the Intersection of AI, Sustainability, and Project Management” — https://hbr.org/2023/10/the-opportunities-at-the-intersection-of-ai-sustainability-and-project-management
IBM (2024), “AI and the future of sustainability: A climate week conversation with IBM experts” — https://www.ibm.com/think/insights/climate-week-qa-ai-sustainability
Financial Times (2025),“SoftBank and OpenAI back sweeping AI infrastructure project in US” — https://www.ft.com/content/48eb53a1-67ca-4509-8c62-401f0cf8b099
UNEP (2024), “Artificial Intelligence (AI) end-to-end: The Environmental Impact of the Full AI Lifecycle Needs to be Comprehensively Assessed” — https://wedocs.unep.org/handle/20.500.11822/46288;jsessionid=99C93C9FC24F1FBD5594EC602AE5571F
Microsoft (2024), “Our 2024 Environmental Sustainability Report” — https://blogs.microsoft.com/on-the-issues/2024/05/15/microsoft-environmental-sustainability-report-2024/#:~:text=In%20aggregate,%20across%20all%20Scopes,semiconductors,%20servers,%20and%20racks.
Google (2024), “2024 Environmental Report” — https://sustainability.google/reports/google-2024-environmental-report/
Google (2021), “How we’re minimizing AI’s carbon footprint” — https://blog.google/technology/ai/minimizing-carbon-footprint/
Nvidia (2024), “Sustainable Strides: How AI and Accelerated Computing Are Driving Energy Efficiency” — https://blogs.nvidia.com/blog/accelerated-ai-energy-efficiency/
IBM (2024), “The State of Sustainability Readiness 2024” — https://newsroom.ibm.com/2024-11-12-new-ibm-report-shows-strong-tailwinds-behind-corporate-investment-in-ai-for-sustainability-but-ambitions-dont-yet-match-actions
US Energy Information Administration (2024), “Data center owners turn to nuclear as potential electricity source” — https://www.eia.gov/todayinenergy/detail.php?id=63304#
McKLinsey & Company (2024), “Sustainability: Sources of value creation” — https://www.mckinsey.com/capabilities/strategy-and-corporate-finance/our-insights/sustainability-sources-of-value-creation
