AI Data Centers Could Boost US Carbon Emissions by 29% Within a Decade
United States, Wednesday, 21 January 2026.
AI systems now match New York City’s carbon footprint and consume water equivalent to global bottled water usage annually. Without policy intervention, data center emissions could surge 19-29% over the next ten years as electricity demand skyrockets 60-80% by 2050.
Research Reveals Alarming Emission Projections
The Union of Concerned Scientists released a comprehensive analysis on January 20, 2026, revealing that artificial intelligence’s explosive growth threatens to derail America’s climate progress [1]. The study projects that without significant policy changes, carbon dioxide emissions from US power plants tied to data centers could increase by 19-29% over the next decade [1]. This stark warning comes as the country faces an unprecedented surge in electricity demand, with projections indicating a 60-80% increase by 2050, driven primarily by data centers that will account for over half of this growth by decade’s end [1]. Steve Clemmer, the lead author and director of energy research at UCS, emphasizes the urgency of the situation, stating that stronger guardrails are essential to ensure data centers don’t compromise electricity access for other customers [1].
AI’s Environmental Footprint Reaches Critical Mass
Research conducted by Alex de Vries-Gao, a PhD candidate at the VU Amsterdam Institute for Environmental Studies and founder of Digiconomist, reveals that AI systems in 2025 generated a carbon footprint equivalent to New York City, producing between 32.6 and 79.7 million tons of CO2 emissions [2]. The environmental impact extends beyond carbon emissions, with AI systems consuming an estimated 312.5-764.6 billion liters of water in 2025 – roughly equivalent to global annual bottled water consumption [2]. This staggering consumption reflects the scale problem inherent in AI operations, where individual prompts may seem negligible but billions of queries across thousands of facilities create massive environmental consequences [3]. The opacity surrounding AI’s environmental impact remains a significant challenge, as data center operators fail to publicly disclose the metrics necessary to assess AI’s true environmental cost [2].
Energy Consumption Reaches Unprecedented Levels
Data center electricity consumption has experienced dramatic growth, rising from 58 TWh in 2014 to 176 TWh in 2023, representing 4.4% of total US electricity consumption [4]. Analysts project this demand will reach 325-580 TWh by 2028, potentially accounting for 12% of total US electricity [4]. The computational intensity of AI operations becomes clear when examining individual usage patterns: a typical GPT-4o prompt consumes approximately 0.34 Wh of electricity, while Google’s Gemini apps use about 0.24 Wh per median text prompt [4]. Ethan Mollick, Professor and Co-Director of Generative AI Labs at the University of Pennsylvania’s Wharton School of Business, contextualizes this consumption by comparing a GPT-4o prompt to roughly 16 seconds of HD Netflix streaming [4]. The cumulative effect is driving US electric power demand to record highs in 2025-2026 [4].
Corporate Investments Signal Clean Energy Transition
Major technology companies are making substantial commitments to clean energy infrastructure to power their AI operations. Meta announced on January 9, 2026, significant deals with Vistra, TerraPower, and Oklo to power its Prometheus supercluster in Ohio, with projects expected to add 6.6 gigawatts of capacity by 2035 [5][6]. These partnerships represent a growing trend among hyperscalers seeking to address the environmental challenges of AI deployment. Amazon and Google have similarly signed clean power agreements, though many companies still plan to build onsite gas plants as backup power sources [1][6]. Solar, wind, and storage technologies comprised more than 90% of new power additions to the US grid in 2025, demonstrating the renewable sector’s momentum [1].
Policy Solutions Could Reverse Emission Trajectory
The UCS analysis demonstrates that strategic policy interventions could dramatically alter AI’s environmental impact. Reinstating tax credits for wind and solar energy could reduce CO2 emissions by over 30% over the next decade while decreasing wholesale electricity costs by approximately 4% by 2050 [1]. The study models that decarbonizing the US grid as AI demand increases would raise wholesale electricity costs by $412 billion through 2050 – a 7% increase – but would avoid up to $13 trillion in climate costs [1]. However, recent policy developments create uncertainty about renewable energy deployment. Three judges ruled on January 17, 2026, that construction could proceed on five East Coast wind farms after the Trump administration issued stop work orders in December 2025, citing national security concerns [1]. Clemmer warns that such actions send a chilling signal to the industry and efforts to meet electricity demand sustainably [1]. Meanwhile, an Interior Department policy has created a bottleneck affecting 22 gigawatts of renewable energy projects [1].