The Growing Energy Demand of AI
The use of artificial intelligence is expanding rapidly month by month, driving an unprecedented surge in electricity consumption. According to a Bloomberg report from December 2025, data center power demand in the US could reach 106 gigawatts by 2035, a 36% jump from the previous outlook published just seven months earlier. One gigawatt can power a medium-sized city of 750,000 homes. To meet this demand, the US is planning to build or expand 74 methane-fired generating stations specifically for AI data centers, as reported by the Environmental Integrity Project (EIP) this month.
Emissions from New Power Plants
The 74 new power plants are expected to release 662 million tons per year of greenhouse gas pollution. This is equivalent to the total annual emissions of Australia and equals the emissions from 140 million conventional cars and trucks. Additionally, these plants could release nearly 160,000 tons per year of other air pollutants, including more than 44,000 tons of nitrogen oxides and nearly 33,000 tons of fine particulate matter.
Behind-the-Meter Plants and Infrastructure
The EIP report notes that these 74 plants are designed as “behind-the-meter” power plants, meaning they will not directly compete with households on regional grids. However, they still compete with utilities and other industrial sites for natural gas and power generating equipment, driving up prices for everyone, according to Oil & Gas Watch. Furthermore, the emissions estimates do not account for additional natural gas pipelines, processing plants, and storage facilities needed to supply the methane. In 2024, S&P Global estimated that data center demand could increase natural gas demand by up to 6 billion cubic feet per day by 2030.
Tech Giants Face Surging Emissions
The rapid AI expansion is causing emissions at major tech companies to rise sharply. Amazon’s emissions rose 16% in 2025 compared to 2024, reaching about 81 million metric tons of CO2 equivalent, roughly equal to emissions from 19 million gasoline-powered cars. Google reported an 18% increase in its overall emissions in 2025, with Scope 1 emissions (from its own operations) up 20% year-over-year. Microsoft and Meta reported emissions jumps of 23% and 64% respectively. Google’s electricity usage increased 37% last year, while Amazon’s emissions from purchasing electricity rose 34%.
Net-Zero Targets Under Pressure
Despite these increases, companies maintain their sustainability goals. Amazon aims for net zero by 2040; Google targets 2030. Google acknowledged that its path to net zero “will not be linear, given our AI infrastructure build out is currently accelerating faster than the grid is decarbonizing.” Microsoft is considering scaling back its pledge to match 100% of hourly electricity consumption with zero-carbon energy by 2030 due to data center expansion. Meta has set a 2030 net-zero target across its entire value chain. Activist investors asked Amazon, Alphabet, and Meta to explain how they will reconcile surging AI electricity demand with climate commitments, but none of the proposals earned majority support.
Sasha Luccioni, co-founder and chief scientific officer of Sustainable AI Group, commented: “We’re essentially in a climate crisis and we should not be having emissions growth at all, arguably, and yet the data centers are going in the opposite direction.”
Historical Contrast: EVs vs. AI
CleanTechnica notes a striking difference in public reaction: a decade ago, concerns about new power plants for the EV revolution sparked widespread protest, but the current plan for 74 new methane-fired generating stations for AI has drawn little criticism. The fossil fuel industry, which previously campaigned against EVs, now sees AI as a major profit boost and is not spreading misinformation about it. Critics argue that capitalism without guardrails is driving the destruction of the environment for the benefit of tech billionaires, asking: “Is AI worth destroying our planet for?”