Artificial intelligence is spreading fast across the American economy—powering everything from customer service to medical diagnostics. But that speed comes with a hidden cost: energy consumption at a scale researchers are only now beginning to measure.
A new study in Environmental Research Letters quantifies what many have suspected: widespread AI adoption across the U.S. could increase annual carbon dioxide emissions by roughly 900,000 tons. That's equivalent to the yearly electricity use of about 300,000 American households.
Before you panic, though, context matters. That 900,000-ton increase represents only 0.02% of total U.S. emissions—a measurable but modest addition compared to sectors like transportation or electricity generation. The researchers behind the study aren't downplaying the problem; they're simply being precise about its scale.
We're a new kind of news feed.
Regular news is designed to drain you. We're a non-profit built to restore you. Every story we publish is scored for impact, progress, and hope.
Start Your News Detox"While the projected emissions from AI adoption are modest compared to other sectors, they still represent a meaningful increase," says co-author Anthony R. Harding. "This underscores the importance of integrating energy efficiency and sustainability into AI development and deployment, especially as adoption accelerates across industries."
The study modeled how AI could expand across multiple economic sectors and calculated the electricity demand—and resulting emissions—that expansion would require. Energy demand could rise by as much as 12 petajoules per year, a significant but manageable figure in a country that consumes roughly 100 times that annually.
What makes this research valuable isn't the alarming headline; it's the specificity. We now have a baseline number. We know that AI's carbon footprint is real but not catastrophic—at least not yet. That gives companies and policymakers something concrete to work with. Some tech firms are already moving: Microsoft has committed to carbon negativity by 2030, and others are investing in renewable energy to power data centers.
The real test comes next. As AI adoption accelerates and models grow more complex, energy demands could climb. Whether that happens depends partly on whether the industry treats efficiency as a core feature, not an afterthought. The researchers are essentially saying: this is manageable—if we start now.
