The Wired article examines a new report from the Union of Concerned Scientists showing that the rapid expansion of artificial intelligence and its data centers in the United States risks boosting national carbon emissions sharply unless policy and energy strategies change. The analysis finds that U.S. electricity demand could increase 60–80% by 2050 because of AI-driven data center growth, with those facilities alone accounting for more than half of that rise by the end of this decade. If current energy and climate policies remain unchanged, carbon dioxide emissions from power plants could increase 19–29% over the next 10 years simply to meet AI’s electricity needs.
Data centers are already a major driver of electricity demand, and the trend toward larger and more numerous AI installations exacerbates this. Estimating future energy use is complicated because utilities and operators release uneven forecasts, and not all announced projects may be built. Technological improvements could make data centers more efficient, but these gains might not keep pace with the sheer scale of demand increases.
The report highlights the role of federal policy. Rolling back renewable energy tax incentives and weakening environmental regulations, particularly on fossil-fuel power plants, increases the likelihood that energy for AI will come from higher-emissions sources. Reinstating credits for wind and solar power could cut CO₂ emissions by more than 30% over the next decade while moderating wholesale electricity costs.
The analysis suggests that deeper actions will be needed to decarbonize the grid and upgrade infrastructure so that AI demand doesn’t lock in fossil-fuel generation. That includes stronger emissions standards and greater investment in transmission and storage for renewable generation. With these changes, the growth of AI could coincide with energy transitions rather than undermine them.