AI Is Pushing Climate Goals Out of Reach, New Reports Say

Without a big increase in investment in renewable energy globally, humanity will not limit global temperature rise to 2 degrees Celsius, but much of the data center boom is powered by fossil fuels.

Surging electricity demand driven by artificial intelligence is putting humanity’s climate goals out of reach, extending the life of fossil fuels and driving up emissions in the U.S. power sector while contributing to deadly extreme weather, according to two new reports published Wednesday.

With power- and water-hungry data centers forecasted to come online at staggering speeds to serve big tech companies’ seemingly bottomless appetite for AI infrastructure, utility companies have turned to fossil fuels to help meet the explosion in demand for power.

It’s a sharp departure from earlier forecasts of only modest, gradual growth in electricity demand, potentially threatening large countries’ commitments to transition away from fossil fuels. President Donald Trump and his administration have spoken glowingly about how AI will reinvigorate U.S. coal and other fossil fuel markets.

“Accelerating from deployment to a deeply decarbonized, resilient energy system is proving far more complex than simply adding megawatts,” said Prakash Sharma, vice president for scenarios and technologies at Wood Mackenzie, an energy consulting firm, in a press release accompanying his company’s new report.

Wood Mackenzie’s analysis concluded that almost no countries—including Canada, France, Germany, Italy, Japan, the United Kingdom and the United States—were on track to meet their 2030 emissions goals. But if countries across the globe show “extraordinary ambition,” according to the report, and make significant, rapid investments in renewable energy, humanity could limit warming to within 1.5 degrees Celsius (2.7 degrees Fahrenheit) by around 2060.

The 2015 Paris Agreement called for holding Earth’s temperature rise below 2 degrees C (3.6 degrees F) above pre-industrial levels, and ideally to just 1.5 degrees C in order to preserve a livable planet. To do that, scientists estimated the global economy would need by 2050 to achieve “net-zero” carbon emissions, in which human activity produces a negligible amount of greenhouse gases that could be absorbed by natural ecosystems rather than persisting in the atmosphere.

Among the world’s largest economies, the U.S. has the biggest gap between current climate transition investments and the spending necessary to reach net-zero emissions. The country would need to increase its spending on reducing emissions by 76 percent to meet the net-zero goal, more than double the increase the European Union would need to make and more than two-and-a-half times the increased spending necessary in China.

“A new climate leadership is emerging,” Sharma said. “As the U.S. doubles down on fossil fuels, pushing allies to buy its LNG, China is seizing the low-carbon mantle through EV and solar dominance, plus aggressive renewables deployment.”

The United States has signaled a willingness to offer tax breaks and open public lands to data centers—warehouses of servers whose computing power drives AI services and much of the internet, many of which will be powered by fossil fuels, according to International Energy Agency estimates.

Data center energy demand “is threatening to sabotage the country’s already faltering climate goals,” wrote John Fleming and Jean Su, with the Center for Biological Diversity, in a report published Wednesday. Fleming and Su found that, if AI data centers powered by fossil fuels grow as forecasted, all other sectors of the U.S. economy would need to cut emissions by 60 percent in order for the U.S. to meet its emissions targets.

“A gas-fed AI boom is going to hurdle us past any chance of keeping to our climate goal or maintaining a safe and healthy future for our planet,” said Fleming, a senior scientist with the Center for Biological Diversity, in a statement. “To the extent that data center buildout is needed at all, it should be powered only by clean, renewable energy.”

McKenna Beck, the Ralph Cavanagh climate solutions fellow at the Natural Resources Defense Council, who was not involved in either report, agreed with that conclusion, and warned that the current demand for AI runs the risk of spoiling climate pledges at the local level, too.

“The reports confirm what we’ve been seeing in states on the ground for the past year—that there’s a real risk of states with stated climate goals backsliding on those,” she said. As an example, Beck brought up North Carolina, which erased its 2030 climate goals this summer in the face of rising electricity demand.

Beck believes that, if given the right guardrails, AI electricity demand is not destined to add a ton of emissions to the U.S. economy. “With the right incentives and requirements, data centers could actually supercharge clean energy,” she said.

But with the Trump administration actively working to stifle renewable energy growth, Beck acknowledged that any good-governance AI policies would need to be implemented on a smaller scale. 

“States are on the front lines right now,” she said.

Cover photo:  A view of Meta’s newly constructed data center on July 18, 2024, in Eagle Mountain, Utah. Credit: George Frey/AFP via Getty Images