Like the world continues to warm and electricity bills take center stage On the national political front, the data center boom will lead to higher carbon emissions and electricity costs in the United States. But a few simple policies could help lower emissions and prices.
This is the message of a new analysis from the Union of Concerned Scientists, released Wednesday, which models a variety of scenarios for how to fuel the next AI boom. The United States is on track to see a 60% to 80% increase in electricity demand by 2050, with data centers alone accounting for more than half of that increase by the end of this decade, according to the analysis. If policies remain the same as today – with attacks on renewable energy embedded in regulatory regimes and few significant national policies limiting carbon emissions from power plants – we could see a 19 to 29 percent increase in CO emissions.2 US power plant emissions are tied solely to data center energy needs over the next 10 years.
There are answers, however: Restoring tax credits for wind and solar, which were policy goals in last year’s One Big Beautiful Bill, would reduce CO emissions.2 greenhouse gas emissions by more than 30% over the next decade, even as data centers absorb a significant share of new electricity demand. They could also lower wholesale electricity prices by around 4% by 2050, after rising slightly over the next decade.
Power plants are the second largest source of greenhouse gas emissions in the United States, accounting for about a quarter of the country’s overall emissions. Last year, emissions from the U.S. power sector increased slightly, marking the first increase since 2023; commercial buildings like data centers, a analysis released last week by Rhodium Group, were the main drivers of this demand.
Predicting how much energy the United States will need in the future for AI is an incredibly tricky project. Most of the public estimates we have are provided by utilities that are facing a number of demands for new capacity from data centers; Data center companies often forward their requests to a number of utilities in order to shop around for the best price, inflating estimates of overall actual needs. Technological advances in the coming years could also make data centers and AI much more energy efficient. Some of the wildest numbers making headlines or trumpeted by executives in the tech and energy sectors are likely exaggerations. (Earlier this month, PJM, one of the nation’s largest regional broadcast organizations, downgraded its projections of how much power the grid will need over the next two years after looking more closely at some data center proposals.) To get an accurate read on this, UCS modelers used medium-term power growth scenarios and assumed that only half of the publicly announced projects underway would actually be built.
But the Trump administration has been so aggressive against renewable energy and climate policies over the past year that the analysis likely underestimates the high level of emissions from data center demand. Although the UCS modeling took into account some policy changes, including rolling back regulations on coal-fired power plants, eliminating tax credits for renewable energy and postponing some offshore wind projects, it did not account for others. An Interior Department policy that mandated review of all wind and solar projects on federal lands, for example, created a huge neck of 22 gigawatts of projects, enough to power more than 16 million homes. In December, the administration issued stop-work orders for five wind farms under construction on the East Coast, citing national security concerns. (On Friday, three different judges ruled that construction could continue.)
“The Trump administration is doing this with projects already approved and under construction,” says Steve Clemmer, lead author of the analysis and director of energy research at UCS. “This sends a frightening signal to the industry and efforts to power data centers and meet electricity demand. We need to build as much as we can, as fast as possible.”
The Trump administration has made clear that it wants to use fossil fuels — particularly coal, the dirtiest form of energy — to fuel the AI boom. Energy Secretary Chris Wright has ordered that at least two coal-fired power plants remain in service beyond their retirement dates, while the White House and the Environmental Protection Agency have pushed a number of policies which benefit the coal industry.
Despite the administration’s clear preference for fossil fuels, they are not always the cheapest or easiest option, especially as demand for gas turbines has soared and experts predict years of waits to build new natural gas plants. Some electricity suppliers are starting to oppose the government’s anti-renewable energy measures. Earlier this month, PJM filed a brief supporting a Virginia offshore wind developer who is suing the Trump administration for blocking the project. In its submission, PJM argues that the wind project, which will provide enough electricity to power more than 600,000 homes, “is important to meet a rapidly increasing demand for electrical energy.”
Big Tech, for its part, has made various promises in recent years to reduce its emissions and be more climate-friendly; many of these commitments were quickly thwarted by the development of AI. It’s unclear whether these companies are willing to directly oppose the Trump administration to make the case for increased renewable energy production. Last week, with encouragement from the White House, Microsoft rolled out a set of commitments so that its data centers are “better neighbors” to the communities where they are located. There was no mention of emissions or climate policies in the set of commitments provided by Microsoft, which did not respond to a request for comment.
UCS estimates do not include estimates of power generation produced in the data centers themselves, which is becoming an increasingly popular option for projects awaiting connection to the grid. These installations can include renewable energy production, generally solar and batteries; different forms of nuclear energy – from restarting entire power plants to investing in small modular reactors– are also popular choices for technology companies. But many of the largest hyperscalers have also announced plans to build their own gas plants on site.
The prices modeled in the UCS analysis are only for wholesale electricity – the cost of delivering electrons to the grid – and not necessarily what Americans see on their bills. Consumers’ electricity bills are determined by a range of factors that go beyond just the price utilities pay for the energy itself, including equipment upgrades and transmission line construction. Renewable energy, which is intermittent and often located further from where electricity must be supplied, requires more transportation than traditional fossil sources, which involves more financial investment.
“What’s so crazy about renewable energy is [that] both policy arguments are true,” says Pier LaFarge, co-founder of Sparkfund, a utility company. “They are the cheapest energy at the source of generation, but they also increase rates because of downstream upgrades to the distribution network.”
Simply re-introducing tax credits for wind and solar would not be enough to stave off the worst impacts of climate change. The UCS study also modeled the costs of policies that would more seriously decarbonize the U.S. grid as demand for AI increases. This includes stricter regulations on power plants and more investment in the transportation improvements that renewable energy needs. This scenario, according to the analysis, would increase wholesale electricity costs slightly through 2050, by about $412 billion, an increase of 7%. However, according to the analysis, doing so would prevent up to $13 trillion in climate costs: damages from floods, wildfires, droughts and other extreme weather around the world, as well as local health costs associated with dirty power plants. (Earlier this month, the EPA announced that it would no longer consider the costs of lives saved from excessive pollution when considering pollution policies around power plants.)
Much of the U.S. grid is in serious need of modernization, especially if the country is serious about moving away from fossil fuels. Part of the challenge in the coming years will be ensuring that the improvements the grid needs – with or without more renewables – are not unfairly imposed on consumers.
“There really needs to be much stronger guardrails in place for the data centers themselves, as well as ensuring that we have enough capacity and power generation to power those data centers, and that it doesn’t take away from other customers,” Clemmer says.
Despite the Trump administration’s aggressive attacks on renewable energy and AI’s eye-popping energy demand figures, there is reason for hope. LaForge believes that utilities’ increasing deployment of batteries, coupled with contracts that require data centers to pay for infrastructure and other associated costs, will help drive down electricity rates for regular consumers. (Unlike credits for wind and solar, tax credits for batteries were mostly won through One Big Beautiful Bill negotiations.) In this scenario, the United States could be more like Texas: tons of cheap wind and solar power on the grid, a few gas plants, and the installation of lots of batteries.
“The good news is that, just as the Biden administration cannot “If you can’t control the fate of the universe, the Trump administration can’t either,” he said. emphasizing that solar, wind and storage accounted for more than 90% of new energy put on the grid last year. “We are building more renewables, faster and in more places, for purely economic reasons.”