These 15 Coal Plants Would Have Retired. Then Came AI and Trump.

By Joe Fassler, a writer and journalist whose work on climate and technology appears in outlets like The Guardian, The New York Times, and Wired. His novel, The Sky Was Ours, was published by Penguin Books. Cross posted from DeSmog.

Since the second Trump administration took power in January, at least 15 coal plants have had planned retirements pushed back or delayed indefinitely, a DeSmog analysis found.

That’s mostly due to an expected rise in electricity demand, a surge largely driven by the rise of high-powered data centers needed to train and run artificial intelligence (AI) models. But some of the plants have been ordered to stay open by the U.S. Department of Energy (DOE), despite significant environmental and financial costs. Energy Secretary Chris Wright, a former fracking executive, has frequently cited “winning the AI race” as a rationale for re-investing in coal.

The fossil fuel facilities are located in regions across the country, from Maryland to Michigan and Georgia to Wyoming. Together, their two dozen coal-fired generators emitted more than 68 million tons of carbon dioxide in 2024. That’s more than the total emissions of Delaware, Maryland, and Washington, D.C. combined.

Nearly 75 percent of the coal plants were on track to shutter in the next two years.

The delays buck the overall trend in the U.S., where coal’s importance as an energy source has diminished rapidly over the past two decades. Coal’s critics say this broad-based phaseout is an urgent matter of public and environmental health. Often called the “dirtiest fossil fuel,” coal creates more climate emissions per gigawatt-hour of electricity than any other power source. And the human impacts of its pollution have been profound: A 2023 study in Science attributed 460,000 extra U.S. deaths between 1999 and 2020 to sulfur dioxide particulate pollution belched out by coal plants.

Cara Fogler, managing senior analyst for the Sierra Club, called the recent spate of delayed closures “unacceptable.”

“We know these coal plants are dirty, they’re uneconomic, they’re costing customers so much money, and they’re polluting the air,” said Fogler, who co-authored a report showing many utilities have backtracked on climate commitments, including coal phaseouts, often citing data centers as a cause. “They need to be planned for retirement, and it’s really concerning to see utilities becoming so much more hesitant to take those steps.”

DeSmog identified the 15 plants by examining changes to the planned retirement dates listed by the U.S. Energy Information Administration (EIA), a DOE agency that compiles data on energy providers, as well as public statements from utilities and the Trump administration. Some of the voluntary delays appear to directly contradict previous net-zero pledges made by several companies.

Neither the Department of Energy nor American Power, a trade association representing the U.S. coal fleet, responded to requests for comment.

What Led to Coal’s Decline?

Not long ago, coal really did keep the lights on. In 2005, it provided roughly half of America’s electricity, making it by far the dominant power source nationwide. But in the past two decades, coal’s market share has rapidly waned. No new coal plants have come online since 2013. These days, its footprint has dwindled, with just 16 percent of the overall energy mix.

In March 2017,  President Trump appeared to blame environmental regulations for coal’s poor fortunes — a trend he promised to reverse.

“The miners told me about the attacks on their jobs and their livelihoods,” Trump said at U.S. Environmental Protection Agency (EPA) headquarters. “I made them this promise … My administration is putting an end to the war on coal.”

But environmental regulations didn’t kill coal. Instead, its demise became inevitable mostly thanks to the rise of a competing fossil fuel: natural gas.

Gas has both economic and technological advantages over coal, said David Lindequist, an economist at Miami University who co-authored a recent paper on the environmental impacts of the shale gas boom.

As new fracking technologies helped to flood the U.S. market with cheap gas in the mid-2000s, utilities began a broad coal-to-gas pivot that’s still underway today. Abundant, often less expensive gas flowed into power plants that operate more efficiently and nimbly than coal plants. This combination of price, efficiency, and flexibility made ditching coal an easy calculation for many utilities.

“The fact that we were able to so successfully phase out coal in the U.S. would never have happened without the fracking boom,” Lindequist said.

Today, coal is at an even greater disadvantage, as renewable energies continue to make economic and technological inroads. The International Renewable Energy Agency found that, in 2024, solar and wind routinely delivered electricity more cheaply than fossil sources of energy. That dynamic has helped solar in particular become the fastest-growing source of power in the U.S.

Meanwhile, America’s newest coal plant — the Sandy Creek plant near Waco, Texas, built way back in 2013 — is currently sitting idle after another catastrophic failure. It isn’t set to resume operations until 2027. The average U.S. coal plant is more than 40 years old, a factor that’s contributed to their decreasing reliability.

“These [coal] plants are so old that at this point there’s very little that could really revive the fleet,” said Michelle Solomon, manager in the electricity program at the nonpartisan think tank Energy Innovation. “I’ve been using the analogy of an old car: Nothing is going to bring my car that has 200,000 miles on it back to being a brand new, efficient car.”

During the Biden years, as technological advancements and historic subsidies made renewables even more attractive, observers broadly believed that coal’s days were numbered. The “writing was on the wall” for coal, Lindequist said.

“Coal may retain a grip in U.S. politics, but its actual role in the generation system is shrinking annually,” researchers for the Institute for Energy Economics and Financial Analysis wrote in a 2024 report. “It is a trend we believe is irreversible.”

Yet even before Biden left office, a new dynamic began emerging: As tech companies started proposing billions in data center build-outs to feed the AI frenzy, utilities started to take a fresh look at their coal plants.

Data Centers Changed Coal’s Trajectory 

In 2020, Dominion Energy, a utility that provides electricity to millions of customers across Virginia, North Carolina, and South Carolina, announced non-binding plans to retire the Clover Power Station by 2025. Running the plant — an 877 megawatt (MW) coal-fired facility near Randolph, Virginia — would be uneconomical under any future scenario, the company found. It just didn’t make financial sense to keep it going.

It reversed course just three years later. Under its 2023 plan, Dominion projected that its energy demand from data centers would nearly quadruple by 2038. That’s an astonishing rise, considering that Virginia already leads the U.S. in data center development by a wide margin. Known as Data Center Alley, the state is home to more than one-third of the world’s largest-scale data centers. Today, Dominion says it doesn’t anticipate retiring any of its existing coal plants — including Clover — until at least 2045, the year that Virginia law stipulates its economy must be carbon-free.

Dominion wasn’t the only utility to cite data center growth as it backtracked on coal. In an August 2024 earnings call, executives of the Wisconsin-based utility Alliant Energy said that the company was “proactively working to attract” data center projects. A few months later, Alliant announced it would delay retiring the Columbia Energy Center, a coal-fired plant near Madison, from 2026 to 2029. The plant’s retirement had already been pushed back once.

Utilities have delayed the retirements of at least 15 U.S. coal plants since President Trump took office in January 2025. Data source: U.S. Energy Information Administration. Credit: Joe Fassler/DeSmog

The trend became notable enough to attract the attention of analysts at Frontier Group, an environmental think tank. In January 2025, Frontier analyst Quentin Good published a white paper showing that utilities had already cited data center growth as a rationale for delaying the phaseout of seven fossil fuel power plants across the U.S.

“We were concerned about the potential for all of this new electricity demand from data centers to slow down the transition to clean energy,” he told DeSmog. “In that report, we discovered it was basically happening already.”

But two other dynamics also began playing out in January: AI hype started to reach new levels of intensity, and power changed hands in Washington.

AI Hype Highs, New Coal Lows 

Data centers aren’t the only reason for the recent upswing in electricity demand. Building electrification, industrial growth, and increased electric vehicle ownership all play roles, too. But nothing has quite caught utilities’ attention like data center projects, which are cropping up with highly localized impacts across the U.S. at a historic rate. Filled with stacks of high-powered computing equipment, the facilities are projected to account for about half of new electricity growth between 2025 and 2030.

On January 21, 2025 — one day after President Trump’s second inauguration — he revealed a new AI infrastructure joint venture involving ChatGPT parent company OpenAI called the Stargate Project, which would spend up to $500 billion on data center build-outs in the next four years. Tech executives announced the initiative’s details alongside Trump during the unveiling at a White House event.

Days later, Meta CEO Mark Zuckerberg said he planned to spend $65 billion on data center build-outs in 2025 alone, including one project “so large it would cover a significant portion of Manhattan.” These announcements followed a similar one from Microsoft in January: a pledge to spend $80 billion on data centers this calendar year.

As the world’s largest tech companies raced to outdo each other, a wave of delayed coal plant retirements followed.

On January 31, Southern Company, a utility serving over 9 million customers across 15 states, announced plans to delay the retirement of generators at two of the largest coal plants in the U.S., both in Georgia. The massive, coal-fired units — two at the Bowen Steam Plant outside Euharlee, and one at the Robert W. Scherer Power Plant in Juliette — had been scheduled to go offline between 2028 and 2035. Under its revised plan, the company pushed retirement back to as late as January 1, 2039(though both plants would be 40 percent co-fired with natural gas by 2030 in that scenario).

In legal documents and public statements, company spokespeople point to data centers as a key rationale for the delays. Last month, at an industry conference in Las Vegas, Southern Company CEO Chris Womack cited data center growth as a key factor keeping fossil energy online, according to the trade publication Data Center Dynamics.

“We’re going to extend coal plants as long as we can because we need those resources on the grid,” he reportedly said.

Next door in Mississippi, Southern Company also delayed the closure of a 500 MW generator at the Victor J. Daniel coal plant in Jackson County. It pushed the retirement back from 2028 until “the mid 2030s.” In documents filed with Mississippi’s Public Service Commission, the state’s utility regulator, Southern appeared to cite a 500 MW Compass Datacenters project as a reason for the change. Southern has pledged to be net-zero by 2050.

As the months passed, the same dynamic unfolded in other states. Alarmed, Good, the Frontier Group analyst, started to track the delays. By October, he published an update to Frontier’s report that found data centers had pushed back at least 12 coal plant closures in the past few years.

“The data center boom has shown no signs of abating,” he wrote. “Even more fossil fuel plants that had been scheduled to retire have been given a new lease on life.”

In its own analysis, DeSmog found that at least 15 coal plant retirements have been delayed since January 2025 alone. Together, those plants emitted nearly 1.5 percent of America’s total energy-related carbon dioxide emissions from 2024.

This comes at a time when the world’s nations need to cut their climate emissions roughly in half to avoid the worst impacts of global heating, according to a recent United Nations report.

But not all the delays can be attributed directly to data center growth. Some have stayed open for a different reason: top-down orders from the Trump administration.

The Department of Energy Steps in 

The J.H. Campbell Generating Plant, a 1.5 gigawatt coal plant in Ottawa County, Michigan, was scheduled to close May 31. The plant even held public tours to give a rare, behind-the-scenes look at aging fossil infrastructure, before it shut its doors for good.

“Now we know cleaner, renewable ways to generate electricity,” a Campbell employee told members of the public on a September 2024 tour.

But just eight days before scheduled to shutter, Department of Energy Secretary Chris Wright ordered Campbell to stay open another 90 days, citing an “emergency” shortage of energy in the Midwest.

Keeping the plant open cost its owner, Consumers Energy, almost $30 million in just five weeks, the company said. Though the plant’s closure was projected to save ratepayers more than $650 million by 2050, Campbell was costing more than $615,000 a day as of September. Yet Wright has since extended his order twice. Campbell now is scheduled to stay open until at least February 2026.

“The costs to operate the Campbell plant will be shared by customers across the Midwest electric grid region,” including customers serviced by other utilities, Matt Johnson, a Consumers Energy spokesperson told DeSmog by email.

Michigan Attorney General Dana Nessel is challenging DOE’s order to keep Campbell open, calling the orders “arbitrary.”

“DOE is using outdated information to fabricate an emergency, despite the fact that the truth is publicly available for everyone to see,” Nessel said in a November 20 press release. “DOE must end its unlawful tactics to keep this coal plant running when it has already cost millions upon millions of dollars.”

Meanwhile, DOE is telling a very different story.

“Beautiful, clean coal will be essential to powering America’s reindustrialization and winning the AI race,” Wright said in September, as the Department of Energy announced $350 million in funding for coal plant upgrades, along with other incentives.

Energy Innovation’s Solomon called the funding “a waste of taxpayer dollars.”

“We’ve been calling it a ‘cash for clunkers’ program where you don’t trade in the clunker,” she said. “Trying to build a modern electricity system using the most expensive and least reliable source of power is really not the answer.”

However, the Trump administration said in September that it plans to feed the AI boom — with an estimated 100 gigawatts of capacity in the next five years — by keeping more old coal plants open. “I would say the majority of that coal capacity will stay online,” Wright said.

Executives from Colorado’s Tri-State Generation and Transmission Association confirmed to DeSmog that they also expect an order to keep a 421 MW coal-fired generator at Craig Station open past its December 2025 decommissioning date.

In late October, Colorado Congressman Jeff Hurd sent a letter to the Trump administration, urging it to extend the life of a 400 MW coal generator at the Comanche Power Station near Pueblo as the owner, Xcel Energy, works to repair the plant’s chronically troubled main reactor. The smaller unit was slated to go offline in December — but, in its case, the administration never needed to act. Last month, Xcel, with the help of Colorado Governor Jared Polis, began to lobby to keep it open at least another 12 months. The state utility regulator appears to have granted that request, according to an agreement with Xcel and other stakeholders.

This delay wasn’t just due to data centers, though their numbers are growing in Colorado. Xcel spokesperson Michelle Aguayo said the delay was “due to a convergence of issues,” including rising electricity demand, “supply chain challenges,” and the continued outage at the main generator. “We continue to make significant progress towards our emission reduction goals approved by the state which would require us to retire our coal units by 2030,” she said.

Delaying the Inevitable

Whether the data center boom will play out as projected is still a matter of speculation.

Last month, power consulting firm Grid Strategies reported that utilities may be overestimating electricity demand from data centers by as much as 40 percent. That’s due in part to the many hypothetical projects, and a widespread practice of double- and triple-counting. Tech companies tend to pitch utilities in multiple regions as they shop around for incentives, creating the appearance of demand from many more data hubs than actually will be built.

Experts have a name for this growing phenomenon: “phantom data centers.”

At the same time, a growing chorus of critics are warning of an AI bubble, arguing that runaway costs can’t justify the kinds of investment being floated. Even the head of Google’s parent company has acknowledged the “irrationality” of the boom.

Critics also say contradictory actions taken by the Trump administration — citing an “energy emergency” while canceling billions in funding for renewable projects — are making the problem worse.

Yet even with all the unknowns, one thing’s certain: Coal’s role in America’s power push can be extended, but it can’t last forever.

Seth Feaster, an IEEFA analyst, says even AI hasn’t changed the big picture: Eventually, coal will die, and it will be killed by other, cheaper forms of energy.

He called the current phenomenon a “period of pause and delay.” In his view, the technological and economic rationales for quitting coal remain undeniable.

“The policy changes here may have a delaying effect on the decline of coal, but they are certainly not changing the direction of coal’s future,” he told DeSmog.

The questions for now are, how long the delays will continue — and at what cost.

Print Friendly, PDF & Email

18 comments

  1. The Rev Kev

    I suppose that all this should come as no surprise. I seem to recall several months ago how Trump or one of his top officials tweeted an image of a lump of coal to say that coal is back again, baby. The way that they are shutting down renewable facilities even if they are complete and profitable shows that this whole seems to be some sort of ideological crusade. Bonus points because they pick up all those votes of miners to the Republican cause. The tech bros won’t care about all that pollution and the extra costs as none of the will be effected by it and won’t be out a penny. Better get use to coal being the boss for the next three years at least.

    Reply
    1. Jan Boudart

      to Rev Kev: Thanks for your comment. Do you have documentation for “…shutting down renewable facilities even if they are complete and profitable…”?

      And what are we going to have to get used to for the next 3 years. If we sit here “getting used to it” for 3 years, what is the outcome? How can we get rid of it instead of getting used to it?

      Reply
    2. Jan Boudart

      to Rev Kev: Thanks for your comment. Do you have documentation for “…shutting down renewable facilities even if they are complete and profitable…”?

      And what are we going to have to get used to for the next 3 years. If we sit here “getting used to it” for 3 years, what is the outcome? How can we get rid of it instead of getting used to it?

      Reply
      1. MicaT

        I will wait to see documentation about that statement of shut down operating solar projects by Trump.

        There certainly are like all power plants times when they have to curtail output. That’s being mitigated with battery storage because batteries are now so inexpensive.

        In the meantime solar is going in all over the country because it’s the least expensive. Yes change to financial incentives have reduced installs going into 2026.
        Trump only has control sort of, of the federal lands permits. He has blocked some with big fanfare only to have many be approved a bit later.

        He also has the ability of stopping new offshore wind leases. His stopping of already permitted and being built ones has been overruled.

        So like our polarized world, if you read the actual solar data install and proposed numbers it’s lower but not zero.
        Also battery BESS are increasing rapidly primarily due to a big drop in prices over the last 2 yrs and continuing to drop.

        Reply
    3. redleg

      You identified one of the specific reasons data centers contract with existing power (and water) utilities- they aren’t technically responsible for any pollution, resource depletion, etc. The utility holds the bag for any adverse effect that might appear, plus all the costs of regulatory compliance, operations and maintenance, etc.
      Meanwhile, data centers are constructed with enormous amounts of “emergency” generators, which effectively power the site. As I understand things (my experience here is limited to backup power for pumping stations), emergency generators are allowed to be constructed to lower environmental standards because they are only supposed to be used in an emergency. So emergency generators have little to no emissions controls, noise abatement, permits, environmental review, etc.
      The use of NDAs mean that identifying problems and collecting data is 100% reactive- complaints are the first indication of problems, and have to be investigated without meaningful information or access.
      Data centers simply must be stopped.

      Reply
      1. Arthur Williams

        I believe it’s Meta who has said that they will require their 350 MW gas turbine power “backup” power plant to run 24/7, 365 days a year, given the power constraints of their location. Some commentators have disparaged their choice of turbine, saying those things are unreliable and cannot be run full power like that all the time.

        Reply
  2. Arthur Williams

    Most of these data centers will never be built because there isn’t enough money. OpenAI isn’t building anything, they’re using their partners, like Oracle, to pay for things. The problem is Oracle and Softbank are out of money, so they won’t be building anything. Once OpenAI dies, likely mid 2026, I expect the AI bubble to pop.

    Reply
    1. 123abceng

      I would give them a chance to survive up to 2027, but I still don’t see any promising scientific approaches, which give us something more reliable than just brute force scaling computational resources. Selling nVidia chips to China is good for market, but it undermines the idea of the world domination.

      Reply
      1. Arthur Williams

        Possibly. The whole AI thing as currently envisaged just doesn’t work. LLM’s are not getting any better, they hallucinate just as much as ever. That particular branch is a dead-end currently. All this talk about agents replacing humans is pure fabrication. There are no “agents” what there are is chatbots that respond to certain triggers and in some cases can run a script. AI doesn’t think, it doesn’t “know” anything, it is merely the best pattern recognizer ever built to date. In testing, an AI given a single task, as in do one thing, fails 20% of the time. In a multi-step task, they fail around 80% of the time. I’ve been using Claude the last 24 hours for a small project, and while the code is fiendishly clever it’s been hopelessly wrong at -every- step requiring constant correction (it makes typos for FFS, how can it misspell “ELSE”). What it is very good at is debugging itself, you tell it it’s wrong and it can figure it out fast.
        These things will never outperform a human. I could have done better than Claud in less time, because I can see where I could merge functions, whereas Claud will repeat “if char>’A and char<'Z+1 a hundred times to see if a char is uppercase.
        It's a bit of a race to see which of Coreweave, CoreScientific and OpenAI dies first. Might be CoreWeave, they are so far out over their skis (as Yves likes to say) that the minute OpenAI fails to pay them for data center construction they may very well turn turtle in a month. Unless CoreScientific goes mams skyward first trying to build what it is contracted to CoreWeave for, since it has never built a data center for gpus before, it's a bitcoin miner.
        Go long bankruptcy law firms.

        Reply
        1. 123abceng

          I have approximately the same impression. Code assistance fails miserably and quickly for non-standard projects. Gems, produced by AI are very rare. I can recall:
          1. Grok gave me a very good paring session just once, during trial period.
          2. Gpt5.2 surprised me yesterday – I would spend couple of days coding by myself. What was noticeable: unusually short names (classes, functions, variables), so they are trying to optimize at that level.
          Who dies first, probably Tab9 (in 2025), and that unfortunate, because suggestions without prompts were useful, in approx 5-10% of cases.
          What was good with 4o – approaching some problem solving, at least at the level of possible variants to consider. Early 2024 was the most productive from those perspectives.
          The most important marker is general health of repositories, driven by companies who promotes their own AI solutions. It’s not getting better in 2025, the amount of reported issues is bigger than it was.
          https://www.githubstatus.com/history – shows, that in 2022 the amount of service disruptions was lower, and it doesn’t fit well into the idea, that Copilot makes code more reliable.

          Reply
  3. 123abceng

    Coal popularity is another one red flag, regarding AI bubble: cheap to invest, cheap to shut down if no need. Investment in nuclear plants requires long term awareness, but everything is temporary in Washington. Whatever Trump says, it’s a compromise between his perceptions and market demand. In his mind, “AGI” is more Adjusted Gross Income than anything else.

    Reply
    1. John Steinbach

      New coal plants aren’t cheap, they’re very very expensive, comparable to nukes. That’s the big reason on new plants have opened since 2013. Like nukes, they take many years to build and cost $10 billion or more.

      As the author notes, fracked natural gas turbines have displaced coal & nuke plants, however many analysts predict the imminent demise of the fracking boom leaving major questions about meeting future energy demands- intermittent renewables mainly supplement & not replace baseline electricity generation.

      Perhaps a return to to “conservation” ethic at the turn of the 20th century is a requirement for dealing with the end of cheap energy & resources in a rational and equitable manner?

      Reply
      1. MicaT

        According to what I find online.
        Coal is $1-6 billion per Gw ( generally more on the lower side
        Nuc is 6-10 billion plus

        Time is much shorter to build coal.
        But you have to have a rail line to the power plant.
        For NG you need pipelines

        I don’t think anyone would invest in a new coal plant without a lot of guarantees

        I’m still not able to find why it’s so expensive to extend a working coal
        Plant. If anyone has info
        I’d like to see it.

        Reply
        1. John Steinbach

          Modern “clean coal “ plants with scrubber and capture technology are comparable in cost and construction time to nukes. That’s is why there are no new coal plants currently under construction.

          Reply

Leave a Reply

Your email address will not be published. Required fields are marked *