Powering the AI Race: Ready, Set, Coal?

The Department of Energy’s Speed to Power initiative aims to accelerate the buildout of grid infrastructure to “win the AI race.” Unfortunately, DOE’s artificial intelligence energy strategy so far has been dangerously backwards. The plan pursues outdated forms of energy generation to the detriment of faster and cheaper ways to power the AI industry.

Under the guise of meeting AI energy demand, the Trump Administration is pushing to delay coal plant retirements, loosen environmental protection rules for toxic coal ash disposal, and preempt state regulations. This strategy could cost tax- and ratepayers over $3 billion.  Since early 2025, the Federal Government has overridden retirement plans set by six utilities and approved by state and energy regulators.

Forcing reliance on19th-century fossil fuels is an expensive and shortsighted response to the real challenge of rising energy demand. In this post, we’ll explore why the real solution to the AI power surge lies in fixing interconnection queues, upgrading transmission, and prioritizing clean energy over costly coal bailouts.

But first, let’s take a minute to unpack the forecasts of data center energy demand.

How much energy do we really need?

The Trump administration justifies its actions by claiming that electricity demand from data centers will skyrocket over the coming decade — and that failing to meet the demand will have disastrous consequences for the United States.

In reality, the details of when and where this demand will materialize are highly speculative. Load forecasts are consistently overestimated, and estimates of data center demand are known to be inflated from duplicative project requests. Furthermore, most forecasts of AI energy demand do not account for the uncertainty of consumer demand for AI, energy efficiency improvements within the sector, practical constraints such as chip manufacturing and supply chain capacity, or  widespread local opposition to data center development.

In a recent reality check, the largest grid operator in the U.S announced that the power demand for 2027 would be “appreciably lower” than current demand forecasts.

However, while there should be skepticism over the scale and assumptions behind load growth forecasts, energy demand is rising, and large load demand is increasing electricity prices for consumers.  Much of that is driven by data center development, and solutions are needed to meet that demand sustainably and protect everyday consumers from footing the bill for technology companies’ power.

Unfortunately, the Trump administration and allies are actively sabotaging the energy sources best suited to meet demand increases by trying to halt clean energy projects.

Coal is not the answer.

Reviving coal-fired generators and delaying planned coal plant retirements is not only harmful for the environment and public health, but expensive and inflexible. Retiring coal plants are being phased out for a reason: They are costly to operate, and increasingly inefficient as they age.

Electricity production from coal has declined steadily since 2000. Coal power plants with higher operating costs are retiring, and the ones still in-service are being utilized less. These units also require long startup, shutdown, and ramp times, making them poorly suited for managing new loads that require flexible energy sources able to respond quickly to sudden changes in demand like data centers, whose demand can be dynamic and unpredictable.  Coal units are also increasingly unreliable, with forced outage rates rising significantly over the past decade.

Reviving retired coal units comes with huge costs, which often fall on utility customers. When plants are retired, utilities reduce maintenance spending, meaning that bringing a unit back online requires significant capital investment to address years of deferred maintenance and degradation.  For example, reviving the Cholla Power Plant in Arizona from its March 2025 retirement would cost utility customers an estimated $1.9 billion.

There are faster and cheaper solutions.

1. Add more clean energy

Earlier this year, the chief executive of the North American Electric Reliability Corporation stated that “[t]o the extent that we’re going to unleash abundant energy in North American in the near term, it’s going to mostly be wind and solar.”

Speed is key, and clean energy projects can be assembled much quicker than fossil fuel power plants. On average, solar, wind, and battery projects are all developed in under 2 years. In contrast, a natural gas power plant can take from 2.6 years (for natural gas combustion turbines) to 4 years (for natural gas combined cycle) to develop, while coal is even slower, taking 6.7 years. Natural gas is further constrained due to supply chain constraints for new gas turbines, resulting in years long delays. Estimates for the backlogs vary, with some reporting that delays can range from 3-4 years or as long as 5-7 years.

Wind and solar are also cheaper to build, which is necessary to address the increasing electricity prices across the country. According to an analysis by financial advisory firm Lazard, wind and solar have a lower levelized cost of energy — which measures the revenue required to build and operate a generator over a specific cost recovery period — than new builds of gas and coal. With rising prices and unprecedented demand on the horizon, the Department of Energy should be looking to renewables to address load growth quickly and affordably.

2. Fix the interconnection queue

There are generators waiting in line to meet incoming large load demand, but regulatory and administrative barriers have made the interconnection process too slow. The average median duration of the interconnection process was approximately five years for projects completed in 2022-2023, with most of that time spent during the interconnection study phases and on the development of network upgrades. In 2024, DOE published a roadmap with 35 solutions to improve the process for connecting generators to the bulk power system. DOE should resume this process and use its technical assistance programs to advise Regional Transmission Organizations, Independent System Operators, and other Balancing Authorities on practical approaches to implement the recommendations in the roadmap to improve the interconnection process.

3. Expand and improve transmission capacity

Transmission expansion, particularly at the regional and interregional level, is a well understood solution to increase the capacity, reliability, and efficiency of the U.S. grid. In 2024, DOE’s National Transmission Planning Study estimated that meeting future growth reliably would require expanding the total transmission system by 2.1 to 3.3 times the 2020 system. The study found that expanding transmission could lower system costs by $270-490 billion, while also improving reliability.

In the near term, Grid Enhancing Technologies (GETs) are a cost-effective solution for rapidly integrating new energy sources without needing to build capital-intensive infrastructure. Examples of GETs include:

• dynamic line rating, which factor real-time weather conditions into line ratings to enable operators to utilize the line’s full carrying capacity;
• advanced power flow control devices, which can increase capacity by 10-25% by enabling the redistribution of power to lines with available capacity;
• and topology optimization software, which distributes power flows across the grid more efficiently.

4. Build with local buy-in

To address the pace and scale of large-scale electricity demand, policymakers must prioritize practices that support inclusive, community-forward solutions. Pursuing rushed development for energy infrastructure projects under the guise of urgency is counterproductive. Not only does it erode trust and harm communities, but it is likely to undermine the success of the project.

Research demonstrates that early engagement with affected communities improves the timeline for implementing energy infrastructure projects. A study evaluating permitting challenges for 37 transmission projects found that early engagement was a key characteristic of projects that were implemented in a timely manner, while other factors, like ineffective interagency coordination and personnel shortages, were the root cause of project delays.

Learn more:

• Read our response to DOE’s Speed to Power RFI here.
• Read about our solutions for accelerating transmission buildout here.
• Read about local and state-led strategies for managing load growth here.

Thanks to the following individuals for their excellent contributions: Caroline Weinberg for analysis of load forecasting; Aaliyah Adkins for analysis of clean energy and fixing interconnection queues; Callie Rabinowitz for research about coal, as well as the authors who contributed to Earthjustice’s Speed to Power RFI comments. This blog is largely based on Earthjustice’s RFI comments, which were authored by numerous other Earthjustice staff.