Shares

We Have What’s Needed to Reliably Power Data Center Boom, and It’s Not Coal Plants

A range of clean, resilient solutions can help us meet the electrical needs of our growing digital economy while saving Americans money.

By Gabriella Tosado, Ashtin Massie, Joe Daniel

After decades of relatively flat electricity demand, the US power sector is expecting demand to grow due, in large part, to new data centers. These energy-intensive facilities are reshaping the grid, with some utilities now projecting over 20 percent load growth by 2035. In places like Virginia, which constitutes 13 percent of all reported data center capacity globally and 25 percent of the data center capacity in the United States, data centers already account for over a quarter of some utilities’ total electric demand, and their footprint is only growing.

The myth

Utilities are struggling to maintain accurate forecasts and identify resources that can meet this growth. There is a high-profile effort to keep coal plants that are set to retire online and run them at unprecedented levels, ostensibly for reasons of reliability. But the truth is, coal-fired power plants, far from being a reliable backbone for this new era of electricity demand, are a brittle, outmoded technology that threatens to undermine the very grid resilience they’re being proposed to protect.

Coal-fired power plants, far from being a reliable backbone for this new era of electricity demand, are a brittle, outmoded technology that threatens to undermine the very grid resilience they’re being proposed to protect

Reality #1: Aging coal plants are failing to consistently deliver under stress

Coal plants face a fundamental constraint: they are aging and increasingly unreliable. Most of the coal fleet was built in the 1970s and 1980s, and years of wear and tear have led to a rise in unplanned outages. In many cases the sheer cost to maintain and modernize these plants did not make sense with the availability of more reliable and affordable alternatives — and that’s still the case.

According to the Energy Systems Integration Group (ESIG) Ensuring Efficient Reliability report, a coal plant’s capacity accreditation, or the amount of time it can contribute to peak demand, is only 83 percent when adjusted for real-world performance. PJM also has capacity accreditation of coal plants at 83 percent and some plants fare even worse. Gridlab’s reliability study found Colstrip, a large regional coal plant in Montana, operating with a capacity accreditation of only 54 percent — meaning it’s effectively unavailable nearly half the time it’s needed.

Extreme weather exacerbates these vulnerabilities. Cold snaps, heat waves, and storms have all exposed coal’s fragility during grid stress events. Reliability is not just about being dispatchable, it’s about delivering performance under stress. Coal plants struggle to do that consistently. For coal plants to truly meet the constant demands of data centers, they would need to run at high-capacity factors and avoid major outages, all of which fly in the face of current performance trends. If a large coal plant trips offline while supporting a cluster of data centers, the sudden loss of supply could lead to cascading failures across the grid. This is because generation must equal load at all times, datacenter or no datacenter. As a result, relying on coal plants to support these high-density digital loads doesn’t enhance reliability, it endangers it. And it’s not a matter of if the coal plant will fail, but when.

Reality #2: The inflexibility of coal plants risks grid stability

Coal boosters often point to the “always-on” nature of coal plants as evidence of their reliability. But that characteristic is a liability, not a strength, when it comes to supporting large, fast-changing loads like data centers. Coal units are inherently inflexible: they ramp slowly, respond poorly to sudden load shifts, and are difficult to turn on or off quickly. This rigidity is a poor match for the dynamic and often unpredictable nature of data center demand. Further, inflexible coal plants can worsen grid congestion; by occupying limited transmission capacity with inflexible generation, they prevent cheaper or cleaner resources from being delivered. This issue has already been flagged by independent market monitors in regions like MISO — which covers 15 US states and a Canadian province — where congestion-related market distortions have cost over $1 billion a year. Coal plants displace faster-responding resources that are better suited to follow load. And the stakes are high.

As noted by the North American Electric Reliability Corporation (NERC), large, voltage-sensitive loads like data centers require flexible, responsive grid solutions, not slow-ramping generators that can take 12 or more hours to come online. NERC’s recent Incident Review and Guidance on Voltage-Sensitive Large Load Integration describes an event in 2024 where a transmission fault triggered a sudden disconnection of 1,500 megawatts of voltage-sensitive data center load, leading to sharp frequency and voltage spikes that required operator intervention. The incident exposes the system’s vulnerability to instability when inflexible generation cannot respond to large load fluctuations.

 If a data center either loses access to load or goes offline rapidly, a grid’s generation needs to respond at sub-second speeds. The average coal plant ramp rate is 4 percent per minute which translates to spending over 20 minutes to respond to a large load event. From a cold start, the average coal plant would take over 12 hours to reach max capacity. Coal plants simply can’t respond fast enough to support the reliability needs of modern data centers. Whether it’s the hours-long startup time from a cold state or sluggish ramp rates to turn off, these plants are too slow to provide the real-time flexibility required during sudden load changes or outages.

Reality #3: Clean resources are available now that can better respond to and support data centers for less

The good news is that we don’t need to rely on brittle coal plants to meet the needs of a digital economy. A range of cleaner, more resilient solutions is already available — and scalable. For example, we recently found that more than 95 percent of future demand can be met with fast, scalable, and clean solutions:

Coal plants are a legacy technology, not a solution for the future. Coal plants’ operational characteristics make them less suited to meet the scale and speed of these new challenges. The path forward is not about discarding the past, but about building on it with cleaner, more adaptable resources that can reliably serve evolving grid needs.

Technologies like battery storage, demand flexibility, and clean energy portfolios offer practical, cost-effective options that align with modern load dynamics. As we noted in a recent article, by running coal plants only when it is economical to do so and using the extra transmission headroom that creates to reinvest with clean energy upgrades, our grid can support the next wave of economic growth with the flexibility it demands. There are reasons to manage the shift to new, clean resources thoughtfully and intentionally, but propping up coal plants that are not suited for the job is a step in the wrong direction.