Why Coal-Fired Power Plants Can't Meet Data Center Demand

The power grid wasn't built for this.

Across the United States — and increasingly around the world — utility operators are making a decision that would have seemed unthinkable five years ago: keeping aging coal-fired power plants online specifically to feed the insatiable appetite of data centers. Not because coal is cheap. Not because it's clean. Because there's simply nothing else ready to take its place fast enough.

That's not a short-term blip. It's a structural problem, and it's getting worse.

The Scale of What Data Centers Actually Consume

To understand why traditional energy infrastructure is buckling, you need to sit with the numbers for a moment. A single hyperscale data center — the kind that Amazon, Microsoft, or Google operates — can draw anywhere from 100 to 500 megawatts of power continuously. That's equivalent to powering a small city, running 24 hours a day, 365 days a year, with zero tolerance for interruption.

Now multiply that by hundreds of facilities planned or under construction across the country, fueled by the artificial intelligence boom that has sent compute demand through the roof. AI model training and inference workloads are dramatically more energy-intensive than traditional cloud computing. Running a single ChatGPT query consumes roughly ten times the electricity of a standard Google search. Aggregate that across millions of daily users, and you start to see the math become genuinely alarming.

The AI revolution isn't just a software story — it's an energy story, and the grid isn't ready.

Utilities that once planned capacity growth in modest annual increments are now fielding requests from data center developers that would strain entire regional grids. Some utilities have reported multi-gigawatt queues of interconnection requests showing up in a matter of months. The energy supply chain, designed for gradual, predictable load growth, is being asked to sprint.

Why Coal Is the Wrong Answer to the Right Problem

Here's where the situation gets genuinely uncomfortable. Facing unprecedented demand, some grid operators have resorted to extending the life of coal-fired power plants that were scheduled for retirement. Reports indicate that at least 15 power plants — most of them coal-fired — have been kept running specifically to meet data center demand. Diesel generation is also being pressed into service as a stopgap.

This is the energy supply chain under stress: falling back on whatever is available, not whatever is appropriate.

The limitations of coal as a data center energy source aren't just environmental — though those are real and significant. Coal plants are operationally inflexible. They take hours to ramp up or down, making them poor partners for the kind of dynamic load management that modern data centers require. They're expensive to maintain, particularly aging plants that were meant to be decommissioned. Fuel logistics are complicated and subject to supply disruptions. And increasingly, they represent a legal and regulatory liability as emissions standards tighten.

Keeping a coal plant alive to power a data center is like patching a highway with duct tape — it buys time, but it doesn't solve the problem.

The environmental math is equally stark. Coal-fired generation produces roughly 2.2 pounds of CO2 per kilowatt-hour — approximately twice the emissions of natural gas and orders of magnitude more than solar or wind. For technology companies that have made aggressive public commitments to carbon neutrality or 100% renewable energy, relying on coal — even indirectly through the grid — creates serious reputational and ESG exposure. Microsoft, Google, and Amazon have all published ambitious clean energy targets. None of those targets account for "we kept a coal plant running to serve our Ohio data center."

Clean Energy for Data Centers: The Gap Between Promise and Reality

Renewable energy is the obvious answer. Solar and wind costs have collapsed over the past decade — utility-scale solar now routinely comes in under $30 per megawatt-hour in competitive markets, cheaper than operating an existing coal plant in many regions. Corporate power purchase agreements (PPAs) have become a standard tool for large tech companies to lock in long-term clean power supply and hit their sustainability metrics.

The problem isn't cost. It's timing and physics.

Solar generates power when the sun shines. Wind generates when the wind blows. Data centers run constantly. Bridging that gap requires either massive battery storage infrastructure, grid-scale long-duration storage solutions that are still maturing commercially, or a hybrid approach that draws from multiple sources simultaneously. None of these come together overnight.

Some operators are finding workable solutions. Meta's data center in DeKalb, Illinois, is powered significantly by wind energy from long-term PPAs. Google has pursued a more sophisticated "24/7 carbon-free energy" matching strategy, attempting to pair every hour of consumption with a corresponding hour of carbon-free generation in the same grid region — a much harder standard to meet than simple annual renewable matching. Microsoft has invested in nuclear power agreements, including a high-profile deal to restart power procurement from the Three Mile Island facility in Pennsylvania, rebranded as the Crane Clean Energy Center.

Nuclear is worth dwelling on here. It provides the one thing renewables currently struggle to guarantee: always-on, carbon-free baseload power. A single nuclear unit can generate 1,000 megawatts continuously, with capacity factors exceeding 90%. For data centers demanding reliable gigawatt-scale clean power, nuclear may be the most serious long-term answer on the table. Small modular reactors (SMRs), while still several years from commercial deployment at scale, have attracted significant investment from data center operators precisely because they could be co-located with facilities and eliminate transmission constraints entirely.

The Economic Reality Nobody Wants to Discuss

There's a persistent myth that clean energy transitions are purely altruistic — that companies pursuing renewable power are sacrificing economic efficiency for ESG optics. The long-term financial picture tells a different story.

Coal plants carry escalating operational costs, particularly as they age and as carbon pricing mechanisms expand. A plant that costs $45 per megawatt-hour to operate today may cost $70 or $80 in five years once compliance costs, maintenance on aging infrastructure, and potential carbon penalties are factored in. Locking into long-term renewable PPAs at $25-35 per megawatt-hour provides both cost certainty and insulation from fossil fuel price volatility.

The capital calculus is shifting too. Institutional investors are increasingly pricing climate risk into infrastructure valuations. Data center assets tied to dirty power sources face growing exposure to stranded asset risk as regulations tighten. Conversely, facilities with clean, firm power supply — nuclear, geothermal, or well-structured renewable-plus-storage arrangements — command premium valuations and attract the broadest pool of capital.

For the energy supply chain itself, the data center boom is creating new market structures. Independent power producers are signing data-center-specific generation contracts that bypass traditional utility structures entirely, offering direct power delivery at negotiated rates. It's a faster path to new capacity, but it requires sophisticated counterparties on both sides of the deal.

What Has to Change — and Fast

Regulatory frameworks are starting to catch up, but slowly. The Federal Energy Regulatory Commission (FERC) has been pushing reforms to the grid interconnection queue — a process that has become so backlogged that new clean energy projects can wait five to seven years for approval. Streamlining that process is essential, and recent rule changes have attempted to introduce more discipline into how projects enter and progress through the queue.

State-level policy is equally important. Data center development tends to cluster in states with favorable tax treatment — Virginia, Texas, Arizona — but energy policy in those states varies enormously. Virginia has aggressive renewable portfolio standards. Texas has an abundant but occasionally fragile wind and solar market. Arizona offers sun but faces water constraints that complicate both renewable cooling and data center operations.

Emerging technologies — offshore wind, enhanced geothermal, long-duration storage using iron-air or hydrogen-based systems — could dramatically change the supply picture within a decade. But a decade is a long time when hyperscalers are building hundreds of megawatts of new capacity every quarter.

The operators who will be best positioned aren't the ones waiting for the grid to solve the problem for them. They're the ones acquiring land near transmission infrastructure, signing creative bilateral power agreements, and building energy procurement into their development strategy from day one — not as an afterthought when the facility is already designed and sited.

Coal plants were never meant to power the AI economy. The real question is whether the clean alternatives can be deployed fast enough to make that clear before the grid makes the decision for everyone.

Explore the InfraSale Marketplace for clean energy solutions.

[INTERNAL LINK: data center energy consumption]

[INTERNAL LINK: renewable energy solutions]

[INTERNAL LINK: nuclear energy for data centers]