The Common Sense Case on Data Centers
A fact-based look at the data center debate: real EPRI data on electricity costs, real PJM price increases, and what Homer City's own power plant could mean for both.
The debate over data centers has turned into one of the loudest fights in American politics, and it landed squarely in our own backyard this year with Amazon’s plans at Homer City. The Economist recently argued that much of the national backlash is driven by exaggerated fears rather than facts. We think that is worth taking seriously, but it deserves real numbers, not just an opinion from London.
Start with electricity, since that is what people worry about most. A study by the Electric Power Research Institute, a nonprofit funded by government, academia and utilities alike, found that average residential electric rates would have been about 6 percent higher between 2019 and 2024 without the data centers built during that stretch. The mechanism is simple. Every electric grid carries fixed costs, the wires, substations and administration that exist whether you use a little power or a lot. Spread those costs across more electricity sold, and everyone’s share goes down.
That is the good version of the story. The honest version includes the bad one too. In the mid-Atlantic grid known as PJM, which covers Pennsylvania, demand from data centers has grown faster than new power plants have been built to serve it. The result has been the opposite of cheaper. PJM’s most recent capacity auction hit the federal price cap, and independent monitors found data centers responsible for roughly 63 percent of that increase, a bill of $9.3 billion passed on to ratepayers across thirteen states. Two thirds of Americans, according to a recent Harvard poll, already expect their bills to rise because of data centers. Given what has happened in our own grid region, that fear is not irrational.
The difference between the two outcomes comes down to one thing: whether new power supply arrives alongside new demand, or after it. PJM has caught on to this. Its newest rules require large new customers built after 2027 to bring their own power rather than simply plugging into the existing grid and letting everyone else pick up the tab.
This is exactly the arrangement taking shape at Homer City. The 4.5-gigawatt natural gas plant rising on the old coal site is not incidental to Amazon’s data center plans, it is the mechanism that could keep this project from becoming another PJM price shock. If the new generation comes online alongside the data centers it is meant to serve, rather than after, the region gains new supply instead of just new demand. That is the test to watch, not a guarantee, but a real one, and it will play out over the next several years as the project moves from paper to construction.
Water tells a similar story of real progress alongside real growth. Older data centers cool themselves by evaporating water into the air, which is efficient but consumes enormous volumes. Newer closed-loop systems circulate the same water again and again, cutting consumption by 40 to 70 percent compared to the old method. The technology has genuinely improved. At the same time, national data center water use is projected to more than double by 2028 as the industry keeps growing, so better technology has not yet meant less total demand, only less demand per facility.
What data centers cannot do is create many jobs for the money spent. They are capital heavy and labor light by design, a warehouse full of machines needs far fewer people than a warehouse full of workers. The jobs that do come with them tend to pay well. Amazon has told Indiana County officials the Homer City campus would bring 1,500 to 2,000 permanent positions at $80,000 to $100,000 a year, real money in this part of the state, even if the headcount is modest next to what an old-line factory once employed.
None of this settles the argument, and it should not. But the common sense version of the story is not that data centers are a menace or a miracle. It is that the outcome depends on the specifics: whether new power comes with new demand, whether cooling is designed to use less water, whether the jobs promised actually materialize. Homer City will be a real test of all three, playing out a few counties over instead of in a policy paper. We will be watching, and reporting, as it does.