On May 3, NERC (the North American Electric Reliability Organization, the non-profit corporation charged with ensuring the reliability of the North American power grid) issued a Level 3 Essential Action Alert, NERC’s highest level of alert. It bears the cumbersome title “Computational Load Modeling, Studies, Instrumentation, Commissioning, Operations, Protection, and Control”. The term “computational load” is how NERC refers to data centers.
The Alert requires organizations that operate parts of the grid on a day-to-day basis to start planning for some of the unique electrical risks that have been brought about by the huge increases in the number and size of data centers in recent years in recent years. This mostly unconstrained growth has led to new risks that are qualitatively different from risks in previous years; these new risks include the serious risk I discussed in this post in early March.
However, the Alert doesn’t even mention a) the likely exponential increases in data center numbers and size coming in the next few years – in other words, if you think this is bad, you ain’t seen nothing yet. How is that going to be handled? What it also never mentions is b) the trivial question😊 of how the needed grid build-out, including generation and transmission, will be paid for. Perhaps you noticed that your electricity bill is higher in the last year or two? Again, you ain’t seen nothing yet…
It’s completely understandable that NERC isn’t addressing either of these issues. Regarding a), they have enough issues to deal with just making sure the grid can handle its current challenges, let alone future ones. Regarding b), NERC isn’t responsible for answering questions about payments – that’s ultimately FERC’s job. However, the public in general can’t afford to let these issues go unaddressed, since they need both a reliable grid and affordable electricity.
This is why I was – at first – pleased to see on May 13 that the Washington Post had put out an editorial titled “The right response to the data center backlash”; it has the subheading “Electricity costs are rising in key markets, but the loudest fears are badly overstated.” I was less pleased when I read the editorial, especially when I remembered that WaPo’s new owner, Jeff Bezos, isn’t exactly a neutral observer when it comes to the subject of massive data center rollouts.
The second paragraph of the editorial doesn’t minimize the problem:
Last year, most people said they supported data centers, even in their town. Now, a poll by Gallup shows that 7 in 10 Americans oppose building a data center in their community. Nearly half are strongly opposed, with only 7 percent strongly in favor. Even nuclear power plants, the Great Satan of not-in-my-backyard activists, face less opposition.
The editorial goes on to point out, in a masterpiece of understatement, “An understandable concern about resource constraints drives much of the trend.” Yep, that sure sounds like what the problem is. People are concerned about paying for food and medical care, and don’t like the idea of having to settle for an unreliable grid as a tradeoff for affordable power.
The next paragraph states the affordability problem quite elegantly:
Data center demand is pushing the capacity of the grid. Residential electricity prices are expected to rise 5 percent nationally in 2026, on top of an 11.5 percent increase in 2025. And those are just averages. In February, electricity rates were up 26 percent year-over-year in Virginia — the world’s largest data center market — as well as 22 percent in Ohio and 20 percent in Pennsylvania.
So far, so good. The editorial was honest about the level of opposition to data centers, and it was also honest about how serious the affordability problem has become. I read on, expecting to hear how that issue could be addressed without compromising grid reliability (as I discussed in this recent post).
The answer to that question is in the next paragraph:
But energy doesn’t have to be a zero-sum game. Many data centers are motivated to build out their own energy capacity, which would help insulate nearby residents from price shocks while also ensuring a steady supply for themselves. Government can help by getting out of the way of energy producers and embracing an all-of-the-above approach to production.
The good news is that the WaPo editorial board has a solution to the problem of power affordability: We’ll just require the new data centers to practice BYOP – bring your own plant. Each data center will have its own power plant, so it won’t have to compete with the power needs of the general public. Thus, prices won’t go up any more than they already have (which is a lot, in some areas, as described earlier) and the public will still have access to the (usually) highly reliable power grid that it’s used to.
What does the public need to sacrifice in return? No biggie: just their health and the future of the planet. You see, in practice, the P in BYOP refers to pollution, not plant. The last sentence of the paragraph makes it clear that we need to let the data center developers come up with their own power generation solutions, unencumbered by pesky regulations on pollution, grid reliability, cybersecurity, etc. We just need to get out of their way.
As is often the case, Elon Musk (and his company xAI) is writing the book (in Memphis and Mississippi) on how to power a huge data center using a large number of heavily polluting gas turbines (50 in Mississippi – up from 27 a month ago - and 35 in Memphis), while pretending these are just a temporary solution and not a power plant that should be subject to regulation, or even permitting.
If you’re keeping score at home, the Memphis turbines constitute a plant of over 400 MW, while the Mississippi turbines total over 900 MW. For reference, Google AI says the average natural gas-fired plant generates 92 MW annually, so the turbines in Mississippi constitute a plant that’s ten times the size of the average gas plant. But since these turbines are just “temporary” (although there don’t seem to be any plans underway to replace them), xAI says they’re not subject to NERC and FERC regulation – let alone EPA regulation.
The Memphis and Mississippi data centers are just the tip of the iceberg. There’s simply no way that enough legitimate generation can be built within the next five or so years to power the data centers that are being proposed right now. Is the solution to impose a nationwide ban on new data centers? I considered that idea last year, but I no longer think it’s a good idea. Unlike fentanyl, new data centers provide benefits as well as drawbacks; they shouldn’t be banned outright.
The problem is that, as my former professor Milton Friedman regularly pointed out, when corporations, while pursuing a legal business, imposes relatively small costs on huge numbers of people that can’t be directly traced to the corporation (such as pollution or high electricity prices), there’s no good way for those people to get full and speedy resolution through the legal system.
In cases like these, the best course is to impose a tax on the class of corporate entities responsible for the problem and utilize the proceeds of the tax for the common good (which might include at least some direct restitution to affected individuals, like electric bill rebates to make up for the increased costs due to the data center buildout). It should probably be an excise tax targeted at data center developers, which they could pass on to the ultimate beneficiary of the data center, e.g. Google, Meta, Microsoft or Amazon.
How much should the tax be? I think it should be a lot. Increased pollution and higher electricity bills are just two of the costs imposed by the data center buildout. The excise tax needs to be high enough to cause the data center developers to rethink their plans and only build the data centers that are the most likely to pay off. Meanwhile, the proceeds of the tax can go to healthcare, education, or even some federal deficit reduction.
Tom Alrich’s Blog, too is a reader-supported publication. You can view new posts for two months after they come out by becoming a free subscriber. You can also access all my 1300 existing posts dating back to 2013, as well as support my work, by becoming a paid subscriber for $30 for one year (and if you feel so inclined, you can donate more than that or become a founding subscriber for $100). Whichever option you choose, please subscribe.
If you would like to comment on what you have read here, I would love to hear from you. Please comment in my chat or email me at [email protected].