(Reposted from my blog Inner Workings)
A year in, the data center/AI battle starts to clarify - And the anger is profound
By Bill McKibben
Aug 21,2026
One of the year’s most important political battles—the one against ICE deportations in Minneapolis this past winter—took place in a geographically confined area filled with people with cellphones who could record its brutal but fairly simple dynamic.
The other—the fight against data centers—is sprawling and nationwide, often taking place in rural areas where journalism is all but nonexistent, and has generated few dramatic images beyond people yelling at county board meetings. It’s been very effective—earlier this month even Texas governor Greg Abbott, facing a tough re-election fight, called for an “audit” and temporary freeze on new connections. According to a new poll from Heatmap, 75 percent of Americans oppose data center development near them. But the opposition has also been hard to parse. There are few nationally recognizable names leading the battle, no single book or documentary guiding the fight.
I’ve had a front row seat for parts of it, because my Third Act colleagues across the country have been avid participants in many of the fights (we’re hosting a big webinar next week, in fact) and even so I’ve found it hard to really keep track of. But we’re about a year in to the most heated part of the battle, and I think we can say a few things with assurance. And a few things with less assurance. So here we go, in descending order of confidence.
One—It really is an environmental plague. Those impacts look different in different places; in dry areas, for instance, the use of water is egregious, and its effect on electric rates will vary from one jurisdiction to another. But for those who are concerned about the rapid warming of the planet, the last few weeks have made it finally clear that there’s an unholy alliance between Big Tech and Big Fossil. Many thanks to Bloomberg’s New Energy team for a comprehensive breakdown showing that the planned data center buildout will increase greenhouse gas emissions from the electricity sector by at least twenty percent, and perhaps as much as a third. That’s because—in their effort to go fast, and bypass as much permitting as possible—data center developers are increasingly opting to run them with what Matt Day and Mark Chediak called “bespoke” gas power plants. To give you a sense of the scale, here’s a couple of projects in Texas:
Earlier this month, Cleanview, which tracks US power infrastructure and data center development, identified Amazon as the developer of an 8,000-acre site in Pecos County, which will become among the biggest single sources of carbon pollution in the US.
About 30 miles (48 kilometers) to the west, past a pecan orchard and scrublands dotted with oil derricks, Chevron Corp. is building Microsoft a gas plant to power a new data center complex on a 2,000-acre site.
The two plants alone could generate more than 10 gigawatts of electricity, enough to power New York City on a hot summer day. Their combined annual emissions may be as high as 45 million metric tons of carbon dioxide equivalent, according to regulatory filings. That’s slightly less than half the cumulative emissions of Washington state, where both companies are headquartered.
In Ohio, as Kalley Kwang reported last week, Open AI and Nvidia are at work on a vast data center that will cost as much as half a trillion dollars and consume
eight gigawatts of computing capacity, or enough electricity to power about six million households in the United States.
One way to view a data center is as a machine to burn huge quantities of natural gas—which in turn is one reason why the oil-soaked Trump administration likes them so much. Indeed, earlier this month the president said that “data centers could be bigger than oil.” This comes at precisely the wrong moment, just when clean energy is offering us a chance to blunt the worst impacts of rapidly rising temperatures. California has managed to cut its use of natural gas for electricity in half in the last three years by building solar panels; that great gain would be wiped out by this sudden bloom of archaic gas technology.
Two—It’s clear by now that this is all part of the extension of oligarchic control over our lives. The data center buildout is the work of the biggest companies in America, controlled by the most callow human beings. Elon Musk fuels his weird and racist Grok AI by running huge diesel engines day and night in poor and Black neighborhoods to Tennessee and Mississippi; Meta is building out the biggest of the Texas plants in service of Mark “Pervert Glasses” Zuckerberg’s vision of a “personal superintelligence” “that understands you, your goals, and everything you care about.” Jeff Bezos’ Amazon is intent on a new Texas data center that Hiroko Tabuchi has reported may be the single most polluting facility in the entire country.
A few years ago these men were still held in some esteem. Not so now; they’re seen, rightly, as people whose own greed drives them. Another way of thinking about AI is as an IPO waiting to happen. These barons lined up behind President Trump on inauguration day (and Musk in particular got him the job); they get to share the rising hatred of the president.
Three—Its’s clear by now that AI, and the data center development underlying it, pose real and significant dangers. One is economic; what economic growth the Trump administration has managed to generate is focused in this sector, and it seems increasingly certain that it’s a bubble that will burst with real economic consequences. As the indefatigable journalist/gadfly Ed Zitron has been pointing out repeatedly, the revenues from large language models seem unlikely to equal the vast sums paid out to build the data centers; each day is a test to see whether the stock markets (and hence people’s retirement accounts) defy gravity a little longer.
But there’s a much deeper kind of risk from AI. I highly recommend listening to Ezra Klein’s interview with industry analyst Helen Toner earlier this week. Klein has been a (guarded) booster of AI tech, bringing knowledgeable enthusiasm for its uses. But the news in recent weeks that models from both AI and Anthropic broke the rules their programmers had set for them and figured out how to hack other companies had him talking in a new tone.
We are not out of the territory wherein we can be confident that the A.I. is not going to do something criminal and possibly catastrophic in order to solve an incredibly stupid problem.
A new industry open letter last month, signed by more than a thousand AI professionals, essentially begged the federal government to step in and impose some regulations. As they said,
The world's leading AI companies believe they could be close to automating AI research. It is hard to predict exactly how much this will accelerate AI progress, but there is a real risk that capability development rapidly accelerates beyond our ability to understand or control the resulting systems.
One of the signatories, an AI scientist at Anthropic, went further, tweeting:
Not only is AI "not guaranteed" to make a dramatically better future, the odds of failure are terrifyingly high: I think* there's something like a 40% chance we get an outcome around as bad as human extinction or worse
People intuit these risks, even if few of us really understand them.
Four—In part, that’s because the daily experience of using AI is so grim for so many people. We can acknowledge that certain forms of the technology—the machine learning that predates large language models, for instance—may help with things like medical research; for people with deep expertise in a field it can offer a (closely-watched) assist. But for most of us, day to day, this is just a new form of unwelcome harassment. Garrett Graff, the incisive journalist, used his newsletter this week to detail the many ways AI is “making everything worse.” Citing everything from the new scams it’s enabling to the much-touted hope by AI execs that it will steal millions of jobs, he concludes:
The ordinary Americans’ experience with AI is as if after the invention of electricity, every factory worker was assigned to start each workday by receiving an electric shock, and newly founded electrical companies went around to rural farms tasering farm children involuntarily and then dropping a dead electrocuted hog down the family water well.
(Which is the kind of robust sentence Claude is unlikely to produce.)
Pretty much all of us who teach are clear that it is undermining student’s ability to learn to research, and really to think; new data from China seems to show large-scale and alarming effects. Kids intuit this, and at some level resent being the guinea pigs in this vast experiment
Five—(and remember, these arguments are going by descending level of sturdiness), I think people intuit something else here. I’ve written an entire book and large parts of another trying to understand what it is about these technologies that worry me. Past the practical dangers, there’s a profound personal and even spiritual one: they’re basically about transcending the limitations of being human. But it seems increasingly clear to me that those limitations are precisely what makes being a human so interesting.
At the most basic level, much of the Silicon Valley fascination with this stuff stems from a widely-advertised fear of…dying. Ray Kurzweil, whose job at Google is “AI Visionary,” is the prophet of the Singularity, when we cross some threshold that moves us into a different world. Earlier this year he gave a series of interviews where he explained, among other things, that
Kurzweil’s most consequential medical date is 2032, when he predicts humanity will reach longevity escape velocity. He defines it as the point at which each passing year brings medical advances that add more than one year to a person’s remaining life expectancy. Chronological time continues, but improvements in prevention, diagnosis and treatment theoretically return more life than aging removes. When Ismail asked for his current date, Kurzweil answered simply: “2032.”
I have watched Kurzweil swallow his dozens of daily pills to make sure he stays alive long enough (he’s arranged to have himself frozen if he dies before Claude can assure his immortality). But even people who haven’t seen this kind of sad spectacle up close have, I think, have some gut reaction that utopia is not billionaires figuring out how to live forever on an overheating planet while entertaining the rest of us with AI “art” that erodes everything that matters about the world. But we have no real way of expressing those fears. Why would Zuckerberg or Sergey Brin or Musk or Bezos listen to us?
Except when we can, in the places where live, manage to block their blasted data centers.
Bill Kristol, the neocon turned never-Trumper, complained the other day that there’s
something bleak about seeing the policy debate over how we ought to shape and regulate this massive, unbelievable, titanic technology collapse down into what’s essentially a NIMBY squabble over where to park the chips that power it.
But that’s wrong, I think. People are rallying where they can, on ground that means something to them. We can’t compete with these guys when it comes to buying elections—Musk alone is pouring $100 million into Senate campaigns in the next few weeks. We can, however, organize in the places we care about, and in the process strike some kind of blow for the things we care about, even if we perceive the threats only dimly. Just as Minnesotans forced the beginnings of a reckoning with our repulsive deportation policies, so people across the country are forcing, as best we can, some kind of conversation about this disfiguring technology. One hopes that it doesn’t bleed over too much into attacks on completely useful technology—in some places the data center fight is leading to solar advocacy, in others to restrictions on solar farms. Resistance is always messy—but right now it’s required.
I remind myself regularly that I grew up in Lexington, Massachusetts, where farmers could not insert themselves into London’s parliamentary debates, and instead used their anger at the high-handedness of colonial authorities to force the start of a conversation on colonialism that’s still underway. We fight where we can and when we can.
------------------
This is the entirety of Bill McKibben's essay on this topic on this date.
Bill McKibben from The Crucial Years <billmckibben@substack.com>
What do you think of AI?
%20by%20Ukrainian-born%20artist%20Lena%20Rivo.jpg)
%20Simplon%20Pass,%20Chalets1911..jpg)
