The Whole Big Data Center Fight Is Not Really About AI
We ran this experiment before. The results are sitting in a book.
Although we feel unknown, ignored
As unrecorded blanks,
Take heart! Our vital selves are stored
In giant data banks…—Felicia Lamport, “Deprivacy,” 1970, from Sarah Igo’s The Known Citizen, p.221
On Tuesday, Kathy Hochul signed the nation’s first statewide moratorium1 on data centers. The politics seem cut and dried: New Yorkers back it 46–21. Not just blue New Yorkers, either; Republicans also favor it by thirteen points. Nationally, seven in ten Americans oppose a data center being built in their own area. When was the last time seven in ten Americans agreed on anything?
There are just a couple problems with the way people seem to want to tell the story about what’s happening.
Everyone — supporters as well as opponents — seems to agree that this fight is about AI.
I don’t. I think the data center fight is at best tangentially about AI, for a few reasons. First of all, that same Gallup report I just cited about how 70% of Americans oppose data center buildouts suggests at maximum 41% of that opposition is driven by AI-related concerns, with only 14% pertaining directly to negative views about AI itself.
Second, literal Kathy Hochul has said she likes AI and wants to see it succeed.
More generally, the anger and the technology just don’t share a constituency. Data centers are absolutely one of the most animated fronts in American local politics right now. In the first three months of 2026 alone, local opponents blocked or delayed at least 75 projects worth some $130 billion. The number of organized data center opposition groups more than doubled to 833 across 49 states, and statehouses fielded over 300 data-center bills in the year’s first six weeks alone.
But AI itself barely registers on voters’ radar2. In Gallup’s long-running open-ended “most important problem” series — where government, immigration, and prices soak up the mentions — it doesn't chart. And when you ask people to list out the problems they think government should be working on, AI doesn’t even crack the top ten.
The intensity runs in exactly the opposite direction you’d expect for something that’s fundamentally about frontier models themselves, as distinct from e.g. a bunch of creepy-sounding aesthetically-ominous buildings owned by faraway entities of whatever sort.
Okay, but that’s a bunch of analysis. Might I instead sway you with some anecdotal and historical reasoning?
Meet my Mom. She is a bit of a legendary asskicker. She was also bigly up in arms about Elon Musk’s Colossus buildout in Memphis back in 2024 (along with about a zillion other people). And with reason! The community’s first contact with the project, in June of 2024, was learning that a decision had already been negotiated under NDAs over the preceding months, with City Council members finding out from the local news, and the facility’s location initially withheld from the public due to “global security concerns.” Within weeks the thing was humming along on dozens of allegedly unpermitted gas turbines.
So she was mad about the smog3, and mad about the water and the power4, and mad about Elon Musk in his capacity as Elon Musk, and mad about whether one dollar of the multibillion-dollar bounty would ever reach an actual Memphian.
What she wasn’t mad about was AI.
My Mom loves AI. She sends me AI-generated tattoos she thinks I ought to emblazon on my body, and posts AI-generated makeovers of our whole family as Christmas cards, and writes “sharing again — it’s so cool” on the most AI-slop-possible rendering of a bunch of kids contorting their bodies into something that maybe sorta looks like the face of Jesus. She is not out there railing against her God-given right to post the following image publicly on the internet for all to see:
And I just don’t think this is some wild, out there, profoundly anomalous opinion.
But the biggest reason I don’t think this fight is about AI is that we already had a version of it sixty years ago, decades before computers themselves became mainstream. In an era of mainframes, the country approached full-blown national revolt over something they called “data banks”.
Before I tell that story, though, let me tell you how I happenstanced upon it.
A few years ago my friend Sherry introduced me to an incredible woman named Li Schorr, who had been on the founding staff of the Office of Economic Opportunity. I was mid-spiel about the importance of cross-partisan efforts to get policies to the finish line by bringing together counterintuitive combinations of people in common purpose across divides — this was right after we had launched Welcome.US — when Li mentioned she had explored the topic previously.
“Oh yeah?” I asked — expecting, like, a take.
“Well, I wrote a book called Common Purpose, about two decades of cross-partisan efforts to get policies to the finish line by bringing together counterintuitive combinations of people across divides.”
Ah. I see.
“Perhaps you should read it.”
Yes. Perhaps I should5.
So here is a maxim I have now paid for several times at retail: when a situation feels unprecedented, go check whether someone else has thought about it. Often, someone smart will have been sitting with your “brand-new” problem for fifty-plus years.
Which is how I ended up reading Sarah Igo’s The Known Citizen, a thorough academic history of privacy in America.
Lo and behold, there’s a chapter called “The Record Prison” about giant physical structures, built at speed, over the objections of the communities around them, housing mysterious data those communities couldn’t see. And lo and behold, a major national movement rose up to fight them.
In 1965, the federal government proposed something called the…drumroll please…National Data Center. The idea grew out of a study concluding that the government’s statistical information was scattered too widely to be useful. Consolidating it all in one place, Igo writes, “struck advocates as eminently rational.” The benefits were “self-evident.”
In Igo’s magnificently dry summary: “None of these planners seem to have anticipated any public outcry.”
Well, there was an outcry.
The planners had neglected privacy so completely in their early proposals that one of them later admitted it had been a “gigantic oversight.” Congress, journalists, scholars, and ordinary citizens pilloried the thing as a “monster,” an “octopus,” and — my personal favorite — “a great, expensive electronic garbage pail.”
The proposal died in 1965. It was revived and killed again in 1967, and revived and killed once more in 1970.
The privacy scholar Arthur R. Miller6 supplied the line that explains the last sixty years of this unease better than anything published in the last sixty days. The National Data Center proposals were, he observed, a “lightning rod for the vague feelings of discontent generated by the computer revolution.”
And I just think both this observation and this metaphor are basically perfect.
A lightning rod doesn’t cause the storm. It certainly doesn’t contain the storm. It’s just the tallest conductive object around, so all the charge accumulating in the atmosphere can’t help but zap right through it.
Those ‘vague feelings of discontent’ didn’t go away just because the National Data Center project faltered. Soon, arrays of “data banks” proliferated, public and private, a vast web of menacing information leveraged against you by unaccountable forces beyond your control.
In the coming decade, the privacy-invading technology that would surpass all others in Americans’ imaginations was the data bank. A steady drumbeat of congressional hearings, scientific commissions, print and broadcast exposés, scholarly studies, science fiction accounts, and social protests would conspire to make record systems, especially computerized ones, the new face of privacy fears.
-Sarah Igo, The Known Citizen, p.226
The danger posed by these data banks — instantiated in the forms of row upon row of depersonalized, rectangular mainframes — wasn’t just hypothetical. “Many consumers’ first brush with the potency of their files came when they were denied insurance or had difficulty applying for a job or a loan,” Igo points out.
And the specter of such eldritch power loomed large in the cultural zeitgeist. “Hidden monitoring devices, vast warehouses of private information, and menacing bureaucracies loop through the cultural and political texts of the period,” Igo says. “Films like The Parallax View (1974), The Conversation (1974), and Three Days of the Condor (1975) took the perspective of lone men…trapped in the labyrinthine plots of shadowy but powerful organizations.”
Backlash ensued in full force. Igo notes that “data centers on university campuses that held big mainframe machines were the target of thirteen direct protest actions between 1968 and 1972.” And these weren’t just fringe figures, either. The President even got involved!
President Gerald Ford voiced the threat posed by computerized government records in vivid, nightmarish terms. In a speech at Stanford Law School, he declared, ‘We must protect every individual from excessive and unnecessary intrusions by a ‘Big Brother’ bureaucracy,’ portraying the vulnerable citizen as a ‘faceless set of digits in a monstrous network of computers’.
-Sarah Igo, The Known Citizen, p.248
There are plenty of distinctions between then and now, of course. The sixties feared the exposure of “secret lives” — the “Dossier,” as one magazine headline put it. The nefarious constructor of data centers then was the government, where today it’s nominally private companies7. But Igo’s summary of the era’s conclusion — that “neither the security of personal data nor the trustworthiness of its keepers could any longer be assumed” — could be pasted into any 2026 op-ed without edits.
So here is what I think is actually going on. We already ran this experiment: the data-bank panic of the 1960s subsided not when the machines stopped coming — as we know by now, they never stopped! — but when Americans got a working pact with the people who ran them: an uneasy, imperfect, but concrete assurance of trust.
We know a moratorium can’t produce trust by itself, because the sixties tried one of those too.
That’s what blocking the National Data Center was, after all: the moratorium to end all moratoriums. The backlash won. The building was never built. But the panic didn't abate — it intensified for nearly a decade. Congress passed the Fair Credit Reporting Act in 1970, but Watergate poured gasoline onto fears of large-scale federal snooping. The Privacy Act followed in 1974.
Those laws didn’t stop the record systems from being built — as we know all too well, our information lives everywhere. What they did was give people rights against those systems: to know your file existed, to see it, to correct it, to have norms of “fairness, accuracy, and security” govern the whole arrangement. Even more than conferring such rights in practice, these laws helped engender both the feeling of a more secure relationship and an impression that someone was actually exerting effort to earn your trust.
Why does any of this matter, though?
Well, for one, it explains why debunking doesn’t work. A lot of the data center discourse is grounded on numbers that don’t hold up. The water numbers in particular are mostly fake news, as Andy Masley has demonstrated at heroic length, and many other environmental impact claims simply don’t survive scrutiny. Debunking hasn’t moved opinion much at all, though. Why? The ‘lightning rod’ metaphor is useful again here: any given statistic is less the storm and more the conductor the charge happens to be traveling through today. Knock it down and the charge finds the next ‘tallest’ thing.
But mainly, I just don’t buy that the animus against data centers proves there’s a massive popular backlash against AI. If that’s what was going on, why did such a similar phenomenon emerge sixty years before AI existed?
Now, what I absolutely do buy is: there’s a backlash against outsize personal and data-security risk; against colossal, omnipresent surveillance; against the raw, rank power of big tech companies enabled by big governments; against the physical imposition of those forces on communities that never chose them; and against the felt powerlessness of having all of it presented to you — behind an NDA and after the fact — as inevitable.
These grievances represent some of the most salient issues in society today — just as they represented some of the most salient issues facing society in 1968. They predate the chatbot by more than half a century. And they’ll last another half a century, too — chatbots or otherwise — until the roots of the problem are dealt with.
Which brings me back to Governor Hochul. Parts of her order, I think, have the right shape. The order directs the state to stand up a Community Investment Framework, and to consider a fund making data centers pay into the grid they strain. Those are pact-shaped instruments — terms, obligations, a relationship you can inspect. They represent baby steps towards trust. If the year gets spent building that, good.
But notice what the pact on offer covers: your utility bill and your water table. Those are the complaints, not the grievance. The 1970s pact worked — to the uneasy, partial extent it worked — because it addressed the grievance itself: it made the record keepers legible and accountable to the people they kept records on.
A pact about kilowatt-hours, negotiated while construction is frozen, is less a clearing of the sky than a rain delay.
Sixty years ago we learned how this fight ends. Not when the technology stops marching forward — it never does. It ends when the people behind the technology offer the rest of us an assurance we can believe.
Until then looms the thundercloud, crackling to strike the nearest rod.
Technically, what’s targeted are new hyperscale builds, 50 megawatts and up, for up to a year.
When you ask people about it directly, for sure, people will say they are concerned. But that’s different than the kind of encompassing day-to-day salience that leads people to stop scrolling Bubu and Dudu videos and get off their couch and take some kind of action.
With good reason.
With (debatably) less good reason, depending.
The book is legit awesome, by the way. One thing you realize is how much more seriously the act of policy analysis qua analysis was taken in an era before everyone had already decided what they thought and were supposed to think.
Though even here, I’m not sure how live the distinction is. “Few distinguished between state and commercial uses of citizens’ information,” Igo notes; what mattered was that an organization “collected and housed information about millions of citizens, using it to make any number of consequential decisions about them.”
Utilities and governments are UNDOUBTEDLY using the extra power demand from AI to both prolong the life of coal and build new fossil plants, ensuring grids don't get cleaner even with massive new clean energy build. So it's entirely reasonable for those who view the rapidly warming climate as an integral issue that the cost benefit for the proliferation of AI data centers doesn't match up.
And that's not even to say the other negative externalities issues from projects that are being fuelled by massive debt and leveraging, on a premise (LLMs will lead to AGI) that is likely bullshit.
> Watergate poured gasoline
And since it was the 70s, that gasoline was full of lead!