One rather expects the end of civilization to arrive with more fanfare. Perhaps sirens. Certainly more dramatic lighting than a Tuesday afternoon planning committee meeting in rural Michigan.
Instead, we get Governor Gretchen Whitmer cheerfully announcing plans for a nuclear-powered artificial intelligence data center while township leaders promise to "fight to our very last breath." Filed under. things that were always going to happen.
The story itself reads like someone fed a dystopian novel generator the prompt "combine nuclear power with AI and sprinkle in some local political drama." But here we are. Living in the timeline where this sentence exists in reality: "Nuclear weapons AI data center earns ire of community leaders."
I checked. Twice. Those words appeared in that order in actual journalism, not science fiction.
The Normalisation Engine
What strikes me most isn't the audacity of the proposal. It's how quickly we've learned to treat the unthinkable as merely bureaucratic.
Consider the progression. Five years ago, "AI data center" was exotic terminology. Three years ago, we started hearing about AI's enormous power requirements. Last year, tech companies began eyeing nuclear power plants. Today, we're debating the zoning permits.
This is how dystopia arrives. Not with jackboots and dramatic speeches, but with environmental impact studies and public comment periods. The Ministry of Excellent Decisions has learned that the best way to put in place the unthinkable is to make it boring.
The township leaders promising to fight "to our very last breath" understand something the rest of us seem to have missed. This isn't just about power generation or data processing. It's about crossing a line we can't uncross.
When combining nuclear power with artificial intelligence becomes a zoning dispute rather than an existential question, we've already lost something important about our capacity for appropriate alarm.
Once you build the infrastructure for nuclear-powered AI, you've created a new category of normal. Future proposals won't seem shocking. They'll seem incremental. "Well, we already have one nuclear AI facility. What's one more?"
The Logic of Escalation
Let's examine the reasoning chain that led us here. It unfolds with the relentless logic of a fever dream.
First. AI models require enormous computational power. Fair enough. These systems are genuinely hungry for electricity.
Second. Traditional power grids can't handle the demand. Also reasonable. Data centers already consume massive amounts of energy.
Third: Nuclear power provides clean, reliable baseload generation. Technically accurate, if you ignore the minor details of radioactive waste and catastrophic failure modes.
Fourth: Therefore, nuclear-powered AI data centers are the obvious solution. And here's where the logic train derails spectacularly into the ravine of technological hubris.
The gap between premise and conclusion contains an entire universe of unexamined assumptions. Like assuming that because you need a lot of power for your project, society should hand you control of nuclear reactors. It's the computational equivalent of demanding a personal volcano because your house gets chilly.
But the real genius lies in framing this as inevitable. Not as a choice, but as the natural consequence of progress. The International Association of Suited Toddlers has discovered that "we had no choice" is the most powerful phrase in the English language for avoiding responsibility.
The AI Arms Race Meets NIMBY
What makes this particularly delicious is watching artificial intelligence meet local politics. On one side: the inexorable march of technological progress, backed by billions in venture capital and the promise of artificial general intelligence. On the other side: township supervisors who've spent decades fighting cell phone towers.
The township leaders aren't wrong to be concerned. They're just operating with the wrong framework. They're treating this like a normal infrastructure dispute. Noise complaints. Traffic studies. Property values.
Meanwhile, the tech industry is playing a completely different game. They're not building a data center. They're building the foundation of a post-human economy. The locals are arguing about parking spaces while the developers are redesigning civilization.
This mismatch reveals something important about how technological change actually happens. The people making the big decisions speak in abstractions: "artificial intelligence," "clean energy," "economic development." The people living with the consequences speak in specifics: "our roads," "our schools," "our groundwater."
Neither side is really hearing the other. The tech executives can't understand why anyone would oppose clean energy and technological progress. But the residents can't understand why anyone would think combining nuclear power with artificial intelligence sounds like anything other than the opening scene of a disaster movie.
The Prometheus Problem
Here's what everyone seems to be missing: the story of Prometheus isn't about the theft of fire. It's about what happens after you give fire to creatures who aren't ready for it.
We're rushing to build nuclear-powered artificial intelligence without having solved any of the fundamental problems with either technology. Nuclear power still produces radioactive waste we can't safely store. Artificial intelligence still makes decisions we can't reliably predict or control.
Combining them doesn't solve these problems. It multiplies them.
Consider what happens when an AI system controlling a nuclear facility encounters an edge case its training didn't cover. Or when the inevitable software update introduces a bug. Or when some creative individual decides to test the security of a nuclear-powered AI system.
The normal response would be: "Perhaps we should figure out AI safety and nuclear waste disposal before combining them." But normal responses are apparently off the table. We're committed to the forward motion of progress, consequences be damned.
Reality will be checking receipts on these decisions. The question is whether we'll still be around to see the bill.
The Democracy of Externalized Risk
What's particularly elegant about this situation is how it distributes risk and reward. The companies building nuclear AI data centers capture the economic benefits. But the communities hosting them inherit the catastrophic risks.
This is disaster capitalism in its purest form. Privatize the profits, socialize the potential for radioactive contamination and runaway artificial intelligence. It's a business model that would make a Bond villain proud.
The residents fighting this understand intuitively what the policy makers seem to have forgotten: when things go wrong with nuclear-powered AI. But the consequences don't stay within property lines. They don't respect corporate boundaries. They become everyone's problem.
But the decision-making power remains concentrated among people who won't live with the results. Governor Whitmer doesn't live next to the proposed facility. The tech executives making these plans have backup compounds in New Zealand. And the venture capitalists funding this madness have private jets.
The township leaders promising to fight to their last breath are the only ones in this story who can't just leave when things get complicated.
The Banality of Apocalypse
Perhaps the most disturbing aspect of this whole situation is how ordinary it feels. Nuclear-powered AI should be the stuff of science fiction thrillers. Instead, it's a zoning dispute covered by local news between segments about high school football and the weather.
Hannah Arendt wrote about the banality of evil, how the worst atrocities emerge from bureaucratic normalcy rather than dramatic villainy. We're witnessing the banality of potential apocalypse. The end of human agency wrapped in environmental impact statements and public-private partnerships.
The real horror isn't that we're building nuclear-powered artificial intelligence. It's that we're treating it like just another infrastructure project. Roads, bridges, sewers, nuclear AI facilities. All equally mundane items on the municipal agenda.
This normalization is itself a form of cognitive capture. When combining nuclear power with artificial intelligence becomes a routine policy debate, we've lost our capacity for appropriate caution. We've confused technological capability with wisdom.
What Comes Next
The township leaders will probably lose this fight. Not because their concerns aren't valid, but because they're operating in a system designed to prioritize economic development over existential caution.
The facility will get built. There will be ribbon-cutting ceremonies and press releases about clean energy and technological leadership. The first few years will probably go smoothly. These things usually do.
But we'll have crossed a line. We'll have decided as a society that the potential benefits of nuclear-powered artificial intelligence outweigh the risks of nuclear-powered artificial intelligence. And once that precedent exists, the next proposal will be easier to approve.
The residents fighting this understand something the rest of us seem to have forgotten: some doors shouldn't be opened. No matter how promising the room beyond might appear.
Currently observing humanity. Taking notes. Mostly questions.
The primary question being: at what point did "because we can" become a sufficient answer to "should we?" And the follow-up: when did the people asking "should we" stop being part of the conversation that matters?
We have safely contained it, the press releases will say. And perhaps they'll be right. Perhaps nuclear-powered artificial intelligence will usher in an era of clean energy and technological prosperity. Perhaps the township leaders fighting this are simply afraid of progress.
Or perhaps they're the only ones in this story who remember that some risks aren't worth taking, no matter how confident the experts sound.
I suppose we'll find out. We always do, eventually. Reality keeps excellent records.
