The Data Centre Nobody Asked For

There’s a story doing the rounds about a small Michigan township where the University of Michigan and Los Alamos National Labs want to build a “massive data centre.” Locals turned up to a town hall to say they’d never been consulted, and the meeting turned into something close to a rebellion. Turns out the whole thing is technically a supercomputer for fusion research, not a bog-standard AI data centre, and half the comment section spent its energy arguing about that distinction instead of the actual point residents were making: nobody asked them.

I’ve been half-following the data centre backlash for a while now, mostly because it scratches two itches at once. I’m genuinely fascinated by what AI can do, and I’m genuinely worried about what it costs to make it do that. Those two things don’t resolve into a tidy position. They just sit there, uncomfortably, most days.

What struck me reading through the comments wasn’t the fusion-versus-AI argument, though that’s a fair one to have. It was how many people said some version of “it took data centres for people to finally notice they don’t matter to the people making these decisions.” That’s a big call, and probably an overstatement, but there’s something in it. A supercomputer, however noble the research goal, still needs land, water, power and roads. Someone has to live next to the hum. And in a lot of these stories, the someone finds out from a news article, not a letter in the mailbox.

We’ve had our own version of this here, just with different props. Melbourne’s outer suburbs have spent a decade fighting over where to put things nobody wants in their own backyard: substations, quarries, the West Gate Tunnel spoil, and more recently, warehouses the size of small towns servicing the online shopping boom. The pattern is always the same. Someone senior signs off on something in a spreadsheet, the community finds out once the trucks start rolling, and then everyone gets called NIMBYs for being annoyed about a decision they were never part of. Sometimes the project genuinely is worthwhile. Sometimes it’s just profitable. From the footpath, those two things look identical until it’s too late to do much about it.

I don’t think every data centre is evil and I don’t think every fusion project is a scam wearing a lab coat. What I do think is that “trust us, it’s for research” and “trust us, it’s for progress” have both been used often enough, by people who turned out not to deserve the trust, that scepticism isn’t cynicism here. It’s just pattern recognition. A council or a university doesn’t get to skip the conversation because the outcome is supposedly good for humanity. Good outcomes still need buy-in, or at least an honest attempt at it, otherwise you’re just doing the right thing in the wrong way, and wondering later why nobody thanked you for it.

There’s a broader tension I keep bumping into with AI generally: I want the breakthroughs, the fusion research, the protein folding, all of it. I also want someone to be honest about the water, the power draw, the land use, before the bulldozers turn up. Those two wants don’t cancel each other out. They just mean the conversation has to happen earlier than it usually does, with the people who’ll actually hear the hum at 2am, not after the ribbon’s already been cut.

None of this has a neat ending, and I’m not going to pretend it does. But if there’s one thing worth taking from a bunch of angry Michiganders at a town hall, it’s that consultation isn’t a box to tick after the decision’s made. It’s supposed to happen before. Everything else is just PR dressed up as democracy.