Responsible Opposition

This one’s for the people who agree with me.

The anger here is earned. A billion-dollar facility is being dropped into a township that gets no vote on it, by an institution that told our supervisor about a land purchase by text message after the closing, and that still hasn’t answered basic questions after nearly two years. Nobody needs talking down from that. It’s a reasonable reaction to what’s actually happened.

But anger isn’t a strategy, and right now it’s feeding a rumor mill that’s costing us more than it’s buying.

The math of this fight

On one side: a university with a communications office, a government relations department, outside counsel, and a federal laboratory partner. On the other: a township full of people doing this after work.

We’re not going to outspend them or outstaff them. The only advantage we’ve got is being right.

And it’s a fragile advantage, because it works in one direction. We have to be right every time. They need us wrong once, in public, in front of a reporter — and then every careful, sourced, documented thing we’ve said gets filed under “the group that claimed there’d be a nuclear reactor.”

That’s not hypothetical. It’s the most common way community opposition dies — not beaten on the merits, just waved off as unserious.

Where the rumors come from

Worth understanding, because it isn’t stupidity and it isn’t bad faith.

When an institution withholds information, people fill the gap. That’s a normal human response to being kept in the dark, and the University’s refusal to publish basic operating figures is what created the vacuum. Give this community the megawatt number, the water number and the cooling design and there’d be a lot less to speculate about.

But knowing why a rumor started doesn’t make it true, and it doesn’t make repeating it free.

Take the one about the facility generating its own power with nuclear reactors. It isn’t sourced to any document or any statement by anyone involved. It assembled itself out of three true things sitting near each other: Los Alamos is a nuclear weapons laboratory, tech companies really are signing nuclear power deals, and nobody will tell us how this thing gets its electricity. Put those in a room and the conclusion forms on its own.

Half an hour of checking shows it’s not just untrue but not possible on this timeline — no commercial small modular reactor operates in the United States, federal licensing alone takes longer than this project has, and the published budget has no generation line in it at all.

Say it at a public meeting and you’ve handed the panel a gift: a wrong thing to correct, on the record, instead of a right thing to answer.

Six habits worth having

1. Ask where it came from before you pass it on

One question does most of the work: what document or statement is this from? If the answer is “someone at the meeting said,” that’s worth knowing and worth checking. It isn’t yet worth repeating as fact.

2. Keep four things apart

What they said. What a document shows. What follows from those. What we think about it. All four are fair game. Collapsing them is how a reasonable inference becomes a false claim in three retellings.

3. Say “I’ve heard” when that’s what you mean

Nothing wrong with raising something unverified as long as you label it. “I’ve heard X and I’d like it confirmed or denied” is a perfectly good question at a public meeting. “They’re doing X” is a claim, and you own it.

4. Give ground where they’re right

Some of what the project says is accurate and some of its choices are sound. The roughly 300 union construction jobs are real and will largely go to Michigan trades who’ve earned them. Limiting backup generation to about 10% of load is a legitimate engineering decision for research computing — I’d have specified it the same way. No plutonium or nuclear material will be stored on that site.

Saying so costs nothing and buys everything. A room that watches you concede three points believes you on the fourth.

5. Keep it about the institution, not the people

The engineers, researchers, facilities staff and contractors on this project are doing their jobs. Elected and appointed officials answer for their conduct in office — their families, homes and private lives never.

Beyond being the right thing, it’s practical. The moment this gets personal, the story stops being about a township with no vote and becomes a story about people being harassed. We’d lose that story, and we’d deserve to.

6. Correct yourself out loud

This is the one people resist, and it’s the one that matters most.

Admitting a mistake feels like handing the other side a win. It’s the opposite. Someone who corrects themselves unprompted is someone whose other claims you can trust, and everyone in the room knows it without being told. Being willing to be publicly wrong about a small thing is exactly what earns you the benefit of the doubt on a big one.

The alternative — defending a claim you’ve privately realized is shaky — is how a credible community turns into a noisy one.

Mistakes I’ve made

I’m not writing this from above the problem. Three I caught in my own work.

The construction cost argument. I was close to publishing a claim that dividing $1.25 billion by 100 megawatts proved this was a fully built-out, liquid-cooled AI facility. Turns out the industry benchmarks I was using cover shell and core construction only — they exclude the servers, networking and storage. The inference didn’t hold. I cut it before it went out.

The staffing number. I said permanent on-site staffing would be “a few dozen.” Published estimates are wider than that, and some industry sources put hyperscale staffing considerably higher per megawatt. The honest answer is a range, and the site now gives the range instead of the number that suited me best.

A statistic that would have helped me. A widely circulated data center policy brief — published by the University of Michigan’s own Ford School, which makes it rhetorically handy for us — has an error in it confusing megawatts with megawatt-hours. I could have quoted it. I don’t, and I flag the error wherever I cite that brief for anything else.

None of those were fun. All three left the argument stronger.

What to say instead

If you’re speaking at a meeting, writing a legislator, or posting about this, you don’t need anything speculative. The documented case is stronger than the rumors, and all of it is sourced on this site.

Roughly 90% of the computing capacity belongs to Los Alamos, not the University — confirmed by the project leader at a public meeting and in the University’s own filing to the State.
The signed agreement with Los Alamos describes a facility in Ann Arbor and never mentions Ypsilanti Township.
The State was offered up to $400 million in 2023 for a laboratory at a different site — the American Center for Mobility.
The board was told to expect over 800 permanent jobs in October 2024. Three weeks later the figure was 200. Nothing released explains why.
The larger job numbers come from a state model that treated this as a software company.
No figure for water use has ever been published — and in May 2024 water and utility figures were designated confidential in an agreement with the State.
The facility will pay no local property tax, and no payment in lieu of taxes has been offered.
Asked how many permanent jobs would go to township residents, the panel said they weren’t sure they could legally say any would.
Almost none of the commitments made to this community are enforceable by anyone.

Every one of those is documented. Not one needs embellishing.

At the microphone

Ask one question. A stacked question gets the easiest part answered and the rest dropped. One specific question, then stop talking and let them answer.

Ask for a number or a date. “How many gallons a day?” is answerable. “Why are you doing this to us?” isn’t, and what you’ll get back is sympathy instead of information.

Have the follow-up ready. When the answer is a reassurance instead of a figure, the useful reply is: that’s not a number — by what date will you publish it? A date is something you can hold them to.

Write down what they say. Who said it, what they said, when. A record of unanswered questions is evidence in itself, and it piles up.

Last thing

None of this asks anyone to be less angry or to want this less. It asks us to be the people in the room who did the reading — because that’s the role actually available to us, and because a community that’s consistently, boringly accurate becomes very hard to dismiss.

If you think something on this site is wrong, tell me. I’ll check it, and if I got it wrong I’ll fix it and say so.

Sources for everything in “what to say instead” are on the timeline and sources pages.