“Very Few Jobs Is the Whole Point of Data Centers”

As data centers spread across the country, Cedar Rapids, Iowa, approved two, one from Google and another backed by Blackstone, with a combined investment exceeding $1.3 billion. In exchange, the developers are contractually obligated to create just sixty-one permanent jobs. In northeastern Ohio, a $136 million data center promises ten.

This week Futurism seized on those numbers, roughly $21 million of investment per job in Iowa, and declared that the economics "sound like a joke." The arithmetic is right.

But the joke isn't the investment. It's the yardstick.

Nobody calls a substation a failure

Judging a capital investment by the headcount it creates is the broken-window fallacy run backwards. The old fallacy praised destruction because repairing it "created jobs." This one condemns construction because it doesn't. Both mistake labor for the prize, when labor is the cost.

A modern electrical substation runs with no one inside it. A gas pipeline moves billions of dollars of energy past a handful of technicians. A container crane lifts a global supply chain with one operator in the cab. Nobody calls those failures. We call them productivity, more output from less labor, which is almost the only thing that has ever made a place richer.

By that measure a data center is the same kind of machine. Few permanent jobs isn't the flaw in the model. It is the model.

The boosters and the critics share one number

The economic-development office that promises the town "thousands of jobs" and the columnist who mocks the project for delivering around sixty are not really opponents. Both have agreed that a billion-dollar computing facility is graded on payroll, so the boosters wield the biggest number and the critics the smallest, and everyone argues about a figure that was never the reason to build.

Critics have legitimate complaints about these deals. The permanent job count is not one of them. A data center is not a factory built to maximize employment; it is a capital-intensive machine built to produce computing power. The question that matters is whether the computing power and the capital behind it justify the deal. If they do not, no jobs promise can rescue it. If they do, a low headcount should not sink it.

The real scandal is the subsidy

If you want a scandal, here is the real one. Those Iowa and Ohio job promises came wrapped in tax abatements, and nationally the giveaways are enormous. Georgia, Virginia, and Texas each now forgo more than a billion dollars a year in data-center tax breaks, and at least fourteen states will not disclose what they hand out. Then there is a subsidy that never looks like one: the cost of new transmission and generation gets spread across everyone's electric bill, so ordinary ratepayers finance a private computing boom.

That is the distortion, and it has nothing to do with the job count. The cure is not to write more jobs into the contract. It is to end the abatements and make each center pay the undisguised cost of what it draws, power first, but water and grid capacity too. The rule is simple: the company that runs up the cost pays it, not the household down the road.

Some breaks may pay for themselves in later taxes. But if a center's returns are that good, it needs no break to come, and a project that pencils only with one was never worth a town's money to rescue.

Do that, and the deal tells the truth about itself. Some centers will still pencil out; some will not get built, and the scarce grid will go to whoever values it most, a data center or the factory next door. That is the market deciding, not a tax office picking winners.

The job count was never the problem. The subsidy always was.