Is AI Taking Your Job? Part 1: The Fancy Computers Are Here. Where Are the Jobs?

AI is eliminating some jobs and reshuffling budgets in ways that make others disappear. So, if companies are pouring hundreds of billions of dollars into AI infrastructure, does any of that spending create jobs — and if so, whose? This is all money spent on really fancy computers, rooms full of chips, cooling systems, and power infrastructure. So what does building the world's fanciest computers get the rest of us, in terms of permanent jobs?

A Brookings study across roughly 770 U.S. data centers matched to BLS county employment data found real local gains, with some caveats attached. Counties that get their first large data center see total private employment rise 4 - 5% over five to six years, with construction employment up 11%, information-sector employment up 22%, and wages up 3 - 4%. Data centers create local jobs, but estimates that ignore existing growth trends can overstate the impact threefold. That 22% figure is county-wide - jobs at surrounding businesses, not inside the data center itself. It turns out not all data centers produce it. A typical county with 98,000 workers could gain about 2,000 to 4,000 jobs within six years, depending on the data center type.

The Brookings researchers split facilities into two types. Hyperscale campuses are built by companies like Amazon, Google, Microsoft, and Meta to run their own workloads, while colocation facilities are built by landlords like Equinix or Digital Realty that lease server space to outside tenants. Only hyperscale counties see the real information-sector gains. A hyperscale campus creates local demand for fiber installers, network technicians, and managed-service providers who set up nearby to serve it. Colocation facilities may create fewer local jobs because remote tenants can lease server space without hiring workers in the area. Scale matters too. A single facility barely moves the jobs needle, but counties with four or more facilities see a 23% jump, suggesting the spillover effect needs a cluster to develop, not just one building.

Narrow the question further to permanent jobs at the fancy computer itself, and the numbers get smaller regardless of type. Virginia data centers generate roughly one permanent job for every $13 million invested. Meta's specifically work out to one permanent job per $33 million invested, worse than TSMC's Arizona chip plant, which manages one job per $14 million. Economist Michael Hicks came to a bleaker conclusion. Studying data center development in Texas, he found close to zero net new jobs at the county level once you account for workers who simply moved over from other local industries rather than being newly hired. 

To be fair to the fancy computers, none of that rules out job creation happening somewhere else in the company entirely. A data center itself might run on a skeleton crew, while the company that owns it grows its workforce elsewhere - new roles in engineering, sales, or operations funded by whatever the infrastructure eventually produces. 

That's a different, company-wide question from "how many people work at the data center.” So, do the fancy computers themselves employ people permanently? Barely. Does the company that owns them grow its overall permanent headcount because of the investment? That's a live, contested question for Part Two of this article.

Written with AI assistance; edited and fact-checked by the author.

Photo by Igor Saikin on Unsplash