The companies building data centers have decided they can’t wait for the labor market to fix itself. Meta, Google, and BlackRock are together putting more than $265 million into recruiting and training electricians, carpenters, and other tradespeople to build AI infrastructure. It’s a private answer to a shortage that used to be Washington’s problem.
How the academies work
Meta’s piece is the biggest, about $115 million for the first year of what it calls an ongoing effort. The model is aggressive: enroll roughly 5,000 people in a month-long course, cover their transportation and housing, then move them straight onto job sites with Meta’s contractors. It’s less a scholarship than a hiring pipeline the company owns end to end. Google and BlackRock are funding parallel efforts aimed at the same trades.
Why the hyperscalers are paying for it
Follow the bottleneck. A data center campus needs thousands of electricians, and there aren’t enough. Builds like the Paducah AI Data Center Campus project 8,000 construction jobs each, and the electrical trade is the tightest link in the chain. When your $100 billion buildout stalls for lack of licensed electricians, spending a few hundred million to make more of them is cheap insurance.
The shift underneath is the real story. For years the trades pipeline ran on federal apprenticeship dollars and community colleges. Now the largest single checks are coming from tech balance sheets, aimed narrowly at the skills those companies need. That solves the immediate crunch. It also raises a question the industry hasn’t answered: what happens to workers trained for one company’s buildout when that buildout slows, and who trains the trades that don’t happen to serve a hyperscaler.