The July jobs report landed this week with a split most people scrolled past: US employers cut 23,000 jobs overall, while construction firms added 22,000, almost all of it nonresidential and infrastructure work, not homebuilding. The cuts came out of local government, leisure and hospitality, and retail. Residential construction lost 44,000 jobs over the past year. Nonresidential and infrastructure construction gained 126,000 in the same stretch.
The cuts and the gains aren't the same story — government and hospitality payrolls have their own troubles, and none of them are about AI. But the gain side has one obvious driver, and it lands on a payroll that looks nothing like the one you budget for.
The Bureau of Labor Statistics doesn't tag any of this "AI hiring." It doesn't have to. Nonresidential specialty trades added nearly 78,000 jobs over the past year. Heavy and civil engineering added 21,000. Durable-goods manufacturing added 18,000 in July alone, architecture and engineering another 4,600. Every one of those categories is where you'd expect hiring to show up if the thing actually being built is a data center: concrete, power equipment, copper, chips, servers, and the labor to put them together.
Joe Brusuelas, chief economist at RSM, has described a "historic cap-ex super cycle" propping up demand for construction and goods-producing workers. The numbers behind it aren't small: the two-year run rate on AI infrastructure investment is already $1.6 trillion, with $4.5 to $5 trillion projected over the next five years. That capital isn't going into software payrolls. It's going into steel, cooling systems, and the electricians who wire them — which lines up with what Jensen Huang has been saying since January.
"It's wonderful that the jobs are related to tradecraft, and we're going to have plumbers and electricians and construction and steelworkers," Huang said at Davos. He added: "Everybody should be able to make a great living. You don't need to have a PhD in computer science to do so."
— Jensen Huang, Nvidia CEO, Fortune
He's not wrong, and the pay backs him up. Journeyman electricians on Northern Virginia data-center jobs are clearing $120,000-plus a year through their union local; foremen working overtime approach $200,000. North Carolina data-center professionals average $207,000. The industry needs 300,000 more electricians over the next decade and is losing 20,000 a year to retirement, with roughly 30% of union electricians already between 50 and 70. Put a number on the total gap and estimates run from 349,000 to nearly 500,000 construction workers short in 2026 alone.
"The electrician shortage is quite dire. Those people are in short supply all across the country, and this has become a leading barrier to data center construction."
— Darrell West, senior fellow, Brookings Center for Technology Innovation, Fortune
Here's where this stops being a labor-market curiosity and starts being your problem. If you're a CTO or VP who's been reading "the AI boom is creating jobs" as evidence to lean into your 2026 or 2027 software hiring plan, you've been reading the wrong sector's data. The jobs the AI boom is actually creating right now aren't the ones on your req list. They're the ones competing for anyone on your team who touches power, cooling, facilities, or physical infrastructure — and they're paying data-center wages to get them.
That's not hypothetical for most technical orgs. Every company running its own data center, colo footprint, or on-prem GPU cluster employs facilities engineers, power and reliability leads, and site ops people who look, on paper, exactly like the talent this boom is bidding six figures for. If you haven't checked what a Northern Virginia data-center contractor is offering your facilities lead this quarter, assume someone else already has.
The mistake isn't ignoring the trades boom because it's not your industry. It's assuming your industry's AI hiring story and the actual AI hiring story are the same one. They're not.
One shows up in a press release about your agent rollout. The other shows up in a resignation letter from your only power engineer, three weeks before your next capacity buildout. July's numbers already told you which one is hiring. Read the right column.
VC5 Consulting places the technical talent that keeps physical infrastructure running — facilities engineers, power and reliability leads, the roles a software hiring plan never budgeted to compete for against data-center wages. If your infrastructure team is quietly getting recruited by the AI buildout, let's talk.