> This is the markdown version of https://www.classet.ai/blog/data-center-trades-shortage > Learn more at https://www.classet.ai ![Low-poly illustration split down the middle: an overflowing resume inbox on one side, a lone electrician wiring a data center on the other](/_next/image?url=%2Fimages%2Fblog%2Fdata-center-trades-shortage.png&w=3840&q=75) # Big Tech Is Drowning in Resumes and Short on Electricians The same AI companies burying recruiters in resumes are begging people to become electricians. Two opposite hiring problems, one lever that wins both: speed. [![Paul Jones](/_next/image?url=https%3A%2F%2Fassets.basehub.com%2Fe0b5701f%2F6599306507912123f90f150a8bfaaf6c%2Fscreenshot-2026-01-28-at-10.53.16-am.png%3Fwidth%3D100%26height%3D100%26quality%3D100&w=96&q=75) Paul JonesHead of Growth at Classet ](/blog/authors/paul-jones) July 29, 2026 Skilled Trades, Hiring Tips Here is a sentence that should not be true: the same five companies whose white-collar hiring is defined by too many applications are, this month, desperately short of workers. The [New York Times reported](https://www.nytimes.com/2026/07/29/business/economy/data-center-electricians-training.html) that Microsoft, Google, Meta, Amazon, and Oracle are recruiting electricians and carpenters by the thousands to build data centers. Not filtering them with an algorithm. Not drowning in applications. Straightforwardly, painfully short-handed. Microsoft president Brad Smith has called the skilled-trades shortage the company's number-one obstacle, ahead of chip supply or power. Same companies, same quarter, two hiring problems that are exact opposites. That contrast is worth sitting with, because it exposes something most hiring advice gets wrong. Quick answer Big Tech has two opposite hiring problems at once. On the software side, too many AI-assisted applications for too few recruiters, so every tool is about subtraction: narrow the pile. On the trades side, no pile at all, so companies like Meta are building the workforce themselves years before they need it. The tactics look nothing alike, but the lever that decides both hires is the same: speed. Whoever reaches a good candidate first with a real offer wins, whether you are sorting 250 applicants or chasing the one you cannot lose. ## Two hiring problems, same companies, same month Start with the trades side, because the numbers are hard to believe. The U.S. needs an estimated 300,000-plus additional electricians this decade, against roughly 81,000 annual openings and about 20,000 electricians retiring every year. Northern Virginia's electricians' union has doubled its membership to 14,700 over seven years and still cannot keep up with data-center demand. Now the software side, which anyone in talent acquisition already lives with. One job post pulls hundreds of applications in a day, most of them AI-written, and the tools built to sort them are themselves AI. Ghosting is at a three-year high in the middle of it. When candidates outnumber roles by that much, the entire job of recruiting becomes subtraction. 300K+ Additional electricians the U.S. needs this decade $115M Meta's year-one spend on free trade bootcamps with a job offer attached 35,000 Applications Meta's predecessor training program drew in its first week ## When you have too many candidates, you filter This is the problem most of you reading this actually have. A req goes live, and within hours the pile is unmanageable. So the toolkit is all about narrowing: knockout questions, resume parsing, screening bots, anything that shrinks an overwhelming stack into a list a human can work. The trap is that filtering harder feels like progress while quietly costing you your best people. Aggressive automated screens reject strong candidates on keyword mismatches. Long application forms lose the exact people who have three other offers and no patience. I wrote about how [the application flood breaks down](/blog/ai-application-flood) into lazy AI slop on one end and real fraud on the other, and the honest takeaway is that no filter reliably catches either, because both are built to satisfy an automated reader. Subtraction is necessary at this volume. It is not the same thing as winning the hire. ## When you have too few, you build the pipeline yourself On the trades side there is nothing to narrow, so the tactic stops being recruiting and becomes workforce creation. Microsoft's Datacenter Academy now runs with more than 50 community colleges and had roughly 800 seats against 12,000 applications in its last cohort. AWS runs a Workforce Accelerator with the nonprofit Per Scholas. Google's STAR Pathway is active in Virginia and Ohio, backed by a separate $15 million commitment to the IBEW's training arm. Meta went furthest. Its new America's Workforce Academy, [launching across Louisiana, Indiana, Ohio, and Texas in 2026](https://bricks-bytes.com/news/big-tech-training-electricians-data-center-labor/), hands out a conditional job offer up front, then pays for a free five-week bootcamp with tuition, travel, lodging, tools, and a daily stipend included, ending in a recognized trade credential. Meta is calling it the largest private-sector skilled-trades program with a job guarantee in U.S. history. The predecessor version, a fiber-training program built the same way, pulled 35,000 applications in its first week. The piece that gets lost is that these are the same companies. Recruiting technology, AI or otherwise, is a tool for choosing among abundance. When the problem is scarcity, the only lever left is building the workforce yourself, years early, betting on people before you have seen them do the job even once. The broader [skilled trades labor shortage](/blog/overcome-skilled-trades-labor-shortage) has been pushing employers toward that kind of pipeline thinking for a while; the data-center boom just made it impossible to ignore. ## The lever that wins both is speed, not volume tactics Here is where I land, and it is the part I care about most. Most hiring managers reading this are not competing with Microsoft for electricians, so do not read the trades story as proof that everyone is suddenly short on candidates. Read it as a control experiment. Two companies, same month, opposite volume conditions, and the thing that decides the hire in both cases turns out to be identical. A data-center project loses a good tradesperson to whichever employer reaches them first with a real offer. A software team loses a good candidate to whichever recruiter responds first, before that person accepts somewhere else. Volume changes your tactics. It does not change what wins. When candidates are scarce, speed is obvious because the person is gone in days. When candidates are abundant, speed hides behind the pile, so teams spend all their energy on better filters and let strong applicants cool for a week. Same failure, different disguise. That is the whole reason we built [Joy](/demo), our AI screener. Joy gets applicants into a real phone, text, or email conversation within minutes of applying instead of days later. Teams using it have cut time-to-hire by as much as 80 percent, not because it filters harder, but because it moves faster, whether you are working through 250 applicants or trying to close the one candidate you cannot afford to lose. If you want the mechanics, I broke down how [voice AI recruiting works](/blog/voice-ai-recruiting) end to end. ## What this means if you're not sitting on a trillion-dollar balance sheet You cannot copy Meta's $115 million bootcamp. You can copy the mechanism underneath it, which is cheaper than it looks. Meta's academy works because it removes the single biggest reason someone will not attempt a career change into a trade: the risk of spending five weeks on a maybe. It flips the usual order. Instead of training people and then deciding who is good enough to hire, it offers the job first, contingent on finishing a short, paid, structured onboarding. If you are chronically short on a specific role, borrow that sequence. Offer the job contingent on completing a brief paid onboarding, rather than asking people to prove themselves first and hope you notice. It costs you a little certainty on day one in exchange for a lot more applicants, and it works at any budget. Pair it with a screening process that actually moves, and you have stopped losing hires to whoever got there first. For teams running hundreds of reqs, the same logic scales through a real [high-volume hiring](/blog/high-volume-hiring-guide) system rather than more headcount. ## Common questions Why are AI companies short on electricians if they're flooded with software applications? Different labor markets. Software roles at Microsoft, Google, and Meta pull far more applicants than there are jobs, made worse by AI-written applications. Skilled trades like electrical work have the opposite math: the U.S. needs an estimated 300,000-plus additional electricians this decade against roughly 81,000 annual openings and 20,000 retirements a year. The data-center construction boom collided with a shrinking trades workforce, so the same company can be buried in resumes for one role and unable to staff another. Is building your own training pipeline realistic for a company that isn't Big Tech? The nine-figure version is not, but the mechanism is. The reason Meta's academy works is that it offers a job first and removes the risk of training with no guarantee. Any employer can borrow that: make the offer contingent on completing a short, paid, structured onboarding instead of asking people to prove themselves first. It trades a little day-one certainty for a much larger applicant pool, and it does not require a trillion-dollar balance sheet. If I have too many applicants, why does speed matter more than better filtering? Because filtering and speed solve different problems. Filters shrink the pile; they do not keep your best candidate from accepting elsewhere while they wait. Strong applicants with other offers are gone in days regardless of how good your screening is. Reaching applicants within minutes, then filtering, means you are fast and selective instead of only selective. See how [instant candidate screening](/blog/voice-ai-recruiting) changes that math. ## Key points - Big Tech has two opposite hiring problems at once: too many software applicants to read, and too few tradespeople to build data centers. - The U.S. needs an estimated 300,000-plus more electricians this decade against roughly 81,000 annual openings and 20,000 retirements a year. - When candidates are abundant, recruiting is subtraction. When they are scarce, it is workforce creation, and the same companies are doing both. - The lever that decides the hire is identical in both conditions: whoever reaches a good candidate first with a real offer wins. - You cannot copy Meta's $115 million bootcamp, but you can copy the mechanism: offer the job first, contingent on a short paid onboarding. ## Next steps If your problem is a full inbox or a single hard-to-fill role, the constant is the same: the moment someone applies is the moment you should be talking to them. [See how Joy works](/demo) and reaches applicants within minutes, inside your ATS or without one. [![Paul Jones](/_next/image?url=https%3A%2F%2Fassets.basehub.com%2Fe0b5701f%2F6599306507912123f90f150a8bfaaf6c%2Fscreenshot-2026-01-28-at-10.53.16-am.png%3Fwidth%3D100%26height%3D100%26quality%3D100&w=128&q=75) Paul Jones Head of Growth at Classet Paul comes from an operator background running an Alpine-owned company, and brings firsthand experience with the hiring challenges Classet was built to solve. He's driven by a belief that the right technology can make meaningful work more accessible. ](/blog/authors/paul-jones) Follow Classet in Google Add classet.ai as a preferred source and our hiring research is more likely to show up in your Top Stories and AI Overviews. [Add as preferred source →](https://www.google.com/preferences/source?q=classet.ai) ## Explore More ### Use Cases - [RPO / BPO Recruiting](/use-cases/call-centers-bpo) - [Healthcare Recruiting](/use-cases/healthcare) - [Hospitality Recruiting](/use-cases/hospitality) ### Integrations - [Greenhouse](/integrations/greenhouse) - [Bullhorn](/integrations/bullhorn) - [Lever](/integrations/lever)