Data center construction spending hit $98.5 billion through the first ten months of 2026, spread across 153 projects, according to Construct Connect’s October 2026 report — and data centers now account for 24% of all nonresidential building starts in the United States, more than any other commercial building type. The same report flags something buyers should sit with: costs are rising faster than square footage. The average project now runs $1.18 billion, more than double the $507.4 million average from 2025, even though floor area per project has barely grown. Most of the industry conversation about why blames power equipment, permitting, and grid interconnection. Those are real constraints. But ask any general contractor currently running a hyperscale job what’s actually slipping their schedule, and increasingly the answer isn’t a transformer or a utility queue — it’s that they can’t find enough electricians.
The number that should be in every delivery-date conversation: 499,000
Industry labor analyses now put the 2026 shortfall in data-center-relevant construction trades at roughly 499,000 workers, up from an estimated 439,000 shortfall in 2025. The Associated Builders and Contractors puts the broader construction industry’s 2026 net-new-hire need at 349,000 workers just to keep supply and demand in balance, before accounting for the disproportionate pull of hyperscale and AI campus work. The Associated General Contractors of America’s own survey work shows 92% of firms report difficulty finding qualified workers, and 45% say workforce shortages have already caused project delays this year. Construction unemployment sat at a record-low 3.2% as of mid-2025 — there is, in a very literal sense, no idle labor pool left to absorb a building boom of this size.
The demographics make the gap structural rather than cyclical. Roughly 20% of the current skilled-trade workforce is already over 55, and industry estimates put retirements at 41% of today’s workforce by 2031. In electrical specifically, an estimated 20,000 electricians retire every year — 200,000 over a decade — against a projected need for more than 300,000 new electricians industry-wide over the next ten years. A data center campus that used to need 750 peak on-site workers now routinely needs 4,000 to 5,000 at the height of construction, competing for the same shrinking pool of licensed journeymen as every other electrification project in the economy: EV plants, battery factories, grid upgrades, and reshored semiconductor fabs.
Wages are the clearest signal, and they’re not subtle
When demand outstrips supply this sharply, the market clears on price. Data center construction roles now carry roughly a 30% wage premium over standard commercial construction. In Northern Virginia — still the densest data center market in the world — IBEW Local 26 electricians are earning around $59.50 an hour, pushing total annual compensation above $120,000, with foremen working overtime in peak-demand stretches approaching $200,000 a year. Even apprentices in those markets are commanding roughly $26 an hour. The local union has responded by nearly doubling its membership, from roughly 7,000 in 2018 to about 14,700 by January 2026, but new electricians take years to train to full journeyman status; a hiring surge today doesn’t relieve a schedule crunch next quarter.
This is showing up outside the traditional hotspots too. Average data center trade compensation is now reported around $207,000 a year in North Carolina’s Raleigh-Durham market and the $180,000s in California, Texas, and Arizona, with Raleigh-Durham wages climbing 9–11% in a single year as projects compete for the same specialized pipefitters, sheet-metal workers, and controls technicians. As Brookings’ Darrell West put it bluntly in recent commentary on the sector: the electrician shortage specifically “has become a leading barrier to data center construction.” That’s a harder claim to dismiss than a generic “skilled labor is tight” line that could describe any construction sector — it’s naming the exact trade that gates the critical path on every data hall build.
What this actually costs a buyer, in schedule and in dollars
For a capacity buyer, the labor shortage doesn’t show up as a line item — it shows up as a pushed delivery date, and pushed delivery dates have a dollar value. Industry estimates put the revenue impact of delaying a 60 MW facility at roughly $14.2 million per month once you account for lost colocation or compute revenue against the capital already committed. Construction backlogs tied to labor and equipment constraints together are running 8.5 to 12 months in some markets. That compounds with the equipment lead times covered in our piece on power transformer shortages — a 12-to-18-month wait for a transformer doesn’t help if there’s no crew available to install it the week it finally arrives, and a fully staffed MEP crew sitting idle waiting on switchgear is its own form of burned capital.
Need Expert Guidance?
Talk to a Data Center Expert
21+ years of hands-on experience in data center design, operations & infrastructure. Book a quick discovery call to discuss your project.
📞 Book a Discovery CallElectrical work alone represents 45–70% of total data center construction cost, which is exactly why this trade’s labor market is the one worth watching most closely in any build-vs-colocate-vs-buy decision. If you’re evaluating a developer’s or GC’s delivery promise, the right diligence question isn’t just “what’s your interconnection timeline” — it’s “what’s your committed electrical workforce for this specific job, and is it unionized, directly employed, or dependent on the same regional subcontractor pool three other campuses nearby are also drawing from.” A contractor who can answer that specifically, with named crews and backlogs, is a meaningfully different risk than one who waves at “we have relationships with labor providers.”
The mitigation playbook buyers are actually seeing work
The industry’s most credible response isn’t waiting for more electricians to graduate — it’s redesigning construction to need fewer of them on-site. Prefabricated, modular construction — where electrical and mechanical systems are built and partially commissioned in a factory setting with stable crews of 20 to 50 workers, then shipped and assembled on-site — is cutting project timelines from a typical 24–36 months down to roughly 16–20 months in early adopter projects, largely by removing the exposure to local labor-market volatility. Factory crews also have far lower turnover than on-site construction, where specialized-trade turnover is running around 73% industry-wide, a number that itself creates rework, safety risk, and schedule slip independent of headcount.
On the workforce-supply side, hyperscalers are now funding their own pipeline rather than competing for someone else’s graduates: Microsoft’s STAR program and Google’s data center technician program are both placing people directly into operations and construction-adjacent roles, and community colleges are picking up matching funding — College of DuPage in Illinois received a $425,000 grant in late 2025 specifically for a data technician certificate track. None of this closes a 499,000-worker gap by itself, but it’s the difference between a buyer whose contractor has a staffing plan and one who’s simply hoping the regional labor market loosens up. For readers weighing operations staffing rather than construction staffing specifically, our technician career path and salary guide breaks down where that separate, equally tight hiring market stands.
The honest counterargument
It’s worth being fair to the other side of this: power and permitting constraints are not a myth, and in several markets — parts of Virginia, Georgia, and the PJM footprint — grid interconnection timelines of 24 to 36 months are a harder, more binary gate than labor is. You can throw money at a labor shortage and partially solve it with wage premiums, modular construction, and overtime; you generally cannot pay a utility to finish a multi-year transmission upgrade faster. Labor is the bottleneck that’s easiest to underestimate precisely because it’s solvable at the margin, which makes it tempting for a project pro forma to assume it away. The projects actually hitting their delivery dates in 2026 are the ones treating skilled-trade availability as its own line item in the risk register, not an afterthought behind power and water.
For a buyer, the practical takeaway isn’t to panic about labor — it’s to interrogate it with the same rigor you’d apply to a power purchase agreement or an interconnection study. Ask your developer or GC for their actual electrical crew commitment, whether it’s modular or stick-built, and how their wage structure compares to the regional premium; a contractor quoting 2024 labor assumptions on a 2027 delivery date is quoting you a schedule they can’t keep. Build the diligence into your capacity-buying process with our Data Center Buyer’s Toolkit.
Written by
Raajeev Ratra
Data Center Infrastructure Expert | 15+ Years in DC Design, Operations & Project Management
Raajeev is a seasoned data center professional with hands-on experience in hyperscale facilities, colocation design, power & cooling infrastructure, and global DC operations. He shares practical insights to help engineers and IT leaders build better infrastructure.