Skilled trades wages on data center construction sites now run roughly 32% above non-data-center construction work, with specialty roles routinely clearing $250,000. The same crafts on a healthcare project earn meaningfully less. The same crafts on an education project earn even less than that. The labor market has decided the build types aren’t interchangeable for staffing purposes, and the pricing reflects it.
Table of Contents
Construction’s workforce conversation tends to flatten into one number: how short is the industry of workers? The Associated Builders and Contractors says the answer is 349,000 net new workers needed in 2026, rising to 456,000 in 2027, and AGC reports 92% of firms can’t fill craft positions. The numbers are accurate but they hide the operational reality. The shortage isn’t uniform. Some sectors are running 40-point net positive on AGC’s outlook while others are flat or shrinking, and a workforce plan that treats labor as one pool will keep getting blindsided by how the actual demand is distributed.
Bridgit’s 2026 Construction Workforce Benchmark Report, drawn from 233 contractors and 114,000 people, makes the variance visible. Industrial and manufacturing work grew 68.1% year-over-year in the dataset. Data center work grew 41.7%. Hospitality declined 0.5%. Education grew 9.4%. Same contractors, same customer base, very different sector pressures.
The labor market is pricing each build type differently
The wage premium on data center work is the cleanest signal. Specialty electricians and pipefitters can clear $250,000 on hyperscale projects, while the same trades on traditional commercial sites operate inside more conventional ranges. AGC’s 2026 Construction Hiring and Business Outlook Report puts the data center sector net positive at 57 percentage points, the highest of any sector and up from 42 last year.
Pricing diverges because the work has diverged. A hyperscale data center pulls roughly 1,500 workers at peak across electricians, pipefitters, ironworkers, HVAC technicians, and concrete crews. A community hospital tops out with a tenth of that crew over twice the schedule. Both projects compete for the same skilled tradespeople in the same regional labor market, and the bidding war the data center wins shows up as schedule slippage on the hospital.
A contractor who treats data center pursuits as marginal additions to a commercial portfolio is treating one labor profile as if it were a different one. The salaried project leadership might transfer cleanly. The craft labor likely doesn’t, which is why workforce planning for specialty trades increasingly has to be built around the specific build types those trades flow through.
Team size and project duration as distinct planning variables
The benchmark report surfaces an insight that doesn’t get attention in standard forecasting models: team size and project duration are independent variables, and the relationship between them changes by build type.
Solar projects in the dataset run a median team size of 28.5 people over a median duration of 28 months, a moderate-sized team holding steady for a long stretch. Data centers compress much larger crews into shorter durations, while mixed-use work spreads smaller teams across longer ones. Each profile demands a different forecasting cadence.
The table below maps the team-size and duration profiles surfaced in the benchmark report against the forecasting cadence each one calls for.
| Build type | Team size profile | Duration profile | Forecasting implication |
|---|---|---|---|
| Solar | Moderate (~28.5 median) | Long (~28 months median) | Steady drumbeat, sequenced phases |
| Data center | Large peak | Short | Sharp ramp, fast wind-down |
| Mixed-use | Smaller | Long | Sustained moderate demand |
| Healthcare | Moderate | Long | Phased build with coordination peaks |
| Industrial / manufacturing | Variable | Moderate | Heavy specialty trades, megaproject pipelines |
Healthcare and industrial profiles inferred from sector growth and project-duration patterns in the benchmark dataset; not all sectors include published medians.
Modeling these as one workforce-planning template produces predictable failures. A large team held for a long stretch needs different recruiting cadence than a large team needed for six months. A smaller team running for years pulls retention investments that a moderate-team, moderate-duration project doesn’t. The sector mix in a contractor’s pipeline determines which staffing problem they actually have.
Data center construction workforce demand in 2026
Data centers are the headline pressure point for 2026. US data center construction spending is projected at $86 billion, driven by hyperscaler investment from Amazon, Microsoft, Google, and Meta. Roughly 40% of AI data center construction sites show signs of schedule slippage tied to electrician and pipefitter shortages, and CBRE data shows data centers under construction in the US dipped to 5.99 GW at end of 2025, the first decline in years, driven by power, permitting, and labor constraints together.
A contractor pursuing data center work faces a recruiting timeline that runs months ahead of the construction timeline. The crew can’t be assembled the week the project breaks ground. It has to be visible in the workforce plan eight to twelve months earlier, with clarity on which specialty trades are scarce and where they’re willing to travel from.
Solar construction project staffing profiles
Solar runs the inverse rhythm. The benchmark report’s 28.5-person team and 28-month duration map to a labor model where utility-scale solar moves slower than data centers but maintains a steady drumbeat of demand. The US solar workforce stood at 279,447 in 2023 and needs to reach roughly 355,000 by 2026 to support 60 to 70 GW of annual installations, with current hiring trends suggesting a shortfall of about 53,000 workers. Solar accounted for 56% of new US grid capacity added in the first half of 2025.
A utility-scale solar build sequences across rural sites over multiple months, often phased to match power purchase agreement deadlines and tax credit windows. A contractor planning solar work needs to model crew availability across long arcs, not peak surges, and workforce planning that captures phase-by-phase staffing needs makes the difference between meeting a tax-credit deadline and missing it.
Industrial and manufacturing growth and labor demand
The industrial and manufacturing sector grew 68.1% year-over-year in Bridgit’s benchmark dataset, the highest growth rate of any category. Reshoring and nearshoring private-sector commitments now total $1.595 trillion across semiconductors, electric vehicles, and battery manufacturing, anchored by megaprojects like TSMC’s $165 billion Arizona fab, Micron’s $200 billion New York investment, Intel Ohio, and Samsung Texas.
These projects don’t stagger in evenly across regions. The Dallas Fed projects Texas employment will grow 1.9% in 2026, led by construction and manufacturing tied to the megaproject pipeline, after adding 30,100 construction jobs from January 2025 to January 2026, the largest numeric construction gain in the country. Sun Belt and Mountain West states are absorbing most of that activity alongside Infrastructure Investment and Jobs Act spend. The Top 50 ENR 400 contractors in the benchmark report hold 21.3% of their portfolio share in industrial and manufacturing work, well above the broader industry, because the bigger contractors are following the megaproject pipeline.
The contrast: hospitality, education, and the steady performers
Not every sector is overheating. The benchmark report shows hospitality work declining 0.5% year-over-year and education growing only 9.4%. Healthcare retains a positive outlook, holding 17% Top-50 ENR portfolio share, with major 2026 megaprojects including IU Health Indianapolis at $4.3 billion and Tutor Perini’s UCSF Children’s Hospital at $960 million.
The contrast is what makes the case for sector-aware staffing. A contractor with mixed-portfolio exposure has internal flexibility. People finishing a hospitality project can rotate onto an industrial pursuit if the workforce planning system surfaces the match. A contractor concentrated in one sector doesn’t have that flexibility, and either staffs up for the boom and risks overstaffing in the bust, or staffs for the steady state and loses schedule when demand spikes. With Industrial and Manufacturing at +68.1%, Data Center at +41.7%, and Hospitality at -0.5% in the same year, treating them as one pool means missing the signal in every direction.
The structural workforce shortage isn’t easing
The 349,000-worker gap ABC projects for 2026 widens to 456,000 by 2027, and the retirement wave compounds it: Deloitte estimates 41% of the current construction workforce will reach retirement age by 2031. Total US construction starts are forecast to grow 1.6% in 2026 after 14.2% growth in 2025, with nonresidential starts dropping 6.6% in December 2025. Modest aggregate growth on top of an already-stressed labor base doesn’t give contractors room to wait the shortage out.
The only available answer is smarter allocation of the people already on the bench. A contractor with 200 people who can map every person to completed build types, willing travel geographies, and role progressions has a meaningfully different planning capability than a contractor whose data lives in three spreadsheets across two offices. Same headcount, very different operational reality.
Forecasting workforce capacity by build type
The operational shift is from forecasting headcount in aggregate to forecasting capacity by build type. A contractor doesn’t have one labor pool. They have a healthcare pool, a data center pool, an industrial pool, and a solar pool, and the size of each is the question that actually matters.
That changes how planning meetings run. Instead of asking “how many PMs do we need next year,” the question becomes “how many PMs with healthcare experience do we need in Q3, against the pursuit pipeline we’ve committed to, given the typical durations of those projects?” That kind of demand forecasting in construction requires more variables to weigh, but it produces a meaningfully better planning decision.
A live view of capacity by build type also changes pursuit decisions. A pursuit that would push the data center pool past its limit while leaving the healthcare pool underutilized is a different pursuit than a quarterly headcount review would describe. Forecasting that connects pursuit pipeline to workforce capacity by sector makes the trade-off visible at the moment go/no-go decisions are being made.
What the data is telling planners
Bridgit’s benchmark report and the external labor market signals point at the same conclusion from different angles. Demand has concentrated, the wage premiums and schedule slippage are the receipts, and project type is now the most reliable predictor of staffing pressure. The workforce pool that delivers a hyperscale data center is not the workforce pool that delivers an elementary school renovation.
If your workforce plan still treats labor as one pool, the next two years will keep producing the same surprises. The data center pursuit gets won and the crew doesn’t materialize on time. The solar contract gets signed and the staffing curve doesn’t match the phased deployment. The hospitality work that was supposed to be steady declines instead. The visibility that prevents those surprises lives in connecting build-type experience, project duration patterns, and pursuit pipeline data into one planning view.
