Over the past several years, growing demand for construction services has created a supply crunch in the U.S. labor force. Experienced tradespeople are aging out , and underinvestment in vocational education has left fewer younger workers prepared to replace them. The shortage is highly visible: from April 2025 to April 2026, unfilled jobs as a percentage of construction employment plus openings rose from 2.4% to 3.0%. To meet demand, the U.S. construction industry will need to attract 349,000 net new workers in 2026, and another 456,000 in 2027.
Now, the race to build out AI infrastructure is putting increased pressure on this skilled-trades gap. To analyze where the potential data center construction skills gap could be worst, Grace Technologies reviewed modeled data center projections from the Pacific Northwest National Laboratory and employment data from the Bureau of Labor Statistics.
We found that the next wave of AI infrastructure could put severe pressure on labor markets, especially in smaller metro areas where modeled data center growth is exceptionally large compared with their existing pool of skilled tradespeople. However, for workers entering or already employed in these trades, this boom could mean surging demand, higher wages, and vital pathways into technical work.
Key Takeaways
- The data center boom is adding pressure to a real construction labor gap. Unfilled construction jobs rose from 2.4% to 3.0% of construction employment plus openings from April 2025 to April 2026, and the industry will need 349,000 net new workers in 2026 and 456,000 more in 2027.
- The most important skills are electrical, mechanical and cooling-related. The analysis identified 11 skilled-trades occupations tied to data center construction, including electricians, HVAC mechanics, plumbers, pipefitters, equipment operators and construction supervisors.
- Potential strain is highest in smaller Virginia and Mid-Atlantic metros. Harrisonburg, Virginia is projected to build 23.0 new data centers per 1,000 skilled workers, the most of any metro area.
- The pressure is not limited to traditional tech hubs. Several high-ranking metros are smaller or mid-sized markets, including Blacksburg, Roanoke, Winchester, Clarksville, Ames, Prescott Valley and Kingsport-Bristol, where even a modest buildout could be large relative to the local trades workforce.
- Large markets may be better positioned to absorb the buildout. Dallas-Fort Worth has the most projected future data center sites, but a larger skilled-trade workforce to bring down demand pressure. New York, Denver, Boston and Minneapolis also have deeper labor pools relative to forecast sites.
Data center demand is expected to rise sharply; BloombergNEF projects U.S. power demand could reach 106 gigawatts by 2035, up from about 25 gigawatts in 2024.
While major tech companies keep siting decisions closely held, PNNL has modeled where future facilities are most likely to emerge based on growth scenarios, market pull, and infrastructure access. The model screens out unsuitable locations like high flood-risk areas, steep slopes, or regions lacking electric substations and municipal water.
Forecasted sites are heavily concentrated in states like Texas, Virginia, California, and Illinois, alongside specific clusters in mid-sized markets where infrastructure access and market pull overlap.
But many of the most suitable locations for data centers could face skilled-trades constraints that raise construction costs, extend project timelines or force developers to draw from wider regional labor markets. Data center construction requires workers who can install and maintain high-capacity electrical systems, cooling equipment, piping, fiber, mechanical systems and backup power infrastructure. Yet many of the existing occupations with skills overlap for data center construction, including electricians, HVAC mechanics, plumbers, industrial machinery mechanics and construction equipment operators, are already in high demand.
Where Could the Data Center Skills Gap Be Worst?
Skilled-trades workers tend to be concentrated in metros with large existing construction, manufacturing, or logistics economies. To identify where the AI boom could create the most pressure, we compared modeled future sites with each metro area's existing skilled-trades workforce. The resulting metric – modeled future sites per 1,000 skilled-trades workers – highlights potential skills gaps and labor-market pressure.
Harrisonburg, Virginia, ranked first by this measure, with 79 modeled future sites and about 3,430 skilled-trades workers. That equals 23.0 projected sites per 1,000 skilled-trades workers. Chambersburg, Pennsylvania, ranked second, with 54 modeled sites and about 2,620 skilled-trades workers, or 20.6 sites per 1,000 workers.
These smaller metros ranked higher than some of the country’s largest data center markets because the metric adjusts for workforce size. Dallas-Fort Worth-Arlington had the most modeled future sites overall, with 181, but also had about 179,360 skilled-trades workers. That works out to roughly 1.0 modeled site per 1,000 skilled-trades workers.
Conversely, some large markets boast significant projected growth alongside labor pools better positioned to absorb the demand.
These metros are not necessarily immune to shortages. In many large markets, the same workers are also needed for housing, transportation, utilities, manufacturing and other forms of development. But the ratio suggests that projected data center growth is smaller relative to the local skilled-trades labor pool than in places like Harrisonburg or Chambersburg.
State Labor Pools May Ease the Skilled Trades Gap
A state-level view suggests that some of the metro-level pressure may be absorbed by broader labor markets. Because skilled-trades workers can travel across metro boundaries or be drawn from elsewhere in the state, the potential gap looks less severe when measured against statewide labor pools. Still, the pattern does not disappear entirely. Virginia remains the clearest pressure point, with 420 modeled future data center sites and about 168,220 skilled-trades workers, equal to 2.5 sites per 1,000 workers.
Other states with relatively high projected site counts compared with their skilled-trades workforces include North Dakota, Wyoming, Oregon, Nebraska and Iowa. Even so, state-level ratios are much lower than the highest metro-level figures, reinforcing that the sharpest strain is likely to be local rather than statewide. State labor pools may provide some cushion, but they may not fully offset bottlenecks in the specific metros where projects are concentrated.
Which Skilled Trades Matter Most for Data Center Construction?
The occupations behind the data center buildout are concentrated in the skilled trades. We identified 11 skilled-trades, construction and maintenance occupations likely to be involved in data center construction, power systems, cooling systems and facility maintenance.
These jobs are not interchangeable, and not every occupation will be needed equally at every site. But together, they reflect the workforce most likely to be involved in building and maintaining the physical infrastructure behind AI and cloud computing.
The key skills are less about software development and more about troubleshooting, equipment maintenance, electrical systems, mechanical systems, construction management, operations monitoring and repair. In other words, the data center boom is a digital growth story that still depends on the physical economy.
What Does This Mean?
For companies, the skilled-trades gap could become another constraint in the data center boom. Developers are already competing for land, power, water access, fiber and grid capacity. In some markets, they may also have to compete for the workers needed to build and maintain the facilities.
That could mean higher labor costs, longer construction timelines and a greater need to draw from regional or national contractor networks. It could also push companies to invest more directly in workforce development.
That is already happening. Meta and CBRE have launched the LevelUp Fiber Technician Pathway, a free four-week training program designed to prepare workers for fiber technician jobs at U.S. data centers. The program is part of a broader push to build the workforce needed for AI infrastructure.
For workers, the trend could create a rare source of leverage. Data centers need people who can work with electrical systems, mechanical systems, cooling systems, cabling, equipment and facility operations. Many of these jobs are accessible through apprenticeships, technical training, community colleges or experience in adjacent trades.
The data center boom may be powered by AI, but it will be built by workers in the physical economy. In places where projected data center growth is large relative to the existing skilled-trades workforce, the next wave of AI infrastructure could intensify competition for some of the most important workers in construction.
Methodology
We analyzed projected data center locations from Pacific Northwest National Laboratory’s IM3 Projected U.S. Data Center Locations dataset. The analysis used the higher-growth, 50% market-gravity scenario. The dataset models future data center facilities in the contiguous U.S. through 2035 and is intended to represent feasible future siting locations under different growth and market-gravity assumptions. In the scenario used for this analysis, each modeled future site is assigned 36 megawatts of IT power and 1 million square feet of campus size.
Modeled future sites were spatially joined to 2025 Census CBSA boundaries. Of the 1,870 modeled future sites, 1,758 were matched to CBSA boundaries. Sites outside CBSA boundaries were excluded from the metro-level ranking. Some sites fell in micropolitan areas that did not have matching metro-level OEWS labor data and were also excluded from the main ranking.
Skilled-trades employment data came from the 2025 Bureau of Labor Statistics Occupational Employment and Wage Statistics metro file. The selected occupation basket included electricians; heating, air conditioning and refrigeration mechanics and installers; plumbers, pipefitters and steamfitters; operating engineers and construction equipment operators; construction laborers; first-line supervisors of construction trades and extraction workers; construction managers; industrial machinery mechanics; electrical and electronics repairers, commercial and industrial equipment; stationary engineers and boiler operators; and maintenance and repair workers, general.
The main metric is modeled future data center sites per 1,000 skilled-trades workers. This metric should be interpreted as a measure of potential labor-market pressure, not confirmed job vacancies or worker shortages.
Navigating the Future of Data Center Safety
As rack densities scale, traditional safety protocols are no longer enough to protect your team and uptime. High-density power demands a shift to engineered solutions like continuous monitoring and closed-door diagnostics.
To explore these challenges further, watch the full webinar recording on demand—High Density, High Stakes: Electrical Safety in the Modern Data Center—featuring Jim Phillips, P.E., BScEE, IEEE Life Sr. Member, MIET, P.E. of Brainfiller and Grace Technologies CEO Drew Allen.
Zero Harm. Zero Downtime.
