Labor Crisis: The Real Barrier to Data Center Expansion

The money is there. The land deals are getting done. Power purchase agreements are being signed. And yet, the great data center buildout of the 2020s is running into a wall that no amount of capital can immediately fix.

It's people.

William Self, Mercer's chief workforce strategist, put it plainly during a March 2026 Marsh webinar focused on the data center investment boom: "I believe the single biggest constraint of this entire buildout is the labor needed – and not capital, land, or even energy." That's a striking claim given that grid capacity and site availability have dominated industry headlines for two years. But Self's framing deserves serious attention — because it reflects something that project timelines are already confirming on the ground.

The Scale of What's Being Built Creates a Workforce Problem No One Planned For

Consider the velocity of commitments. NTT DATA just landed 115 MW worth of U.S. data center deals as part of a $10 billion AI push. Hyperscalers are signing contracts for campuses that didn't exist 18 months ago. When you add up the announced capacity across the industry, you're talking about tens of millions of square feet of critical infrastructure that needs to be built, commissioned, and staffed — all within overlapping timelines.

That's not a minor surge in demand for skilled tradespeople. That's a structural shock to a labor market that was already tight.

Electricians, ironworkers, HVAC technicians, low-voltage specialists — these aren't workers you can train in a weekend boot camp. Journeyman electricians typically complete four to five years of apprenticeship. Ironworkers need certifications and experience that take years to accumulate. The data center industry is essentially trying to schedule a construction supercycle on top of a labor pool that was already fully employed.

Two Phases, Two Completely Different Talent Problems

Here's where the workforce challenge gets more complicated — and where Self says many companies are making a critical planning error.

Data center labor demand doesn't arrive as a single wave. It comes in two distinct phases, each requiring a fundamentally different talent strategy.

Phase One: Building the Thing

The first phase is construction. This is the world of hard hats, union halls, and skilled trades. Electricians to run the power distribution. Ironworkers for structural steel. HVAC and mechanical crews to install cooling infrastructure. These workers are project-based by nature — they move from site to site, and their availability is governed by regional union density, prevailing wage rates, and competing construction projects in the same geography.

The problem is that data center construction is now competing with semiconductor fab construction, battery gigafactory buildouts, EV manufacturing plants, and federally funded infrastructure projects — all of which are chasing the same licensed trades workforce. In regions like the mid-Atlantic, Texas, and the Southeast, where data center density is highest, construction labor shortages are already causing schedule slippage and cost overruns on active projects.

Phase Two: Running the Thing

The second phase begins the moment a facility goes live — and it's where Self warns that planning often breaks down entirely.

Operating a hyperscale or even mid-tier data center requires a different kind of expertise: critical facilities engineers, electrical and mechanical maintenance technicians, DCIM (data center infrastructure management) specialists, and increasingly, professionals with experience managing liquid cooling systems as AI workloads push power densities into territory that traditional air cooling can't handle.

This isn't a workforce you can hire off the street the week before commissioning. Recruiting, onboarding, and training a qualified critical facilities team takes months — and retention is a persistent challenge because experienced operators command premium salaries and face constant recruiting pressure from competing facilities.

Too many organizations are building for Phase One and assuming Phase Two will sort itself out. It won't.

Strategies That Actually Move the Needle

The industry isn't sitting still, but the solutions being deployed vary widely in their effectiveness.

Upskilling existing employees is the most immediately actionable lever. Several large operators are partnering with organizations like AFCOM and 7x24 Exchange to create structured career pathways that convert general IT staff or facilities personnel into certified critical environments professionals. This is smart — it taps an existing population of workers who already understand the operational culture.

Community college and trade school partnerships are gaining traction, particularly in markets where major operators have established long-term footprints. Northern Virginia, which hosts the densest concentration of data center capacity anywhere on earth, has seen Loudoun County Public Schools and local community colleges develop curricula specifically targeting data center operations. These programs don't produce journeyman electricians overnight, but they build a pipeline of entry-level technicians who can grow into critical roles.

Apprenticeship programs tied to union halls remain the most reliable path for construction trades. The challenge is that apprenticeship programs take years to yield results — which means the workers needed for projects breaking ground in 2025 and 2026 needed to start training in 2021. The industry's explosive growth was somewhat predictable; the failure to invest proportionally in workforce development pipelines was not inevitable.

There's also a geographic dimension that doesn't get enough attention. Data center operators increasingly have flexibility in where they site new capacity — and workforce availability is becoming a legitimate site selection criterion alongside power, land cost, and network connectivity. States and municipalities that have invested in technical education infrastructure are gaining a competitive advantage in attracting investment.

What Delays Actually Cost

Schedule slippage in data center construction isn't an abstract inconvenience. It has direct financial consequences.

A hyperscale facility running 12 months behind schedule represents hundreds of millions of dollars in deferred revenue for the operator, breach-of-contract exposure on pre-signed leases, and cascading effects on the AI infrastructure timelines of the tenants waiting for that capacity. At a moment when AI model development and inference deployment are moving at a pace measured in months, not years, a one-year delay in bringing capacity online is a meaningful competitive setback.

For smaller colocation operators and edge deployments, the math is similarly brutal. Construction labor cost overruns driven by shortage-induced wage inflation compress margins that were already thin. Difficulty staffing operational roles creates either a safety and compliance risk (if facilities are understaffed) or an overhead burden (if operators resort to expensive managed service arrangements to fill gaps).

Data center labor shortages aren't just an HR problem — they're a financial risk that belongs on the balance sheet.

Where This Goes From Here

Automation will eventually reshape some of the operational workforce picture. AI-assisted monitoring tools, predictive maintenance systems, and increasingly autonomous facility management platforms will reduce the ratio of human operators to megawatts under management. But that evolution is measured in years and decades, not quarters — and it doesn't touch the construction trades problem at all. Robots are not pulling cable in raised-floor environments at scale.

The more immediate variable is immigration and visa policy, which affects the availability of workers across both construction trades and technical operations in ways that are politically volatile and hard to forecast.

What's clear is that the organizations that treat workforce development as a strategic priority — not an afterthought — will have a decisive advantage in the data center expansion race. Capital is fungible. Experienced critical facilities engineers and licensed electricians are not.

The investors, developers, and operators who understand that distinction will be the ones who actually deliver the capacity the market is demanding. The ones who don't will be explaining schedule delays to their LPs for years.

Call to Action: Ready to tackle the labor crisis in data centers? Explore solutions and opportunities at InfraSale Marketplace.

[INTERNAL LINK: workforce development strategies]

[INTERNAL LINK: data center labor market trends]

[INTERNAL LINK: construction trade challenges]