The United States is short roughly five hundred thousand electricians. A single hyperscale AI campus in its peak construction phase pulls fifteen hundred to three thousand trades workers from within a two hundred mile commute radius and holds them there for eighteen to thirty months. The arithmetic does not close. By 2028 the binding regional variable on AI campus delivery is no longer megawatts. It is the trades labor shed.
Megawatt scarcity is the conversation of 2024 and 2025. It is the conversation that financial press has learned to repeat. But megawatts are a contractual problem with contractual solutions: behind-the-meter generation, gas peakers, co-located storage, transmission upgrades, brownfield substation reuse. The capital stack is willing. The regulatory framework, however slow, is converging. The technology exists. The constraint that has no equivalent escape hatch is the human one. You cannot import a journeyman electrician on a ninety day timeline. You cannot synthesize an ITAR-cleared substation technician. You cannot accelerate the four to five year apprenticeship clock by writing a check.
Two regions with identical megawatt availability, identical signed power purchase agreements, identical water rights, and identical permits will deliver campuses on radically different timelines because of how many trades workers live within a two hundred mile arc and how many of those workers are already committed to automotive, aerospace, refinery, or competing data center work. This essay is about the math behind that divergence.
What A Campus Actually Consumes In Labor
The headcount profile of a hyperscale AI campus is not uniform across its construction life. It is a steep curve with a long, fat peak. Site prep and earthwork employ two hundred to four hundred operators and laborers for six to nine months. Vertical concrete and steel ramps to six hundred to nine hundred workers for nine to twelve months. The electrical and mechanical fitout is where the curve breaks. Switchgear installation, busway, transformer placement, medium-voltage cable pulling, fiber backbone, cooling distribution, and chiller plant assembly require fifteen hundred to three thousand trades workers concurrently for twelve to eighteen months.
Of that peak headcount, roughly forty to fifty percent is electricians and electrical apprentices. Fifteen to twenty percent is pipefitters and HVAC sheet metal. Ten to fifteen percent is mechanical insulators, riggers, and crane operators. The remainder is general construction, controls technicians, and specialty trades.
The two hundred mile commute radius matters. Per diem economics break down past that arc. Workers will travel two hundred miles for a six month contract. They will not relocate families for an eighteen month construction phase that ends with no resident demand. So the question is not whether the national trades workforce can supply a campus. It is whether the specific two hundred mile arc around a specific site can supply fifteen hundred to three thousand workers who are not already committed elsewhere.
The Retirement Cliff
The median age of a journeyman electrician in the United States is approximately forty six. The median age of a journeyman pipefitter is approximately forty eight. Bureau of Labor Statistics occupational projections through 2032 show electrician demand growth of roughly six percent and pipefitter demand growth of roughly five percent. The projections do not adequately capture the retirement curve.
The International Brotherhood of Electrical Workers and the United Association run the dominant apprenticeship pipelines for these trades. A first year apprentice does not become a journeyman for four to five years. The pipeline does not respond to spot demand. An hyperscaler announcing a campus today places no marginal new electrician into the field before 2030. Community college HVAC, electrical, and pipefitter programs nationally produce approximately twenty five thousand to thirty five thousand completions per year against annual demand of sixty thousand to eighty thousand. The gap is not closing. It is widening because the retirement cohort is the largest in trades history and the entry cohort has not been rebuilt since the 1980s deindustrialization.
Between 2026 and 2032 a generational retirement cliff hits the journeyman population. The apprentices to replace them are already in the pipeline or they are not. New entrants past 2026 do not reach the field in time.
The Named Regional Shortages
Phoenix metro is the canonical case. Intel Ocotillo, TSMC Arizona, and the Mesa data center cluster have absorbed Maricopa County electricians at a rate that has driven journeyman wages up an estimated thirty to thirty five percent between 2022 and 2025. Per diem demands routinely add another fifteen to twenty percent. Overtime premiums on critical-path electrical work exceed double time. Local contractors report eighteen to twenty four month lead times on commercial electrical bids.
Columbus Ohio is the second canonical case. Intel Ohio One, Anduril Arsenal-1, Honda EV battery, and the broader central Ohio manufacturing buildout have created a trades vacuum that pulls workers from Cincinnati, Cleveland, Pittsburgh, and Indianapolis. The IBEW Local Six hundred eighty three apprenticeship class has been at capacity for three consecutive cycles. Wage inflation tracks Phoenix.
Northern Virginia, particularly Loudoun and Prince William, has documented electrician shortages that predate the AI buildout. The data center alley density has already absorbed the available pool. Incremental campuses now compete directly with each other for the same crews. Dallas-Fort Worth shows the same pattern with the addition of refinery and petrochemical poaching. Atlanta carries Plant Vogtle nuclear adjacency, EV manufacturing, and warehouse density layered on top of data center demand. Charleston South Carolina pulls Boeing aerospace and BMW Spartanburg adjacency against compute. Wenatchee and Quincy Washington have a pool measured in hundreds, not thousands, and Microsoft, Sabey, and adjacent operators have already committed it. Tucson layers Raytheon defense work against compute construction.
In every named metro the local pool is committed. The marginal campus draws from a national itinerant pool that is itself depleting and aging.
What Automotive And Aerospace Take From The Pool
The Inflation Reduction Act and the CHIPS and Science Act simultaneously activated automotive electrification, semiconductor fabs, battery gigafactories, and grid hardening. Stellantis, Ford, and General Motors EV battery plants pull from the same electrician and pipefitter pool as Intel Ohio, TSMC Arizona, and Microsoft Wisconsin. Aerospace, particularly Boeing in Charleston and Lockheed in Fort Worth and Marietta, pulls the cleared subset of the same pool.
The wage differential matters. An EV battery plant fitout pays competitive but not exceptional rates. An aerospace cleared electrical job pays a meaningful premium for clearance overhead. A hyperscale data center fitout in a constrained metro pays the highest rates in the documented history of US commercial electrical work. The bidding war is not theoretical. It is observable in real-time wage data and in the documented project delays at automotive plants in the Midwest and Southeast where electrical contractors were poached mid-project by data center general contractors offering thirty percent premiums.
The Cleared Trades Sub-Category
A narrower constraint sits inside the broader one. Sovereign and quasi-sovereign tenants, defense-adjacent compute, classified workloads, and federal site reuse all require ITAR-cleared and in some cases higher-cleared construction labor. The cleared engineering pool is approximately eighty thousand nationally. The cleared trades pool is approximately thirty thousand. These workers concentrate in defense corridors: Huntsville Alabama, the Greenville-Spartanburg arc, Charleston South Carolina, Fort Worth, Tucson, the Norfolk-Hampton Roads area, Augusta Georgia near Fort Gordon, and Wichita.
A campus targeting sovereign-tenant absorption must site within reasonable commute of one of these corridors or accept that its cleared workforce will be drawn from a pool small enough that any single competing project, a new aerospace program, a fab expansion, a base modernization, will reset the schedule. By 2028 cleared trades availability becomes the single highest-margin labor category in US construction.
The Substation Technician Choke Point
The utility-side trades constraint deserves separate attention. Substation technicians, transmission line crews, and high-voltage electricians at investor-owned utilities are reporting eighteen to twenty four month staffing gaps. Brownfield substation reuse, the fastest path to interconnect at scale, requires utility crews to perform commissioning and tie-in. Those crews are scheduled out two years in advance at most major utilities in the Mid-Atlantic, Southeast, and Texas. A campus with a signed interconnect agreement and full equipment delivery still waits in a utility crew queue. This is not a contractual problem. It is a headcount problem at the utility itself.
The crew chief and superintendent layer compounds the issue. A general foreman on a hyperscale electrical project requires five to seven years of journeyman experience past initial certification. The supervisory bench is even thinner than the journeyman bench. Parallel campus construction in a single metro is limited not by total trades workers but by the number of qualified general foremen available to lead concurrent crews.
What Becomes True By 2028 To 2030
Trades-shed mapping becomes a first-order site selection criterion alongside transmission, water, and gas. Site selection memos that today list megawatt availability, water draw, gas pressure, fiber routes, and entitlement timeline will add a labor-shed analysis showing two hundred mile journeyman counts by trade, current commitment to competing projects, retirement curve over the construction window, and apprenticeship pipeline depth.
The premium for sites within one hundred miles of a major automotive or aerospace cluster compresses. The proximity-to-talent logic that today favors these locations inverts once the poaching dynamic dominates. Sites in mid-tier metros with active community college trades programs and no competing automotive or aerospace buildout will trade at a structural premium.
Regional workforce co-investment becomes a standard line item. Community college partnerships, IBEW and UA pipeline funding, apprenticeship sponsorship, and direct contribution to vocational program capacity will appear in campus development budgets at one to three percent of total project cost. This is not philanthropy. It is supply chain investment.
The Inflation Reduction Act prevailing-wage tie-ins shift competitive economics toward jurisdictions with deeper apprenticeship infrastructure. The political economy of Davis-Bacon and prevailing-wage rules, which today is read as a cost burden, becomes a quality screen. Jurisdictions with strong union apprenticeship programs deliver projects on schedule. Jurisdictions without them do not.
Cleared-workforce mapping becomes a sovereign-tenant underwriting requirement. Any campus marketed for defense, intelligence, or sensitive federal workloads is underwritten against the cleared labor shed of its location, not against its general construction labor shed.
The binding constraint on AI campus delivery in 2028 is not the grid. It is the population of forty six year old journeyman electricians who will choose which projects to take and which to refuse. That population is finite, aging, and already committed. The site selection question is no longer where the power is. It is where the people are.