Data-Centre Construction Has Produced a Skilled-Trades Shortage Nobody Budgeted For

A large data centre is, from a construction standpoint, an electrical project with a building around it. Switchgear, transformers, distribution, backup generation, uninterruptible supply, and the terminations connecting all of it demand licensed electrical trades in quantities that dwarf what a comparably sized commercial building requires. When several such projects are underway in one region simultaneously — which is the current condition in a number of markets — they draw from a labour pool that was sized for the previous decade's demand. The shortage behaves differently from a materials shortage, and worse. Equipment with a long lead time can be ordered earlier; the delay is knowable and can be planned around. Licensed trades require years of apprenticeship, and the pipeline cannot be accelerated by paying more — higher wages move workers between projects rather than creating additional workers. The observable effects are wage escalation, schedule slippage that propagates across every project competing for the same crews, and the quieter cost of experienced people being pulled from maintenance work at existing facilities onto new construction. The regional dimension makes this a genuine competitive factor between locations, and one that rarely appears in site-selection material. Two candidate sites with comparable power availability, land cost and connectivity can differ substantially in how quickly a project actually gets built, purely on trades availability and on whether the local training infrastructure has been producing electricians at a rate matched to demand. Jurisdictions that invested in vocational training a decade ago are, without having planned for this specifically, holding an advantage that money cannot quickly replicate. The general lesson extends past data centres. Every infrastructure buildout is eventually rate-limited by a workforce that takes years to produce and is invisible in planning documents until it binds. Grid upgrades, port expansion and rail electrification have all met the same wall. Compute capacity has simply reached it faster than most, because the demand behind it arrived faster than any of them.