For most of the past two decades, switchgear was a late-stage procurement decision. Engineers specced the servers, designed the cooling architecture, priced the power contract, and then ordered the electrical distribution equipment in the final construction phase. That sequence worked when lead times were predictable and demand was steady. Adhum Carter Wolde-Lule, Director at Prism Power Group, described what has changed: switchgear is fast becoming one of the biggest hidden constraints in commercial construction, and data center development is where the constraint is most acute.
Three forces are converging on the same supply chain simultaneously. Data centers are consuming massive quantities of low and medium voltage switchgear as AI workloads push power draw per facility well beyond what the previous generation of hyperscale builds required. Government-backed digital infrastructure initiatives in the UK and elsewhere are adding public-sector demand on top of private hyperscale procurement. And the manufacturers who produce switchgear, particularly the medium voltage equipment that feeds the transformer-to-distribution path in large data center campuses, have finite production capacity that does not scale quickly.
Carter Wolde-Lule's prescription is precise: organizations must now treat switchgear as a critical path risk, which means early design freeze and early manufacturer engagement. In practical terms, this means the electrical distribution architecture needs to be finalized before the cooling design is locked, before the structural work is completed, before the lease is signed at some sites. The sequence that worked in low-demand periods, where switchgear was treated as a near-commodity available on reasonable notice, is now a schedule risk that can slip a project by six to twelve months.
This is the same dynamic that reshaped CDU procurement over the past three years. The liquid cooling supply chain tightened as hyperscale demand exceeded manufacturing capacity, and operators who treated CDUs as late-stage decisions found themselves commissioning facilities with cooling infrastructure gaps. Switchgear is following the same path, but with the added complication that electrical distribution equipment has longer manufacturing lead times than most cooling components and fewer substitute vendors at the medium voltage tier.
Data center development has historically been organized around the technology procurement cycle, where the servers and networking equipment set the schedule and everything else followed. That organizational logic breaks down when the infrastructure supporting the compute, electrical distribution, cooling, and colocation capacity, all face independent supply constraints that compress against each other. The operators who are executing consistently on AI infrastructure timelines are running parallel procurement tracks, not sequential ones. Switchgear, CDUs, chillers, and transformers are all being specified and ordered at design freeze, well before the construction schedule would traditionally require them.