Site selection teams are good at comparing tax abatements, labor pools, and land costs down to the decimal point. Power access has historically gotten a checkbox: available, yes or no. That habit is now an expensive one. A facility that clears every other diligence hurdle but lands in a congested interconnection queue can sit fully built and unable to generate a dollar of revenue for a year or more, and the costs that accumulate during that wait rarely make it into the original capital model.
The exposure looks slightly different depending on what gets built, but it shows up in the same queues everywhere. A battery plant, a semiconductor fab, and a data center campus increasingly compete for capacity on the same substation. The clearest public numbers on delay costs come from the data center sector, where lease pricing is tracked closely: one 2026 industry benchmarking analysis estimated that a delay at a 60-megawatt (MW) facility can cost approximately $14.2 million a month in lost revenue once financing, staffing, and opportunity costs are factored in, and CBRE's tracking shows the national average wholesale data center asking lease rate reached $194.95 per kilowatt per month in 2025. A manufacturing CFO will not find an equivalent lease-rate benchmark for a chemical plant or a battery gigafactory. Debt service accrues on a facility that isn't producing. Customer contracts carry penalty clauses for missed delivery dates. Workers get paid to stand around a building that can't yet run, and inventory sits with nowhere to go.
Interconnection Delays Are Now Measured in Years, Not Utility Cycles
Large-load projects across the country, and especially in parts of Northern Virginia, are increasingly planning around multi-year interconnection timelines rather than the single-year utility service requests that used to be standard. The U.S. Energy Information Administration (EIA), drawing on PJM's 2026 load forecast, reports that new generation in the region now takes four to seven years to move from application to operation, and PJM has told regulators that the Dominion zone, which covers Northern Virginia, will see the largest absolute increase in summer peak demand of any zone in the system through 2030. A company that picked a Northern Virginia site on land price and labor availability alone, without pricing in queue position, is now exposed to a multi-year wait it never budgeted for.
The cost of that congestion does not stay contained to the companies waiting in line. PJM's capacity auction price for the Dominion zone rose from $28.92 per megawatt-day in the 2024/2025 delivery year to $329.17 per megawatt-day for 2026/2027, a roughly tenfold increase, according to the Institute for Energy Economics and Financial Analysis (IEEFA). Capacity costs pass through to every large electricity customer in that zone, whether the facility in question is a hyperscale campus or a manufacturing plant with nothing to do with the congestion driving the price up. Virginia's own legislative watchdog, the Joint Legislative Audit and Review Commission (JLARC), has projected that the state's unconstrained power demand could double within a decade, a forecast that tracks closely with PJM's own outlook for the region.
Bridge Power Is a Capital Expense Few Site Selection Models Include
When grid power will not arrive on schedule, the fallback is temporary generation, and that fallback has its own price tag. Virginia has permitted nearly 10,500 diesel generators statewide, representing roughly 27 gigawatts (GW) of backup capacity. Those generators serve a range of customers, including hospitals, wastewater treatment plants, telecom infrastructure, and conventional manufacturing backup, but JLARC has tied a meaningful share of the recent growth in permitted capacity directly to data center construction timelines. One prominent example of gas-fired bridge power shows the scale involved: INNIO and VoltaGrid structured a 2.3 GW order, delivered as 92 power packs of 25 MW each, to support Oracle Cloud Infrastructure campuses while permanent grid connections catch up. A manufacturer facing a similar delay is unlikely to sign an order that size, but the same logic applies at a smaller scale. Temporary generation is no longer a contingency line item. It is a capital commitment sized to bridge a gap the utility created.
Carrying Costs Pile Up While the Meter Stays Dark
Every month a facility sits complete but unpowered, interest on the construction debt keeps accruing, and skeleton crews still have to be paid just to maintain a building that isn't running. Equipment depreciates the whole time without producing anything. CBRE's 2026 briefing on energy and construction cost volatility urges finance teams to review energy contract structures and reassess construction budgets against current assumptions rather than the numbers baked into a program approved a year or two earlier. In the data center sector specifically, per-MW build costs run $8 million to $12 million for standard capacity and $15 million to $20 million or more for AI-optimized builds, and a facility carrying that kind of capital while sitting idle is carrying a much larger number every month than the figure on the original pro forma. Manufacturing projects rarely publish comparable per-unit benchmarks, but the same math, capital deployed against a completion date that keeps slipping, applies just as directly.
The Hidden Costs Rarely Included in the Original Pro Forma
The costs that get missed are rarely exotic. Delayed revenue generation is the most obvious one: a facility that cannot operate cannot bill anyone. Interest on construction loans keeps compounding whether the building is producing anything or not. Temporary generation, when it becomes necessary, shows up as an unplanned capital outlay rather than an operating expense anyone modeled. Equipment commissioning delays push out the point at which a facility can even attempt full production, and idle workforce costs accrue for every week between hiring and actual output. Customer commitments made on the original opening timeline get renegotiated, sometimes with penalty clauses attached, and construction phases deferred to wait for power tend to cost more the second time around as materials and labor prices move. None of it shows up on a standard site selection scorecard, and by the time it does, it is already locked in.
The Sites That Avoid This Cost Lock In Power Before the Land Deal Closes
The "energy-first" era, where power access alone determined a project's fate, is over. Projects increasingly need what the firm calls a social license to operate, layering utility relationships and regulatory durability on top of raw power access. The clearest evidence of that shift is on the generation procurement side rather than site selection itself: Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI signed a White House Ratepayer Protection Pledge in March committing to build, bring, or buy their own generation rather than assume the grid will simply provide. Few manufacturers have that kind of leverage or balance sheet, which makes the underlying lesson more urgent, not less. A utility's informal assurance that power will be available is not the same as a signed agreement for delivery by a specific date, and companies that treat the two as equivalent are the ones discovering the hidden cost after the opening date has already slipped.
