Canceled data center projects leave utilities with stranded infrastructure costs that are often passed on to ratepayers, while capacity reassignment can take months or years to resolve.
When a large data center project s or walks away, it leaves behind more than an empty site. Connecting a facility to the power grid typically requires securing a place in a utility’s interconnection queue, posting a deposit to hold that place, and often having the utility build or expand substations or transmission lines tailored to the project’s specific capacity needs. If a project stalls, these commitments don’t vanish; they’re reassigned, clawed back, or stuck in limbo, and the resulting costs surface in places few people think to look.
Nearly one in five data center interconnection requests never become real load, according to the Capgemini Research Institute’s 2026 report, “AI Meets the Grid: Shaping the Data Center Power Play.” The report, based on a January 2026 survey of more than 600 electricity executives, found that 67% described these as “phantom” load requests, with roughly 19% never materializing. The gap points to a structural problem in how utilities are being asked to plan and invest ahead of uncertain, fluid demand.
Stranded Capacity: Who Pays, and When? #
It’s easy to assume the developer eats the loss and everyone else moves on. That’s only partly true.
Developers do lose real money on sunk costs – mainly engineering, permitting, and early construction that can’t be recovered – said Sam Tabar, CEO of WhiteFiber. But utilities are exposed as well. By the time a project falters, they may already have begun building substations and transmission lines meant to serve it. Those investments typically remain in place as part of their rate base, whether the load shows up or not. Tabar calls that outcome the least examined part of the story. “It is this cost socialization that is largely unaccounted for in the public discussion around data center projects,” Tabar told Data Center Knowledge.
Timing determines who ultimately bears the cost, according to Ildi Telegrafi, a Policy Fellow at the Alliance for Innovation and Infrastructure. If a project is pulled out before any utility construction begins and before customers are paying for it, the cost remains with the developer. If construction is underway or finished and the expected load then disappears, the bill shifts to other customers.
Jim Tyler, CEO of Erthos, frames the outcome as a genuine regulatory puzzle with no default answer, only options negotiated case by case. “This creates highly complex regulatory dilemmas about whether these costs should be reassigned to other projects, sit with the utility, or enter a rate case to be paid by ordinary customers,” Tyler told Data Center Knowledge.
The Interconnection Queue Reality #
There’s a common assumption that when one project cancels, the next one in line simply moves up. It rarely works that way.
“Interconnection studies are done by location and for a specific load,” Tabar said. Swapping in a different project at the same point usually triggers a fresh study because the original assumptions (load size, delivery point, and timeline) may no longer apply. That restudy is not a formality. “Restudies take time, sometimes six months to over a year, and the project inherits updated assumptions about grid conditions that can change its economics significantly,” Whitaker Irvin Jr., founder and CEO of Q Hydrogen, told Data Center Knowledge.
How Cancellations Cascade: Three Case Studies #
Three recent cases show how differently a cancellation or reassignment can play out. Importantly, replacing a project (even at the same site) often requires new environmental reviews, fresh permits, and updated community engagement, which can introduce different timelines and levels of local support. These factors compound the timing and cost considerations that ripple through utilities and markets.
Georgia. Atlanta-based T5 Data Centers proposed a project near Fort Gordon in Augusta in 2022, then withdrew before the year’s end. Eagle South LLC later filed plans to build “Project Eisenhower,” a $2 billion, 2.1 million-square-foot data center campus on the same site, drawn by the same nearby Georgia Power substation that made the location attractive to begin with, according to the Augusta Chronicle. The project is in site development but is delayed relative to its original schedule, which was to be completed in the second quarter of 2026.
Virginia. QTS and Compass Datacenters both walked away from Prince William County’s “Digital Gateway” corridor, a 2,139-acre plan that could have drawn up to $30 billion in investment, after courts invalidated the county’s rezoning approvals. Compass withdrew in April 2026 after spending $40 million, according to Bisnow’s reporting. QTS ended its remaining appeals in July, according to Data Center Knowledge’s reporting. The corridor sat atop Dominion Energy’s transmission infrastructure, which the county had spent years marketing as the project’s core selling point.
Cloud leases. Not every reassignment involves a utility filing. Microsoft walked away from roughly 2 GW of data center leases and projects across the US and Europe in early 2025, according to a TD Cowen analyst note reported by Bloomberg in March 2025. Some capacity in Europe was reportedly reassigned to Google and Meta.
“The biggest shock has been the duration of the ripple effect.” – Sam Tabar, CEO of WhiteFiber
What the Ripple Effect Misses #
These cases point to divergent outcomes for a shared problem. In Georgia, a successor data center project directly reused the site and power access left behind by a canceled one, though years passed between the two. In Virginia, hard-won capacity and transmission access have no confirmed successor yet. In the cloud-lease scenario, a competitor simply stepped into the space a rival had given up – no site or substation involved at all.
“The biggest shock has been the duration of the ripple effect,” Tabar said.
Most people assume that when a project is canceled, the capacity it would have used simply frees up and the market adjusts. Tabar sees it differently: we pay outsized attention to splashy announcements of proposed developments, but far less to cancellations and what happens to the committed infrastructure costs in between.