Every data center RFP eventually asks the same three questions: how much does it cost, how fast can it be built, and how big can it get. In 2026, a fourth question has quietly become the one that kills more deals than the other three combined: how long until the utility can actually turn the power on? JLL’s 2026 Global Data Center Outlook puts a blunt number on it — seven years to secure a 100 MW connection in Northern Virginia, a figure the firm says is now enough to stop conversation cold at preconstruction meetings. Buyers who are still underwriting site selection around land price and fiber access are underwriting the wrong variable. The binding constraint has moved upstream, into the interconnection queue, and it is reshaping where capacity actually gets built.
The queue is now the project
Lawrence Berkeley National Laboratory’s “Queued Up: 2026 Edition” is the closest thing the industry has to an authoritative count of what’s actually waiting for grid access. As of the end of 2025, roughly 8,200 projects representing about 1,312 GW of generation and 749 GW of storage — some 2,060 GW combined — were sitting in interconnection queues across the seven ISO/RTOs and more than 50 balancing areas LBNL tracks, covering about 98% of U.S. generating capacity. The median time from interconnection request to commercial operation for projects that reached service in 2025 was over five years, and LBNL notes that duration is still climbing.
The historical odds are worse than most buyers assume. Of the interconnection requests filed between 2000 and 2020, LBNL found only 13% had reached commercial operation by the end of 2025. Seventy-five percent were withdrawn outright. That matters directly to a capacity buyer: a developer’s marketing deck showing “queued” megawatts is showing you a number with roughly a one-in-eight historical chance of ever energizing on the original timeline. Data Center Watch, a research group that tracks project opposition and delay, estimated more than $150 billion in U.S. data center projects were blocked or delayed in 2025 alone — not from lack of capital, but from power, permitting, and local pushback. FERC has started reforming large-load interconnection rules in response, which is worth understanding before you sign anything (see our breakdown of FERC’s new interconnection rules), but reform is a multi-year process layered on top of an already multi-year queue.
Northern Virginia, Phoenix, and Dallas-Fort Worth are now cautionary tales, not default choices
The three markets that built the hyperscale industry are the three markets where the math has turned against new entrants. In Dominion Energy’s territory, data centers already account for roughly 25% of Virginia’s electricity load, and Mordor Intelligence’s analysis of Dominion’s own filings projects that could reach 46% by 2030. Dominion’s 2024 Integrated Resource Plan says the utility needs roughly 27 GW of new generation by 2039 just to keep up, and PJM has reportedly fielded 92 separate transmission upgrade bids tied to the buildout, with one analyst estimate putting the total cost near $51 billion. CBRE says the region still absorbed more than 1 GW of new capacity in 2025 — demand hasn’t slowed, the queue has just gotten longer behind it.
Texas tells a similar story with different mechanics. ERCOT’s large-load interconnection queue hit roughly 410 GW in April 2026 — about five times the grid’s current generating capacity — and data centers account for an estimated 73% of that backlog, according to ERCOT data reported by the Texas Tribune. The state passed Senate Bill 6 in June 2025 specifically to let ERCOT manage large-load curtailment, and the grid operator’s “Batch Zero” cluster-study process, approved in June 2026, is an attempt to triage that queue into something buildable. It’s a serious effort, but it’s a response to a problem that already exists, not a reason to assume Texas timelines look like they did in 2023.
Need Expert Guidance?
Talk to a Data Center Expert
21+ years of hands-on experience in data center design, operations & infrastructure. Book a quick discovery call to discuss your project.
📞 Book a Discovery CallPhoenix shows what happens once a region’s committed capacity turns into a planning ceiling. Arizona Public Service has roughly 4.5 GW of committed large-load demand in queue, and Pinnacle West CEO Ted Geisler told investors on an August 2025 earnings call that APS sees another 19 GW of potential load behind that. A joint Energy Promise Task Force review of APS and Salt River Project pipelines in April 2026 put combined proposed data center load at roughly 29,000 MW — an order of magnitude beyond what either utility is built to serve on any near-term timeline. CBRE’s assessment is unambiguous: new projects in Phoenix without an existing power commitment now face timelines comparable to Northern Virginia’s.
Secondary markets are absorbing the overflow — with real tradeoffs
This is where the picture gets more useful for buyers who still need capacity this decade. Grid-tracking firm DC Hub’s Data Center Power Index shows tier-1 markets’ aggregate “excess power” score falling 11% in 2026, while secondary markets — its ranking covers roughly 332 scored metro areas — gained 18% on the same measure. Markets like Cheyenne, Wyoming; Columbus, Ohio; Mt. Pleasant, Tennessee; and San Antonio, Texas are the names DC Hub flags as having moved from “ready to build” afterthoughts to genuine options, even as it flags Northern Virginia, Phoenix, and Silicon Valley as markets that moved the other direction — from “ready to build” in 2023 to “avoid” or “caution” today. The same data shows what that repricing looks like in M&A: hyperscale-grade campuses of 50–200 MW traded at $12–18 million per megawatt in 2024–2026, up from roughly $5 million per megawatt in 2019. As one market participant quoted in DC Hub’s reporting put it, “buyers are paying for the queue position, not the concrete.”
The counterargument matters, and buyers shouldn’t skip it. 451 Research’s framing for why a market succeeds is still land, reasonable power cost, dense fiber, and proximity to hyperscaler demand — together, not power availability alone. A secondary market with a short queue but thin fiber routes, no bench of experienced electrical contractors, or a two-hour drive to the nearest major peering point can still blow a build schedule. iMasons’ State of the Digital Infrastructure Industry 2026 report puts the skilled-labor gap in concrete terms: more than 80,000 electrician job openings annually in the sector, a shortage that doesn’t disappear just because a site has grid headroom. Large power transformers — the component that actually closes the loop between substation and facility — still carry two-to-four-year lead times industry-wide, regardless of how fast the interconnection agreement gets signed.
What this changes about how buyers should diligence a site
The practical shift is sequencing. Price per kilowatt and contract terms used to be the first filter; time to power now has to come before both, because it determines whether the other numbers are even real. Before evaluating a developer’s pro forma, ask for the actual interconnection queue position and request date, not just the advertised “available” megawatts — and treat anything without a signed power letter of intent from the utility as a timeline estimate, not a commitment; Dominion explicitly tells prospective customers not to rely on a site without one. Ask what a developer’s fallback is if the utility slips: behind-the-meter generation has become a real bridge option, with iMasons pricing a 350 MW gas-fired bridge installation at roughly $500 million in capital, but that only works if gas turbine and emissions-permitting timelines are also in your model. And weight the labor and materials side of the equation as heavily as the electrons — a transformer on order today and an electrician workforce that exists locally are just as load-bearing as a queue position.
None of this means avoid Northern Virginia, Phoenix, or Dallas-Fort Worth outright — absorption numbers show real demand is still closing in all three. It means treat them as markets where you need an existing relationship, an existing power commitment, or a multi-year runway, and treat the Cheyennes and Columbuses of the data center map as live options rather than fallback plans. The buyers who get burned in 2027 won’t be the ones who picked the wrong chip generation. They’ll be the ones who signed a lease on a queue position that was always more likely to be withdrawn than energized. Before you commit capital to a site, get the toolkit built for exactly this kind of diligence at TechInfraHub’s Data Center Buyer’s Toolkit.
Written by
Raajeev Ratra
Data Center Infrastructure Expert | 15+ Years in DC Design, Operations & Project Management
Raajeev is a seasoned data center professional with hands-on experience in hyperscale facilities, colocation design, power & cooling infrastructure, and global DC operations. He shares practical insights to help engineers and IT leaders build better infrastructure.