Texas Froze Its Data Center Queue — What Buyers Need to Know

In early August, Texas Governor Greg Abbott did something no governor of a major data center market had done before: he told the state’s grid operator to stop connecting new data centers until it could prove the backlog wasn’t a threat to the grid itself. The number that triggered it was staggering. ERCOT’s interconnection queue had swelled to roughly 474 gigawatts of requested capacity — about five times the grid’s all-time peak demand record of 91.3 GW, set during a July heat wave. Roughly 90% of those requests, by Abbott’s own office’s estimate, came from data centers. Three months later, the pause is still in effect, the audit has a real deadline, and BloombergNEF has put a dollar figure on what it could cost buyers: up to $15 billion in delayed AI compute revenue. If you’re sourcing capacity anywhere in ERCOT territory, or watching Texas as a bellwether for how other states might react to runaway interconnection queues, this is the story to understand before you sign anything.

What Actually Happened, and When

The sequence matters because it shows this wasn’t one order but an escalating series of them. On June 10, 2026, Abbott’s office first directed the Public Utility Commission of Texas (PUCT) and ERCOT to find ways to keep data center infrastructure costs off residential bills. On August 3, he went further, ordering a “comprehensive verification and audit of all data centers” in the ERCOT interconnection queue, with a blunt warning: any project that fails will be denied connection to the grid. ERCOT responded the same day by suspending the first scheduled deliverable of Batch Zero, its new large-load transmission planning process, and by August 20 had paused approvals to energize any data center or crypto-mining site of 75 MW or more.

The audit itself covers five areas: whether a project pays its own infrastructure costs or leans on public incentives; whether it brings its own power or depends on the grid, including any on-site generation plans; whether it brings and reuses its own water rather than competing with local communities for it; what it’s doing to limit impact on neighbors; and who actually owns and controls the project. On September 14, Abbott pulled the Texas Water Development Board into the process to compel water-use surveys, with noncompliant projects facing civil and criminal liability and denial of environmental permits. On September 21, he directed the Texas Commission on Environmental Quality (TCEQ) to stop issuing any data center permits at all until ERCOT’s audit wraps.

The Pause Isn’t as Clean as the Press Releases Suggest

Here’s the part buyers should sit with: an October 20 report from FracTracker and Public Citizen found that TCEQ took 35 permitting actions on data center projects after the governor’s directive — 14 fast-tracked or engineer-assigned permits, 11 stormwater approvals, 7 emissions-estimate certifications, and 3 public notices, including preliminary approvals tied to power plants meant to serve a Nexus Data Centers campus in Hubbard. Named recipients of stormwater approvals included projects from Crusoe and EdgeConneX. The governor’s press secretary disputed that any permits had been “approved,” and TCEQ’s interim executive director said in an October 19 letter that staff are screening and pausing applications covered by the directive, with the stormwater permits since suspended.

The disagreement over what counts as “paused” is the real lesson here, not a gotcha against either side. A directive that sounds absolute in a press release can still leave room for staff discretion, grandfathered applications, and adjacent permit types to keep moving. If you’re underwriting a Texas project right now, don’t take “the state paused data centers” at face value in either direction — get your own written confirmation from TCEQ and ERCOT on exactly which of your project’s pending approvals are actually frozen versus still processing.

The Price Tag and the Calendar

BloombergNEF’s analysis, led by senior policy associate Derrick Flakoll, estimated the pause could delay 49.8 GW of new ERCOT data center demand — nearly 20% of the roughly 253 GW U.S. development pipeline it tracks. Modeling a three-month slip for projects scheduled to come online between Q3 2026 and Q1 2027, BNEF put cumulative revenue at risk at around $8 billion by Q1 2027 if 60% of the delayed load is AI-related, rising to roughly $15 billion in a full-delay, all-AI-compute scenario. The gap between those numbers comes down to one thing: AI compute rents for vastly more than traditional cloud capacity. BNEF, citing Silicon Data’s GPU rental benchmarks, pegs AI compute revenue at roughly $1.76 billion per GW per month, versus CBRE’s estimate of up to $175 million per GW per month for a traditional colocation shell in Dallas — a tenfold gap that explains why even a short delay on AI-heavy projects moves the dollar figure so much.

The dates that matter for planning: ERCOT’s September board update showed 362 large loads qualified to move forward in Batch Zero, split between 204 projects (66.4 GW) conditionally included as base load and 158 as studied load — while 373 projects totaling 302.2 GW were excluded outright. ERCOT is due to file eligibility and community-impact reports with the PUCT by December 10, 2026, with the commission discussing them on December 17, and study results for passing loads due April 9, 2027 (a date ERCOT says it isn’t trying to move). Underneath all of this sits the PUCT’s new SB 6 interconnection rule, adopted September 18: a flat $100,000 study fee for all large-load customers plus financial security of $50,000 per MW of gross contracted peak demand — calculated on a project’s full load, so co-located or backup generation doesn’t reduce the deposit required.

Texas Isn’t Alone, and That’s the Bigger Signal

It’s tempting to read this as a Texas-specific political story ahead of Abbott’s November re-election, and BloombergNEF itself raised that possibility. But treating it as a one-state anomaly misses the pattern. New York’s Executive Order 62 created a temporary moratorium in July on new hyperscale data centers of 50 MW or more. Illinois paused new Data Center Investment Program agreements starting July 1. Florida’s SB 484 now bars cost-shifting of large data center costs onto residential and small-business ratepayers. And outside the regulatory world entirely, PJM’s independent market monitor told federal regulators in September that 4,017 MW of forecast data center load — 42% of the growth PJM had assumed would materialize — has already been canceled or delayed, with the Department of Energy now pushing PJM to formally track which large projects are delayed, reduced, or canceled going forward.

The common thread across all of these is that grid operators and state governments are no longer taking data center load forecasts at face value, and buyers shouldn’t either. A queue number like 474 GW is a list of requests, not a demand forecast; ERCOT’s own July operations report showed only 46% utilization against already-approved energization capacity, which is as much a story about timing as about inflated asks. Our look at time-to-power as the real 2026 site-selection metric flagged this shift before Texas made it official policy: the binding constraint on a data center deal is increasingly the queue, not the chip order.

What This Means If You’re Buying Capacity Right Now

A few concrete steps are worth taking regardless of where your project sits in Texas’s process. First, get in writing whether your specific site falls inside Batch Zero’s 204 conditionally-included base-load projects, its 158 studied-load projects, or the 373 excluded ones — that classification, not the headline pause, determines your real timeline. Second, price the SB 6 security deposit into your capital schedule now: $50,000 per MW on gross contracted demand is a real cash commitment before a study has even started, and it doesn’t shrink if you plan behind-the-meter generation. Third, don’t assume self-generation is a clean workaround for grid delays — on-site power still needs TCEQ air permits, which fall under the same pause. Fourth, document your water sourcing and community-impact plans before the December 10 filing, not after; projects that can answer the audit’s questions cleanly are the ones best positioned when classifications resume. And finally, if your timeline genuinely can’t absorb a slip into the April 2027 study window or beyond, it’s worth pricing out a comparable site outside ERCOT, because the PJM data suggests queue discipline is becoming the rule rather than the Texas exception.

None of this means Texas stops being a top-tier data center market — it remains the country’s second-largest after Virginia, and the Data Center Coalition’s Dan Diorio has argued a well-run audit could ultimately separate serious developers from speculative ones, reinforcing the state’s position. But “second-largest market” and “fastest path to power” are no longer the same claim, and buyers who conflate them are the ones most likely to get caught by a December filing deadline they didn’t know existed. For a structured way to run this kind of due diligence across any market you’re evaluating, start with the TechInfraHub 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.

Connect on LinkedIn →