Texas Just Moved the Data-Center Bottleneck From Zoning to the Interconnection Queue

For two years, this has been a local story. A county board votes. A township passes a 180-day pause. A planning commission adds a setback nobody expected. Developers got good at reading that map — find the friendly jurisdiction, and the project moves.

Texas just changed which map matters, and it did it in a single letter.

On August 3, 2026, Governor Greg Abbott sent a letter to the Public Utility Commission of Texas and ERCOT directing an audit of every data center project currently advancing through ERCOT's interconnection process — and instructing that no additional data centers be approved to move forward until it is done. Any project that fails to comply with the audit is to be denied connection to the Texas grid. (Texas Tribune; KERA News)

The size of what just stopped

The audit covers roughly 250 to 300 projects, most of them data centers, together accounting for about 200 gigawatts of future demand. That is more than twice ERCOT's all-time peak demand record, set the month before. (KERA News)

The procedural consequences landed immediately. ERCOT suspended the Batch Zero Large Load classification notifications that had been scheduled for August 7, 2026, and moved to seek a good-cause exception at the PUCT's August 20 open meeting. ERCOT has indicated the audit itself will take several months to complete, and that no data center can be approved for grid connection until it is finished. (Gibson Dunn; Troutman Pepper Locke)

This was not a first move. On June 10, 2026, Abbott had already directed the PUCT and ERCOT to act on the cost that rapid data-center expansion was pushing onto Texas ratepayers. (MultiState) The August order is the escalation of that same concern from price to access.

Why it lands hardest here

Texas is the last place a developer expected to find a gate. Look at what our own data says about it.

• Texas has more counties than any other state, and the SitePath county dataset puts its median permitting-risk score at 34.6 — the third-lowest median in the country, behind only Montana and Vermont, and this in a state with no active RPS mandate.
• No Texas county carries an active utility-scale solar moratorium. Nationally, a dozen or more do.
• Texas is SitePath's densest battery-storage build-out: dozens of counties, nearly all of them classified major-deployment, and exactly one carrying an enacted ban.
• Texas holds dozens of data-center county records — more than half emerging-market, a sizeable minority major-market — and not one of them is in a restrictive posture. A minority of all mapped data-center project points nationwide sit in Texas.

County ordinances and commissioners-court orders.

Put those together and the picture is unambiguous. By every land-use measure we track, Texas is the most permissive large market in the country. The county boards were never the obstacle there.

Which is exactly why this matters more than any local vote of the last two years. The state reached past the layer developers had learned to manage. All that hard-won skill at working a county board is worth nothing at an interconnection queue.

The local layer did not stand still

Underneath the state order, Texas jurisdictions have been building their own rules, and the pattern is instructive:

• Austin County passed a countywide moratorium in July 2026 on new AI data center and battery energy storage development — one instrument covering both technologies.
• Hill County adopted a one-year moratorium on data-center construction in unincorporated areas in May 2026, then rescinded it in June 2026 in the face of a $100 million lawsuit, replacing the pause with a checklist of requirements for developers.
• El Paso (July 2026) set a 300-foot minimum distance from residential areas plus noise mitigation and special-use permitting.
• Forney (April 2026) confined data centers to light industrial zoning with a 1,000-foot residential buffer.
• Lewisville (June 2026) required a special use permit with two public hearings.
• Mesquite (July 2026) established a dedicated regulatory framework covering data centers and BESS together.

(MultiState)

Hill County is the one to study, and it cost $100 million to teach the lesson. A moratorium is the fastest tool a county has and the most legally exposed. A requirements checklist is slower to write and far harder to sue. Notice that the jurisdictions who moved second — El Paso, Forney, Lewisville, Mesquite — all wrote rules rather than pauses. Rules are what survive.

What it means for you

The binding constraint moved upstream. Land use asks may you build here. Interconnection asks may you draw power at all — and in Texas that question is now answered by a state audit, not a county board. Diligence that scores a site on zoning alone is scoring the wrong layer.

Storage is being pulled into the data-center fight. Austin County and Mesquite both wrote data centers and BESS into a single instrument. That coupling is new and it is a risk to storage developers who have spent three years building a permitting record entirely separate from the AI buildout. A battery project that would sail through on its own merits can now inherit a data center's politics.

Speed was always the asset, and it just repriced. Roughly 200 GW of requests sit behind a process expected to run for months — in the market that was supposed to be the fast one. If you hold a queue position anywhere else in the country, it is worth more this morning than it was three weeks ago. That is not a happy accident; it is the whole game now.

Expect the model to travel. A grid-side audit is cheaper for a state to run than a legislative moratorium, harder to challenge than a local ban, and it reaches every project at once. Texas is not the last state that will notice this.

What this analysis does not tell you

It does not tell you which specific Texas projects will clear the audit, how ERCOT will sequence Batch Zero once notifications resume, or whether the PUCT granted the good-cause exception sought at its August 20 meeting — that meeting is the same day this piece publishes. It does not model rate impacts. And SitePath's county-level classifications describe land-use posture only; they carry no interconnection-queue status, which is precisely the layer that just became decisive.

Figures drawn from SitePath's own county records are computed when this page is built, so they move when the data does. Everything else is linked to its primary source.