Call It What It Is: Texas Didn’t Order an Audit. It Ordered a Pause.

By Joseph Hoefer

Texas isn’t an outlier. It’s an early signal of where AI infrastructure policy is heading.

Gov. Greg Abbott didn’t call it a moratorium. He called it a “comprehensive verification and audit.” But for the roughly 2,000 data center projects now sitting in the Electric Reliability Council of Texas’s interconnection queue, the practical effect is much the same: projects cannot meaningfully move forward until the review is complete. Whatever label Texas chooses, that’s a pause in all but name.

This matters beyond Texas because other states are likely to watch what happens next. Six weeks earlier, New York Gov. Kathy Hochul imposed a year-long pause on construction of the largest data centers and took a very public hit from President Trump for doing it. Abbott, a Republican governor in the country’s second-largest data center market, has now arrived at a remarkably similar place without ever using the word “pause.” That a blue-state governor and a red-state governor have converged on the same instinct tells you this is no longer a partisan fight over whether AI infrastructure should exist. Instead, the debate is maturing. States increasingly want more visibility into what they’re approving before they approve the next wave of projects. The question is shifting from whether to build AI infrastructure to how states should govern it.

That instinct is legitimate. Texas has concluded it doesn’t have enough information to evaluate the cumulative impact of what’s sitting in the queue. Regulators have struggled to get consistent information about projected power demand, water use, and the interaction between tax incentives and infrastructure planning. That isn’t necessarily because developers are withholding information. It’s because permitting and interconnection processes were built for an era of individual industrial facilities, not hundreds of gigawatt-scale projects arriving within a compressed period of time. That’s a systems problem, not an indictment of any developer or operator. Closing that information gap is a reasonable goal, and an audit is a reasonable way to pursue it.

But reasonable goals still have costs, and that’s the part of the conversation many states are skipping.

Start with the communities waiting on these projects. Many of the towns and counties with facilities in the queue are counting on the tax base, jobs, and infrastructure investment tied to a specific development timeline. A review process without a clearly defined endpoint doesn’t just delay approvals. It delays those benefits as well, creating real costs for communities that expected investment to arrive on a predictable schedule.

Then consider the capital behind these projects. Financing commitments, land options, power purchase agreements, and orders for transformers and other long-lead equipment don’t simply wait while regulators deliberate. Every month of uncertainty gets priced into the next deal, either through higher financing costs or through decisions to deploy capital in jurisdictions that offer greater regulatory certainty.

This is where an open-ended audit can run counter to its own purpose. The developers most exposed aren’t speculative operators with little to lose. They’re the companies that have already invested significant capital, secured land, worked with local communities, and committed to long-term development. States often say they want to reward responsible developers while discouraging speculative projects. But without a timely review process, an audit can have the opposite effect, imposing the greatest costs on those with the deepest commitments while allowing less-invested operators to simply wait or move elsewhere. If the goal is to encourage orderly, community-engaged development while discouraging speculative proposals, the timeline matters just as much as the audit itself.

The ripple effects extend well beyond developers. Utilities plan transmission upgrades around expected demand. Construction firms allocate labor months in advance. Equipment manufacturers schedule production years ahead. When project timelines become indefinite, that uncertainty propagates throughout the infrastructure ecosystem. For time-sensitive projects, a prolonged pause can have many of the same practical consequences as a canceled one.

None of this is an argument against scrutiny. If anything, a well-designed statewide review process could prove more durable than the patchwork of county-level fights Texas has already experienced…from Hood County’s failed effort to Hill County’s moratorium and the litigation that followed its rescission, to San Marcos’ local restrictions. Those approaches have been ad hoc, politically volatile, and frequently challenged in court. A clear statewide framework could replace that uncertainty with something more predictable for everyone involved.

But could is doing a lot of work in that sentence, and Abbott’s directive doesn’t yet clear that bar. A durable audit regime needs a defined scope, published timelines, transparent evaluation criteria, and clear treatment for projects already substantially underway. Without those guardrails, the distinction between an audit and a moratorium becomes largely semantic.

The real test isn’t whether Texas audits these projects. It should. The test is whether the state uses that audit to build a predictable regulatory framework or simply creates an open-ended permission slip for delay. Investors can price regulation. Communities can plan around regulation. What neither can price nor plan around is uncertainty. And that’s ultimately what Texas risks creating if this pause remains a process without a predictable end.

Joseph Hoefer is Chief AI Officer & Principal Monument Advocacy. 

This article was originally published by RealClearEnergy and made available via RealClearWire.

The climate data they don't want you to find — free, to your inbox.
Join readers who get 5–8 new articles daily — no algorithms, no shadow bans.
4.2 9 votes
Article Rating
17 Comments
August 14, 2026 10:20 pm

A pause to review and establish procedures is necessary. It is obvious that data centers can be built faster than generation capacity. Moving to bring these two into alignment is the correct course.

Chemman
August 14, 2026 11:13 pm

Assuming that the 2000 data centers in cue are 1 gigawatt each that is 2 terawatts of energy. If 70% is used by the compute that’s 1.4 terawatts of compute needed for Texas to fully utilize the data centers. Where is that coming from because every State is going to be competing for the compute.
AI and data centers can be a good thing is the data centers get used. If the compute isn’t there then a lot of people will be hurt financially from the fallout.

August 15, 2026 4:06 am

“Start with the communities waiting on these projects. Many of the towns and counties with facilities in the queue are counting on the tax base, jobs, and infrastructure investment tied to a specific development timeline.” Wrong, Joe.

It’s the communities that do NOT want the projects and politicians who are trying to balance what they perceive as mana from heaven against the public backlash. That’s your first mistake.

Your second mistake is on the facts. A billion dollars investment actually strips money from the county — it doesn’t inject it into the local economy.

These project don’t ’produce’ sustainable jobs — only during construction are there meaningful jobs, and those are imported to the smaller towns, overwhelming local infrastructure, policing, grocery and liquor store supply chains. After build-out, there’s perhaps 20 jobs at a typical hyper-scale facility — and that is trending down, ironically, as AI develops.

Taxes? Likely wrong again. Most are funded by PILOT schemes that have wrongly calculated the but-for revenue. Millions of dollars of lost taxes are the result, and when combined with the couple dozen permanent jobs created, counties are paying a million dollars or more per job created.

Infrastructure? No, most counties — by which I mean the tax payers of the counties — end up double-subsidizing development because of infrastructure maintenance and build-out costs outside the fence line of the project. Think roads, transmission lines, water; then there’s noise and light pollution that is unpriced and much more than inconvenient.

And the main point of contention, Electricity Costs? In counties that host data centers, bills increase by an average of five percent because of the data centers. Five percent increases on bills to the people who have already double-subsidized these projects.

Don’t get me wrong, there is benefit overall by having hyper-scale data centers. But there are no benefits to rural communities, none. Some cities and states can benefit because they have the market scale and infrastructure to not only attract a single project, but multiple projects — and that is where the benefits tend towards the positive, in the agglomeration affects. One-offs are disasters for local communities.

Sweet Old Bob
Reply to  Willy
August 15, 2026 5:44 am

“In counties that host data centers, bills increase by an average of five percent because of the data centers.”

so why does the Virginia county with 200 centers have the lowest electricity cost in the state ???

Reply to  Sweet Old Bob
August 15, 2026 6:45 pm

That’s why it’s called an average, Bob.

Reply to  Willy
August 15, 2026 6:56 am

I don’t know where you are getting your numbers but, they seem off to me.
Large datacenters require a substantial maintenance and repair crew besides a crew to do hardware and software updates – both crews total over 100 at a hyper-scaler facility.

The major hurdles to datacenter placement are power and water availability. The biggest, in my opinion, is water. Power generation can be from several sources (that should be substantially provided by the developer, in my opinion) but water, has to be local.

The worst place to put a datacenter is where it is dependent on ground water rather than a river or lake. Even with closed loop cooling systems, millions of gallons of water are required to start and maintain levels.

Reply to  Brad-DXT
August 15, 2026 6:50 pm

I agree, with cooling towers, water is a huge deal — quantity required and quality impacts.

Labor is a vapor. A newer hyperscale center will not need more than 50 FTE. These facilities are parasitic on rural host counties — extracting resources from, not adding to, the local economies.

Reply to  Willy
August 16, 2026 6:27 am

I’m about 40 miles away from the latest hyperscale facility being built. The developers state over 100 full time jobs.
I’ve worked on two datacenters. One hyperscaler and one only stretching a city block. The jobs are there.

Reply to  Brad-DXT
August 17, 2026 4:23 am

Your number might be true for that h-s center. Good for you. My number (20) comes from the official submission of the developer for a proposed facility 20 miles from me. We each have a data point. I have also read assessments (most recent by Ga Tech, June 2026) looking at hundreds of operations. Labor demand is shrinking as technology changes. The take-away, Brad, is not that no people are employed — it’s that these facilities do not ‘bring jobs and revenue’ to rural counties. They aren’t net positive. The jobs are not there — if you mean anything at all like, “this will be great for the local job market and economy.” That’s a vapor. Period.

Tom Johnson
August 15, 2026 4:08 am

It seems to me that far more “Data Centers” are being planned than will ultimately be successful. Though quite useful, AI may not produce the $trillions in revenue that would be necessary to make all of the planned centers profitable. The next question should be “What will happen to the ones that fail commercially?” Given the history of rapid advancement in computing technology it’s not wise to predict that will end now.

Scissor
Reply to  Tom Johnson
August 15, 2026 4:37 am

I’m reminded that Paul Krugman wrote that the internet’s economic impact would be “no greater than the fax machine’s.”

George Thompson
Reply to  Scissor
August 15, 2026 5:50 am

Krugman is and always has been a fool, an idiot, and wrong about just about anything economic one could name. Truly a joke…and a bad one at that.

Tom Halla
August 15, 2026 4:37 am

Unfortunately, this is following the Green Blob playbook of no due process and delay

2hotel9
Reply to  Tom Halla
August 15, 2026 5:43 am

Actually, one thing Abbott is about is due process and holding people accountable. THAT scares the hell out of lots of people, especially leftists.

2hotel9
August 15, 2026 5:40 am

Instead of AI data centers they need to build MORE gas fired electricity generation plants.

Dave Andrews
Reply to  2hotel9
August 15, 2026 7:58 am

“Orders for new natural gas fired power plants surged to 130GW in 2025, a 25 year high, with US data centre demand a major factor”

IEA ‘World Energy Investment 2026’ (May 2026)

2hotel9
Reply to  Dave Andrews
August 16, 2026 4:05 am

Saying you want to build something and actually building something are two entirely different things. Consumer demand drives construction of electric generation plants, not fantastical AI crap.