Texas Froze Its Data Center Boom. The Scarcity It Confirms Is Firm Power.

Texas Froze Its Data Center Boom. The Scarcity It Confirms Is Firm Power.

The backlash against data centers has hardened from neighborhood protest into official state policy, and in Texas it carries the weight of the governor's office. In early August, Greg Abbott ordered the state's utility commission and ERCOT — the nonprofit that operates the Texas grid and its wholesale power market — to audit every data center sitting in the queue to connect, a freeze that blocks new grid hookups while a review regulators expect to last several months runs its course. Defending the move on a Sunday morning news show, he said the industry basically dug their own grave for the problem that's been caused for them and that's why they got the backlash they deserve, because the facilities "pop up in places that nobody ever heard of before" rather than working with the communities around them.

There is real anger behind that verdict, and it is easy to see why. Polls show Texans broadly oppose data centers in their own communities, and a tax break the industry has enjoyed since 2014 now runs past a billion dollars a year. It is, after all, the same governor who less than a year ago was branding Texas the epicenter of AI development. But the conclusion the political class and a good part of the market have drawn from the anger — that data centers are the problem and slowing them down is the fix — is, in my opinion, a false narrative. It mistakes a demand problem for a supply problem. The thing Texas is short of is not data centers; it is firm power, meaning generation that can be turned on when the grid calls for it. And a permit freeze does not create a single megawatt of that.

Consider the scale of what the freeze is sitting on. ERCOT's interconnection queue — the list of projects that have asked to plug into the grid — holds more than 474 gigawatts of requests, roughly 90% of it from data centers, against an all-time record peak of 91.3 gigawatts set in late July. That is more than five times the most Texas has ever drawn from the grid at one moment. The audit will pull in about 250 to 300 of those projects representing some 200 gigawatts of load, according to ERCOT officials. Abbott's directive arrived after the state's first attempt to count what these facilities actually use — a water-and-power survey where, of 377 companies notified, only 28 submitted responses. Not all of the queue is real; both the grid operator and the developers admit much of it will evaporate as financing and leases fall through, and ERCOT has conceded it cannot produce a reliable demand forecast out of the pile. But the one thing nobody can wave away is the direction of travel: load arrived faster than the grid operator could even study it. That is the working definition of a supply constraint.

The Supply Gap

Texas runs what the industry calls an energy-only market: generation is paid only for the power it delivers, and scarcity prices are allowed to spike — up to $5,000 per megawatt-hour — to drag every available unit online. Its marginal fuel is natural gas. Yet the pipeline of new firm capacity is startlingly thin: roughly 3 GW of new dispatchable gas and diesel capacity is under development against 23 GW of battery storage, and gas turbines themselves are backed up in supply chains. In practical terms, a region drowning in cheap natural gas is still bottlenecked on converting that gas into kilowatts fast enough. University of Houston energy fellow Ed Hirs adds the retail-side corollary: data center and crypto load has been pushing up electricity prices by at least 5% a year while the grid cannot serve the volume of new requests.

The federal energy office's own modeling shows what the imbalance does to prices. In its high-demand scenario, average wholesale power at the ERCOT North hub runs roughly 79% above its baseline forecast in 2027, or about $85 per megawatt-hour against a baseline near $47, with gas and coal burned harder to keep up. The Dallas Fed's researchers compress the cause into a single line: AI data centers are reversing two decades of flat U.S. electricity demand. In an energy-only market that hides no cost, load that grows faster than firm supply does not negotiate politely. It reprices the entire forward curve.

The Freeze Creates Nothing

None of this is changed by the freeze, because a pause on demand does nothing to accelerate supply. BloombergNEF figures the order will delay 49.8 GW of U.S. data center load, roughly a fifth of the national development pipeline, and, in the worst case, cost developers on the order of $15 billion in machine revenue that cannot go live. Deferred demand is not destroyed demand. Most of it will wait, and some of it will go where it can get power faster — which is how a state that wants to be the AI epicenter manages to hand the buildout to someone else.

The freeze does not even reach the fastest-growing escape hatch: generation built behind the meter, on the data center's own site, which never touches the ERCOT grid at all. Texas already leads the U.S. with roughly 40 GW of announced behind-the-meter capacity, much of it gas-fired and stacked in the Permian Basin where natural gas has nowhere else to go. This is where the New Age of Energy Abundance meets the AI buildout — the same fracked gas that caps the upside in oil prices is becoming the fuel cell for compute, and the chronic negative streaks in Waha natural gas prices have been erased by exactly this demand. The perverse effect is that the more onerous Texas makes its grid connection process, the more load pushes behind the meter, outside ERCOT pricing, outside ratepayer rules, and largely outside the state's new standards.

The politics is meanwhile starting to fight the physics on the other flank. The 765-kilovolt transmission lines that would carry West Texas power to the load centers have ignited a rural land fight; Lt. Gov. Dan Patrick is pressing regulators to deny the lines until the legislature rewrites the approval process — the same lines ERCOT warns could bring rolling blackouts to the Permian Basin if they are not built. Abbott's own standards add a requirement no utility-scale project could satisfy: that data centers actually lower the cost of electricity for consumers. Bolt hundreds of gigawatts of new load onto a grid and charge it for the infrastructure, and you get many things; falling bills for every household is not one of them. The openly sensible part of this agenda is the "who pays" fight — a 2025 Texas law now directs regulators to make the largest load customers carry a reasonable share of the interconnection and transmission costs they trigger, rather than socializing it across every household under the grid's old peak-hour formula. That is the correct, pro-ratepayer reform. The freeze is the theater surrounding it.

The Trade

Which is where this stops being a Texas story and becomes positionable, because the market briefly embraced the false narrative. Vistra is the cleanest pure-play on Texas merchant power — a generation fleet that sells into the ERCOT market rather than serving a captive customer base — and it has been sold hard. The stock sits near its 52-week low after giving back roughly 29% over the trailing year, including about 17% in the month of the freeze, as the near-term forward curve reset on deferred demand. That selloff is, in my opinion, an irrational overreaction. Deferring a plug-in date changes the quarter; it does not change the equation the queue already documented — 474 GW of load standing against a 91.3 GW grid.

The fundamentals now line up at a price the selloff stopped respecting. Vistra trades around 10 times trailing EV/EBITDA — enterprise value over earnings before interest, taxes, depreciation, and amortization, a rough cash-earnings multiple — against roughly 13 to 16 times for the rest of the complex, with CenterPoint, Constellation, NRG, and AEP all richer on the same basis. Free cash flow rose 36% over the past year; the dividend has grown for six consecutive years and the payout sits near 15% of earnings. Be clear about what this is: at a yield under 1%, it is not an income stock, and the balance sheet is the honest caveat — about $19 billion of net debt against $5.1 billion of operating cash flow is a lot of leverage to carry into a power-price cycle. That is exactly why this belongs as a moderate, volatility-tolerant position rather than the anchor of an income portfolio.

Of the alternatives, I would rather own the existing generation than the construction bets. GE Vernova sells the turbines and grid equipment the buildout will demand whether the freeze holds or not, but that story is no longer cheap. CenterPoint, the Houston-area regulated wires utility, offers a safer roughly 3% dividend on the same load growth — a denser expression of the same thesis for investors who want less torque. Vistra is the pure scarcity trade: the cheapest multiple in the complex, leveraged to exactly the thing Texas cannot produce fast enough.

That being the case, I rate Vistra a Buy on the pullback — for investors who can live with the balance-sheet leverage and power-price volatility, sized as a moderate slice of energy exposure, not as a stand-in for a utility income holding. The conditions that would break the thesis are knowable: ERCOT forward prices falling durably beyond what the market has already reset, or anchor data center customers walking away from their firm-power commitments in volume. Neither has happened, and the freeze on its own does not do it. The state was angry enough at the data centers to freeze them, and the industry earned its share of that anger — but the market has translated a political settling of accounts into a repricing of the one asset that is about to grow scarcer rather than less scarce. That, in my opinion, is the false narrative, and it is why I would buy the dip.