Texas did something this week that every state chasing artificial intelligence infrastructure should study before the next county zoning hearing turns into another all-or-nothing brawl. Gov. Greg Abbott told the Public Utility Commission of Texas and ERCOT to protect ordinary Texans from carrying the grid, water, and neighborhood costs of data center expansion. That is not an anti-compute position. It is the beginning of a build-it-right doctrine.

That distinction matters because the data center debate has been getting stupid in both directions.

On one side, industry boosters sometimes talk like every substation, transmission upgrade, tax break, water withdrawal, backup generator, gas plant, road burden, noise wall, and public-service strain is just the price of progress. On the other side, some opponents talk like compute itself is the enemy, as if the United States can stay competitive in AI, defense, medicine, logistics, education, manufacturing, cyber defense, and energy systems while refusing to build the physical infrastructure that modern computing requires.

BadPD's frame is different: build the compute, but do not make the public quietly subsidize the shortcuts. Bring your own power. Pay for your own interconnection. Use water-efficient or waterless cooling where local water stress requires it. Put real numbers into annual public reports. Do not hide behind tax incentives while residential customers see the infrastructure bill. If a project needs new power generation, storage, transmission, noise mitigation, setbacks, or water protections, put those costs in the project ledger, not on families who did not sign the development agreement.

Abbott's June 2026 move is important because it shows that this is no longer just a left-environmentalist complaint or a rural NIMBY fight. Texas is one of the most pro-growth, pro-business, pro-energy states in the country. If Texas is saying the data center boom needs guardrails, the industry should hear the warning clearly: the political center of gravity is moving from whether data centers are allowed to exist to whether they are allowed to externalize their costs.

What Abbott Put On The Table

Business Insider reported on June 11 that Abbott proposed rules to keep the rapid growth of data centers from burdening everyday Texans with infrastructure costs. The package described in the reporting includes requiring data centers to pay for their own electric infrastructure, requiring water-efficient technologies, and repealing long-standing sales tax exemptions that have become increasingly expensive.

Chron reported that Abbott directed the PUC and ERCOT to ensure data center interconnections reduce residential electricity costs and that data centers pay all electric infrastructure costs tied to their development. The same report says Abbott wants state regulators to review what they can do under existing authority to safeguard Texans, property, and resources, and wants a joint PUC/ERCOT memo by July 17.

Axios San Antonio framed the same move as a notable shift for one of the country's friendliest data center states. Axios reported that Texas has hundreds of existing or planned data centers and that the cost of the state's data center sales tax exemption is projected to grow from $14.6 million in the 2014-15 budget cycle to an estimated $3.3 billion for 2028-29, citing Texas Senate figures. That is not pocket change. That is a public finance question.

The key policy pieces are straightforward.

First, data centers should pay for their own electric infrastructure. If a hyperscale facility needs new substations, transmission upgrades, distribution changes, or local grid hardening, that cannot become a hidden residential rate increase unless the public has deliberately chosen that subsidy in daylight.

Second, data center interconnections should be structured to reduce or at least not increase residential electricity costs. That is a tougher standard than merely saying a project brings jobs or tax base. It asks whether the power system effects are net-positive for households.

Third, new facilities should use water-efficient cooling and report electricity and water consumption. This is basic accountability. If a facility wants public incentives, utility cooperation, or zoning approvals, the public deserves a yearly ledger of how much power and water it uses and what it is doing to reduce local stress.

Fourth, future facilities should reduce neighborhood impacts through things like setbacks and noise-reduction technology. Data centers are not invisible. Their diesel backup generators, cooling systems, substations, transmission corridors, security perimeters, and traffic impacts land somewhere.

Fifth, Texas lawmakers may revisit tax incentives. This is where the political fight gets real. A state can decide that a tax break is worth it. But once the break grows into the billions, the burden of proof should move back to the companies and officials defending it.

The BadPD Build-It-Right Test

The question is not whether data centers are good or bad. The question is whether the deal is honest.

A serious data center approval should answer seven questions before the ribbon-cutting photo.

Who pays for the grid upgrades?

Who pays for new generation or storage?

What happens to residential bills?

How much water is used, where does it come from, and what happens during drought?

What cooling system is used and why?

What incentives, abatements, exemptions, or public infrastructure commitments are attached?

What binding obligations survive after the press release?

If those questions are answered clearly, then build. If they are buried in consultant memos, nondisclosure agreements, utility filings, late agenda packets, or economic-development booster language, then the public should slow the deal down.

This is where Abbott's move is most useful. It pressures the industry toward better projects without pretending that America can ban its way into technological strength. The goal is not to give China, Russia, or any other competitor a free strategic lane because local governments got burned by bad incentive deals. The goal is to stop weak deals from poisoning public trust in good infrastructure.

The build-it-right model says a data center can be pro-American and still be required to bring its own power strategy.

It can be pro-AI and still be required to use water-efficient cooling.

It can support national competitiveness and still be required to pay for the substation.

It can create jobs and still be required to publish its annual water and electricity ledger.

It can be welcomed by a state and still be told no if the project depends on a cheap-public-water shortcut or a free-grid-ride.

The Water Ledger Is Not Optional

The Guardian reported on June 8 that a large share of planned AI data centers in the United States are slated for drought-hit land. Its analysis found that many proposed facilities are landing in places already facing water stress, while industry growth is pushing demand for cooling and power. The industry has answers, including closed-loop cooling, water restoration programs, and claims that data centers can use less water than some alternative land uses. Some of those answers may be real in specific projects. Some may be sales language. The only way to know is to force project-level disclosure.

Texas is not one water market. The water question in San Antonio is not the same as the water question in West Texas, North Texas, Houston, or a rural county trying to balance aquifers, agriculture, industrial recruitment, and population growth. A statewide rule that says data centers should not drain water needed for communities is a start. The real test is whether local permits, utility agreements, and project filings make that promise enforceable.

Water-efficient cooling is not a magic phrase. It can mean different things depending on the site, climate, server density, power source, and engineering choices. Closed-loop systems may save water at the facility while increasing energy needs. More energy can mean more upstream water use if the power comes from thermal generation. Waterless or low-water cooling can be more expensive. On-site generation can reduce grid pressure but raise air-quality, noise, fuel, or safety questions. Battery storage can help with peak demand but does not create energy by itself. Microgrids can be useful if they are real, not branding.

That is why the ledger has to include the whole system, not just the part that looks good in a slide deck.

A good project should say: here is how much water we use in normal conditions, here is how much during peak heat, here is the drought contingency plan, here is where the water comes from, here is what we recycle, here is the cooling technology, here is the backup system, here is the power source, here is how we handle leaks, here is how we protect local wells, and here is the enforcement mechanism if we miss the target.

If a company cannot answer that before approval, the answer is not ready.

The Grid Ledger Is The Ratepayer Ledger

Electricity is where the hidden subsidy often lives.

A data center may pay its own direct bill and still impose costs on the grid. If the facility's demand requires transmission upgrades, new substations, distribution reinforcement, reliability projects, reserve capacity, or peak-demand changes, those costs can move through utility rate cases and planning processes that normal residents never read. By the time the bill lands, the ribbon-cutting is old news.

That is why Abbott's instruction to make data centers pay for electric infrastructure is the core of the story. It moves the debate from emotional claims to cost assignment.

If a data center is profitable because it shifts infrastructure costs to everyone else, that is not innovation. That is extraction.

If a project lowers residential bills by financing grid upgrades that benefit the broader system, that is a stronger case.

If a project brings its own dispatchable power, storage, demand-response capacity, or ability to curtail during emergencies, that needs to be counted.

If a project quietly increases peak load, forces new utility investment, and then hides behind job numbers, that needs to be exposed.

The same standard should apply everywhere. Texas, Virginia, Georgia, Arizona, Michigan, Ohio, Minnesota, Illinois, Colorado, Nevada, Oregon, Washington, Pennsylvania, Florida, and New York are all going to face versions of this fight. The details vary, but the public question is the same: is the data center paying its way, or is it arriving with a private-profit plan and a public-cost tail?

The Incentive Ledger Needs Sunlight

The tax incentive fight is going to be messy because governors and local officials love ribbon cuttings. Data centers bring big capital investment numbers. They can boost tax bases, especially where land was previously low-value. They can create construction work and some permanent technical jobs. They can anchor broader industrial development. They can also deliver fewer permanent jobs than the public imagines while consuming enormous power and water resources.

That does not mean incentives are always bad. It means incentives need a scorecard.

How many permanent jobs are created?

What is the average wage?

How much local tax revenue is generated after abatements?

What public infrastructure is required?

What is the cost per job?

What is the cost per megawatt of load?

What is the water cost per unit of public benefit?

What clawbacks apply if promises are missed?

What documents are public before approval?

If the sales tax exemption is projected to cost billions, the legislature should not treat it as background noise. Every exemption is a policy choice. Every policy choice has an opportunity cost. Money not collected from one favored sector is money that has to be made up somewhere else, cut somewhere else, or justified by benefits that can survive audit.

Texas is right to ask whether the old incentive structure still fits the new scale of AI infrastructure.

Do Not Let The Opposition Become Anti-Compute Slop

There is a trap here. Some activists and politicians will hear Abbott's move and use it as a permission slip to oppose every project. That would be a mistake.

America needs compute. It needs AI infrastructure. It needs cloud capacity. It needs cyber defense, health research, logistics systems, grid modeling, advanced manufacturing, education tools, and local economic development. The choice is not between data centers and virtue. The choice is between honest infrastructure and sloppy cost shifting.

A serious opposition movement should not be anti-data-center by default. It should be anti-secret-subsidy, anti-water-grab, anti-ratepayer-bailout, anti-noise-dump, anti-rubber-stamp, and anti-paper-microgrid.

Likewise, a serious industry should not treat every question as backward NIMBYism or foreign-influence conspiracy. If residents ask who pays for the substation, answer them. If farmers ask about aquifers, answer them. If neighbors ask about generator noise, answer them. If regulators ask about curtailment and peak demand, answer them. If watchdogs ask about tax exemptions, answer them.

The companies that can answer should want this fight. A build-it-right standard separates the serious operators from the subsidy tourists.

What Other States Should Copy

Texas should not be the only state moving in this direction. Every state with a data center pipeline should adopt a public package before the next wave of AI load hits the grid.

Require project-level power and water disclosure before approval.

Require a ratepayer-impact analysis for large-load interconnections.

Require data centers to pay direct infrastructure costs created by their projects unless a public benefit is proven and voted on in daylight.

Require drought contingency plans and cooling-technology explanations.

Require annual electricity and water reporting.

Require noise, setback, backup-power, fuel-storage, and emergency-response plans.

Publish all incentives, abatements, and infrastructure commitments in one public file.

Make clawbacks automatic if job, investment, water, power, or mitigation promises are missed.

Create a bring-your-own-power or grid-support credit for projects that add reliable capacity, storage, demand response, or power back to the grid during emergencies.

Do not ban good projects because bad projects made residents angry.

That is the balance. Build the infrastructure. Protect the public. Make the ledger public.

The Texas Signal

The Abbott move is not the end of the data center fight. It is the beginning of a better fight.

The old fight was too often a shouting match: jobs versus environment, AI versus farmers, economic development versus homeowners, tech future versus local control. The better fight asks for receipts.

Show the grid ledger.

Show the water ledger.

Show the incentive ledger.

Show the neighborhood ledger.

Show the emergency plan.

Show who pays if the project is wrong.

If the answer is the company, the project may be worth building. If the answer is the ratepayer, the resident, the aquifer, or the next city council after the tax break has already been signed, then the deal is not ready.

Texas just gave the country a useful phrase: data centers must not pass the bill to everyday people. BadPD would sharpen it further. Build the compute. Bring the power. Protect the water. Pay the cost. Publish the ledger.

That is not anti-AI. That is adult supervision.

The Follow-Up Clock

The immediate follow-up is not a vibes check. It is a calendar check. Chron reported that the PUC and ERCOT have been directed to send the governor a joint memorandum by July 17. Axios reported that the Texas Senate Committee on Finance is scheduled to discuss the data center sales tax exemption on July 27. Those dates should now be on every ratepayer, utility, water, and local-government desk in Texas.

The July 17 memo should not be allowed to disappear into agency language. It should say what regulators can do now, what they cannot do without legislation, how interconnection costs will be assigned, how residential bill impacts will be measured, and what information data centers must disclose before connecting to the grid. The July 27 tax-exemption hearing should not become a generic tech-growth pep rally. It should put a dollar figure next to each claimed benefit and ask whether the exemption still makes sense at AI scale.

This is the moment to convert political attention into binding rules. Otherwise, the same public officials who say data centers should pay their own way in June can quietly approve exceptions by winter.

Reader Safety And Source-Status Note

This article is an infrastructure-accountability receipt, not an anti-compute ban argument. It supports American AI and cloud buildout while requiring public ledgers for grid costs, water use, incentives, and neighborhood impacts.

Source Trail

• Business Insider: Greg Abbott wants data centers to stop passing the bill to Texans (June 11, 2026) – Current report on Abbott proposal, pay-your-own infrastructure, water-efficient technology, and tax-exemption repeal.
• Chron: Abbott tells Texas regulators to make data centers pay their own way (June 10, 2026) – Local/regional report on PUC/ERCOT directives, July 17 memo deadline, water reporting, grid capacity, setbacks, and noise mitigation.
• Axios San Antonio: Gov. Greg Abbott calls for broad data center rules (June 10, 2026) – Texas data-center scale, incentive-cost numbers, and July 27 Senate Finance hearing context.
• Houston Chronicle: Abbott tells PUC, ERCOT not to pass new data center costs to customers (June 10, 2026) – Additional Texas ratepayer and data-center regulation context.
• Guardian: Majority of new U.S. AI datacenters to be built on drought-hit land (June 8, 2026) – National water-stress and AI data-center cooling context used to test Texas water-ledger questions.