Texas Has A Hidden Resource That Could Power The AI Boom

There is a peculiar kind of blindness that afflicts policy conversations about infrastructure. We look at a problem and a resource that sit side by side, each one the obvious answer to the other, and somehow we manage to see neither. Texas has this problem right now, and the cost of not correcting it will be measured in billions of dollars of investment that flow to other states, other countries, and other futures that are not ours to claim.

The problem is water. The resource is also water. And the opportunity sitting between them, if Texas acts with the clarity and speed the moment demands, is nothing less than becoming the dominant geography for artificial intelligence infrastructure in the world.

Let us be precise about what we are talking about. Every day, oil and gas production in Texas generates what the industry calls produced water, the salty, hydrocarbon-tinged brine that comes up with petroleum from underground formations. It is not clean water. It is not water anyone drinks. It is, in the language of oilfield operations, a waste stream, something to be managed, moved, and injected back underground at the lowest possible cost. The numbers involved are almost incomprehensible. Conservative estimates place Texas produced water volumes at roughly 22 million barrels per day, a figure that translates to approximately 932 million gallons per day. Some analyses put the statewide figure closer to 33 million barrels per day, or about 1.4 billion gallons per day. These are not projections or aspirations. They are descriptions of what is already happening, every day, across the Texas oilfield.

Now consider the other side of the ledger. While AI data centers can be designed to use very little water, availability of abundant, cheap water can greatly reduce the amount of electricity they require. A large hyperscale facility that uses water for cooling, the kind that houses the compute clusters powering the next generation of AI models, can consume anywhere from 1 million to 5 million gallons of water per day, with some major projects allocated as much as 8 million gallons per day under peak conditions. That water is used primarily for cooling, to manage the extraordinary heat generated by racks of GPUs and CPUs running at full load around the clock. This is not a design flaw. It is physics. Heat must go somewhere, and water is the most efficient and least expensive medium for moving it.

The arithmetic here is not complicated. If Texas is generating somewhere between 932 million and 1.4 billion gallons of produced water per day, and a hyperscale data center requires roughly 2 million gallons per day, then the theoretical gross supply of produced water could, in volumetric terms alone, support hundreds of such facilities. The Texas Produced Water Consortium has estimated that 2 to 4 billion barrels per year, roughly 230 to 460 million gallons per day, could realistically be treated for beneficial reuse outside the oilfield as costs and regulatory frameworks mature. Even that more conservative, realistic subset of produced water could supply on the order of 115 to 230 hyperscale facilities, a number that would make Texas the single most important concentration of AI compute infrastructure in the world.

The question is not whether the water exists. It does. The question is whether Texas has the imagination, the regulatory architecture, and the infrastructure to convert that waste stream into a strategic asset before the opportunity passes.

To understand why this matters so much, we need to understand the water problem that data centers are already facing elsewhere. In Northern Virginia, the country’s largest existing data center market, infrastructure upgrades are increasingly contested and expensive, and the density of facilities is straining local utilities and community patience. In Arizona, Nevada, and Utah, AI infrastructure investment has run directly into groundwater constraints and the political backlash that follows when residents of already-dry communities watch industrial cooling towers consume millions of gallons of water that could otherwise serve agricultural or municipal needs. In Georgia and the Pacific Northwest, similar tensions have emerged in varying forms. Water is becoming a chokepoint for AI investment, and the companies building these facilities know it.

Texas has power. Texas has land. Texas has a business climate that is, by any honest measure, among the most hospitable to large-scale industrial investment anywhere in the country. If Texas can also credibly offer a water supply for data center cooling that does not depend on municipal systems, stressed aquifers, or slow multi-jurisdictional permitting, it changes the calculus for every major technology company making trillion-dollar infrastructure decisions over the next decade. The OpenAI “Stargate” project in Abilene, described as a roughly 900-megawatt facility, is an early signal of what this kind of investment looks like. It will not be the last such project if Texas moves correctly.

What would moving correctly look like? It requires thinking clearly about three interlocking components: infrastructure, economics, and regulation.

Start with infrastructure. The good news is that the foundation already exists. West Texas already has a water midstream industry, companies that build pipelines, storage, and recycling systems to handle produced water at scale, analogous to the oil and gas midstream sector but for water. Aris Water Solutions, for example, operates roughly 790 miles of pipeline and handles approximately 1.8 million barrels per day of produced water, with recycling capacity of about 1.4 million barrels per day. That is not a speculative vision. That is existing infrastructure, already built, already operating. Other major operators maintain similarly large pipeline footprints. The question is not whether the water can be moved. It already is being moved, in enormous quantities, every day. The question is whether it can be moved in a direction that is commercially and logistically connected to data center campuses rather than exclusively to disposal wells.

The natural architecture for a produced-water-to-data-center supply chain involves several stages. Produced water is gathered from well pads through existing or new gathering pipelines and moved to regional treatment hubs. At those hubs, it undergoes pretreatment to remove oil, solids, and hydrocarbons, followed by additional treatment, potentially including reverse osmosis or thermal desalination processes, depending on salinity and target water quality. The treated water is then stored and blended to stabilize quality, moved through dedicated transmission pipelines to data center campuses, and subjected to final polishing at a facility-level water plant before entering the cooling system. Blowdown and concentrate from the cooling cycle are managed through a disposal or reuse loop. This is not a novel or exotic supply chain. It is substantially similar to how reclaimed water is used for industrial purposes in various municipalities already, adapted to the realities of produced water chemistry.

The geography of this opportunity is concentrated in West Texas, and that concentration is actually an advantage. Because produced water volumes are overwhelmingly dominated by the Permian Basin, with estimates ranging from 12 million to 24 million barrels per day in that region alone, it is possible to think about large centralized treatment capacity near the fields, paired with dedicated pipelines to nearby industrial loads. Data center campuses co-located with West Texas energy assets, including both large-scale natural gas generation and the region’s substantial wind and solar capacity, could draw from both the power grid and produced-water supply chains in a way that no other geography can match. The sweet spot, economically speaking, is shorter-haul campuses in West Texas that can be underwritten by long-term water offtake agreements tied to large compute loads.

That brings us to economics, which is where honest analysis requires acknowledging the current constraints. Produced water, as it stands, is predominantly managed through injection into disposal wells, and the economics of disposal, while rising, remain cheaper than industrial-grade treatment. Deep disposal has historically cost somewhere in the range of $0.60 to $0.70 per barrel in some Permian contexts, while treatment sufficient for industrial cooling can cost multiple dollars per barrel, with the Texas Produced Water Consortium citing an average treatment cost of roughly $2.55 per barrel to reach a specified salinity target, and wider industry analyses citing ranges from approximately $2.25 to $10 per barrel depending on water quality and end use.

A reasonable interlocutor might pause here and ask: if disposal is cheaper than treatment, why would producers and midstream companies change their behavior? This is the right question, and the answer involves several converging pressures that are already shifting the economics of produced water in Texas.

Disposal is not as cheap or as stable as it used to be. Injection volumes are increasingly linked to induced seismicity, and regulators in Texas and at the federal level are imposing restrictions on some disposal wells in seismically sensitive areas. The EPA has initiated review of Texas oversight of Class II injection wells, creating regulatory uncertainty around the status quo. A 2025 analysis suggested that new requirements could increase producers’ water-handling costs by 20 to 30%, with disposal potentially rising toward $0.75 to $1 per barrel in some contexts. When disposal costs rise and treatment costs fall, the crossover point where industrial reuse becomes economically competitive moves closer.

Furthermore, the economics look different when a large, creditworthy anchor tenant, a hyperscale data center operator with a long-term water offtake contract, is available to help finance the capital-intensive treatment and pipeline infrastructure. Long-term contracts can justify the upfront investment in treatment hubs in ways that speculative or spot-market water sales cannot. This is precisely how energy infrastructure has been financed in Texas for decades: long-term power purchase agreements enable generation investment; the same logic applies to water. If a technology company signs a 15-year agreement to purchase treated produced water for data center cooling at a negotiated price, that agreement becomes the collateral and the justification for the treatment hub. The hub finances itself over time, with the anchor tenant’s payments covering debt service.

The four-sided market that emerges from this structure is worth understanding carefully. Oil and gas producers benefit when produced-water reuse reduces their dependence on disposal wells, decreasing exposure to regulatory and seismicity risk while providing an alternative outlet for volumes that would otherwise require costly logistics. Water midstream companies benefit by developing a new offtake market tied to compute demand rather than drilling activity, diversifying their revenue base and enabling larger and more efficient treatment facilities. Data center developers benefit from access to a large non-municipal water source that reduces permitting friction and community opposition in water-stressed regions. And landowners benefit from the easements, corridor leases, and associated fees that pipeline corridors generate, much as energy midstream investment has historically benefited rural Texas landowners. The interests of all four parties point in the same direction.

The regulatory piece is where Texas has to act with specificity and speed. The policy apparatus around produced-water beneficial reuse already exists in the form of the Texas Produced Water Consortium and sustained legislative attention to the question. Texas legislation has contemplated the Texas Commission on Environmental Quality developing effluent standards and rules for produced-water treatment and beneficial use. The direction of policy is clear. What is needed is the translation of that direction into specific, workable pathways for industrial cooling reuse.

Three regulatory steps matter most. First, Texas needs clear “fit-for-purpose” industrial standards for cooling-water reuse. Cooling water is not drinking water. The relevant standards involve controlling fouling, scaling, corrosion, and biological growth, not meeting potable water quality benchmarks. A regulatory framework that holds produced water to potable standards for data center cooling reuse is unnecessarily burdensome and economically prohibitive. A framework that defines specific parameters for industrial cooling acceptability, covering key measures of dissolved solids, oil and organics content, specific metals, and biological load, gives both water suppliers and data center operators a clear target to engineer toward and a permitting pathway that is actually achievable at commercially viable cost. Second, Texas needs treatment certification and monitoring requirements robust enough to generate public confidence without being so prescriptive that they freeze innovation. The goal is a regime that ensures water quality and provides accountability, not one that substitutes regulatory compliance theater for actual performance. Third, Texas needs defined pathways for concentrate and blowdown disposal, the management of the more concentrated waste streams that result from the treatment process. A reuse system that simply shifts disposal risk from one point to another does not solve the underlying problem. Integrating concentrate management into the regulatory framework from the outset prevents that outcome.

There is a broader argument here that deserves to be made directly. The AI infrastructure buildout now underway is among the largest voluntary concentrations of private capital investment in American history. The decisions being made right now, about where to build data centers, where to connect them to power, and where to source their cooling water, will shape economic geography in Texas and across the country for generations. These are not ordinary infrastructure decisions. They are, in a meaningful sense, decisions about which places will be at the center of the economy that is coming and which places will be at its periphery.

Texas has an extraordinary natural endowment for this moment: land, power, business climate, and now, if we choose to see it clearly, water. Produced water is not glamorous. It does not appear in brochures or economic development pitches. It smells like the oilfield and it travels through pipes that most people never think about. But in volumetric terms, it is already the largest industrial water stream in Texas, and it is pointed in the wrong direction. It goes down, into disposal wells, where it presses against fault lines and costs producers money. It could go sideways, to cooling towers, where it dissipates heat, earns a return, and anchors the compute infrastructure of the 21st century.

Critics will raise objections, and they deserve to be taken seriously. Some will argue that Permian produced water is simply too salty, with total dissolved solids reaching 100,000 milligrams per liter or more in some formations, to be economically treated for industrial cooling at scale. This is a real technical constraint. Reverse osmosis, the most cost-effective desalination technology, becomes substantially more expensive and energy-intensive at higher salinities, and thermal desalination approaches have their own cost and energy challenges. The answer is not to pretend the constraint does not exist but to recognize that the economics are moving, that hybrid membrane-and-thermal treatment trains are advancing, that data centers with well-designed hybrid cooling systems can reduce the water quality demands on their supply, and that the long-term contract structures that would finance this infrastructure can be written to reflect actual achievable water quality rather than idealized specifications. Not every barrel of Permian produced water will be treatable at economically viable cost for every data center configuration. But a substantial fraction will be.

Others will worry about the regulatory complexity of permitting produced-water treatment and transport for a new end use. This concern is legitimate. Permitting is slow, and slow permitting has killed good infrastructure ideas before. The response is not to minimize the regulatory complexity but to argue for regulatory design that creates a specific, expedited lane for produced-water-to-industrial-cooling reuse, with defined standards and timelines, rather than forcing every project to navigate an uncertain general permitting process. Texas has demonstrated the capacity for regulatory innovation in energy. There is no reason it cannot do so in water.

The strategic vision for Texas is not to replace municipal water with produced water everywhere. It is more targeted than that: create a permitting and infrastructure lane for industrial cooling reuse using produced water in the specific regions, primarily West Texas and the Permian Basin, where produced water is most abundant and municipal water is most constrained. Co-locate AI and data center campuses with West Texas energy assets, taking advantage of the region’s grid-scale renewable capacity and natural gas generation, and the industrial land footprint that oil and gas development has already carved out. Use long-term compute offtake and water offtake contracts to finance the treatment hubs and pipeline corridors that make all of it work, converting what is now a cost center and a regulatory liability into a partially monetizable utility input.

The alternative is to watch other geographies figure this out first. It is not a comfortable thought. But it is an honest one. The water is already there, in quantities that would astound anyone who has not looked at the numbers. The question is only whether Texas will act with sufficient clarity and urgency to turn it into something. The answer should not require much deliberation.

If you enjoy my work, please subscribe: https://x.com/amuse.

Sponsored by the John Milton Freedom Foundation, a nonprofit dedicated to helping independent journalists overcome formidable challenges in today’s media landscape and bring crucial stories to you.

READ NEXT: Elite Response Force Activated By Trump In Unusual Deployment