Texas Data Centers Claim Water Efficiency Gains as the State's Supply Crisis Deepens

Thomas Gleeson, chairman of the Texas Public Utility Commission, told a state House committee last week that regulators need a clear and accurate picture of how much water data centers actually consume. The timing of his testimony was not incidental. Texas is simultaneously grappling with decades of water mismanagement and a surging appetite for electricity driven largely by the explosive growth of data center construction across the state. That collision of pressures is forcing a reckoning that goes well beyond kilowatt-hours.

Data centers have long relied on evaporative cooling systems that consume enormous volumes of water, essentially trading one resource for another when they draw power from the grid. A single large hyperscale facility can use millions of gallons per day, comparable to the consumption of a small city. As developers race to plant roots in Texas, attracted by its deregulated energy market, available land, and relatively low costs, the question of what that means for aquifers and river basins is becoming impossible to defer.

Developers, for their part, are playing offense. Several have come forward with claims of significant reductions in water usage intensity, pointing to newer cooling architectures, closed-loop systems, and air-side economization as evidence that the industry has turned a corner. The argument is essentially that the data center of 2025 is a fundamentally different machine than the one built a decade ago, and that regulators and lawmakers should evaluate the sector on its current trajectory rather than its legacy footprint.

The problem is that "huge cuts" in water usage are difficult to verify without standardized, mandatory reporting. Gleeson's call for a clearer picture reflects exactly that gap. Texas does not currently have a unified framework requiring data centers to disclose water consumption in real time or at the facility level. Developers can make efficiency claims, and those claims may well be accurate, but without independent verification tied to actual operational data, regulators are essentially taking the industry at its word.

This is where systems thinking becomes essential. If Texas moves forward granting permitting approvals, tax incentives, or grid interconnection priority based on self-reported efficiency figures that later prove optimistic or misleading, the state could find itself locked into infrastructure commitments that strain water resources for decades. Large data centers are not temporary installations. They are 20 to 30 year bets on the landscape, and the feedback loop between underreported consumption and aquifer depletion runs slow enough that the damage may not become visible until it is very difficult to reverse.

Texas already has a complicated relationship with water scarcity. The Edwards Aquifer, which supplies San Antonio and surrounding regions, has been the subject of legal battles and regulatory fights for years. The Ogallala Aquifer, which underlies much of the Texas Panhandle, is being drawn down faster than it recharges. Adding large-scale industrial water consumers to a system already under stress is not a neutral act, regardless of how efficient those consumers claim to be.

The second-order consequence worth watching closely is what happens to municipal water systems in the communities that host these facilities. Data centers often negotiate directly with utilities or secure water rights independently, which can mean that local governments and residential users are competing for the same diminishing supply without full visibility into what the industrial sector is drawing. In drought years, which are becoming more frequent across Texas, that competition becomes acute.

There is also a political economy dimension that rarely gets discussed. Data center developers bring jobs, tax revenue, and the kind of economic development optics that elected officials find hard to resist. That creates an incentive structure in which local and state governments may be predisposed to accept efficiency claims without demanding the verification infrastructure that would make those claims meaningful. Gleeson's testimony suggests at least some awareness of this dynamic at the regulatory level, but awareness and action are different things.

The broader national context matters here too. As the AI buildout accelerates demand for compute infrastructure, Texas is not alone in facing this tension. Virginia, Georgia, and Arizona are all navigating versions of the same problem. But Texas, with its unique grid structure under ERCOT and its history of resource governance challenges, may be the place where the contradictions surface most visibly.

If the state moves quickly to establish mandatory, facility-level water reporting requirements with independent auditing, it could set a standard that the rest of the country follows. If it does not, the efficiency claims being made today will remain unverifiable, and the gap between what developers promise and what aquifers actually experience will quietly widen.