Data centers are thirsty, and in this basin their thirst tends to land on a river. The usual worry is that a new server farm will draw millions of gallons of clean water a day to keep its machines cool, water that would otherwise go to people, farms, and streams. A new study out of the University of Texas asks a different question: what if we cooled them with the dirty, salty water nobody else wants instead?
Researchers at the University of Texas at Austin modeled whether desalination, the same technology used to make seawater drinkable, could turn brackish groundwater and even oil and gas wastewater into cooling water for data centers, so they would not have to drink freshwater. Their answer, published in the journal Water Research, is that it can be done with equipment we already have, and that the energy it takes is only a fraction of what a data center already burns. There is a real cost to it, but not a technology barrier. For a region staring down both a data center boom and a century of fouled water, that is a question worth sitting with.
The problem: data centers drink freshwater
A large data center can use millions of gallons of water a day, most of it to carry away the heat its servers throw off. The catch is that the water has to be fairly clean to begin with, under about 500 parts per million of dissolved solids, or it corrodes and clogs the cooling equipment. That is why operators reach for the same freshwater everyone else depends on, and why a proposed data center on a river sets off alarms. The fight has also gone political, with public worry about data centers and their water use climbing right alongside the midterm season. Finding another source of cooling water would take a lot of air out of that fight.
What the study found
The team, led by Vishnu Sree Shanthanu Katakam, modeled the cost and energy tradeoffs of cleaning up a range of salty sources, from mildly brackish groundwater to much saltier oil and gas wastewater, using two proven methods. Reverse osmosis forces water through membranes that strain out salt and impurities. A heat-driven process boils the water off as vapor, leaves the junk behind, and condenses the clean steam. Running three hypothetical data centers through the model, they found that a project in Longview, Texas, drawing on brackish groundwater cleaned by reverse osmosis, put the least strain on its local water supply of the options they tested. And even the most energy-hungry version of desalination used only a sliver of the power the data center was already drawing.
“We’re not waiting for a futuristic technology to come along and solve this problem for us. We can solve parts of this problem with the technology we already have; there’s just a cost to it.”
Vaibhav Bahadur, mechanical engineering professor, University of Texas at AustinWhy it matters for the rest of us
The study is built on Texas oil and gas wastewater, but the idea behind it should make anyone in the Upper Ohio Basin sit up. We do not have a Permian Basin, but we do have something like it: a century of abandoned mine drainage, water laced with salts and metals that already has to be collected and treated before it fouls a stream. The study does not claim that a data center could run on treated mine water, and nobody has shown that it can. But it is exactly the right question to ask in a region that has both companies eyeing it for server farms and more impaired water than it knows what to do with. If a data center is going to land here anyway, the better fight may not be only whether it comes, but whether it is made to cool itself on water we would otherwise have to clean up regardless, instead of on the Mon, the Yough, or the Allegheny. The technology, it turns out, is not the thing standing in the way.
Based on reporting by Anthropocene Magazine, “Could dirty, salty wastewater quench data center thirst?”, October 15, 2026, drawing on a study by Vishnu Sree Shanthanu Katakam and colleagues in the journal Water Research (2026) and comments from University of Texas at Austin professor Vaibhav Bahadur. The abandoned mine drainage comparison is our own. Current as of October 21, 2026.
