Researchers at the University of Texas at Austin have modeled the use of desalinated brackish groundwater and oilfield wastewater as cooling agents for AI data centers. The findings, published in the journal Water Research in September 2026, suggest that treating these water sources requires only a small fraction of a facility's total energy budget.
The study introduces a new metric, Saline Water Utilization Intensity, to assess local water stress. Modeling indicates that a proposed project in Longview, utilizing brackish water and a two-stage reverse osmosis system, would be the most efficient configuration.
The Permian Basin generates approximately 8 billion gallons of produced water daily, while a large data center can consume up to 5 million gallons of water per day for cooling. This research arrives as Texas Governor Greg Abbott paused new data center permits in September 2026 amid state investigations into water usage.
While the study offers a potential path for developers to justify projects to regulators, the authors note that the work is currently a model rather than an operational installation. Practical challenges, such as the disposal of concentrated salt waste from desalination, remain to be addressed.
