The rapid growth of artificial intelligence (AI) data centers in West Texas and southeast New Mexico has the potential to significantly boost local economies in the region. It is also creating challenges. Data centers require significant power—facilities that draw more than 75 MW are categorized as “large loads” by the Electric Reliability Council of Texas (ERCOT). 75 MW of electricity is enough to power 15,000-20,000 homes. The total population of Pecos, Texas, for example, is only 10,000 people. With the addition of a significant number of massive AI data centers, and other large loads in the region—including those related to electrification of oil and gas operations and other industrial processes—Texas is bracing itself for a potential doubling of its power demand by 2030.  

In addition to their power demands, AI data centers are also straining water infrastructure. Due to the availability of land and local access to energy resources, many data centers are being sited in Texas and New Mexico in locations that face water scarcity. Depending on their size and cooling strategy, data centers can require 0.5 million to more than 5 million gallons of water per day. Although most AI data centers in West Texas and southeast New Mexico are now employing air-cooled strategies, these can still require significant quantities of water. The bottom line is that local water resources will be strained if not managed properly.  

The University of Texas at Austin (UT Austin) is conducting research to identify water sources in West Texas and southeast New Mexico that can support the development of large load AI data centers without increasing pressure on limited freshwater supplies. This work more broadly supports industrial water-supply planning more generally and identifies priorities for future sampling, modeling, and aquifer testing. The report by Lucy Ko and colleagues assesses the brackish Dockum Aquifer as a viable potential water source for data centers and other large industrial facilities in water-constrained areas of West Texas and eastern New Mexico. 

The region has long relied on the freshwater Ogallala Aquifer, where groundwater withdrawals have exceeded recharge for decades. At UT Austin, we have been working to identify unknown or poorly understood aquifer features (e.g., deep channels, collapsed zones, faults), improve resolution accuracy of aquifer maps and informing strategic groundwater development, particularly for brackish water supply projects, as well as assessing and piloting novel imaging techniques for scalable aquifer exploration.  

The deeper and more mineralized Dockum Aquifer is an underused brackish groundwater resource that may provide an alternative to Ogallala withdrawals.  

The Bureau of Economic Geology in the Jackson School of Geosciences, established in 1909 as the State Geological Survey of Texas and the oldest research unit at UT Austin, assembled a team of world-leading experts in hydrogeology, groundwater geochemistry, water-resource assessment, treatment considerations, and large-load infrastructure analysis to evaluate Dockum Aquifer resources for AI data centers. The team compiled data from 895 groundwater samples from across the Dockum and associated aquifer systems into a three-dimensional hydrogeochemical framework showing salinity classes and major-ion chemistry. The potential of the Dockum depends on local water chemistry, aquifer productivity, treatment requirements, management of treatment residuals, and responsible long-term withdrawal planning.  

Key Takeaways from the Report: