Artificial intelligence may live online, but the infrastructure powering it has a very real environmental footprint. As AI continues to expand across the United States, the massive data centers needed to run it are drawing increasing attention—not only for how much water they use, but also for what can happen to that water afterward.
A recent investigation by The Guardian highlighted wastewater concerns surrounding data centers across the country. One of the most recent cases occurred in Cheyenne, Wyoming, where testing of wastewater associated with the construction of a Meta data center detected a rare bacterium called Cupriavidus gilardii. Although the wastewater was not intended for drinking, local officials were concerned about possible exposure when the water was reused for irrigation and eventually stopped accepting it as a precaution. Meta has disputed the test results and said there was no impact on the public or environment.
The controversy points to a much larger issue. Data centers require enormous amounts of water, particularly for cooling the powerful computer systems operating inside them. According to The Guardian, a mid-sized data center can consume around 300,000 gallons of water each day. Once that water has been used, wastewater may be released for treatment or reuse. Depending on the facility and cooling system, environmental advocates are concerned that discharged water could contain substances such as heavy metals or PFAS, often called “forever chemicals.”
Temperature can also matter. Water used for cooling can leave a facility warmer than when it entered. In New York, for example, opponents of a facility drawing water from Seneca Lake have raised concerns about plans that could allow significantly heated water to be discharged back into the lake. Warmer water can stress temperature-sensitive species and potentially contribute to conditions that favor harmful algal blooms.
These concerns are not limited to one state. Wastewater, stormwater, wetland and water-quality disputes involving data centers have emerged in places including Virginia, West Virginia and New York as communities question how rapidly expanding facilities will affect their local water resources.
The debate highlights an often-overlooked side of the AI boom. Every search, generated image and AI response ultimately depends on physical infrastructure somewhere. As demand for artificial intelligence grows, so will the challenge of powering and cooling that infrastructure responsibly.
Technology and environmental protection do not necessarily have to compete. But as data centers expand, communities will need greater transparency about how much water facilities use, what is discharged afterward and how nearby ecosystems are protected.
The question is no longer simply how far AI can advance. It is how we can make sure that progress does not come at the expense of the water systems we all depend on.
Diya Comar