Image Courtesy: AP
One of the largest data center projects in Georgia reportedly consumed nearly 30 million gallons of water without being properly monitored or billed, exposing major weaknesses in local infrastructure as US communities race to approve AI-related developments.
The facility, operated by Quality Technology Services in Fayette County, Georgia, used the water during construction while nearby residents were simultaneously being asked to conserve water because of drought conditions. Some locals also reported sudden drops in water pressure during the same period. County officials later discovered that two industrial-scale water connections tied to the project were not being correctly tracked. One hookup had reportedly been installed without the utility’s knowledge, while another was never linked to the company’s billing account, according to Ars Technica and reporting from Politico.
After the issue surfaced, QTS paid roughly $150,000 in retroactive charges. However, the county chose not to fine the company or penalize it for exceeding agreed peak water usage limits. Fayette County water officials described the situation as a “procedural mix-up” and partly blamed outdated monitoring systems and staffing shortages for the oversight.
The controversy has frustrated many residents, particularly because local authorities admitted they did not want to damage relations with QTS, one of the county’s largest customers. Fayette County Water System Director Vanessa Tigert told Politico that maintaining a good partnership with the company factored into the decision not to pursue stronger enforcement measures.
The incident is becoming a broader warning sign for cities and counties rapidly approving hyperscale data centers tied to the AI boom. Across the United States, many local water systems are aging, understaffed, or still transitioning from older infrastructure that struggles to detect leaks or sudden spikes in industrial usage.
While companies often highlight energy consumption linked to AI infrastructure, experts increasingly argue that water demand may become just as significant. Large data centers require massive cooling systems, while semiconductor manufacturing and electricity generation also consume huge amounts of water behind the scenes.
Research cited by industry groups suggests that AI-related water use could more than double over the next 25 years. The issue is especially sensitive because many data centers are being built in regions already facing water stress or recurring droughts.
QTS says its facility will eventually use a closed-loop cooling system designed to reduce long-term water consumption once construction ends. The company also stated that all water usage complied with existing regulations and denied claims that its operations caused countywide water pressure problems.
Still, public concern continues to grow in multiple states. Community groups in places like Arizona and Utah have increasingly pushed back against fast-tracked data center projects, arguing that officials are approving major developments before fully understanding the environmental impact.
The Georgia case is now being viewed as a clear example of what can happen when infrastructure oversight fails to keep pace with the rapid expansion of AI-related construction.