The rapid expansion of AI is driving a surge in data center buildout across the US that is prompting a heated debate about the technology’s environmental impact. Even as communities across the US are embroiled in disputes about new data center construction, elected officials and industry leaders are increasingly focused on the resources needed to keep these systems operating.
Data centers have long supported the internet’s plumbing, but the new generation designed to support AI workloads operates at a larger scale. Training and operating advanced models requires clusters of powerful servers that generate vast amounts of heat and need extensive cooling systems.
A typical mid-sized data center can consume roughly 300,000 to 500,000 gallons of water each day, while some of the largest AI-focused sites may require millions of gallons daily during peak operations. Much of that water evaporates during the cooling process.
Electricity demand is similarly voracious. The International Energy Agency estimated that data centers accounted for roughly four percent of US electricity consumption in 2024. AI applications are forecast to dramatically increase that share as computing requirements grow. Producing responses through gen AI systems require far more power than traditional search queries.
Tech and Public Policy Collide
The tech industry acknowledges the scale of these demands but argues that the issue is more nuanced than critics suggest. As reported in Techstrong.ai, OpenAI chief executive Sam Altman pushed back on claims that AI systems consume extraordinary amounts of water, saying they have “no connection to reality.”
At the same time (and more rooted in fact), tech firms are exploring alternative power sources for the massive AI facilities. While the Trump administration has emphasized fossil fuel development and rolled back some renewable policies, some tech companies have expanded investments in renewable energy projects, including solar power and battery storage. Microsoft, AWS, and Google have move to deploy renewable energy. However, many data centers still rely heavily on conventional electricity grids that are largely powered by fossil fuels.
The environmental questions surrounding AI infrastructure are entering the political arena. During a recent campaign appearance in Florida, gubernatorial candidate James Fishback criticized data center development, claiming the facilities sharply increase electricity prices and consume large volumes of water. Independent analysis found that while data centers can influence electricity rates, they do not in fact raise household power bills by as much as 30 or 40 percent.
Regulators in several states are now considering policies to address these concerns. Florida lawmakers recently advanced legislation aimed at monitoring how new data centers affect local utilities and water systems. Similar discussions are happening elsewhere as communities evaluate proposals for large AI facilities.
Debate Continues
Meanwhile, the industry’s infrastructure buildout shows little sign of slowing. Companies including OpenAI, Microsoft and Oracle are planning enormous campuses designed specifically for AI workloads. One project proposed in New Mexico, known as Project Jupiter, could include on-site power plants fueled by natural gas to supply the electricity required by the data center complex.
In sum, the challenge is not the size of any single facility but the cumulative effect of dozens of them being built across the country. Forecasts suggest that the combined water consumption of U.S. data centers could equal the needs of millions of households within the next few years.
For policymakers, citizens and tech leaders, the struggle ahead is balancing the economic and technological promise of AI with the resources needed to sustain it. There will, to be sure, be winners and losers in that fight, and it is unlikely that deep-pocketed tech firms will be denied their new data centers.
In any case, as AI adoption keeps accelerating, this conflict between AI infrastructure and its environmental footprint is becoming one of the defining debates of our era.
