When data centers enter a community, water is often the first concern to surface. Residents worry about wells running dry. Local officials brace for strain on municipal systems. Developers, for their part, arrive with demand projections that can appear outsized at first glance.
Both sides have a point.
“On paper, these data centers are just banks of computers that generate a lot of heat,” said Shannon Markham, vice president and regional manager for water at AECOM, a global infrastructure company. “And water is often used as a cooling component, either directly or indirectly.”
That basic function is straightforward. What’s less clear—and often contested—is how much water is actually required, and under what conditions.
Developers typically present demand based on peak scenarios: the hottest day of the year, at full compute load, when cooling systems are under maximum stress. Those figures are designed to ensure system reliability. They are not meant to represent average daily use.
“They base their demand on the worst possible time,” Markham said. “What those don’t really reflect is reality on an average day basis.”
That distinction is important. But it does not resolve the issue.
Peak demand is not hypothetical. Systems must be built to accommodate it, and in regions where infrastructure is already constrained, those peak scenarios can drive real costs—whether they occur daily or only a handful of times each year.
For site selectors, the takeaway is not to dismiss peak numbers, but to interrogate them. How often will those conditions occur? What infrastructure is required to support them? And who ultimately pays for that capacity?
The answers vary widely by region.
In water-stressed parts of the western United States, large new users—data centers included—face increasing scrutiny. Allocation, long-term supply, and competing demands from agriculture and residential growth are already under pressure. In those markets, water can be a limiting factor in site selection.
In the Midwest, the picture is more mixed.
They base their demand on the worst possible time. What those don't really reflect is reality on an average day basis.
“Compared to the West Coast, where they get a fraction of the rain we do, we’re in a different position,” said John Newsome, administrator of water for the City of Columbus, Ohio. Many Midwestern systems were built to support heavy industrial use and now operate below capacity. That can create an opening for new demand.
“I do see an opportunity for some of the Rust Belt cities that have lost big manufacturing and population,” Newsome said. “They have excess water.”
But excess capacity on paper does not always translate to project readiness.
Water systems are localized. Treatment plants, distribution networks, and permitting frameworks all shape how—and whether—that capacity can be used. In some cases, infrastructure upgrades are required before a project can move forward. In others, competing growth—from residential development to other industrial users—can quickly absorb available supply.
Even in regions that are not water-constrained, water can still become a point of friction.
“Land use, energy demand, and water—they all come up,” Markham said, describing the sources of community opposition.
In practice, water concerns are rarely evaluated in isolation. They tend to be bundled into broader questions about how a project will affect a community—its resources, its costs, and its long-term trajectory.
“My energy rates are going to go up. My well is going to run dry,” said Trace Johnson, president of AI Al, describing how projects are often perceived at the local level. Those concerns are sometimes overstated. But they are not irrational.
Data centers are intensive users of both water and energy, and their demands are highly concentrated. Even if average consumption is lower than peak projections suggest, the infrastructure required to support that peak must still be financed, built, and maintained.
That is where the conversation is shifting.
I do see an opportunity for some of the Rust Belt cities that have lost big manufacturing and population. They have excess water.
Rather than focusing solely on total water availability, communities and developers are increasingly looking at how water is sourced, delivered, and reused. Recycled water systems—long used in parts of the West—are gaining attention as a way to reduce pressure on potable supplies.
“Recycled water lessens our demand on potable water,” Newsome said.
For site selectors, reuse infrastructure can signal a higher level of readiness. But it also introduces additional layers of complexity, from permitting to treatment standards to long-term system management.
The result is a more complicated equation than many initial site screens capture.
Water is not always the primary constraint. But it is increasingly part of the risk profile—one that intersects with infrastructure, policy, and community perception.
For companies evaluating sites, that means moving beyond binary questions of supply. It requires a closer look at system capacity, peak demand assumptions, infrastructure timelines, and the local political environment.
It also requires a more proactive approach to communication.
“I think we should make sure we’re having communications upfront,” Markham said.
That may be the most consistent lesson across markets.
Because in many cases, projects are not delayed or derailed by water itself, but by how water is understood—or misunderstood—by the stakeholders involved.
For site selectors, the implication is straightforward: water is neither a non-issue nor a universal barrier. It is a variable—one that must be evaluated with the same rigor as power, labor, and land.
