Data centers are having a water moment. As AI drives rapid growth in computing infrastructure, data center water use is receiving unprecedented attention.
Headlines increasingly associate large water numbers to data centers, AI workloads and individual digital activities. Yet the seemingly simple question, “How much water does a data center use?” has no single answer.
Water requirements vary widely depending on cooling technology, climate, workload, water source and facility design.
Recent research from Lawrence Berkeley National Laboratory found that water consumption associated with individual data center workloads can vary by more than four orders of magnitude. The researchers concluded that there is no single recipe for minimizing data center water use because the outcome depends on a combination of site-specific and technology-specific factors.
Before comparing data center water numbers, it is worth understanding what is measured, where the water is used, and how the surrounding water infrastructure fits into the picture.
At its simplest, computing requires electricity, and nearly all that electrical energy ultimately becomes heat that must be removed. How that heat is removed varies significantly between facilities. Thousands of servers operating continuously generate enormous thermal loads, and today’s increasingly powerful AI hardware is pushing those loads higher.
Many facilities have traditionally relied on evaporative cooling because water is highly effective at rejecting heat. In a cooling tower, water absorbs heat and a portion of the water evaporates into the atmosphere. As evaporation occurs, minerals become increasingly concentrated in the remaining circulating water. A portion must therefore be removed as blowdown and replaced with fresh makeup water.
Others use combinations of evaporative cooling, hybrid systems, air-cooled chillers, direct air cooling and increasingly, direct-to-chip liquid cooling. But liquid cooling does not necessarily mean high or low water consumption. Liquid cooling describes how heat is transported away from the computing equipment. How that heat is ultimately rejected determines much of the facility’s direct cooling demand. Closed-loop liquid systems, for example, can continuously recirculate coolant rather than continually consuming new water.
Cooling architecture is therefore one of the first things to understand when evaluating a data center’s water footprint.
Water-use figures can describe very different things. A number may represent:
Some studies go further to include water used in semiconductor manufacturing and other parts of the technology supply chain.
These are all useful measurements, but they describe different boundaries. Comparing them without understanding what is included can produce very different conclusions.
One common data center metric is Water Usage Effectiveness or WUE, typically expressed as liters of water used per kilowatt-hour of IT energy. It is useful for comparing facility water efficiency, but even WUE does not describe the entire relationship between a data center and the local water system. Cooling decisions can involve trade-offs between water and energy, so neither should necessarily be evaluated in isolation. Therefore, it is worth understanding what is measured and how.
Water use also needs local context. A gallon consumed in a water-abundant region does not carry the same implications as a gallon consumed in a water-stressed watershed. Climate also affects cooling demand and the performance of different cooling technologies.
Timing matters too. Utilities must plan for peak demand, not just annual consumption. On hot days, cooling-water requirements can increase at the same time community water demand is rising. Recent research from University of California Riverside and Caltech highlights the potential infrastructure implications of these peak demands as data-center development accelerates.
For site planning, annual consumption is therefore only part of the picture.
Source availability, water quality, local conditions and peak capacity all matter.. This makes water not only a resource-consumption question, but also an infrastructure-capacity and reliability question.
For a new project, some of the most important questions may be:
Reliability, peak capacity, and redundancy can be just as important for facilities designed around continuous operation.
Understanding a data center’s water strategy means looking beyond the volume entering the facility and considering the entire water cycle, from source to discharge or reuse.
Water may come from a municipal potable supply, reclaimed municipal wastewater, groundwater, or surface water, depending on local availability, infrastructure, permitting, and project requirements. Each source has unique water-quality characteristics that can influence treatment requirements and overall system performance.
Groundwater may contain elevated hardness, iron, manganese, or dissolved minerals. Surface water quality can fluctuate seasonally, while reclaimed water may contain constituents that require additional treatment before reliable use in cooling systems. As a result, source-water chemistry plays an important role in determining treatment needs, operating efficiency, maintenance requirements, and long-term water-management strategies.
Once used within the facility, water quality remains a consideration on the discharge side of the system. In evaporative cooling systems, a portion of the water is lost through evaporation, while cooling-tower blowdown is used to control the buildup of dissolved solids in recirculating water. Depending on its composition, blowdown may require treatment before reuse, discharge, or acceptance by a municipal wastewater system. The U.S. Department of Energy has identified treatment and reuse of cooling-tower blowdown as one strategy for reducing freshwater demand.
One of the most important questions in data center water planning is not simply how much water is available, but what water quality is required for the intended application.
Potable water is not always necessary for every use. In some regions, reclaimed municipal wastewater can provide a reliable alternative for non-potable applications when water quality, infrastructure, economics, and long-term availability align. Matching the appropriate water source to the application can help reduce pressure on potable-water supplies while supporting facility operations.
Viewed through this lens, water planning becomes a broader infrastructure decision rather than solely a cooling-system consideration. For developers, engineers, utilities, and communities evaluating new projects, four questions provide a useful starting point:
Addressing these questions early can influence site selection, cooling strategy, treatment requirements, utility planning, discharge management, and opportunities for water reuse. By evaluating the full water cycle from source through treatment, use, and discharge, stakeholders can make more informed decisions that balance operational reliability, resource stewardship, and long-term growth.
The rapid expansion of AI infrastructure will undoubtedly put greater attention on data-center water use. But scale alone doesn’t tell us whether an individual data center’s water strategy is responsible, resilient or appropriate for its location.
That requires understanding the whole system:
Source → Treatment → Use → Cooling → Wastewater → Treatment → Reuse or Discharge
And increasingly, the decisions made at one point in that cycle affect everything downstream. A project may need to evaluate an alternative water source, condition water for its intended use, manage or treat wastewater streams, create opportunities for reuse, or work with a community whose existing wastewater infrastructure must accommodate new development.
There will not be one water strategy that works for every data center. The goal is to understand the complete water cycle early enough to make the right decisions for the project and the community around it.
At Axius Water, we bring the full water cycle perspective to the data-center water conversation. Our companies bring experience that addresses the needs arising throughout the water journey, from treatment and conditioning to wastewater management, filtration, polishing and opportunities for reuse. That breadth allows us to look at more than one point in the process. We work with developers, engineers, utilities, and communities to evaluate how decisions made upstream affect treatment requirements and options downstream.