Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Data centers use water mainly to remove heat from servers, but the amount varies dramatically by cooling design, climate, workload, water source and reporting boundary. Evaporative systems consume water as it changes to vapor; air-cooled and closed-loop systems can require little or no ongoing cooling water. A facility’s “data center” label therefore says almost nothing about its actual local demand.
To evaluate a proposed or operating site, ask for measured withdrawals and consumption, peak and seasonal demand, water sources, the cooling system in operation, its WUE (water usage effectiveness), discharge details and the records supporting each figure. WUE is useful for comparing intensity, but it cannot replace absolute volumes or local water-supply analysis.
What water does in a data center
Servers turn electricity into computation and heat. Cooling equipment transfers that heat from the server room to a heat-rejection system, which releases it outdoors. Water may be involved in one or more parts of that chain:
- Heat transfer inside the facility: Chilled-water loops or direct-to-chip liquid systems carry heat away from processors and other equipment.
- Heat rejection outdoors: Cooling towers and evaporative condensers release heat by evaporating some of the water.
- Humidity control: In dry climates, water can be added to supply air so static electricity does not damage equipment.
- Maintenance and startup: Piping, tanks and treatment systems may need initial fills, flushing and periodic top-offs.
Water that evaporates is generally considered consumed at the site because it is no longer available for immediate return to the supplying watershed. Water withdrawn, used and then discharged is not the same as water consumed, so residents should request both numbers.
Electricity also has an indirect water footprint. Power stations can withdraw or consume water, depending on their generation technology and location. A facility that reports only water entering its property is not reporting the water associated with producing its electricity.
Cooling designs and their water implications
Evaporative cooling
Cooling towers and evaporative condensers spray or distribute warm water across a surface while fans move air through it. A portion of the water evaporates, carrying heat away. The remaining water becomes concentrated with minerals and is periodically drained as “blowdown” and replaced with make-up water. Hot, dry weather usually increases evaporation, so annual averages can hide high summer or peak-day demand.
Evaporative designs can deliver efficient cooling with comparatively low electricity use, but their consumptive water demand is ongoing. The source may be treated municipal water, reclaimed wastewater, surface water or groundwater; the source and its local scarcity matter as much as the volume.
Air-side or dry cooling
Air-cooled chillers and dry heat exchangers reject heat directly to outdoor air. They can reduce operational cooling-water consumption to near zero, although a site may still use water for humidification, domestic needs, fire protection, cleaning or occasional maintenance. Dry systems often require more electricity during hot periods, when fans and compressors work harder.
Closed-loop liquid cooling
A closed loop circulates the same treated fluid through equipment and heat exchangers. Once filled, it normally needs only leak repairs, treatment and small maintenance additions. The heat still has to leave the site, so a closed loop may be paired with an air-cooled or evaporative heat-rejection system. “Closed loop” therefore does not automatically mean zero water use; ask which part of the system is closed and how heat is rejected.
Project-specific statements illustrate why design details matter. For example, Oracle’s Project Jupiter explanation describes its own planned water approach and figures. Those are company statements for that project, not an independently verified industry benchmark. The same distinction applies to any operator’s marketing material.
How much water do data centers actually use?
Two measurements answer different questions:
- Withdrawal: the volume taken from a utility, river, aquifer or other source.
- Consumption: the portion not returned to the source, commonly including evaporation and water incorporated into products or waste.
Ask for annual totals and peak-day values for both. A facility can have large withdrawals but return much of the water after treatment, or modest withdrawals with a high share consumed through evaporation.
Recommended Free Tools
WUE: a useful intensity metric
Water usage effectiveness (WUE) is typically expressed as liters of cooling and humidification water per kilowatt-hour of IT energy. Lower WUE means less water per unit of computing energy, but it does not reveal the facility’s total size, local scarcity, seasonal peaks, source, discharge or off-site electricity water.
| Published figure | What it covers | How to interpret it |
|---|---|---|
| 0.58 m³/MWh (equivalent to 0.58 L/kWh) | European Commission communication for 2024; weighted average of 458 EU facilities with reliable nonzero-water reports | The population excludes facilities reporting zero consumption, including dry-cooling sites. It is an EU reporting-period average, not a universal facility benchmark. European Commission source |
| 0.27 L/kWh | Microsoft FY25 global WUE for data centers fully owned and controlled by Microsoft and operational for 12 months at calculation time; FY25 ran July 1, 2024–June 30, 2025 | A company-reported value with a defined corporate scope and methodology, not a figure for every Microsoft site or the industry. Microsoft methodology |
To convert a WUE value into an approximate annual cooling-water volume, multiply liters per kilowatt-hour by the site’s annual IT energy in kilowatt-hours. Confirm that the operator’s denominator includes the same buildings, backup systems and reporting period you are evaluating. A low-intensity site can still consume more water in total if its computing load is very large.
Why a low WUE can still matter locally
WUE does not tell residents whether water comes from a stressed aquifer, a drought-limited reservoir or a plentiful reclaimed-water system. It also omits several questions that determine local impact:
Rank #3
- Is the number an estimate, a meter reading or an independently audited result?
- Does it include only cooling towers, or also humidification, initial fills, maintenance top-offs, backup generation and construction?
- Are contractor-operated buildings and leased capacity inside the boundary?
- Is the figure annual only, or are monthly and peak-day values available?
- What volume is discharged, at what temperature and quality, and under which permit?
- Does the operator report water used to generate purchased electricity?
The UC Berkeley Center for Law, Energy & the Environment’s February 2026 California report emphasizes that site-level data are incomplete and that transparent, location-specific reporting is needed. It does not provide a single number that can be used as a universal facility estimate.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
What determines a site’s water demand
Climate and weather
Evaporative systems generally consume more water in hot, dry conditions. A design that looks modest on an annual average can produce substantial summer peaks. Ask for a month-by-month profile and the maximum day or week, not just an annual total.
IT load and operating schedule
Water demand follows heat production. A partially occupied building may use far less than its permitted capacity; later expansion, artificial-intelligence workloads or higher server utilization can change the profile. Separate current operations from build-out projections.
Water quality and treatment
Hard or mineral-rich water may require more blowdown. Treatment chemicals, filtration and discharge limits can affect how much make-up water is needed. A proposal should identify treatment equipment and the assumed cycles of concentration.
Source and local scarcity
One cubic meter of reclaimed water is not equivalent, in community impact, to one cubic meter of potable water from a drought-stressed basin. Identify every source—potable, reclaimed, non-potable, surface water or groundwater—and the supply agreement or water right supporting it.
Reporting boundary
Operators may report a single building, a campus, a legal entity or only facilities they own and control. Compare like with like, and ask whether figures include leased halls, temporary generators, construction use and off-site power production.
A resident’s step-by-step records check
- Locate the project file. Search the planning application, environmental review, utility-capacity documents, water-rights records and drought or watershed plans for the site address and expansion phases.
- Identify the design. Find the proposed and currently operating cooling system. Check whether the design or power supply changed after the original approval.
- Request volumes in consistent units. Ask for estimated and actual annual withdrawals and consumption in gallons or cubic meters, plus peak-day and seasonal requirements. Require the reporting period and meter or modeling method.
- Trace every water input. Determine whether each stream is potable, reclaimed, non-potable or groundwater, who supplies it and whether initial fills, humidification, cooling towers, backup power and maintenance are included.
- Test the WUE calculation. Ask for the exact definition, IT-energy denominator, time period, included facilities and treatment of zero-water or dry-cooled buildings.
- Follow the return flow. Obtain discharge volume, temperature, chemical limits, receiving body, treatment process and permit. Confirm whether any water is trucked away or discharged through a third party.
- Compare demand with competing uses. Put the project’s seasonal requirement beside household, agricultural and industrial demand, drought restrictions, available supply and other approved developments.
Questions that expose weak or incomplete claims
- “How much water will this site withdraw and consume in a normal year, a design year and the highest-demand day?”
- “Which numbers are measured actuals, which are engineering estimates, and who verified them?”
- “What happens to demand during a heat wave, drought restriction or full build-out?”
- “Which water rights, utility contracts or replacement obligations authorize the supply?”
- “What is the water balance for evaporation, blowdown, leaks, discharge and reuse?”
- “What electricity use does the water-saving design require, and what generation mix supplies that power?”
- “Will reports be published regularly, and are confidentiality claims masking the facility total or only security-sensitive details?”
How to compare two designs or projects
Use the same axes for every proposal. A comparison that lists only WUE can favor a large site in a water-stressed area over a smaller site using reclaimed water.
| Comparison axis | What to collect | Why it matters |
|---|---|---|
| Absolute demand | Annual withdrawal and consumption, peak day, monthly pattern and full-build-out forecast | Shows infrastructure and watershed pressure that an intensity metric can hide |
| Source and scarcity | Potable, reclaimed, non-potable or groundwater source; basin condition and supply rights | Same volume can have very different community consequences |
| Cooling boundary | Evaporative, air-side or closed-loop equipment; included buildings and systems | Prevents misleading comparisons between unlike facilities |
| WUE method | Value, IT-energy denominator, period and inclusion rules | Reveals whether reported intensities are comparable |
| Energy trade-off | Facility electricity use and generation-related water | Water savings at the site may increase energy and off-site water use |
| Discharge | Volume, temperature, chemistry, treatment and permit | Returned water can still affect a stream, sewer system or treatment plant |
| Evidence quality | Metered actuals, modeled estimates, audit status and public reporting schedule | Separates accountable commitments from projections |
Utah’s disclosure rule as a current example
There is no single nationwide data-center water-disclosure system. Utah provides one current state example. Under rules effective July 1, 2026, a qualifying new project is generally larger than 10,000 square feet and intends to draw at least 75 acre-feet of water annually. The Utah Division of Water Rights says preconstruction reports include the location, estimated withdrawals, applicable discharge plans and temperature adjustments, and water-reuse or replacement efforts. Annual reports include actual withdrawals and related efforts, and public annual reports are to be posted by September 1, subject to stated confidentiality exceptions.
The Utah law excludes large centers already operating or under construction before its effective date. A resident must therefore check both the project’s status and the records required in that jurisdiction; Utah’s approach should not be treated as a national rule.
Water, energy and siting trade-offs
Reducing on-site water use can increase electricity use. Dry cooling may require more fan and compressor power, especially during hot weather. Conversely, evaporative cooling can lower electricity demand while consuming more local water. The better choice depends on basin scarcity, grid mix, climate, wastewater availability, carbon goals and the scale of the facility.
The California analysis from UC Berkeley stresses that water-efficient cooling strategies often use more energy, and that the trade-off varies by scale and location. Ask the operator to present both water and energy consequences rather than claiming that one metric proves overall sustainability.
Discharge, temperature and water quality
Cooling-tower blowdown contains concentrated minerals and treatment chemicals. A discharge may enter a sanitary sewer, an industrial treatment system or a surface-water body, each with different limits. Temperature can matter even when chemical concentrations comply: warm water can alter receiving-water conditions. Request the permit, maximum temperature adjustment, pollutant limits, monitoring frequency and corrective-action plan.
If the facility sends wastewater off site, identify the receiving treatment plant and whether its capacity and water-quality permits account for the additional load. “Reused” water should be traced through the entire balance so that a reduction in fresh withdrawals is not mistaken for zero environmental impact.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Scan for outdated or missing drivers - takes under a minute3Repair Windows errors before they cause bigger problemsCommon mistakes when reading a proposal
- Treating a sector average as a site estimate: EU or corporate averages combine different climates, technologies and reporting boundaries.
- Confusing withdrawal with consumption: A returned discharge is not the same as evaporated water, and neither number alone describes local stress.
- Ignoring zero-water facilities in an average: The European Commission’s 0.58 m³/MWh figure covers 458 reliable nonzero-water reports and excludes facilities reporting zero consumption.
- Accepting “closed loop” without asking about heat rejection: The internal loop may still feed an evaporative tower or other water-using equipment.
- Using annual totals to dismiss peaks: Utilities and watersheds must meet maximum-day and dry-season demand.
- Assuming a published WUE is audited: Confirm whether meters, estimates or an independent assurance process produced the number.
- Overlooking project changes: A revised server density, cooling plant or power design can invalidate the original water estimate.
A defensible answer to “How much water will this data center use?”
The answer should name the facility, reporting boundary, cooling design, water sources, time period and evidence quality. It should give withdrawal, consumption and peak-season figures, then explain discharge and the electricity-related water footprint. If any element is unavailable, say exactly which one and why; do not substitute an industry average.
Residents who assemble those records can distinguish a genuinely low-water design from a low WUE number, identify who bears the supply risk and test whether projected demand fits local water plans. That is the information needed to judge a data center on its actual effects rather than its label.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
