数据中心水利用：从效率指标转向现实场景下的水韧性

water tower at dusk

Data centers’ license to operate increasingly depends on context-based water resilience.Getty Images

Data centers are facing a new reality: water is no longer a background utility but a core constraint that shapes siting, design, operations, and community relations. The conversation is moving beyond average efficiency metrics to show that facilities can operate under peak stress conditions without exacerbating local water stress or pollution. This roundup of Data Center Knowledge articles connects technical, economic, and policy dimensions of data center water use and highlights where the industry is headed next.

Metrics and Measurements

WUE 101 – and Its Limits

Water Use Effectiveness (WUE) measures a data center’s total water consumed per kilowatt-hour of IT energy (liters/kWh). Tracking WUE supports sustainability goals, cost control, cooling optimization, and transparent community reporting. However, WUE can be misleading if taken alone because it doesn’t reflect water sourcing or consistently measured boundaries, and it’s a relative metric (not total water use). Ambient conditions and metering practices can also affect comparability across sites.

Beyond WUE: Measuring Resilience Where It Matters

Water resilience for data centers can’t be judged by WUE or “waterless” claims alone; it must be proven under peak stress, according to water treatment specialist Dr. Marcello Serrano. Using California’s Imperial Valley dispute over a proposed AI data center as an example, Serrano argues that credible stewardship hinges on site-specific, transparent assessments that address five areas:

Peak operations under extreme heat
Water rights and curtailment risks during shortages
Reclaimed water availability and quality onsite
Discharge constraints (sewer capacity, permits, and thermal/chemical limits)
Water-energy tradeoffs vs. grid and environmental conditions

Applied early in site selection and operations, this framework helps operators and utilities plan, align on capacity and discharge requirements, and avoid escalating resource conflicts into litigation.

Cooling Technologies

Why Cooling Water Is Hard to Reuse

Since most data centers rely on direct evaporative cooling, most intake water exits as vapor, while the mineral-concentrated remainder is periodically flushed as blowdown, which typically requires treatment before any onsite reuse. While facilities can cut potable demand by implementing closed-loop or hybrid heat-rejection systems, treating blowdown, or repurposing residual water as graywater, these approaches add costs, complexity, footprint, and often higher electricity use in many climates. As of now, economics in many markets still favor conventional evaporative designs, though rising water prices and evolving regulations could change that.

Technology Trajectory: Evaporative Today, Liquid Tomorrow

Evaporative cooling remains prevalent because it is inexpensive to install and operate, scales effectively to very large facilities, and uses less electricity than mechanical refrigeration. However, it consumes large amounts of water and can strain local supplies. Alternatives such as liquid cooling (direct-to-chip and immersion), mechanical refrigeration, and free cooling are available, but high installation costs, energy availability constraints, and scalability considerations have slowed adoption.

Water Sources and Treatment

Practical Pathways to Lower Freshwater Demand

Most data centers use municipal water for cooling – about 95% of total data center water use – which can burden communities by contributing to shortages and higher costs. Alternatives can reduce pressure on potable systems, including surface water (rivers and lakes), seawater (with corrosion and distillation considerations), groundwater (where available), wastewater, and graywater. As data center power use grows and groundwater constraints intensify, shifting from municipal supplies to alternative sources is increasingly important for sustainability and community impact.

Turning to the Sea: When Seawater Makes Sense

Standard evaporative cooling systems cannot use seawater without extensive pretreatment to manage scaling and corrosion. Alternative approaches, such as wet surface air cooling, can tap the sea as a heat sink. The main obstacle is financial, not technical, and growing water concerns alongside AI-driven demand may prompt more coastal facilities to evaluate seawater options despite higher upfront costs.

Siting and Infrastructure

AI-Scale Compute Makes Water a Siting Constraint

As dense GPU clusters concentrate unprecedented thermal loads, water is emerging as a siting constraint on par with power. Nearly all server energy becomes heat that must be rejected, and cooling choices create explicit electricity-water tradeoffs. Municipal systems, built for distributed, steady loads, can struggle to meet data centers’ high peaking factors during extreme conditions, and wastewater capacity and reclaimed water access are becoming strategic constraints.

Infrastructure Reality Check: Virginia’s Groundwater and Urban Utilities

A July 2026 Virginia DEQ assessment concluded that, under current permitting standards, no location in the Virginia Coastal Plain can support a major new groundwater withdrawal at the scale typical of some hyperscale data centers. Near the I-95 data center corridor, modeled sustainable capacity is below 30,000 gallons per day, with the highest regional estimate around 360,000 gallons per day in the eastern Northern Neck – still far below hyperscale needs. As a result, future developments in the region would need to rely on municipal or alternative water supplies, making water availability a key constraint alongside power and transmission.

Environmental and Health Impacts

Water Pollution Risks from Cooling Systems

Data centers’ evaporative cooling can introduce biocides, corrosion inhibitors, and metals into local waters, especially at high discharge volumes. To reduce the impact, operators should favor lower- or no-water cooling where possible, use safer chemistry, distribute discharges, and route or treat wastewater before release.

Policy and Governance

The Global Picture: Rapid Growth, Patchwork Governance

Data centers rank among the world’s top 10 most water-intensive industries, per a joint report by BSI Group and Waterwise. A single 1 MW facility can consume over 25 million liters of water annually. Many high-growth markets lack robust water governance and water-efficiency policies, even as the global data center capacity is projected to nearly double by 2030. Without coordinated policy, enforceable standards, and broader adoption of alternative cooling and non-potable water sources, AI- and cloud-driven expansion may deepen water stress in already constrained regions.

Community and Stakeholders

Community Impacts: Alberta’s Debate

Calgary and other Alberta communities have experienced recent water main breaks and restrictions, fueling concerns that data center growth may increase pressure on municipal systems. A large proposed campus near Calgary was rejected, with water use cited alongside environmental, power, and agricultural issues, even as other projects and AI hubs move forward. Given Canada’s cold climate, experts note that low- or zero-water cooling can be economically attractive and help mitigate municipal water dependencies, though it’s not universally necessary or appropriate in every region.

The Bottom Line

Proving water resilience is becoming central to data centers’ license to operate. That entails planning for extremes, matching designs to local hydrology and infrastructure, prioritizing non-potable sources and closed-loop systems where feasible, tightening discharge controls, and engaging transparently with regulators and communities. Metrics such as WUE remain useful, but context-based, site-specific resilience will determine whether facilities can grow sustainably in a water-constrained world.