Every time you interact with a chatbot, train an AI model, or stream high-resolution video, a data center somewhere is hard at work processing your request. While the power demands of these facilities have attracted growing attention, a less visible but equally pressing issue is coming into focus: water consumption.
As AI adoption accelerates and data centers proliferate, particularly in arid regions, water use for cooling is becoming an environmental flashpoint. This piece explores how data centers use water, the challenges their water consumption poses to local communities, and how the industry is working to balance energy efficiency with responsible water management.
Cooling the cloud comes at a cost
Data centers primarily use water for cooling, making it one of the most critical yet underappreciated resources supporting our digital infrastructure. Cooling prevents servers from overheating, which is essential for maintaining reliability, performance, and equipment longevity. As AI workloads become more prevalent, the heat generated by densely packed servers is increasing rapidly. This intensifies the need for effective cooling systems, many of which rely on substantial amounts of water to operate efficiently.
Water usage effectiveness (WUE) is the industry-standard metric for measuring data center water use, calculated by dividing the annual site water usage in liters by the total IT energy usage in kilowatt-hours. A lower WUE indicates higher efficiency. For context, evaporative cooling systems can push WUE well above 1.8, while advanced liquid cooling systems can keep it under 0.3.
Large hyperscale data centers can consume between one and five million gallons of water per day. These numbers vary significantly depending on the cooling technology used and the geographic location of the facility. Microsoft, for example, reported an average WUE of 0.30 liters per kilowatt-hour across its data centers in 2023. Meta has also made significant progress, maintaining a WUE of 0.20 since 2022. Amazon Web Services (AWS) leads the group with a reported global WUE of 0.19 in 2022, improving to 0.18 in 2023. This trend highlights the potential for continued efficiency gains across the sector.
Balancing efficiency with reality
To reduce their water footprints, data center operators are rethinking their approach to cooling. Some facilities have adopted closed-loop systems that reuse water internally rather than continuously drawing new supplies. Others are transitioning to dry or hybrid cooling systems, which significantly lower water use but often come with tradeoffs in energy efficiency and cost.
Optimizing for one environmental goal can make it harder to achieve another. Evaporative cooling, for example, is highly energy efficient but significantly increases water consumption. Conversely, dry cooling systems reduce water use but often require more electricity, especially in warmer climates. In many cases, the most energy-efficient systems are also the most water-intensive, forcing operators to weigh these competing sustainability goals. However, emerging technologies like direct-to-chip and immersion liquid cooling challenge this tradeoff. These systems can deliver both low water usage and high energy efficiency. Despite this advantage, their broader adoption is limited by higher capital costs, infrastructure complexity, and compatibility issues with existing hardware environments.
Exhibit 1. Typical WUE and power usage effectiveness (PUE) ranges for cooling technologies used in data centers
Communities are taking notice
As AI propels exponential growth in data center deployments, water usage is reaching a tipping point. This is particularly true in regions already facing water scarcity. In The Dalles, Oregon, for instance, Google’s facilities consumed over 275 million gallons of water in 2021 alone, representing a 25 percent increase compared to the previous year. With additional data centers planned, residents have expressed mounting concerns about the strain on local water supplies.
Similar tensions are surfacing across the U.S. Southwest, where arid conditions and prolonged droughts have heightened awareness of resource allocation. Communities in Arizona and New Mexico are increasingly questioning the logic of siting water-intensive facilities in areas where water is already scarce. Policymakers are also taking note of these trends. The European Commission, for example, is working on setting binding water use caps for data centers as part of a broader regulatory effort to make digital infrastructure more sustainable.
These pressures are not occurring in isolation. Communities and regulators are increasingly evaluating the total environmental footprint of data centers, weighing both water use and carbon emissions in siting and permitting decisions. Once driven primarily by considerations such as land availability and electricity pricing, site selection for data centers is evolving. Today, long-term water availability and infrastructure resilience are becoming just as critical to planning decisions.
Rethinking water for the AI era
In response to rising scrutiny, leading technology companies are beginning to prioritize transparency and innovation in how they manage water. Google, Microsoft, and Meta now include water usage metrics—like WUE—in their sustainability disclosures. Microsoft has implemented region-specific solutions, such as outdoor air and direct evaporative cooling at its Boydton, Virginia data center, which achieved a WUE of just 0.17 liters per kilowatt-hour.
Other efforts include the use of non-potable water sources, wastewater recycling, and designing infrastructure with the flexibility to adapt cooling methods seasonally or by workload.
While much of the focus is on water used directly for cooling, it’s also important to recognize the indirect water footprint tied to electricity generation. Many data centers draw power from grids that rely on thermoelectric plants, which consume water for steam cycles and cooling. A truly comprehensive water strategy must account for both operational and upstream water impacts. These innovations, while still nascent, signal a shift toward holistic environmental management in data center operations.
Regulators, communities, and investors are demanding more responsible practices, and the industry faces a pivotal moment. Water stewardship is emerging as a defining challenge, requiring not only better engineering but also stronger governance and long-term planning.
Looking ahead, as AI continues to shape the future, it brings with it a responsibility to manage the resources that power it. The spotlight on water use is more than a regulatory concern; it reflects growing public awareness of the hidden costs of our increasingly digital lives. In meeting these challenges, data centers are poised to become testbeds for the next generation of sustainable innovation.
-Piercen Hoekstra
Contact us to learn more about our work on rising U.S. power demand led by data centers, data center power supply options, and developments in data center energy efficiency technology.
ADI Analytics is a prestigious, boutique consulting firm specializing in oil & gas, energy transition, and chemicals since 2009. We bring deep, first-rate expertise in a broad range of markets including energy transition and electric utilities, where we support Fortune 500, mid-sized and early-stage companies, and investors with consulting services, research reports, and data and analytics, with the goal of delivering actionable outcomes to help our clients achieve tangible results.
We also host the ADI Forum, one of Houston’s distinguished industry conferences, to bring c-suite executives from oil & gas, energy transition, and chemicals together for meaningful dialogue and strategic insights across the value chains.
Subscribe to our newsletter or contact us to learn more.
