AI data centers must choose between depleting water supplies or straining electrical grids, with no solution that avoids environmental and community costs. As artificial intelligence infrastructure expands globally, companies like Microsoft, Google, and Amazon are grappling with a fundamental physics problem: keeping next-generation computer chips cool requires either millions of gallons of water or enormous amounts of electricity. Whichever path they choose triggers public opposition and environmental trade-offs that experts say have no easy answers.

Why Is Cooling AI Data Centers So Resource-Intensive?

Computer chips powering AI systems generate intense heat that can cause them to overheat and shut down, much like an overheated laptop. Historically, data centers used simple air conditioning, but as chips became more powerful and climate change pushed summer temperatures higher, cooling large buildings with air alone became prohibitively energy-intensive, sometimes requiring hundreds of megawatts of electricity equivalent to a small power plant.

To address this, most hyperscale data centers switched to water-based cooling methods, particularly evaporative cooling systems similar to those used by power plants. Water absorbs and transports heat far more efficiently than air. According to Google's calculations, using 264 million gallons of water for cooling in central Europe can reduce a data center's power demands by 41,000 megawatt-hours, enough to power approximately 10,000 homes.

However, this solution creates a new problem: two-thirds of U.S. data centers built since 2022 are located in areas experiencing high levels of water stress, raising concerns about depleting local water supplies in drought-prone regions.

What Are the Trade-Offs Between Water and Energy?

The choice between water and electricity is not a simple one. If a company reduces water consumption, its power needs surge, increasing carbon emissions and straining electrical grids. If it reduces electricity demand, water consumption climbs, potentially depleting natural resources in water-stressed communities. This creates what experts call a no-win scenario.

"There's no easy answer, it's a lot of trade-offs," said John Ikeda, chief mission officer at the Water Environment Federation, a technical and educational nonprofit for water professionals that has worked with Amazon on data center issues.

John Ikeda, Chief Mission Officer at the Water Environment Federation

Different tech companies are making different bets. Microsoft and Quality Technology Services have pledged to use zero water in cooling, accepting higher power usage and potential carbon emissions. Google and Amazon say they use water strategically to reduce pressure on electrical grids in wetter climates, but avoid water-based cooling in drought-prone areas.

How Are Companies Trying to Solve This Problem?

Rather than accepting the trade-off as permanent, major tech companies are investing in innovative cooling technologies designed to reduce both water and energy consumption simultaneously. Amazon developed a system that lowered the company's North American water use by 946 million liters in 2024, equivalent to drinking water for approximately 1.3 million people annually, while improving water efficiency by 17 percent that year.

Microsoft is experimenting with technologies that allow AI computer chips to function at higher temperatures, potentially reducing the power demand for cooling. As Steve Solomon, Microsoft's vice president for data center infrastructure engineering, explained, warmer operating temperatures can improve energy efficiency.

Amazon Web Services is planning a facility in Louisiana designed to use outside air for cooling during 87 percent of the year, with evaporative cooling deployed only during the hottest months to reduce electricity demand by 25 to 35 percent precisely when electrical grids experience peak summer loads. In more temperate climates like Ohio, some facilities may require water-based cooling for only 3 percent of the year.

• Evaporative Cooling: Water is sprayed to absorb heat from the data center, then released as vapor into the atmosphere, reducing energy consumption but consuming significant water resources.
• Adiabatic Cooling: Outside air sent through the data center is first sprayed with water to cool it down, balancing water and energy use more efficiently than full evaporative systems.
• Hybrid Systems: Facilities use evaporative cooling only during the hottest times of the year and rely on outside air for cooling during milder periods, minimizing both water and energy waste.
• Higher Operating Temperatures: Allowing chips to run at warmer temperatures reduces the cooling load and improves overall energy efficiency without sacrificing performance.

Why Is Transparency About Data Center Operations So Important?

Despite these innovations, public opposition to data centers is intensifying. Seven out of 10 Americans oppose data centers, citing water use as a top concern, according to a recent Gallup poll. In the first quarter of 2026 alone, at least 75 data center projects valued at $130 billion were disrupted by local opposition, and county-level moratoriums are becoming increasingly common.

A major barrier to informed decision-making is the lack of transparency. Most tech companies do not fully explain the details behind their water-versus-energy trade-offs. There is no federal disclosure requirement, and companies that do share information through sustainability reports use inconsistent metrics and varying levels of detail, making it difficult for local governments and communities to assess the true environmental impact.

"No other major U.S. energy-consuming sector, nor one which is growing so quickly from economic and infrastructure perspectives, suffers from as many public data blind spots as U.S. data centers," testified Eric Masanet, a data center expert at the University of California, Santa Barbara, before Congress in February.

Eric Masanet, Data Center Expert at the University of California, Santa Barbara

Microsoft's approach to this challenge reflects a broader shift in how the company views its environmental responsibility. In its 2026 Environmental Sustainability Report, Microsoft announced that it achieved a major milestone by replenishing more water globally than it withdrew in fiscal year 2025, more than 14 million cubic meters, marking progress toward its goal of becoming a water-positive company.

However, Microsoft acknowledged that global water replenishment is not sufficient. The company is shifting focus to ensure it restores more water to the specific watersheds where it operates than it withdraws, prioritizing projects in water-stressed regions designed in partnership with local communities. This localized approach recognizes that environmental sustainability cannot be solved through global averages alone.

What Does This Mean for the Future of AI Infrastructure?

The tension between AI's resource demands and sustainability solutions is forcing the industry to ask harder questions about how to build infrastructure responsibly. Microsoft stated that "the path forward will not be defined by simple trade-offs or single solutions. It will depend on how effectively we align innovation with stewardship".

Microsoft

As AI continues to reshape economies and accelerate innovation, the infrastructure supporting it must evolve to strengthen rather than strain the systems and communities it depends on. This requires greater operational rigor, stronger integration across sustainability priorities, and sharper focus on durable outcomes for local communities.

The data centers of tomorrow will likely rely on a combination of advanced cooling technologies, renewable energy sources, and localized water management strategies tailored to regional conditions. But success will ultimately depend on whether tech companies can achieve the transparency and community partnership necessary to build public trust in AI infrastructure development.