AI data centers face water-use backlash

Leading technology giants such as Google, Microsoft, and Meta Platforms have begun setting a precedent in environmental transparency by disclosing their AI data centers' water consumption, with Microsoft notably pledging greater openness amid local concerns like those in Columbus, where data centers consume 3% of the city's water. However, transparency remains uneven across the industry, as Amazon lags behind, prompting the United Nations to call for full disclosure of data center footprints to better understand and manage their environmental impacts.

Regulatory responses to the escalating water demands of AI data centers are intensifying globally, with governments implementing measures ranging from Australia's legal mandates requiring large data centers to source new renewables and minimize water use, to New York's moratorium on new data center construction pending legislation. Meanwhile, local authorities like those in The Dalles have resisted disclosing water usage data despite Google's data center consuming about one-third of the city's water, underscoring ongoing transparency challenges that complicate effective oversight.

The regulatory landscape remains fragmented and often reactive, as evidenced by Colorado's repeated failure to pass data center water regulations and the absence of comprehensive federal laws governing water use in AI infrastructure. States and municipalities have adopted diverse strategies including mandatory public reporting of water consumption, watershed-level reviews, and moratoriums to fill policy gaps, yet these piecemeal efforts struggle to keep pace with the rapid expansion of data centers, which now consume approximately 2% of U.S. water and have tripled their usage since 2014.

Experts emphasize that transparency must extend beyond direct water withdrawals to include the indirect water footprint embedded in electricity generation, advocating for integrated water-energy regulatory reviews and performance-based consumption limits tailored to local watershed stress. The federal government is urged to support standardized metrics and data collection without overriding state water laws, leveraging procurement policies and research funding to incentivize disclosure and adoption of low-water cooling technologies, thereby addressing the complex water-energy nexus inherent in AI data centers.

Communities worldwide are grappling with the tangible environmental and social impacts of AI data centers, as seen in Bengaluru where over 30 facilities consume 20 million liters of water daily, intensifying local water stress amidst poor public accountability and regulatory gaps. This strain is mirrored in rural America and Malaysia, where residents express deep concerns over water overuse, farmland loss, and toxic runoff, fueling organized opposition through zoning battles and public protests. The scale of pushback is unprecedented, with more than 140 local groups blocking or delaying billions in investments, reflecting a broad-based demand for transparency and sustainable management.

High-profile advocates like Erin Brockovich have spotlighted the environmental harms caused by major AI data center operators such as Meta, Google, and Amazon, citing instances of water contamination in Cheyenne and risks to the Ashville aquifer. Brockovich emphasizes that secretive deals, NDAs, and weak regulations leave communities blindsided until crises emerge, underscoring transparency as the critical first step toward accountability. This advocacy has galvanized public and regulatory scrutiny, exemplified by the Ohio EPA’s decision to pause permit approvals amid growing environmental concerns.

While the AI industry highlights data centers’ economic contributions—such as property tax revenues funding schools and hospitals, job creation, and investments in grid modernization—skepticism remains high among local residents who question claims that water use is minimal due to closed-loop liquid cooling. The disconnect between industry messaging, which portrays data centers as clean and sustainable, and community experiences of water stress and environmental degradation fuels ongoing opposition and complicates efforts to secure social license for new projects.

The environmental footprint of data centers extends beyond water consumption to exacerbate urban heat island effects, raising health risks in nearby neighborhoods as documented in Phoenix where temperatures near large facilities can spike by up to 16°F. Experts and community leaders advocate for collaborative mitigation strategies, including improved facility design and neighborhood cooling measures, while calling for transparent, policy-guided investments to balance AI’s economic benefits with the protection of critical local resources like water and farmland.

By early 2026, major tech companies like Google and Microsoft are spearheading technological innovations to curb the substantial water consumption of AI data centers, with Google's Nebraska facilities alone using 732 million gallons in 2025. Microsoft’s introduction of a closed-loop cooling system has already cut water use by 25%, targeting a 40% reduction by 2030 amid surging AI and cloud demands. Complementing these efforts, emerging cooling technologies such as direct-to-chip liquid cooling and immersion cooling are rapidly advancing, promising enhanced efficiency and significant reductions in resource consumption, as highlighted by industry trends noting the imminent deployment of 100 to 600 kilowatt racks utilizing these methods.

This technological shift is part of a broader movement to integrate data centers sustainably within local economic ecosystems, minimizing their footprint on power and water resources while delivering community value. However, the adoption of advanced cooling and water reuse strategies—ranging from reclaimed wastewater use to closed-loop systems—is heavily influenced by factors such as cost, facility design, and fragmented regulatory landscapes. As noted in recent analyses, evaporative water-cooled systems, while energy-efficient, consume substantially more water than air-cooled alternatives, underscoring the complex trade-offs operators face in balancing environmental impact and operational efficiency.

Despite concerns about water usage, experts emphasize that water consumption in data centers is largely a function of power usage, with modern closed-loop cooling systems drastically reducing daily water needs. Industry voices argue that it is feasible to operate data centers with minimal consumptive water use by leveraging increased power input for cooling, challenging common perceptions about their environmental footprint. For perspective, one analysis analogizes that a large data center’s water use is comparable to that of a sizable fast-food restaurant, such as a McDonald's, highlighting that water consumption may be less significant than popularly assumed.

Australia is taking a pioneering regulatory stance to ensure that the rapid growth of large AI data centers aligns with environmental sustainability by mandating these facilities to source new renewable energy, pay full grid costs, and minimize water consumption, as announced by Prime Minister Albanese. This approach contrasts sharply with the UK, where the explosive expansion of around 125 data centers in water-scarce regions like Slough is straining water supplies, with some facilities demanding up to 3 million liters daily—equivalent to the peak consumption of 3,500 households. The UK government's failure to incorporate data center water needs into official forecasts and AI growth policies has been called “fatally flawed” by Water UK, highlighting a critical gap in resource planning that threatens long-term water security.

New Zealand is strategically positioning itself to capitalize on the global $2 trillion AI data center boom by emphasizing clean energy and innovative sustainable water management, aiming to avoid the pitfalls seen in other regions. Meanwhile, Malaysia, particularly Johor state, has emerged as Asia's fastest-growing data center hub but now faces growing resource anxieties and community pushback, including unprecedented protests over water use. In response, Johor has implemented strict regulations banning extremely water-intensive data centers and requiring renewable energy sourcing, while operators like China's ZDATA and Japan's NTT are adopting treated wastewater and closed-circuit cooling systems to reduce environmental impact. Despite these constraints, Johor's planned data center capacity is projected to surge eightfold to 7,000 megawatts, underscoring the tension between rapid expansion and sustainability.

Other Malaysian states such as Selangor are adopting a more measured approach by vetting data center proposals against global energy and water efficiency standards and mandating at least 30% local content in key areas like integrated circuit design and cooling systems, reflecting a broader push for sustainable development and local economic benefits. However, as Malaysia tightens scrutiny, investment interest is shifting toward neighboring Southeast Asian countries like Thailand, which offers larger-scale campuses and growing infrastructure, though Johor remains attractive due to its relative stability and government support. This regional dynamic illustrates how regulatory frameworks and sustainability demands are reshaping investment flows within Asia's booming data center market.

The UK faces a daunting challenge in managing the water demands of its AI data center expansion, with current consumption at 6.6 million liters of potable water daily potentially tripling to nearly 20 million liters by 2030 if government targets to triple computing capacity are met. Addressing this will require either long-term infrastructure projects like new reservoirs—which can take up to 15 years to build—or increased environmental water extraction, both fraught with difficulties amid ambitious targets to reduce leaks and consumption. In contrast, New York State has proactively imposed a moratorium on new large data center construction until comprehensive regulatory frameworks are established, signaling a more cautious approach to balancing growth with environmental stewardship.