Version 2

AI data center construction is meeting organized community resistance across the United States, driven primarily by water consumption concerns. A clear split has emerged in the regulatory response: local and municipal actions — permanent bans, moratoria, and zoning restrictions — are succeeding in slowing or blocking projects [5][12][7], while state-level moratorium legislation is spreading but largely stalling in legislatures [13][14]. The Stratos hyperscale project in Utah was cut by 50% after residents successfully protested a water rights application threatening the Great Salt Lake [4], and a Gallup poll found 70% of Americans oppose data center development, with water scarcity as the top concern [19].

Water consumption has become a material constraint on the AI infrastructure build-out, not just an environmental complaint. SpaceX formally identified water scarcity as a potential business risk in its IPO filing [19], and the divergence between effective local actions and stalled state legislation means developers face a fragmented, jurisdiction-specific regulatory environment rather than a predictable national framework. With AI data center water use projected at 9.3 trillion litres annually by 2030 [20], communities holding local authority over land use have demonstrated they can force project changes before construction begins.

AI data centers rely on evaporative cooling systems that consume fresh water at scale, making water-stressed regions sites of growing conflict between developers and communities. As hyperscale facilities have multiplied, residents have learned that water rights applications and local zoning processes offer the most effective legal levers for blocking or scaling back projects. The Stratos project in Utah became the clearest test: the developer sought to transfer 1,900 acre-feet of water from a ranch to a campus originally designed to cover nearly three times the area of Manhattan, raising concerns about the already-stressed Great Salt Lake. The Friends of Great Salt Lake filed formal protests [1][2], a second water rights application was withdrawn [3], and the project was cut by 50% before construction began [4]. Developer Kevin O'Leary publicly acknowledged the failure: "We pissed off a lot of people," and said he regrets not working more transparently with state officials from the start.

The municipal and local response has spread well beyond Utah. Monterey Park, California voted 86% to permanently ban data centers, reportedly the first such ban in the United States [5][6]. Seattle City Council members introduced moratorium legislation in April 2026 [7]. The Seminole Nation of Oklahoma imposed a moratorium on hyperscale facilities [8]. Illinois Governor Pritzker announced a pause on data center tax incentives starting July 2026 [9]. Erin Brockovich launched a national complaint tracking map to aggregate community opposition [10][11], and a dedicated site (datacenterbans.com) tracks active moratoria across jurisdictions [12].

A key distinction has emerged between the local and state levels of government. Local and municipal actions are advancing while state-level moratorium bills are stalling in legislatures despite broad introduction [13][14]. Columbia Law School's climate blog and law firm ArentFox Schiff have both flagged the result as a legally complex and shifting regulatory landscape [15][16], with developers facing inconsistent rules by jurisdiction rather than a unified framework. Not all governments are moving toward restraint: Michigan Governor Whitmer overrode community pushback to preserve data center tax incentives [17], and reports indicate the Trump administration is considering invoking Cold War-era emergency authorities to direct resources toward AI infrastructure [18] — a posture that sits in tension with municipal bans and state pauses.

Public opinion and financial disclosure data both back the direction of local actions. A Gallup poll found 70% of Americans oppose data center development, with water scarcity ranked above energy and land use as the top resource concern [19]. Research from a UN-affiliated institute projects AI data center water use at 9.3 trillion litres annually by 2030 [20][21]. SpaceX amended its IPO filing to formally disclose water scarcity and drought as potential constraints on operations, and market commentary increasingly frames data center water use as investor risk [19][22]. Industry-aligned voices argue the public narrative is based on misleading data [23], but that counter-argument has not yet been substantiated with primary sourcing in the documented coverage.