What is the estimated annual water consumption by data centers in the United States and how has it changed between 2020 and 2025?

Data centers are massive buildings packed with computers that power everything from social media and video streaming to AI tools you might use for homework or fun. They need huge amounts of water mainly for cooling the servers so they don't overheat. Most of this happens through evaporative cooling, where water turns into vapor and isn't fully returned to rivers or lakes.

A 2024 report from Lawrence Berkeley National Laboratory estimates that U.S. data centers used about 17 billion gallons of water directly in 2023 for cooling. That's a big jump since 2020 as AI workloads exploded and more giant facilities came online. Earlier estimates put direct use around 5.6 billion gallons back in 2014, showing steady growth that's likely continued through 2025. On top of that, indirect water use tied to generating electricity for these centers reached roughly 211 billion gallons in 2023—far larger but spread across power plants.

Nationally this is still under 0.2 percent of all U.S. water withdrawals, yet it hits harder in dry spots like Arizona or parts of Virginia where facilities cluster. For you, this matters because the apps and services you rely on daily keep expanding, and water shortages could affect farming, drinking supplies, or even raise costs in your community later.

Views differ on solutions. Tech companies highlight new efficient cooling methods and say their total impact stays small compared to agriculture. Others argue for stricter rules like mandatory water-use reports and efficiency standards for new builds, pointing out that many facilities don't share detailed numbers. Balancing faster AI progress with protecting local water resources remains the core challenge.

The rapid expansion of data centers driven by artificial intelligence, cloud computing, and digital services has intensified scrutiny of their environmental footprint, particularly water consumption. As facilities scale to meet surging computational demands, questions arise about sustainable resource use amid regional droughts and competing agricultural or municipal needs. Estimates of annual U.S. data center water consumption vary by methodology and scope, distinguishing direct withdrawals for cooling from indirect usage tied to electricity generation. Between 2020 and 2025, growth in hyperscale operations has likely accelerated these figures, prompting policy debates on transparency, efficiency standards, and innovation incentives. This analysis draws on reports from Lawrence Berkeley National Laboratory, academic reviews, and policy assessments to examine trends, tradeoffs between economic benefits and ecological risks, and implications for balanced regulation that supports technological progress without compromising water security.

Direct water consumption by U.S. data centers primarily stems from evaporative cooling systems, where water is withdrawn but not fully returned to watersheds. A 2024 Lawrence Berkeley National Laboratory report estimates 17 billion gallons consumed in 2023, projecting potential doubling to around 34 billion gallons by 2028 as capacity expands. Earlier academic assessments indicate substantial growth from prior baselines: one review calculated roughly 21.2 billion liters (about 5.6 billion gallons) in 2014, aligning closely with recent federal and state-level findings of 17 billion gallons in 2023. These figures reflect a marked increase since 2020, coinciding with AI-driven buildouts that boosted data center energy loads and associated cooling requirements.

Indirect consumption, linked to thermoelectric power plants supplying electricity, dwarfs direct use. Federal estimates place the 2023 indirect footprint at roughly 211 billion gallons nationwide, varying by fuel mix—higher for coal or natural gas facilities than renewables. Combined impacts for major operators like Meta, Apple, and Google alone could associate with over 100 billion liters of indirect water use in electricity production. Local perspectives highlight variability: while aggregate national figures represent under 0.2 percent of total U.S. water withdrawals, concentrated siting in arid regions such as Arizona or Virginia amplifies stress on aquifers and rivers, as noted in analyses from environmental groups and utilities.

Policy viewpoints diverge on mitigation. Industry analyses emphasize efficiency gains through advanced cooling like immersion or free-air systems, arguing data centers contribute modestly compared to agriculture or thermoelectric sectors overall. Conversely, transparency advocates and researchers point to inconsistent reporting, with many facilities exempt from detailed disclosures, complicating accurate tracking. Between 2020 and 2025, the shift toward larger AI clusters has outpaced some conservation measures, though renewable integration could curb indirect volumes. Competition concerns also surface, as water-intensive operations may disadvantage smaller firms or strain public infrastructure without updated permitting frameworks. Academic and congressional reviews, including those responding to the Energy Act of 2020, underscore the need for integrated energy-water modeling to inform incentives that reward both innovation and stewardship.