HomeNews › Data Center Water Use and Discharge

What's Actually in Data Center Water Discharge — and Who Regulates It

The coverage of data center water use tends to focus on a single number: how many millions of gallons per day a facility consumes. That framing — data centers as water consumers — is accurate but incomplete. It misses the other half of the story: what happens to the water that comes back out. Cooling systems don't just consume water. They concentrate it, add chemicals to it, heat it, and discharge it — as blowdown, as thermal effluent, and in some configurations as wastewater with a contaminant profile that municipal pretreatment programs were not designed to handle. The contamination incident at Cheyenne, Wyoming made that gap visible in the most public possible way. This piece covers the engineering behind how data centers actually use water, what's in the water when it leaves, and what regulatory framework governs its discharge.

The Scale of the Problem

A medium-sized data center can consume up to 300,000 gallons of water per day for cooling — roughly equivalent to the daily water usage of 1,000 households, according to a June 2026 University of Georgia Extension publication (Saha, UGA CAES TP-121). Larger hyperscale facilities can each use up to 5 million gallons per day, or approximately 1.8 billion gallons annually — equivalent to a town of 10,000 to 50,000 people.

The United States has approximately 4,286 data centers as of May 2026, representing 38% of all data centers worldwide — by far the largest share of any country. Georgia alone has 213, concentrated in the metro Atlanta area. The world's largest data center hub, Loudoun County, Virginia, used approximately 1 billion gallons of water — mostly potable — in 2023, around 2.75 million gallons per day.

Usage data from UGA Extension (Saha, 2026): one AI query uses approximately 16 oz of fresh water. A one-hour video conference call uses approximately 60 oz. Streaming one hour of HD video requires 0.1 to 3 gallons. Browsing or scrolling social media for one hour uses approximately 15 oz. These per-query figures are small; multiplied across billions of daily interactions globally, drawing from a small number of concentrated facility locations, the localized impact on specific watersheds and municipal supplies is substantial.

Five Cooling System Types — and Their Very Different Water Profiles

Not all data centers have the same water footprint. The cooling technology deployed determines how much water is consumed, whether it is discharged or evaporated, and what it contains when it leaves the facility. Understanding this distinction matters for anyone evaluating the water quality implications of a specific facility.

Cooling TypeWater ConsumptionDischarge ProfileEnergy UseTypical Use Case
Evaporative (open-loop)High — most water evaporatesConcentrated blowdown: salts, minerals, biocides, corrosion inhibitors, heavy metalsLowerMajority of large modern data centers; hot climates
Closed-loop water coolingUp to 70% lower than open-loopMinimal — sealed system; small volumes of treated water periodically dischargedHigherWater-stressed locations; premium installations
Air coolingNegligible direct water useNo process water dischargeHighestCooler climates; smaller or older facilities
Direct liquid / immersion coolingVery lowCoolant fluids may contain refrigerants or PFAS compounds; specialized handling requiredLow-moderateEmerging; high-density GPU installations
Free coolingNegligibleThermal discharge to receiving water (river, sea)LowestCold climates; near large water bodies
Source: Saha, U.K. (2026). Understanding How Data Centers Impact Surface and Ground Waters. UGA Cooperative Extension TP-121.
The water/energy tradeoff is real and quantified. Using natural gas to power Texas data center electricity loads would require 50 times more water than solar generation, and 1,000 times more water than wind, according to a Houston Advanced Research Center study cited by the Lincoln Institute (Gorey, 2025). But wind power requires four times as much land as solar and 42 times as much as natural gas. There is no option that optimizes for water, energy, and land simultaneously — every siting and technology choice involves tradeoffs between them.

What's in Cooling Tower Blowdown

The most common data center discharge stream is cooling tower blowdown — the water intentionally removed from an open-loop evaporative cooling system to prevent dangerous concentrations of dissolved solids. As water evaporates in the cooling tower, the minerals and chemicals it carried remain behind, progressively concentrating in the recirculating water. Blowdown is the controlled purge of this concentrated water, replaced by fresh makeup water to keep the system within acceptable operating parameters.

Blowdown is not clean water. According to UGA Extension (Saha, 2026) and EPA guidance on water-related permits for data centers, cooling tower blowdown may contain:

PFAS in data center coolants: the UGA Extension publication specifically identifies that some nonwater liquid coolants used in direct liquid and immersion cooling may contain refrigerants and PFAS ("forever chemicals") requiring special forms of treatment. Fluorinated gases (F-gases) used in air cooling are also a PFAS subcategory. If leaks, spills, or improper wastewater disposal occur, these compounds can enter soil and groundwater. EPA guidance notes that cooling-system blowdown may require permits or wastewater treatment before discharge — but the applicable monitoring requirements have not been broadly updated to reflect PFAS-containing coolant chemistries.

Thermal Discharge: The Less-Discussed Impact

In facilities using free cooling — drawing water from rivers, lakes, or reservoirs and returning it after absorbing heat — the discharge returns to the receiving water at elevated temperature. Warmer discharge water reduces dissolved oxygen, stresses fish and aquatic organisms, promotes algal blooms, and can cause long-term changes to aquatic ecosystems. This is thermal pollution, regulated under the Clean Water Act but subject to site-specific permit conditions that vary considerably between facilities and jurisdictions.

Evaporative cooling avoids thermal discharge to receiving waters — but at the cost of losing most of that water entirely to the atmosphere. "Even if they're using reclaimed or recycled water, that water is no longer going back into the base flow of the rivers and streams," Julie Bolthouse, director of land policy at the Piedmont Environmental Council, told the Lincoln Institute. "Everybody is upstream from someone else." Washington, DC, for example, still loses water supply if Northern Virginia data centers use recycled water, because that water won't return to the Potomac River.

The Regulatory Framework — and Its Gaps

Data center water discharge is regulated under the Clean Water Act through two primary mechanisms: NPDES permits for direct discharge to surface waters, and industrial pretreatment program requirements for discharge to municipal sewer systems.

The EPA specifically identifies cooling tower blowdown from data centers as a wastewater stream that may require permits before discharge (EPA, 2026). Facilities that discharge directly to waterways are subject to NPDES permit conditions. Facilities that discharge to the municipal sewer must comply with local pretreatment standards, including categorical and local limits on specific pollutants.

The critical gap the Cheyenne incident exposed: existing NPDES pretreatment programs were designed for conventional industrial dischargers — metal finishers, food processors, chemical manufacturers — and their characteristic pollutants. Data center fill-and-flush operations during construction and commissioning generate a different contamination profile: large volumes of water that have circulated through new cooling infrastructure where unusual organisms can colonize during the months-long construction period. When Goat Systems LLC discharged fill-and-flush wastewater from Meta's $800 million campus into Cheyenne's municipal system, the monitoring requirements in place were not calibrated to detect Cupriavidus gilardii — an organism that isn't part of the standard pretreatment monitoring panel. Full coverage of the Cheyenne incident here.

The operational phase raises a parallel issue. Most pretreatment programs have well-established limits for conventional pollutants — TSS, BOD, pH, certain metals — but variable coverage for the specific chemical additives used in modern cooling tower chemistry, and essentially no requirements for PFAS monitoring in cooling system discharges. As direct liquid and immersion cooling systems become more common and PFAS-containing coolants enter more facilities, this gap will widen.

Transparency: A Documented Absence

A significant obstacle to managing data center water impacts is the near-total absence of facility-level water use and discharge data in the public domain. The UGA Extension publication notes that public records regarding data center water usage are currently limited, with many operators protected by local nondisclosure agreements that prevent communities from accessing information about specific facilities.

Georgia Senate Bill 421 — the "Data Center Transparency Act" — would prevent local governments from using nondisclosure agreements to conceal data center water or electricity usage data, making it publicly available. As of May 2026, the bill was pending in committee.

Nationally, the EPA monitors aggregate usage but does not publish facility-level data. Some major operators release voluntary sustainability reports — Google reported using more than 5 billion gallons across all data centers in 2023 — but the coverage is inconsistent and the methodology varies between companies, making facility-to-facility comparisons difficult.

57% of data centers use potable water: one analysis estimated that approximately 57% of all data centers draw their cooling water from potable water supplies — treated municipal water or groundwater of drinking-water quality (Mytton, 2021, cited in UGA TP-121). An estimated 80–90% of this water is consumed from watersheds — the same sources providing drinking water to residents through public water systems. The local, concentrated nature of this withdrawal is what drives community concern: nationally, all US data centers combined consumed about 449 million gallons per day in 2021 — a relatively small 0.14% of total US water consumption — but the withdrawal is not evenly distributed. It is concentrated in specific locations, often smaller communities with limited water infrastructure capacity.

Geographic Concentration in Water-Stressed Areas

Roughly two-thirds of data centers built since 2022 have been located in water-stressed regions, according to a Bloomberg News analysis cited by the Lincoln Institute. Arizona has attracted significant data center investment despite being one of the most water-constrained states in the country. Northern Virginia — home to the densest concentration of data centers on Earth, with approximately 300 facilities in a handful of counties — draws from the same Potomac watershed that supplies Washington, DC.

Peter Colohan, director of partnerships and program innovation at the Lincoln Institute, framed the geographic dynamic precisely: "Wherever they choose to put a data center, it is like a giant soda straw sucking water out of that basin. And when you take water from a place, you have to reduce demand or put water back in that same place — there's no other solution."

In Newton County, Georgia, a Meta data center that opened in 2018 uses 500,000 gallons of water per day — 10% of the entire county's water consumption. Newton County continues to receive permit requests for new data centers that would use up to 6 million gallons per day, more than doubling what the entire county currently consumes.

What This Means for Municipal Water Systems and Treatment Professionals

For municipal water utility managers, pretreatment program administrators, and water treatment professionals, data center growth presents a set of specific operational challenges that differ from conventional industrial users:

Key Takeaways

  • Data center cooling generates two distinct discharge streams: concentrated blowdown from evaporative cooling (containing biocides, corrosion inhibitors, heavy metals, and potentially PFAS) and thermal effluent from free-cooling configurations — each with different regulatory treatment requirements.
  • 57% of data centers use potable water for cooling, drawing from the same municipal and groundwater supplies that serve residential users. The impact is locally concentrated even when nationally small.
  • Five cooling technologies have very different water profiles. Evaporative (open-loop) systems dominate and have the highest blowdown discharge volumes. Closed-loop systems reduce consumption by up to 70%. Immersion and direct liquid cooling may involve PFAS-containing coolants requiring specialized discharge management.
  • PFAS in data center coolants is an emerging and underregulated issue. Some nonwater liquid coolants and F-gas refrigerants used in air cooling are PFAS compounds. Current pretreatment program monitoring requirements do not broadly address this chemistry.
  • The Cheyenne incident exposed a construction-phase gap: fill-and-flush commissioning discharges carry biological contamination risks that standard pretreatment monitoring panels were not designed to detect. Other municipalities hosting data center construction face the same gap.
  • Two-thirds of new data centers are sited in water-stressed regions, concentrating large withdrawals in locations with the least capacity to accommodate them. Evaporative cooling fully removes that water from the local watershed — it does not return to the supply even as treated effluent.
  • Transparency is nearly absent at the facility level. Nondisclosure agreements, voluntary-only sustainability reporting, and the absence of mandatory facility-level water discharge reporting leave municipal utilities with inadequate information to manage cumulative impacts.

Sources

  1. Saha, U.K. (2026). "Understanding How Data Centers Impact Surface and Ground Waters." UGA Cooperative Extension Temporary Publication TP-121. University of Georgia College of Agricultural and Environmental Sciences. Published June 10, 2026. fieldreport.caes.uga.edu/publications/TP121
  2. Gorey, J. (2025). "Data Drain: The Land and Water Impacts of the AI Boom." Land Lines Magazine, Lincoln Institute of Land Policy. Published October 17, 2025. lincolninst.edu — Data Drain: Land and Water Impacts of the AI Boom
  3. Hegde, G. (2026). "Myths vs. Reality: Data Centers and Water Usage." Florida Water & Pollution Control Operators Association. Published January 23, 2026. Cited in Saha (2026), UGA TP-121.
  4. U.S. Environmental Protection Agency. (2026). "Water and Data Centers." epa.gov/watersense/water-and-data-centers
  5. U.S. Environmental Protection Agency. (2026). "Water-Related Permits for Data Centers." epa.gov/watersense/water-related-permits-data-centers
  6. Mytton, D. (2021). "Data centre water consumption." npj Clean Water, 4, 11. doi.org/10.1038/s41545-021-00101-w
  7. Siddik, M.A.B., Shehabi, A., & Martson, L. (2021). "The environmental footprint of data centers in the United States." Environmental Research Letters, 16, 064017.
  8. Houston Advanced Research Center (HARC). "Powering Texas Digital Economy: Data Centers and the Future of the Grid." Cited in Lincoln Institute (Gorey, 2025).
Related CWL coverage: The Cheyenne contamination incident — in which Meta's construction entity was found in significant noncompliance with industrial pretreatment regulations after fill-and-flush cooling system discharge introduced a rare bacterium into the municipal wastewater system — is covered in detail in Contaminated by Construction: Meta's Wyoming Data Center and Cheyenne's Wastewater Crisis. The NPDES industrial pretreatment framework is covered in the Industrial Water Waste Treatment guide.
More News