AI operations
2 sourcesWater used or consumed while a computing service runs. The result depends on the facility, cooling system, workload, location, and time period.
Reviewed evidence
See where every water number comes from.
This library includes only sources attached to published water explanations. Open “How we know” for the scope, calculation basis, and caveat behind each number.
First, check what kind of water is being counted.
Two numbers can both be accurate and still be unsuitable for direct comparison. These five lanes keep their boundaries visible.
Electricity
1 sourceWater consumed where electricity is generated. This is an upstream grid boundary, not water used directly at a data center.
Daily water
3 sourcesWater delivered at the point of use for activities such as showering, washing clothes, or running an appliance.
Supply chains
4 sourcesWater counted across a product's supply chain. These footprints may combine rainfall, irrigation, and water-quality estimates.
Big picture
2 sourcesSystem totals for a sector, grid, or country. These figures describe infrastructure at scale, not one AI request.
Evidence library
Read the record in plain language.
Showing 11 reviewed sources
Government source2026Daily water
DOE Energy Saver: Reduce Hot Water Use for Energy Savings
Supports 1 published water explanation on this site.
How we know
Average shower hot water
Direct, everyday water · Absolute Volume · Delivered · Household Delivered
37.9 L
Basis: per shower, hot water only
Important limit: DOE table lists shower at 10 gallons/use. Energy is calculated assuming 65°F temperature rise and no heater losses.
United States · per-event · derived · Non-AI context record
Calculation and assumptions
Normalized from seed research (basis: per shower, hot water only, energyWh=1588.8, waterMl=37854.1).
- DOE table lists shower at 10 gallons/use. Energy is calculated assuming 65°F temperature rise and no heater losses.
Government source2025Daily water
ENERGY STAR: Clothes Washers
Supports 1 published water explanation on this site.
How we know
ENERGY STAR clothes washer load
Direct, everyday water · Absolute Volume · Delivered · Household Delivered
53 L
Basis: per load
Important limit: Full-sized ENERGY STAR washers use 14 gallons/load vs 20 gallons for standard machine.
United States · per-event · derived · Non-AI context record
Calculation and assumptions
Normalized from seed research (basis: per load, waterMl=52995.8).
- Full-sized ENERGY STAR washers use 14 gallons/load vs 20 gallons for standard machine.
Company disclosure2025AI operations
Google Cloud: Measuring the environmental impact of AI inference
Published by Google Cloud
Supports 1 published water explanation on this site.
Look for: Opening summary and footnote 1
How we know
Gemini Apps median text prompt
Operational and cooling water · Absolute Volume · Consumption · Direct Operational
0.26 mL
Basis: per median text prompt
Important limit: Company-estimated May 2025 median using 2024 fleet-average water-use effectiveness. Not independently verified or representative of every model, prompt, modality, location, or future performance.
Google global production fleet · per-event · company-estimated
Calculation and assumptions
No unit conversion; Google reports 0.26 mL per median Gemini Apps text prompt.
- Company-estimated May 2025 median using 2024 fleet-average water-use effectiveness. Not independently verified or representative of every model, prompt, modality, location, or future performance.
Company disclosure2025AI operations
Sam Altman: The Gentle Singularity
Published by Sam Altman
Supports 1 published water explanation on this site.
Look for: Parenthetical average-query statement
How we know
ChatGPT average query
Operational and cooling water · Absolute Volume · Mixed Or Unclear · Mixed Or Unclear
0.32 mL
Basis: per average query
Important limit: Public CEO estimate with no supporting measurement period, geography, accounting boundary, or technical methodology. Stored as 0.32 mL rather than a falsely precise conversion.
Not disclosed · per-event · company-estimated
Calculation and assumptions
0.000085 U.S. gal × 3,785.411784 mL/U.S. gal = 0.32176 mL; normalized to 0.32 mL to match source precision.
- Public CEO estimate with no supporting measurement period, geography, accounting boundary, or technical methodology. Stored as 0.32 mL rather than a falsely precise conversion.
Government source2025Big picture
USGS: Water use across the conterminous United States, water years 2010–20
Published by U.S. Geological Survey
Supports 3 published water explanations on this site.
Look for: Key Points; Tables 1–4; Summary and Conclusions
How we know
U.S. crop irrigation consumptive use
National and system context · Absolute Volume · Consumption · Direct Operational
286.5 billion L
Basis: CONUS average day, water years 2010–2020
Important limit: Modeled mean for the conterminous United States across water years 2010–2020. This is consumptive use, not total withdrawal, and it predates the 2023 data-center estimate.
Conterminous United States (CONUS) · per-day · government-modeled · Non-AI context record
Calculation and assumptions
75,698 Mgal/day × 1,000,000 gal/Mgal × 3,785.411784 mL/gal
- USGS national models estimate use where water is used; some interbasin transfers can therefore be attributed to the demand location.
U.S. public-supply consumptive use
National and system context · Absolute Volume · Consumption · Direct Operational
16 billion L
Basis: CONUS average day, water years 2010–2020
Important limit: Modeled mean for the conterminous United States across water years 2010–2020. Public supply serves many end uses; this is consumptive use, not deliveries or total withdrawal.
Conterminous United States (CONUS) · per-day · government-modeled · Non-AI context record
Calculation and assumptions
4,219 Mgal/day × 1,000,000 gal/Mgal × 3,785.411784 mL/gal
- USGS national models estimate use where water is used; some interbasin transfers can therefore be attributed to the demand location.
U.S. thermoelectric freshwater consumptive use
National and system context · Absolute Volume · Consumption · Direct Operational
11 billion L
Basis: CONUS average day, water years 2010–2020
Important limit: Modeled mean freshwater consumptive use for the conterminous United States across water years 2010–2020. Thermoelectric withdrawals are much larger because most withdrawn water returns.
Conterminous United States (CONUS) · per-day · government-modeled · Non-AI context record
Calculation and assumptions
2,904 Mgal/day × 1,000,000 gal/Mgal × 3,785.411784 mL/gal
- USGS national models estimate use where water is used; some interbasin transfers can therefore be attributed to the demand location.
Academic research2024Big pictureElectricity
Lawrence Berkeley National Laboratory: 2024 United States Data Center Energy Usage Report
Published by Lawrence Berkeley National Laboratory
Supports 2 published water explanations on this site.
Look for: Printed p. 57, Figure 5.10 and the following paragraph
How we know
All U.S. data centers: direct water consumption
National and system context · Absolute Volume · Consumption · Direct Operational
180.8 million L
Basis: all U.S. data centers, average day in 2023
Important limit: This is direct cooling-related consumption for all U.S. data centers, not AI alone. A national annual estimate can hide large differences by site, season, cooling design, and local water stress.
United States · per-day · independently-modeled · All data centers; not AI-specific
Calculation and assumptions
66 billion L/year × 1,000,000,000 L/billion L × 1,000 mL/L ÷ 365 days/year
- Figure 5.9 is a modeled national estimate for all data centers, not a meter total and not an AI-only estimate.
All U.S. data centers: electricity-related water consumption
Electricity-related water · Absolute Volume · Consumption · Upstream Electricity
2.2 billion L
Basis: all U.S. data centers, average day in 2023
Important limit: This is upstream water consumed to generate electricity for all U.S. data centers, not direct facility water and not AI alone. It must not be summed or ranked with direct-operation records.
United States electricity grid mix at data center locations · per-day · independently-modeled · All data centers; not AI-specific
Calculation and assumptions
800 billion L/year × 1,000,000,000 L/billion L × 1,000 mL/L ÷ 365 days/year
- The report describes the result as nearly 800 billion liters; 800 is retained as an explicitly approximate source value.
Government source2023Daily water
ENERGY STAR: Dishwashers Key Product Criteria
Supports 1 published water explanation on this site.
How we know
ENERGY STAR dishwasher cycle
Direct, everyday water · Absolute Volume · Delivered · Household Delivered
12.1 L
Basis: per cycle
Important limit: Energy derived from max 240 kWh/year and 215 cycle/year assumption; water uses 3.2 gal/cycle criterion.
United States · per-event · derived · Non-AI context record
Calculation and assumptions
Normalized from seed research (basis: per cycle, energyWh=1116.3, waterMl=12113.3).
- Energy derived from max 240 kWh/year and 215 cycle/year assumption; water uses 3.2 gal/cycle criterion.
Independent nonprofit2012Supply chains
Water Footprint Network Product Gallery: Beef
Supports 1 published water explanation on this site.
How we know
Beef serving
Embedded and lifecycle water · Absolute Volume · Embedded Mixed · Product Lifecycle
1,746.3 L
Basis: per 1/4 lb serving
Important limit: Derived from 15,400 L/kg global average and 0.113398 kg serving. Mostly green water; local impact varies.
Global product average · lifecycle · independently-modeled · Non-AI context record
Calculation and assumptions
Normalized from seed research (basis: per 1/4 lb serving, waterMl=1746329.2).
- Derived from 15,400 L/kg global average and 0.113398 kg serving. Mostly green water; local impact varies.
Independent nonprofit2011Supply chains
Water Footprint Network Product Gallery: Coffee
Supports 1 published water explanation on this site.
How we know
Cup of coffee
Embedded and lifecycle water · Absolute Volume · Embedded Mixed · Product Lifecycle
130 L
Basis: per 125 mL cup
Important limit: Embedded agricultural water footprint, not direct tap water consumed while brewing.
Global product average · lifecycle · independently-modeled · Non-AI context record
Calculation and assumptions
Normalized from seed research (basis: per 125 mL cup, waterMl=130000).
- Embedded agricultural water footprint, not direct tap water consumed while brewing.
Independent nonprofit2011Supply chains
Water Footprint Network Product Gallery: Cotton
Supports 1 published water explanation on this site.
How we know
Cotton shirt
Embedded and lifecycle water · Absolute Volume · Embedded Mixed · Product Lifecycle
2,500 L
Basis: per 250 g cotton shirt
Important limit: Global average; local irrigation and blue-water share vary widely.
Global product average · lifecycle · independently-modeled · Non-AI context record
Calculation and assumptions
Normalized from seed research (basis: per 250 g cotton shirt, waterMl=2500000).
- Global average; local irrigation and blue-water share vary widely.
Independent nonprofit2011Supply chains
Water Footprint Network Product Gallery: Tea
Supports 1 published water explanation on this site.
How we know
Cup of tea
Embedded and lifecycle water · Absolute Volume · Embedded Mixed · Product Lifecycle
30 L
Basis: per 250 mL cup
Important limit: Embedded agricultural water footprint, not direct tap water consumed while brewing.
Global product average · lifecycle · independently-modeled · Non-AI context record
Calculation and assumptions
Normalized from seed research (basis: per 250 mL cup, waterMl=30000).
- Embedded agricultural water footprint, not direct tap water consumed while brewing.
Why there is no master ranking
Withdrawal, consumption, delivered water, cooling water, and lifecycle footprints describe different water flows. A single ranked list would hide those differences.
Receipt calculation chain
The sources behind everyday activities.
These records feed the Water Receipt directly. Derived and modeled lines combine more than one source, so the original inputs remain visible instead of being replaced by a platform-sounding number.
Receipt evidence reviewed July 9, 2026
derivedreportedHome tapPower grid
DOE Energy Saver: Reduce Hot Water Use for Energy Savings
Published 2026
Used by: Hot shower
Open sourcederivedreportedHome tap
EPA WaterSense: Statistics and Facts
Published 2026
Used by: Household leaks, Running faucet
Open sourcederivedPower grid
ENERGY STAR: Clothes Dryers Key Product Criteria
Published 2025
Used by: Dryer load
Open sourcereportedHome tap
ENERGY STAR: Clothes Washers
Published 2025
Used by: Washer load
Open sourcederivedmodeledPower grid
ENGIE Impact: Electricity consumption of videos viewed on Meta platforms
Published 2025
Used by: Meta videos: allocated, Meta videos: marginal, Short-form video proxy
Open sourcederivedreportedPower gridData center
Google Cloud: Measuring the environmental impact of AI inference
Published 2025
Used by: AI text prompts
Open sourcereportedPower grid
EPA Greenhouse Gas Equivalencies Calculator references
Published 2024
Used by: Smartphone day
Open sourcederivedmodeledreportedPower grid
Lawrence Berkeley National Laboratory: 2024 United States Data Center Energy Usage Report
Published 2024 · Printed p. 57, Figure 5.10 and the following paragraph
Used by: AI text prompts, Dishwasher, Dryer load, Home electricity, Hot shower, Meta videos: allocated, Meta videos: marginal, Phone streaming, Short-form video proxy, Smartphone day, Streaming hour
Open sourcederivedreportedHome tapPower grid
ENERGY STAR: Dishwashers Key Product Criteria
Published 2023
Used by: Dishwasher
Open sourcederivedPower grid
EIA: How much electricity does an American home use?
Published 2022
Used by: Home electricity
Open sourcemodeledPower grid
IEA: The carbon footprint of streaming video
Published 2020
Used by: Phone streaming, Streaming hour
Open sourcederivedSupply chain
Water Footprint Network Product Gallery: Beef
Published 2012
Used by: Beef serving
Open sourcemodeledSupply chain
Water Footprint Network Product Gallery: Coffee
Published 2011
Used by: Coffee
Open sourcemodeledSupply chain
Water Footprint Network Product Gallery: Cotton
Published 2011
Used by: Cotton shirt
Open sourcemodeledSupply chain
Water Footprint Network Product Gallery: Tea
Published 2011
Used by: Tea
Open source