Environment & Energy
How much water do data centers use every day?
Every AI query drinks water - Global data centres consume 12+ billion litres per day to stay cool
Roughly 21K l every second.
litres of water consumed by global data centres today
Source: IEA Data Centres Report 2023; UC Riverside / Li et al. (2023). View on dashboard →
Why does the internet need billions of litres of water?
Data centers need huge amounts of water for cooling. Global estimate: 660 billion liters per year. Google used 29 billion in 2023; Microsoft 6.4 billion. US data centers alone: 64 billion liters for direct cooling. AI workloads generate more heat, so water demand is rising fast. Each ChatGPT conversation uses about 500ml of water for cooling.
Data centre water vs. energy, today
Water and electricity are the twin costs of running the internet. Both are under pressure as AI workloads surge.
What this means for you
Sending 100 emails consumes approximately 50 grams of water in data centre cooling. Streaming an hour of HD video uses around 100ml. A single ChatGPT conversation of 20-50 messages can consume the equivalent of a 500ml bottle of water, according to a 2023 UC Riverside study.
As AI workloads scale, the water intensity of computation is increasing sharply. Microsoft's 2023 sustainability report disclosed that its global water consumption increased 34% year-over-year, driven primarily by AI infrastructure.
Data centres typically draw water from local aquifers or municipal supplies, meaning they directly compete with agriculture and drinking water in drought-stressed regions. Several US states and EU regulators are now requiring data centre operators to disclose water usage and obtain permits, creating new infrastructure constraints for AI expansion.
Water consumption: the key figures
Global data centers consume ~660 billion liters of water annually (2025 estimate)
Google 2023: 29 billion liters of water withdrawn (23B liters evaporated/consumed), a 44% increase from 2022
Microsoft 2022: 6.4 billion liters, a 34% year-on-year increase driven by AI infrastructure expansion
US data centers consumed ~64 billion liters for direct cooling in 2023; projected to double or quadruple by 2028
Training GPT-3 is estimated to have required 700,000 liters of freshwater (UC Riverside, 2023)
Data centers vs. water: the hidden cooling crisis of the AI age
The hidden water footprint of AI
The energy footprint of AI receives significant attention, but the water footprint is less discussed and equally significant. A 2023 study by UC Riverside found that ChatGPT's conversational interface, run at scale, consumes the equivalent of a standard 500ml water bottle per 10-50 responses. Training GPT-3 consumed an estimated 700,000 liters of freshwater. As AI inference scales to billions of daily users, the aggregate water consumption grows proportionately. Microsoft's AI-integrated Office 365 and Azure services drove a 34% jump in its water use in a single year.
Location, location, location
Where a data center is built determines its water stress impact. Many major data centers in the US are located in the desert Southwest (Arizona, Nevada), regions already experiencing severe water scarcity. A data center in a water-stressed region using 15 million liters per day has a much greater environmental impact than the same facility near abundant freshwater sources. Data center operators are increasingly required to disclose water usage data; Microsoft committed to becoming water positive (returning more water to watersheds than it consumes) by 2030.
Data centre water consumption trends: 2020-2025
Data centres rely on evaporative cooling towers that consume 1-5 litres of water per kWh of IT load. Global data centre water consumption reached an estimated 1 trillion litres in 2022, with AI inference queries adding a further estimated 500 ml per 20-50 requests, comparable to a bottle of water for a few minutes of chatbot use.
| Year | Rate (L/s) | GL/day | Context |
|---|---|---|---|
| 2020 | 8K L/s | 0.68 GL | Pre-AI era; steady growth |
| 2022 | 13K L/s | 1.10 GL | AI-driven acceleration; Google +34%, Microsoft +34% |
| 2025 | 21K L/s | 1.81 GL | AI inference at scale; growing demand across all hyperscalers |
| 2028 (forecast) | 48K L/s | 4.11 GL | AI data center expansion; hyperscaler capacity tripling |
When data centre water use became a public issue
- 2012Large hyperscale data centers begin using evaporative cooling at massive scale; water consumption becomes a known issue
- 2020Global data center water consumption estimated at ~250B liters/year; major operators set water efficiency targets
- 2022Microsoft water use +34% in one year; Google: 20B liters; AI workloads begin water demand step-change
- 2023UC Riverside study: ChatGPT uses ~500ml water per conversation; Google: 29B liters; US: 64B liters direct cooling
- 2024Microsoft launches zero-water cooling datacenter design; global estimate reaches ~660B liters/year
- 2030Microsoft commitment: water positive by 2030; projected doubling of data center water demand without efficiency innovation
Industry and academic research on data centre water
| Year | Finding | Value | Source |
|---|---|---|---|
| 2020 | US data centers: ~35 billion liters/year direct cooling water; global ~250B liters estimated | 250.0B global liters/year (2020 est.) | FDM Group |
| 2022 | Google: 20B liters; Microsoft: 6.4B liters (+34%); global estimate ~400B liters; AI workloads begin driving step-change | 400.0B global liters/year (2022 est.) | FDM Group |
| 2023 | Google: 29B liters; US total 64B liters direct cooling; researchers: ChatGPT uses ~500ml per conversation in cooling water | 660.0B global liters/year (2023 est.) | FDM Group |
| 2025 | FDM 2025: ~660 billion liters/year globally; Microsoft launches zero-water cooling datacenter design August 2024 | 660.0B global liters/year (2025) | FDM Group |
| 2028 | Projection: US data center water demand to double or quadruple to 128-256B liters/year by 2028 as AI expands | US projection 2028 | FDM Group |
12 billion litres per day in perspective
660 billion liters per year is equivalent to 264,000 Olympic swimming pools, or enough to supply drinking water to 1.4 billion people for a year at 1.3L/day
Training a single large AI model uses approximately 700,000 liters of water, more than a lifetime of drinking water for an adult
How the number is calculated
660 billion litres/year global estimate ÷ 31,557,600 seconds = approximately 20,912 litres/second. This uses the 2025 estimate from Nature Machine Intelligence (Li et al.) applied to current data centre infrastructure scale. Google's disclosed 29 billion litres in 2023 and Microsoft's 6.4 billion litres serve as cross-checks on total industry estimates. AI-specific water use: each ChatGPT session (20-50 queries) ≈ 500 ml, per University of California Riverside research.
Sources: FDM Group - Environmental Impact of Digitalisation. Methodology →
Frequently asked questions
- How much water do data centers use globally per year?
- A 2025 estimate puts global data center water consumption at approximately 660 billion liters per year. US data centers alone consumed ~64 billion liters in 2023 for direct cooling. Google disclosed 29 billion liters withdrawn in 2023; Microsoft reported 6.4 billion liters in 2022 (a 34% increase year-on-year).
- Why do data centers need so much water?
- Most large data centers use evaporative cooling systems (cooling towers), where water absorbs heat from the building and then evaporates, carrying the heat away. This is highly efficient but consumptive: the water evaporates and is permanently lost. Additionally, thermoelectric power plants that generate electricity for data centers also require cooling water.
- Is AI making the water problem worse?
- Yes. AI workloads require high-density GPU clusters that generate significantly more heat per rack than traditional servers, requiring more aggressive cooling. Researchers at UC Riverside estimated that training GPT-3 consumed 700,000 liters of freshwater. Each ChatGPT conversation is estimated to require cooling water equivalent to a 500ml bottle.
How the data centre water estimate is sourced
The primary academic source is Shaolei Ren's work at UC Riverside and the 2023 Nature Machine Intelligence paper on AI water consumption. Corporate water disclosure comes from Google's Environmental Report 2024, Microsoft's 2023 Sustainability Report, and Amazon's Sustainability report. Global totals are modelled from IEA data centre energy consumption cross-referenced with published water-use effectiveness (WUE) industry benchmarks from the Uptime Institute.
Explore related: Data center energy - E-waste generated - Sand mined for chips, and the live AnythingCounter dashboard.