Global water withdrawals are nearly flat around 3,850–3,900 billion m³ and agriculture takes 71%; data centers consume under 0.05% of that scale, but locally it can matter

Asked:

“I’m worried about water usage. Show the major water usages over the last 10 years and projected through next 10. How do data centers factor in?”

World Bank/FAO annual freshwater withdrawals for the world, by sector, observed 2015–2022 (the latest reported global years), extended 2023–2036 as a transparent baseline scenario at the UN-reported ~1% annual demand growth with 2022 sector shares held fixed. Units are billion m³ per year; 1 billion m³ = 1 trillion litres.

Observed withdrawals 2015–2022 and a 1%-growth baseline scenario to 2036

Solid areas: observed World Bank data. Hatched, lighter areas right of the 2022 line: scenario, not observation. Hover for exact values.

Withdrawals vs consumption — not the same. The chart above shows withdrawals: water taken from rivers, lakes and aquifers, much of it returned after use. Data-center figures below are consumption: water evaporated or otherwise not promptly returned. The comparison is for scale only, never a sector share.

Data centers on their own magnified axis — consumption, in billion m³

IEA estimates of global data-center water consumption. At the withdrawal chart's scale these bars would be invisible: ~0.01% of global withdrawals today, ~0.03% around 2030.

Three water footprints of a data center

1 · Onsite cooling (direct)

Water evaporated in cooling towers or used in evaporative systems at the facility.

65.9

billion litres, U.S. onsite, 2023 (17.4 billion gallons); 38–73 billion gallons projected 2028

2 · Electricity generation (indirect)

Water consumed by the power plants that supply the facility — often the dominant share.

799

billion litres, U.S. indirect, 2023 (211 billion gallons) — about 12× onsite use

3 · Supply chain

Water embedded in chips and hardware manufacturing. Rarely quantified; excluded from the totals above because boundaries and models differ.

—

not directly comparable; do not add across scopes

What decides local risk, and what works

Risk drivers. Siting in dry basins, hot climates driving evaporative cooling, competition with municipal supply, and a water-intensive power mix can make the local burden far larger than the ~0.01% global figure suggests.

Levers. Cooling design (closed-loop and liquid cooling), climate-aware siting, recycled or non-potable water, and a cleaner power mix — because electricity-related water often dominates the footprint.

Global withdrawals were essentially flat: 3,851 billion m³ in 2015 and 3,853 billion m³ in 2022 — api.worldbank.org

Sector shares shifted slightly: agriculture 71.5%→71.0%, industry 16.5%→15.7%, domestic 12.0%→13.3% between 2015 and 2022 — api.worldbank.org

UNESCO's 2024 report independently corroborates the split — roughly 70% agriculture, just under 20% industry, about 12% domestic — and reports demand growth of just under 1% per year — unesdoc.unesco.org

IEA puts global data-center water consumption around 560 billion litres/year now, rising toward 1.2 trillion litres/year by 2030 — small globally, fast-growing locally — msci.com

U.S. data centers directly consumed about 17.4 billion gallons onsite in 2023, but an estimated 211 billion gallons of electricity-related water consumption was an order of magnitude larger — theglobalstatistics.com

The rows behind the chart

Year	Agriculture	Industry	Domestic	Total (bn m³)	Status	Source

Method: annual freshwater withdrawals for the World aggregate from four World Bank API series (total in billion m³; agriculture, industry, domestic shares in %), last updated 2026-07-13; global observations after 2022 are null, so the observed window is 2015–2022. Scenario 2023–2036 compounds the 2022 total at 1%/year (UNESCO WWDR 2024 growth rate) with 2022 shares held fixed — a baseline scenario, not a forecast. Data-center figures are water consumption (IEA global; LBNL U.S.) and are not added to withdrawal totals because metric and boundaries differ. 1 billion m³ = 1 trillion litres. Corroborating sets: 698 UN/OECD/WRI extraction rows, 931 data-center evidence rows, 250 shortlist rows; only headline official estimates shown for space.

This report was generated automatically by Keenable SELECT at a user's request, from publicly available web sources linked herein. Keenable does not review, verify, or endorse its contents and makes no representation as to accuracy, completeness, or timeliness; AI-based extraction may contain errors. Nothing in this report is investment, legal, financial, or other professional advice. All trademarks and referenced content remain the property of their respective owners; no affiliation or endorsement is implied. To report an error, rights concern, or request removal: legal@keenable.ai.