Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Some data centres consume substantial amounts of water, but there is no universal data-centre-to-golf-course ratio. The answer changes with climate, cooling technology, water source, accounting method and the year and region being compared. A national or company-wide total can therefore hide serious local water stress—or make a relatively efficient facility look worse than it is.
The direct answer
Data centres do not all use water in the same way. Facilities with evaporative cooling consume water as it turns into vapor, while air-cooled and closed-loop systems can greatly reduce direct onsite consumption. Electricity production also has a water footprint, but that indirect use occurs at power stations rather than inside the data centre and should be reported separately.
As an Amazon Associate I earn from qualifying purchases.
Golf courses mainly use water for irrigation. Their demand is dominated by turf area, rainfall, temperature, season, irrigation practice and the course’s water source. A course in the arid U.S. Southwest can use many times as much irrigation water as one in the Northeast. That makes “one golf course” an unstable measuring unit.
Free tools Windows power users keep installed
One-click scans. No signup required.
The most defensible conclusion is conditional: a particular data centre may consume more or less water than a particular golf course, and a national data-centre total cannot be fairly ranked against a global golf-course total without matching geography, year and scope.
#1 Best Overall
What “water use” means in this comparison
Withdrawal is not the same as consumption
Withdrawal is water taken from a river, aquifer, utility or other source. Some withdrawn water is discharged back after treatment. Consumption is the portion not returned to the immediate water cycle, often because it evaporates. A withdrawal figure and a consumption figure cannot be compared as if they were the same quantity.
Cooling towers are a common source of consumption. They reject heat by evaporating part of the circulating water; additional water may be discharged as “blowdown” to control mineral concentration and replaced with fresh makeup water. An air-cooled chiller can avoid most evaporative losses but may use more electricity, especially in hot weather.
Direct onsite and indirect electricity-related water
Direct onsite use covers cooling, humidification and related facility operations. The Lawrence Berkeley National Laboratory’s 2024 United States Data Center Energy Usage Report models this category for U.S. data centres.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Indirect use is associated with generating the electricity that powers servers and cooling equipment. Thermal power plants may withdraw and consume water for their own cooling systems. LBNL estimates this category separately and notes that the calculation uses regional grid mixes rather than facility-specific power-purchase agreements or behind-the-meter generation. Adding indirect water to direct water can be useful for a full life-cycle view, but the boundary must be stated.
Key figures, with their boundaries
| Measure | Figure | What it covers | Source and qualification |
|---|---|---|---|
| Google data-centre fleet consumption, 2021 | Approximately 4.3 billion U.S. gallons | Google’s global data-centre fleet; consumption, not a withdrawal total | Google, 2022 company reporting |
| Google’s comparison for that fleet | About 29 southwestern U.S. golf courses | An analogy based on estimated annual irrigation for courses in the U.S. Southwest | Google’s 2022 comparison; not a universal ratio |
| U.S. direct data-centre consumption, 2023 | 66 billion litres | Modeled direct onsite consumption | Lawrence Berkeley National Laboratory, 2024 |
| U.S. direct data-centre consumption, 2014 | 21.2 billion litres | Same modeled category as the 2023 estimate | Lawrence Berkeley National Laboratory, 2024 |
| U.S. indirect electricity-related consumption, 2023 | Nearly 800 billion litres | Water associated with electricity generation, not water delivered to data centres | Lawrence Berkeley National Laboratory, 2024; modeled with regional grid assumptions |
| Average southwestern U.S. golf course | 149 million gallons per year | Irrigation demand | USGA survey-era estimate, 2012 |
| Average northeastern U.S. golf course | 13.8 million gallons per year | Irrigation demand | USGA survey-era estimate, 2012 |
| English courses with direct-abstraction licences | About 10 million litres per day on average; up to 24 million litres per day in summer | Irrigation by a subset of courses holding licences; approximately 820 licences | UK Government, National Framework for Water Resources 2025 |
| England’s estimated course count | About 2,200 courses | All courses, not only those in the licensed-abstraction subset | UK Government, 2025 |
| Google data centres and offices together, 2023 | 6.4 billion gallons | Combined consumption for offices and data centres | Google 2024 Environmental Report; company estimate equivalent to 43 southwestern U.S. golf courses |
These figures are not a single dataset. They combine a company’s operational accounting, a modeled national estimate, historical golf-course surveys and a licensed subset of English courses. Their value is in showing scale and the importance of scope—not in producing a worldwide ranking.
Google’s 2022 comparison used a southwestern estimate of about 459 acre-feet per course per year, or roughly 150 million gallons. Multiplying that regional yardstick by 29 produces a figure close to the reported 4.3 billion gallons. The arithmetic explains the analogy, but it does not make the underlying course estimate representative of every climate.
How much water does a data centre use compared with a golf course?
The U.S. Southwest comparison
Google’s often-quoted comparison uses its global 2021 data-centre consumption and an estimated annual irrigation requirement for southwestern U.S. courses. The analogy treats the southwestern course estimate as a common yardstick, even though Google’s facilities are distributed across many climates and watersheds.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Rank #2
- Ideal for Gifting
- Ideal for a bookworm
- Compact for travelling
Using a southwestern course as the unit makes the result look very different from using a northeastern course. The USGA figures in the table show a more than tenfold regional spread. A reader who sees “29 golf courses” should therefore read it as “29 courses defined by a particular southwestern irrigation estimate,” not as 29 average courses worldwide.
The Northeast illustrates why climate matters
Northeastern courses generally receive more precipitation and have different turf and seasonal conditions. Their irrigation demand in the USGA survey is far below the southwestern estimate. That does not mean every northeastern course is low-water: a drought year, extensive irrigated fairways, high play levels or limited rainfall can raise demand. It means geography must accompany every course figure.
England’s licensed-abstraction data
The UK Government’s 2025 framework reports average and summer irrigation withdrawals for English golf courses that hold direct-abstraction licences. The figures do not represent every course’s total demand, courses using mains water or recycled supplies, or unlicensed sources. They also describe daily rates, so multiplying an average day by 365 would be misleading unless the underlying seasonal pattern and licence conditions were known.
England’s estimate of roughly 2,200 courses provides context for the licensed subset, but it is not evidence that all courses use the same amount. A course’s abstraction licence, storage ponds, rainfall, soil and irrigation schedule determine its actual consumption. The reported 10 million litres per day is an average for the licensed group, while 24 million litres per day is a summer peak; neither is a national total for all English golf.
Why cooling design changes the result
Evaporative cooling
Evaporative systems can be energy-efficient because evaporation removes heat effectively. They usually consume more onsite water than systems that reject heat entirely through air or refrigerant loops. Water quality also matters: hard or mineral-rich supplies can increase blowdown and treatment needs.
Air-cooled systems
Air-cooled chillers and dry coolers use little or no water for heat rejection. Their trade-off is electricity demand. On very hot days, fans and compressors may work harder, increasing operating energy and potentially the facility’s indirect water footprint if the grid relies on water-intensive generation.
Closed-loop liquid cooling
Microsoft describes a newer closed-loop design in which water circulates between servers and chillers after the system is filled during construction. Because the same water is recirculated, the design can avoid ongoing evaporation at facilities that use zero-water cooling.
Microsoft’s December 9, 2024 explanation reported an average fleet water-usage effectiveness (WUE) of 0.30 litres per kilowatt-hour for the prior fiscal year and said next-generation facilities using zero-water evaporation are expected to move WUE toward zero. WUE is a facility-efficiency metric, not a direct statement of total water consumed: a very large site can have a low WUE and still use substantial water. Mechanical cooling can increase power use, however, so “zero water for cooling” does not mean zero environmental impact.
Recommended Free Tools
Google’s reported trade-off
Google says water cooling can reduce energy use by about 10% compared with many air-cooled data centres. That is a company-reported comparison, not a universal performance guarantee. The best design depends on local hydrology, weather, electricity sources, emissions goals and the reliability of reclaimed-water supplies.
“The best approach depends on local factors — there is no one-size-fits-all solution.”
That principle explains why two facilities with similar server loads can have very different water profiles. Cooling selection is normally made during site design, when engineers weigh heat load, climate, water chemistry, available treatment capacity and restrictions on summer withdrawals.
Water source matters as much as volume
A litre of potable water drawn from a stressed aquifer has a different local consequence from a litre of treated wastewater available for reuse. When evaluating a facility, identify whether cooling uses:
- Potable municipal water;
- Fresh surface water or groundwater;
- Reclaimed wastewater;
- Rainwater or stormwater captured onsite; or
- A blended supply that changes by season.
Google reports using reclaimed or non-potable water at more than 25% of its data-centre campuses. It also describes a Douglas County, Georgia project in which treated leftover water is returned to waterways. These are company statements; local permits and watershed conditions determine how much benefit a particular project delivers.
Reclaimed supply does not automatically eliminate impacts. It may require pumping and advanced treatment, and a wastewater stream diverted to cooling may otherwise support a river or downstream user. Conversely, replacing potable withdrawals with reuse can materially reduce pressure on a drinking-water system. The relevant question is what source is displaced, in which season, and where treated discharge goes.
Rank #4
Why national totals can hide local water stress
Water scarcity is experienced in a watershed, not in a national spreadsheet. A facility may represent a small share of national demand while competing with farms, households or ecosystems during a summer shortage. Conversely, a facility in a wet region using reclaimed water may have a smaller local effect than its raw volume suggests.
Assess a proposed or existing site across five dimensions:
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware match- Season: peak cooling demand may coincide with the same dry months when communities restrict irrigation.
- Source reliability: a backup supply can shift pressure to another basin if the primary source is unavailable.
- Return flows: discharged water may be treated and returned, or it may leave the immediate watershed.
- Water quality: treatment, blowdown and thermal discharge affect what can be reused.
- Electricity mix: a low-direct-water design can still have indirect water use through a water-intensive grid.
Permits and operating conditions can matter as much as engineering. A licence may cap abstraction, require reductions during drought, or specify monitoring and reporting. Communities evaluating a project should request peak-day data, not only an annual average, and should check whether a proposed expansion changes the permitted source or discharge.
A fair method for comparing a data centre and a golf course
- Set the geography. Compare facilities and courses in the same watershed or at least the same climate region. Label cross-region comparisons as analogies.
- Match the year. Do not pair a 2023 modeled data-centre estimate with a 2012 golf-course survey without explaining the time gap.
- Use the same accounting unit. Compare consumption with consumption, or withdrawals with withdrawals. Convert gallons and litres transparently if needed.
- Define the boundary. State whether the data-centre number includes offices, humidification, backup systems and electricity-generation water. For a course, state whether the number covers only irrigation or all operational water.
- Separate direct and indirect water. Put onsite cooling in one line and power-sector water in another. Adding them is an analytical choice, not an automatic correction.
- Identify the source. Potable, reclaimed, surface and groundwater supplies have different scarcity and treatment implications.
- Show seasonality. Annual totals can conceal a short period of intense demand. Include summer or drought-period rates when available.
- Report uncertainty. Modeled estimates, company disclosures and licence-based surveys do not have the same precision or coverage.
For a practical site-to-course comparison, first compare annual direct consumption, then compare the highest-demand month or day. Next list each source and return flow, and only then add indirect electricity water if both parties agree to that wider boundary. This sequence prevents a modeled grid estimate from being mistaken for water physically delivered to a data-centre site.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How to read Google’s golf-course analogy
The 2021 comparison is useful as an order-of-magnitude illustration of one company’s global consumption, but it is not a measurement of the average data centre. It uses a regional golf-course irrigation estimate and does not imply that every course or facility has the same annual demand.
Google’s 2023 environmental report uses a different scope: consumption by its data centres and offices together. Its equivalent of 43 southwestern U.S. golf courses should not be relabeled as data-centres-only consumption. Mixing the two years or scopes would create a false trend.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Clear out junk files and repair common Windows errors3Fix the driver behind crashes, sound loss and screen glitchesGoogle also characterizes U.S. data-centre use as less than 1% of the water Americans use on lawns annually. That is the company’s lawn comparison, not evidence that data centres use less water than golf courses.
Best Value
- It can be a gift option
- Comes with secure packaging
- Helpful in various ways
What the U.S. estimate adds
LBNL’s modeled U.S. series shows direct data-centre consumption increasing from 21.2 billion litres in 2014 to 66 billion litres in 2023. The same report’s nearly 800 billion litre indirect estimate demonstrates why the electricity boundary can dominate a water-footprint calculation.
The indirect number should not be interpreted as water flowing through data-centre cooling equipment. It is an estimate of water consumed at electricity-generation facilities, allocated using regional grid assumptions. A data centre with a different power contract, onsite generation or renewable-energy arrangement may have a different indirect profile than the national average.
Questions facility and community decision-makers should ask
- What is the facility’s annual and peak-day direct consumption?
- Are reported figures withdrawals, consumption or both?
- Does the boundary include offices, laboratories and backup-generation testing?
- What percentage of supply is potable, reclaimed or non-potable?
- How does demand change during a drought or heat wave?
- What discharge quality and return-flow arrangements apply?
- Which electricity assumptions were used for indirect water calculations?
- Are efficiency metrics such as WUE reported with their time period and facility boundary?
- For a golf-course comparison, are the course’s irrigated acreage, climate, turf and water source known?
- What local permit limits or seasonal restrictions apply?
Bottom line
There is no scientifically sound global answer that data centres either waste more water than golf courses or use less. The evidence supports narrower conclusions: some cooling systems consume a great deal of water; golf-course demand varies dramatically by region; indirect electricity-related water can exceed direct onsite use; and local scarcity matters more than a national ranking. A credible comparison always states the year, geography, source, accounting boundary and whether it measures withdrawal or consumption.
Frequently Asked Questions
Do data centres use more water than golf courses?
There is no universal answer. Google’s global data-centre fleet consumed about 4.3 billion gallons in 2021, an amount it compared with irrigation for 29 southwestern U.S. golf courses. That analogy uses a regional course estimate and cannot rank every data centre against every course.
What is the difference between data-centre water withdrawal and consumption?
Withdrawal is water taken from a source; consumption is the portion not returned to the immediate water cycle, often through evaporation. A withdrawal total and a consumption total must not be compared as equivalent figures.
Why can a data centre with zero-water cooling still have a water footprint?
Closed-loop or air-cooled systems can eliminate ongoing onsite evaporation, but they may use more electricity. Water can then be consumed at power stations supplying that electricity, so direct and indirect boundaries should be reported separately.
How much water does an average golf course use?
The USGA’s 2012 survey-era figures were about 149 million gallons per year for an average southwestern U.S. course and 13.8 million gallons for an average northeastern course. Climate, irrigated area, turf and rainfall make these regional estimates, not universal averages.
Does Google’s 2023 golf-course comparison cover only data centres?
No. Google’s 2024 Environmental Report says its data centres and offices together consumed 6.4 billion gallons in 2023, which the company equated to irrigation for 43 southwestern U.S. golf courses. It should not be described as a data-centres-only figure.
What should a community check before approving a data centre?
Check annual and peak-day consumption, source type, drought restrictions, permits, return flows, discharge quality and the electricity assumptions behind any indirect-water estimate. Ask whether reported numbers are withdrawals or consumption and whether offices and other buildings are included.
The Bottom Line
Data-centre water use cannot be fairly ranked against golf-course use without matching geography, year, source and accounting boundary. Direct cooling consumption, indirect electricity water and seasonal local scarcity should be shown separately.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
