How Much Water Does a Data Centre Use?

How Much Water Does a Data Centre Use?

Data centres power the internet—from streaming and cloud storage to AI tools—but behind the scenes, they also consume significant amounts of water. 

So how much are we really talking about, and why does it matter?

Why Data Centres Use Water

At their core, data centres are filled with servers that generate heat 24/7. To prevent overheating, operators rely on cooling systems—and many of these use water.

There are two main cooling approaches:

• Air cooling: Uses chilled air (less water, more electricity)

• Liquid cooling: Uses water directly or indirectly (more efficient, but water-intensive)

Water is especially common in evaporative cooling systems, where heat is removed through evaporation—similar to how sweat cools your body.

So, How Much Water Are We Talking About?

The answer varies depending on size, climate, and technology—but here are realistic benchmarks:

A medium-to-large data centre can use 300,000 to 5 million gallons (1–19 million litres) of water per day

That’s roughly equivalent to the daily water use of thousands to tens of thousands of people

For a more relatable example:

A single large facility operated by companies like Google or Microsoft can consume millions of litres daily, especially in hot regions

Even individual user activity adds up:

Training or running large AI models can indirectly consume hundreds of millilitres of water per query, depending on the system and location

Why Location Matters

Water usage depends heavily on where the data centre is located:

• In cooler climates (e.g. Ireland or Finland), less cooling—and therefore less water—is needed

• In hotter, drier regions, water demand increases significantly

This has raised concerns about building data centres in water-stressed areas.

Measuring Water Use: WUE

The industry uses a metric called Water Usage Effectiveness (WUE):

It measures litres of water used per kilowatt-hour (kWh) of energy consumed

Efficient facilities aim for low WUE values

Reducing WUE is now a key sustainability goal across the industry.

Environmental Impact

High water use can strain local supplies, especially during droughts. Key concerns include:

• Competing with communities for drinking water

• Increasing pressure on ecosystems

• Hidden “indirect” water use through electricity generation

As demand for cloud computing and AI grows, so does scrutiny of water consumption.

What Companies Are Doing About It

Major tech firms are investing in ways to reduce water use:

• Google aims to become water positive (replenishing more water than it uses). 

Google’s goal is to replenish around 120% of the water it consumes - a highly unlikely scenario in practical terms using mainly 'offsetting' routes.

• Microsoft is developing closed-loop cooling systems that recycle water

Microsoft, too, aims for the 120% 'water positive' target - by giving funds to projects that put water back into natural systems.

These moves, which are essentially virtue-signalling to the masses tactics and little else, are proof that large multi-national monopolies like Microsoft and Google take no true regard for their impact on the environment. 

To use vast quantities of water and then offer to give funds to organisations to support a 'water positive' project is wholeheartedly without doubt talking loud and saying nothing.

• Amazon is working on water-efficient data centre designs

Although Amazon tout 40% improvement in water use effectiveness since 2021, what this really breaks down to is roughly 0.15 litres of water saved per kWh of energy

All three of these companies say they plan to be 120% water positive by 2030. So, is that even possible let alone likely..?

The short answer: it’s technically possible but practically very challenging. 

Let's break it down more clearly.

What “Water Positive” Really Means

For Amazon (and companies like Google and Microsoft), being “water positive” doesn’t mean each data centre becomes self-sufficient. 

It’s about a global or regional net balance:

• Reduce – use less water in cooling, operations, and processes

• Replenish – fund ecosystem restoration, aquifer recharge, and community water projects

• Offset – ensure total water returned ≥ water consumed

So, they may reduce their already very high water usage slightly, throw money at some projects to look good, and look to offset all the damage they do.

So a water‑stressed region where Amazon draws water might still experience a huge loss of water, but other projects elsewhere “pay it back”.

Viability Challenges

Scale of water use

• A single AWS data centre can use millions of litres per day.

• Replenishment projects must be very large or numerous to exceed this consumption.

Location and type of water

Not all water is equal: recycled, groundwater, surface water, or potable water each have different impacts.

Returning water in one region may not fully alleviate stress in another where it’s taken.

Project effectiveness

Ecosystem restoration or aquifer recharge can take years to show measurable results.

Estimates of water “returned” may be based on models, not precise measurements.

Climate and drought variability

Extreme heat or droughts could increase water use faster than projects can replenish it.

Why It Could Still Work (although highly unlikely)

• Amazon, Google and Microsoft has billions of dollars and operational expertise to deploy large‑scale projects

• Using recycled/non-potable water in cooling already cuts a huge chunk of fresh water consumption

Bottom Line

Technically viable if they hit all efficiency targets, scale projects, and maintain strong water accounting.

Locally debatable – some communities may still experience high water use at individual sites.

Dependent on monitoring – transparency, reporting, and verification will be key.

The Future of Data Centre Water Use

As digital demand accelerates, the industry faces a balancing act:

More data → more servers → more cooling

But also → better efficiency and smarter design

Emerging technologies like AI-optimized cooling and liquid immersion systems could dramatically reduce water dependence in the coming years. 

But do we have years to wait, can we rely on the same companies who create the problem to create the solution, and should any business really be allowed to put our water supplies under so much pressure and future instability..?