When we talk about data centres and cloud infrastructure, the conversation usually focuses on energy efficiency. But there’s a quieter crisis happening that deserves equal attention: water consumption.

Data centres are thirsty. Cooling thousands of servers requires enormous amounts of water, and as cloud computing grows, so does demand on already-stressed water resources. That’s where WUE comes in—a measurement that’s helping the industry wake up to its water footprint.

What is WUE?

WUE (Water Usage Effectiveness) is a metric that measures how efficiently a data centre uses water relative to the IT equipment it consumes.

Put simply,

WUE = Annual site water usage (litres) / Annual IT equipment energy consumption (kWh)

The result is expressed as litres of water per kilowatt-hour of IT energy consumed.

A WUE of 1.8 L/kWh means the data centre uses 1.8 litres of water for every kilowatt-hour of computing power it delivers. A WUE of 0.5 L/kWh is significantly more efficient.

The lower the WUE, the better. It means you’re doing more computing with less water.

Why This Matters

Think about it this way – your email, your cloud storage, your streaming video—all of it runs on servers. Every time you upload a file or stream a movie, water is being used to keep those servers cool. Most people have no idea this connection even exists.

For data centres, water isn’t just a sustainability concern—it’s becoming a serious business risk. Water scarcity is increasing in many regions. Regulations are tightening. Competition for water resources is intensifying. Companies that ignore their water footprint are building infrastructure on increasingly unstable foundations.

How WUE is Calculated

The calculation sounds simple, but in practice it’s more nuanced:

WUE = Total annual water consumption / Total annual IT equipment energy

Here’s what “total water consumption” includes:

• Cooling tower water – The primary component, used to cool equipment
• Recirculated water – Water recycled through cooling systems multiple times
• Make-up water – Fresh water added to replace evaporated water
• Rainwater harvesting – Sometimes included, sometimes not (varies by methodology)

The challenge – different data centres measure and count water differently. Some include rainwater, some don’t. Some count only fresh water, others count recycled water. This inconsistency makes comparing WUE across different facilities tricky.

A Simple Example

Imagine a data centre with:

IT energy consumption: 87.6 million kWh (87,600 MWh) annually
Annual water consumption: 131.4 million litres

WUE: 131.4 million litres / 87.6 million kWh = 1.5 L/kWh

Why Water Consumption Matters

Water is increasingly precious. Here’s why WUE is becoming critical:

Environmental Impact

• Water-cooled facilities have enormous environmental footprints
• In water-scarce regions, data centres compete with agriculture, drinking water, and ecosystems
• Cooling water often returns to rivers and lakes at higher temperatures, affecting aquatic life

Regional Scarcity

• Some data centre regions face severe water stress
• Droughts are becoming more frequent and severe
• Groundwater reserves are depleting faster than they replenish
• Future expansion of data centres in water-stressed areas may be impossible

Regulatory Pressure

• Governments are implementing stricter water-use regulations
• Environmental compliance costs are rising
• Data centres in water-scarce regions face increasing restrictions
• Companies are facing pressure from customers and investors to reduce water consumption

Business Resilience

• Water scarcity threatens data centre operations
• Water pricing is rising, increasing operational costs
• Reputational risk: consumers increasingly care about environmental impact

Sustainability Commitments

• Major cloud providers have committed to sustainability targets
• Reducing water consumption is essential to meeting net-zero goals
• Investors and customers demand accountability

WUE vs PUE – How They Relate

You’ve probably heard of PUE (Power Usage Effectiveness), the more established metric. Here’s how they differ:

PUE (Power Usage Effectiveness)

• Measures: Energy efficiency
• Formula: Total facility power / IT equipment power
• Focuses on: How efficiently the whole facility uses electricity
• Best case: PUE of 1.0 (impossible in practice, but lower is better)
• Industry benchmark: Uptime Institute’s 2024 survey reported an average annual PUE of approximately 1.56. (meaning over 50% energy goes to cooling, power distribution, etc.)

WUE (Water Usage Effectiveness)

• Measures: Water efficiency
• Formula: Total water used / IT equipment power
• Focuses on: How much water is needed per unit of computing
• Best case: Lower numbers are better, no theoretical minimum
• Industry average: Around 1.8 L/kWh

The Relationship

PUE and WUE are related but independent. A data centre can have:

• Low PUE, high WUE – Energy efficient, but comparatively water intensive.
• High PUE, low WUE – Energy hungry but water-efficient (using dry cooling methods)
• Low PUE, low WUE – The ideal scenario, though rare (efficient cooling technology, located in cool climate)

The real world: most data centres have optimised for PUE (energy) because energy costs are immediate and measurable. Water costs were traditionally cheaper, making WUE less of a priority. That’s changing as water becomes scarcer.

The Limitations of WUE

WUE is useful, but it’s not perfect. Here’s what it doesn’t tell you:

It doesn’t account for water quality

• A facility using recycled water looks the same as one consuming fresh groundwater
• WUE doesn’t distinguish between water sources
• Regional impact varies enormously: using recycled water in a wet region has different implications than using groundwater in a desert

It doesn’t reflect drought conditions

• WUE is an annual average
• A facility using water during drought has greater impact than the same facility in a wet year
• Seasonal variation isn’t captured

It ignores water return quality

• Water used for cooling is often returned warmer, potentially harming ecosystems
• WUE doesn’t measure thermal pollution or chemical additives in returned water

It doesn’t factor in location

• Using 1.8 L/kWh in a water-rich region is very different from the same consumption in a water-stressed area
• WUE doesn’t tell you if you’re in a drought-vulnerable region

It doesn’t consider IT workload type

• Some workloads (AI training, cryptocurrency) use vastly more power than others (web serving)
• Two facilities with identical WUE might have different real-world impact based on workload type

It requires standardised measurement

• Different facilities measure water differently
• Some include rainwater, some don’t
• Comparable benchmarking is difficult

Better approach: WUE works best when combined with context—the water source, regional water stress, facility location, and workload profile all matter.

Cooling Architecture & Water Demand

Here’s where engineering gets real: how you cool your data centre fundamentally determines water consumption.

• Evaporative Cooling — can improve energy efficiency but consumes water through evaporation.
• Air/Dry Cooling — can significantly reduce operational water consumption but may involve energy-efficiency trade-offs depending on climate and design.
• Adiabatic Cooling — combines air cooling with evaporative assistance when conditions require it, so water use varies.
• Liquid Cooling — transfers heat from high-density IT equipment more efficiently, but overall facility water consumption depends on how that captured heat is ultimately rejected.
• Immersion Cooling — removes heat directly from submerged IT equipment using dielectric fluid and eliminates the need for server fans; the wider cooling-loop design determines overall facility water consumption.

The Trade-off

There’s usually a tension: water-cooled systems achieve the best PUE (lowest energy use) but the worst WUE (highest water use). Dry cooling and free cooling improve WUE but often increase energy consumption, raising PUE.

The best facilities don’t choose—they use multiple methods. They might use free cooling in winter, transition to adiabatic cooling in spring, and dry cooling in summer. This hybrid approach balances energy and water efficiency.

Beyond WUE – Water Usage Impact

WUE measures how much water a data centre consumes relative to its IT energy use, but it does not account for local water scarcity. Newer approaches such as Water Usage Impact (WUI) add geographical water stress to the assessment, recognising that consuming water in a water-rich region can have a very different environmental impact from consuming the same amount in a water-stressed location.

The Bottom Line

WUE is forcing data centre operators to consider something they long ignored: water is not infinite.

As cloud computing continues to grow and water becomes scarcer, the industry is shifting from optimising only for energy (PUE) to optimising for both energy and water (PUE + WUE).

What Organisations Should Consider

When evaluating the water impact of cloud and data-centre infrastructure, consider:

• WUE — how much water the facility consumes relative to IT energy.
• Water source — whether fresh, reclaimed or alternative water sources are used.
• Location — particularly local water availability and water stress.
• Cooling architecture — and the balance between energy and water efficiency.
• Transparency — whether meaningful environmental performance data is reported.

Your data has to live somewhere. Understanding how that infrastructure uses both energy and water can help you make a more informed choice.