Why the digital world has a thirst problem

A single Google search requires half a millilitre of water, while ChatGPT consumes 500 millilitres for every five to 50 prompts. Multiply these figures by billions of daily interactions, add streaming services and cloud storage, and the result is staggering: data centre water consumption has reached crisis levels, with some facilities using more water annually than entire cities. 

This hidden cost of our digital addiction is now triggering protests from Uruguay to Holland, as communities fight tech giants for access to their most precious resource. Data centres are essentially the backbone of our digital infrastructure – massive warehouse-like facilities packed with thousands of servers that store, process, and transmit the data we use every day. 

When you stream a film on Netflix, access files on Dropbox, shop on Amazon, or video call via Zoom, you’re relying on these facilities to deliver that service seamlessly. Major tech companies operate enormous data centres worldwide. Google runs facilities that power Gmail, YouTube, and Google Drive. 

Microsoft’s Azure cloud services depend on data centres spanning multiple continents. Meta (formerly Facebook) operates facilities supporting Instagram, WhatsApp, and Facebook itself. Even seemingly simple services like online banking, weather apps, or GPS navigation all rely on these digital powerhouses working around the clock.

The scale of these operations is remarkable. Google operates data centres across 24 regions globally, from Virginia to Singapore, each containing thousands of servers. Microsoft runs over 200 data centres worldwide, whilst Amazon Web Services operates facilities in 84 availability zones across 26 geographic regions. 

These facilities range from modest buildings housing hundreds of servers to hyperscale complexes spanning millions of square feet.

Why data centres need cooling

The fundamental challenge with data centres lies in heat generation. Thousands of servers running continuously 24/7 produce enormous amounts of heat – similar to having multiple industrial ovens operating simultaneously. 

Without proper cooling, this equipment would overheat within minutes, causing system failures and potentially destroying expensive hardware worth millions of pounds. According to the World Economic Forum, even a small-scale data centre can have a substantial thermal footprint. 

The computational demands of modern applications, particularly artificial intelligence and machine learning have intensified this cooling requirement significantly. Advanced AI models like GPT-3 require immense computational resources, which directly translates to increased heat generation and, consequently, greater cooling demands.

Data centres traditionally employ two primary cooling methods: air cooling and water cooling. Air-cooled systems use fans and air conditioning units to manage temperature, but they’re less efficient for high-density computing environments. Water-based cooling proves far more effective at removing heat, which explains why hyperscale operators increasingly favour this approach despite its substantial water requirements.

The scale of data centre water usage

Data centre water consumption occurs primarily through cooling systems, which include cooling towers, chillers, and liquid cooling systems. Water proves remarkably efficient at absorbing and dissipating heat compared to air-based alternatives, but the volumes required are staggering.

According to industry data, a 1-megawatt data centre can consume up to 25.5 million litres of water annually just for cooling – equivalent to the daily water consumption of approximately 300,000 people. To put this in perspective, a medium-sized 15-megawatt data centre consumes as much water annually as either three average-sized hospitals or more than two 18-hole golf courses.

Hyperscale facilities operated by companies like Google demonstrate the true scale of the challenge. Google’s data centres average 550,000 gallons (2.1 million litres) daily, totalling approximately 200 million gallons (760 million litres) annually per facility. Microsoft’s global operations consumed nearly 6.4 million cubic metres of water (approximately 1.69 billion gallons) in their most recent reporting year – a 34% increase from the previous year.

The process itself involves multiple stages where water is lost. In typical chilled water systems, water is cooled in central chillers, and then circulated through cooling coils that absorb heat from data centre air. 

The heated water then passes to cooling towers where it interacts with outside air, allowing heat to escape. During this evaporative cooling process, significant amounts of water are permanently lost to the atmosphere.

Amazon Web Services employs direct evaporative cooling systems where hot outside air is pulled through water-soaked cooling pads. The water evaporates, reducing air temperature before it enters server rooms. Whilst efficient, this method results in substantial water consumption, with AWS reporting a global Water Usage Effectiveness (WUE) metric of 0.19 litres per kilowatt-hour.

The AI water crisis

The AI boom has dramatically exacerbated data centre water consumption. In a World Economic Forum article titled”Why circular water solutions are key to sustainable data centres“, Wesley Spindler, Managing Director of Global Sustainability Leadership at Accenture, notes that GPT-3 consumes an estimated 500ml of water per 10-50 responses. When multiplied across billions of users globally, the total water footprint becomes enormous.

In the same article, Luna Atamian Hahn-Petersen, Senior Manager of Sustainability Strategy at Accenture, points out that AI models require immense computational power for training complex data models. 

When energy is used at those levels, water becomes essential for cooling the machines processing AI workloads. By 2027, global AI demand is expected to account for 1.1 to 1.7 trillion gallons of water withdrawal – more than four to six times Denmark’s total annual water consumption.

This represents a fundamental shift in data centre water consumption patterns. Traditional computing workloads generated predictable cooling demands, but AI training involves intensive computational bursts that can dramatically spike cooling requirements within individual facilities.

Where data centres source their water

Data centres primarily obtain water from municipal or regional water utility companies. For cooling purposes, they mainly use potable water suitable for drinking, though some operators are transitioning to alternative sources. 

Google employs reclaimed or non-potable water in over 25% of its data centre campuses, whilst alternative water sources typically contribute less than 5% of the total supply across the industry.

These alternative sources include on-site groundwater, surface water, seawater, produced water from oil and gas extraction, and rainwater harvesting systems. 

However, regulatory restrictions and treatment costs often limit their viability. Meta Platforms reports that over 99% of their water withdrawal comes from third-party municipal supplies, with less than 1% from groundwater sources.

Water reuse and treatment challenges

Data centres do attempt to reuse water through circulation within cooling systems. Google reports this method can save up to 50% compared to traditional “once-through” systems. However, water reuse faces significant limitations due to scale formation and conductivity issues.

During evaporative cooling, scale-forming minerals such as calcium, magnesium, and silica become increasingly concentrated. Eventually, this necessitates water replacement to prevent equipment damage. Additionally, wastewater often becomes contaminated with dust, chemicals, and minerals, hampering cooling efficiency if recirculated without treatment.

Some facilities employ stormwater retention ponds to collect rainwater for treatment and reuse in cooling systems. However, effective water treatment requires substantial infrastructure investment and ongoing operational costs that many operators find prohibitive.

Why it’s contentious

The controversy surrounding data centre water consumption stems from multiple factors, particularly competition for scarce water resources in drought-prone regions. According to the United Nations, by 2025, 50% of the world’spopulation is projected to live in water-stressed areas, making data centre water usage a critical environmental priority.

The socio-economic implications are significant. When data centres increase reliance on local water supplies, farmers face reduced irrigation access, leading to lower crop yields, whilst water prices often increase for residents. The socio-economic well-being of regions becomes imperilled when data centres compete with essential human needs for scarce water resources.

Real-world conflicts have already emerged globally. In early 2023, plans for a large hyperscale data centre in Uruguay sparked substantial protests. Residents, already suffering from severe drought conditions, opposed the development, fearing it would further threaten their limited access to safe drinking water and worsen agricultural losses. 

Similar tensions have arisen in Holland, Chile, and other water-stressed regions where data centre developments compete with local water needs. The measurement challenge compounds the controversy. Less than a third of data centre operators actively track water usage metrics, according to industry research. 

This lack of transparency significantly undermines efforts to understand the full environmental impact, especially in water-stressed regions. While Water Usage Effectiveness (WUE) was introduced as a metric – similar to Power Usage Effectiveness (PUE) for energy – it only accounts for on-site water use. 

This ignores substantial indirect consumption from electricity generation, which often relies on water-intensive processes like steam production in thermoelectric power plants. By focusing solely on direct usage, operators fail to capture their true water footprint.

Industry response and solutions

Environmental advocates argue that data centre water consumption represents a hidden environmental cost of our digital lifestyle. Unlike carbon emissions, which receive significant attention, water usage remains largely invisible to consumers who stream videos or store files in the cloud.

However, the industry is responding with ambitious commitments. Major operators including Amazon Web Services, Microsoft, Google, and Meta have pledged to become “water positive” by 2030, meaning they’ll replenish more water than they consume. 

Amazon aims to replenish 3.9 billion litres annually through water restoration projects, whilst Microsoft has committed to reducing water used in evaporative-cooled data centres globally by 95% by 2024.

Companies are investing in circular water solutions, including closed-loop cooling systems, wastewater recycling, and rainwater harvesting, which can reduce freshwater use by 50-70% when implemented. Microsoft is leveraging adiabatic cooling methods that use outside air instead of water when temperatures fall below 29.4 degrees Celsius.

Advanced cooling technologies offer additional promise. Liquid cooling systems, which use liquid coolant to efficiently dissipate heat directly from components, provide more efficient heat management compared to traditional air-cooling methods. However, these technologies require significant capital investment and technical expertise.

The integration of circular water management principles represents a critical step towards ensuring data centres remain capable of supporting technological advancement whilst minimising environmental impact. 

As Sadaf Hosseini, Head of Growth, Partnerships and Innovation Ecosystems at UpLink notes in the World Economic Forum’s article, incorporating these solutions into standard operations helps mitigate environmental impacts whilst supporting long-term operational efficiency.

The tension between our growing digital demands and finite water resources represents a critical challenge requiring urgent attention. Every cloud upload, AI query, and streaming session now carries an invisible water cost that communities worldwide are beginning to feel acutely. 

As governments grapple with water scarcity and climate change intensifies drought conditions, the tech industry faces a stark choice: innovate towards truly sustainable cooling solutions or risk becoming the villain in water-stressed regions globally. 

The next decade will determine whether our insatiable appetite for digital convenience can be reconciled with the fundamental human need for clean water – or whether the cloud’s hidden thirst will force us to choose between technological progress and environmental survival.

(Photo by Taylor Vick)

Want to learn more about cybersecurity and the cloud from industry leaders? Check out Cyber Security & Cloud Expo taking place in Amsterdam, California, and London.

Explore other upcoming enterprise technology events and webinars powered by TechForge here.