When it comes to reducing water and energy footprints, many organizations see only obstacles. At Ecolab, we see only opportunities.

The AI revolution has increased energy and cooling demand on an unprecedented scale. This is driving a migration in data center cooling architectures to mitigate the heat generated by today’s advanced-computing applications and chips. 

To put demand into perspective, a typical query answered by a leading generative AI system requires 2.9 watt-hours (Wh) of electricity, or roughly 10 times the amount of electricity consumed for the standard search engine query¹. When multiplied across millions of daily queries, this consumption grows substantially. Current forecasts indicate that computing infrastructure for advanced large-language models could, by 2030, require 327 gigawatts (GW) of electricity² to keep up with demand. That’s approximately 70% of the total electricity used in the United States in 2024.³

To stay competitive now and in the future, data center providers must invest in state-of-the-art cooling infrastructures and digital monitoring that allows them to conserve resources, optimize performance, and quickly adapt to changes in demand.

Energy requires water

Today, power is the number-one source limiting data-center growth. Depending on geography and computing demand, increased water requirements can also strain local watersheds. Conserving power and water is paramount for data centers to gain a competitive edge and improve operational efficiency.

In an effort to reduce their water footprint, some organizations have adopted waterless cooling methods. Without a comprehensive analysis, waterless cooling methods can overlook two fundamentals that may actually increase the hydro footprint of the data center: 1) water is a much more efficient means of cooling than air and requires less energy; and 2) energy production itself requires the use of water.

For example, it takes 570–1,100 liters of water⁴ to create 1 megawatt-hour (MWh) of electricity by burning natural gas, creating a significant water footprint to start. Depending on the local climate and the design of the liquid-cooled system, water-cooled data centers typically use 10–30% less energy⁵ than air-cooled chiller applications. This is due to the thermodynamics of water itself: water is denser than air, giving it nearly 3,500x the heat-carrying capacity of air and transferring heat 23.5x faster⁶.

Even with water’s superior cooling efficiency compared to air, organizations can ensure the use of water in their systems is strategic, using water where it makes sense and only as much as needed.

Infrastructure and digital monitoring

The process begins with site selection: evaluating climate, water availability, source of power, and local watershed impacts, because those factors materially influence the most sustainable cooling choices. To meet the cooling demands of digital infrastructure, data-center cooling topology must then be designed with versatility and flexibility in mind, not only for today but for the future. This planning relies on a holistic approach, leveraging a combination of cooling methods from site to chip. When direct-to-chip cooling is in place, for example, the load for facility cooling systems (e.g., cooling towers and chillers) can be eased, helping data centers reduce water and power consumption. 

Once the system topology is up and running, it needs to be consistently monitored and managed to ensure each dynamic system within the cooling infrastructure continues to perform at the high levels for which they're designed.

Consider that, on average, the optimized water usage of a mid-sized, 100-MW data center is less than the annual water consumption of an 18-hole golf course⁷. 

Using the same analogy, demand for computing power and cooling requirements can wax and wane the same way that weather conditions can affect the amount of water needed to maintain a golf course. But data-center water consumption can fluctuate hourly, regardless of weather, driven by factors such as cooling load, water quality, equipment efficiency, ambient temperature, and humidity. Adapting to these demand fluctuations is the key to maximizing system efficiency.

Currently, the pace of data-center development exceeds the capacity of any one developer to think through every cooling system decision from design to operation to maintenance. To stay competitive, data centers can benefit from a supplier that plans and manages cooling from top to bottom, leveraging performance data to identify areas for improvement and to help mitigate risks and challenges. This not only takes the strain off individual site operators but has measurable benefits to both the environment and the bottom line.

Strategic approach to water usage in data centers

The most efficient cooling systems require a forward-thinking, holistic design approach that contextualizes the local community, power grid, and watershed to inform decisions and maximize resource efficiency and sustainability.

Such an approach begins long before a data center is built. Site selection is critical when it comes to optimizing free cooling as well as sourcing reliable water sources. In the facility design stage, efficient cooling designs (including adiabatic and direct-to-chip liquid cooling) can ease a data center’s demand for electricity.

Once a data center is up and running, water reuse and recycle projects can be implemented to reduce the need for potable water. Digital monitoring can further support water reduction opportunities by helping operators detect and address cooling system upsets before they escalate to become big problems, and by enabling those systems to adapt to fluctuations in demand.