As digital and virtual as the world of AI may seem from a distance, on the ground it is just as—or perhaps even more—dependent on natural resources. Did you know that it takes a lot of water to cool a data center? A recent study sheds light on this issue, detailing how much water a medium-sized AI data center uses in a year, and how many gallons of water a large data center uses annually compared to a medium-sized data center. Data from the Environmental and Energy Study Institute has revealed a shocking truth: Data centers are using a lot of water.
Data centers use so much water in a year
According to the Environmental and Energy Study Institute (EESI), a medium-sized data center can use approximately 110 million gallons of water annually for cooling purposes—an amount equivalent to the annual water use of approximately 1,000 homes. Large data centers can use up to 5 million gallons of water per day, or about 1.8 billion gallons annually—an amount equivalent to the water needs of a city with a population of 10,000 to 50,000 people.
Why do data centers need water?
Training and running any AI model requires graphics cards (GPUs) and thousands of powerful servers. These servers generate so much heat that if they are not kept cool, they are at risk of overheating and damage. To maintain the temperature of these spaces, a method called “evaporative cooling” technology is used. This technology depends on the use of water. The fact that AI consumes so much water each year raises serious questions about sustainability—especially in areas that already struggle with water scarcity.