Data Centers & Water Use

My last post included my first AI-built image. This post is going to discuss the issue of water use by data centers, including those fueling generative AI applications like the one that created my sad pharma-polluted fish.

This is an emerging and hot topic. I’m sure many of you have been in conversations about AI where someone has expressed concern about the environmental impacts of AI. There are absolutely a multitude of environmental impacts, but for several reasons, it’s also a complex and unclear topic without a firm base of factual information anchoring it.

In August 2025, The Alliance for the Great Lakes published a report titled A Finite Resource: Managing the Growing Water Needs in the Great Lakes Region. The report details growing demand for water in the region due to several drivers. One is data centers, but there is also demand growing from critical minerals mining and agriculture. 

I’m going to share several headlines from the section of the report about data centers. If you would like to learn more, I encourage you to read the full report which can be trusted as evidence-based non-partisan information. 

One thing not under debate is that data centers use a lot of water. Direct water use is mostly for cooling the machinery. Because this is most commonly an evaporative process, the water is not recycled or returned to source like the human wastewater cycle. Indirect water use is due to the electricity generation needed to power the data center. However that electricity is generated – coal, natural gas, or nuclear power plants – that process also uses water. 

However, what is not known is how much a lot is. The amount of water used will obviously vary between data centers depending on many variables – size, scale, cooling technology, location and climate. But to give some sense of it, there is data showing that a hyperscale data center of around 10,000 square feet in size with 5,000 servers may use between 1 and 5 million gallons of water per day. If a data center uses 365 million gallons in a year, that would be equivalent to the water use of 12,000 people in that same time. 

However, data centers are not being transparent about their water use because they are not required to by law and because thus far, many states and municipalities have not integrated the issue into their vetting and planning processes. Beyond not being transparent though, there is broad indication that many companies may not even be tracking water usage. According to AGL’s report, potentially less than one-third of data centers are tracking water usage. This obviously leaves an enormous gap in the data and information needed for responsible resource planning.

Demand for data centers is increasing as AI technology scales and expands rapidly. For economic development reasons, Great Lakes states have thus far incentivized the building of data centers. As of July 2025, Illinois had the 4th largest number of data centers in the US and Ohio the 5th largest. Without more data, better planning and increased accountability, this sector may become a liability for the region resulting in water shortages and groundwater conflicts.

Let’s continue to ask questions and try to establish the facts needed for awareness, advocacy and decision-making.