Connecting the Pieces: AI, Data Centers, and the Environment

Hitika's Substack

SubscribeSign in

Connecting the Pieces: AI, Data Centers, and the Environment<br>Five questions about how data centers run AI, what they consume, and where the realistic paths to sustainability lie.

Hitika<br>Jun 03, 2026

Share

Whenever I tell someone I'm pursuing AI, they question me about its impact on the environment. That led me to write this article, hoping to build more awareness and understanding of how data centers power AI, and how much environmental impact this has.<br>We all understand that AI has a large scale impact, but understanding the thought process behind these data centers can help us understand what actual steps we can take to mitigate the risks. It is often difficult to stop people from using or building something if we do not redirect them to a more sustainable and efficient alternative.<br>Before I proceed further, I know this is a vast and sensitive topic and it requires extensive research. This article is a starting point to think, connect, and correct our understanding of AI and its environmental impact.<br>Now let us dive in. When I think of data centers, I think of the following questions.<br>What is the requirement of water, and what happens if sufficient water is not provided?

Where are the data centers located, and what is the criteria for selecting the land?

Where to run AI: Edge vs Data Center

Regulation around data centers?

How does scaling and inference impact AI performance and data centers?

1. What is the requirement of water, and what happens if sufficient water is not provided?

According to the American Rivers report, data centers use water for two purposes. The first is direct use, which is the water that serves the cooling systems. The second is indirect use, which occurs at the power plants that generate electricity for the data centers, and the amount of water involved varies depending on the type of power plant in the region supplying that electricity. In total, this accounts for 300,000 gallons to five million gallons of water per day , ranging from a small scale to a large scale single data center.

Schematic of a Typical Data Center Evaporative Cooling System, U.S. Department of Energy.<br>Source: Rivers and Data Centers, American Rivers

Being mindful about the quality of water is an important factor as well. A data center can be designed to use 50% less water, but if it inadvertently pollutes a river or overloads a sewage plant, the harm to the community can be far worse than the volume of water consumed, according to a report by Florida Water and Pollution Control.<br>The reason cooling of these data centers is important is because when a facility does not receive enough water, it risks overheating. To prevent equipment failure and service outages, the center is often forced to throttle computing power, shift workloads to other regions, or rely on less efficient, energy intensive air cooling.<br>2. Where are the data centers located, and what is the criteria for selecting the land?

While there is a lot of confidentiality around the data center approval requirements, according to the Procore article, major factors include accessibility to the electric grid. To mitigate this, companies like Amazon have started pursuing data center projects adjacent to existing nuclear power stations, where electricity would be purchased directly from the plant operator.<br>Zoning laws also play an important role, and different states have different laws for sanctioning land. On one hand, local politicians and economic development officials in a growing number of communities have data centers on their radar. They are aware of the tax revenue these facilities can generate, and in many cases are working proactively to attract development by creating tax incentives. On the other hand, many data center hotbeds that once rolled out the red carpet for developers have become increasingly hostile towards the industry, particularly as data centers encroach on residential and rural communities.<br>Its accessibility to water is another factor, since the cooling systems require steady and substantial water supplies.<br>Transport and accessibility is another factor. According to ESRI, while data centers do not generate high levels of daily traffic, they do require reliable access for construction, maintenance crews, and emergency services, in addition to the workforce. Planners must evaluate roadway capacity and connectivity to ensure access without disrupting surrounding land uses.<br>Data centers are typically located in major technological hubs, such as Northern Virginia (”Data Center Alley”), Phoenix, Dallas, Chicago, and Portland, as well as near heavy fiber routes and intercontinental subsea cable landing stations. [1, 2]<br>3. Where to run AI: Edge vs Data Center

We are able to access data in some form or another, whether through a simple search engine, a transaction, or even storing someone’s contact information. We are becoming...