The Environmental Fears Surrounding AI Data Centers Are Misplaced

The construction of artificial-intelligence data centers in America has sparked concerns over their environmental impact. However, these worries are largely unfounded and based on misinformation. According to AI researcher Andy Masley, by 2030, the total amount of land occupied by data centers will be roughly the size of Rhode Island. This may seem like a significant area, but it's comparable to other uses that provoke little public outcry.

For example, government-subsidized corn farming for ethanol production occupies an area roughly the size of New York State. Meanwhile, the U.S. Department of Agriculture pays farmers to leave farmland the size of Kentucky unplanted, yet there is no public outrage about these practices. The land-use argument against data centers is thus wildly out of proportion.

Regarding water use, critics argue that data centers consume large amounts of water for cooling. However, most data centers rely on closed-loop cooling systems, which recycle water and reduce the strain on local resources. While this can affect areas with water scarcity or drought, responsibility for mitigating these risks falls on local governments and their negotiations with data-center companies.

In contrast to the panic over AI water use, suburban sprinklers across the country use more water in two days than all U.S. data centers use in a year. Newer cooling systems can bring on-site water use close to zero, making the concerns about water usage even weaker.