Artificial intelligence (AI) data centers in the US are placing increasing strain on water security across the US, according to a recent report from Bloomberg.

The report found that close to two-thirds of new data centers constructed or in development since 2022 have been in locations with high levels of water stress. This is especially concentrated in five states, which currently account for 72 percent of all new data centers in locations of acute water stress.

The five states pinpointed by the report are: California, with 17 data centers, Arizona with 26, Texas also with 26, Illinois with 23, and Virginia with 67.

“Every part of the state is facing this water-energy nexus crisis,” Amy Bush, a hydrologist for RMBJ Geo Inc, told Bloomberg, while reflecting on the Texan market. Texas in particular has seen a huge increase in planned AI data centers, due to its significant energy reserves and lack of state taxes.

The report highlights that this is not a trend unique to the US, with other regions also experiencing a similar move. Notably, arid regions such as Saudi Arabia and the UAE, which suffer from severe water scarcity due to their desert climate, have seen a significant uptick in data center developments.

Bloomberg highlights that there is a correlation between regions with ample energy resources and favorable regulations with a lack of available water, which has the effect of placing significant stress upon local supplies crucial for other sectors such as agriculture.

The report argues that water has become one of the last considerations for data center operators when they decide where to site their operations, with the cost of real estate and power being the major drivers. However, with water scarcity becoming an increasing issue, this may begin to change.

An April report from the International Energy Agency (IEA) revealed that in the US an average 100MW data center also consumes around two million liters of water every day. Globally, the IEA report estimates that the sector consumes more than 560 billion liters of water annually, with projections as high as 1,200 billion liters by 2030.

Concerns over water use have led to a number of data center developments facing opposition from local communities. Last September, Google was forced to delay a data center in Chile after an environmental court ordered the company to revise its application, amid local pushback and water concerns over its use of the capital's strained aquifer.

The major factor behind this huge water use is the reliance on evaporative cooling, which results in around 80 percent of the water used in the cooling process being evaporated, with the remaining 20 percent sent to a wastewater facility.

The majority of the major hyperscalers have committed to reducing the use of water as part of their data center operations. This has taken the form of both the deployment of less water-intensive cooling technologies and water stewardship programs. For example, in March, Google signed four partnerships aimed at supporting water stewardship in Chile, California, Taiwan, and France.

Next generation cooling technologies such as cold plates, one-phase immersion, and two-phase immersion have also been touted as potential solutions for data center water use. In a recent study by Microsoft researchers, it was found that cold plates and two forms of immersion cooling can water usage by 31 to 52 percent over their entire life cycles when compared to traditional air cooling in data centers.

In addition, data center firms have begun to explore more novel solutions to water use, including rainwater harvesting and reuse, river and geothermal cooling, and chilled water systems.