Artificial intelligence is drying up the earth: how AI has brought third world problems to the first world
8 May 16:01
Artificial intelligence is depleting water resources: data centers spend millions of liters of water to cool servers to process requests. As Bloomberg found out, AI development significantly exacerbates the problem of water shortage, especially in regions where it is already in short supply.
According to the publication, almost two thirds of new data centers built or planned in the US since 2022 are located in regions with high water stress. Five states – Arizona, Texas, California, New Mexico, and Nevada – concentrated 72% of new facilities in such arid zones.
The reason is simple: large tech corporations such as OpenAI, Google, Microsoft, and Amazon need more and more powerful data centers to run AI, which encourages their construction in places with cheap electricity or favorable regulations. However, these areas often suffer from water shortages, which creates competition between companies and local communities for access to water resources.
Traditionally, resource-depleting corporations have been less sensitive to the concerns and interests of local populations in the Third World than in developed countries. We have seen many reports of a company causing environmental disasters in places where resources are being developed in developing countries. So there is a certain irony in the fact that now residents of wealthy and developed American states are facing similar problems.
Читайте нас у Telegram: головні новини коротко
Water consumption is going through the roof
Every day, data centers consume millions of liters of water, mostly for evaporative cooling systems. According to a study by the International Energy Agency (IEA), a typical 100 megawatt data center, which consumes more electricity than 75,000 homes, uses about 2 million liters of water per day. This is the equivalent of the water consumption of about 6,500 households.
Globally, data centers consume 560 billion liters of water annually, and this figure could rise to 1.2 trillion liters by 2030. The situation is particularly alarming in China, India, Saudi Arabia, and the UAE, where more and more centers are appearing in arid areas.
The problem has already led to protests in the Netherlands, Uruguay, and Chile. In Chile, the authorities even temporarily revoked Google’s permission to build the center due to public outrage. In the United States, a wave of outrage is rising in Arizona and Texas, where large AI centers threaten local agriculture and domestic water supply.
What is the solution?
Tech companies are looking for solutions. Microsoft has developed a closed cooling system that circulates water without evaporation and plans to implement it in centers in Wisconsin and Arizona in 2026. OpenAI, through its partners in Texas, also plans to implement such systems, but they consume more electricity than evaporative systems.
Despite new developments, most AI data centers still use evaporative cooling. Amazon, for example, uses direct evaporative cooling, although it tries to minimize the impact by using treated wastewater instead of drinking water.
A separate challenge is non-transparency: large companies do not disclose detailed data on water consumption by each center. In the United States, the city of Dulles sued to prevent the publication of information about Google’s water use, citing trade secrets. Only after a 13-month battle was the data made public.
Experts emphasize that water resources should become one of the key factors when planning the development of AI infrastructure.
“This is a crisis that is only growing and spreading,”
– warns Newsha Ajami from Lawrence Berkeley National Laboratory Center.
Against this backdrop, the statements of IT giants about “water positivity” by 2030 (a commitment to return more water to the environment than is used) look less and less realistic, environmentalists believe. After all, technological progress requiring more and more AI power endangers not only water but also the environmental promises of the companies themselves.
