According to a research report by JPMorgan Chase, The high energy and water consumption issues in AI data centers are gradually emerging, especially in the past when they were relatively overlooked.

With the rapid development of AI technology, the demand for water in data centers has sharply increased. It is expected that by 2030, the daily water consumption will reach 450 million gallons, equivalent to 681 Olympic standard swimming pools. This not only poses enormous pressure on areas with scarce water resources, but may also pose challenges to the operation of data centers. Therefore, while promoting the development of AI, it will become crucial to address water resource management issues in data centers.

Energy and water consumption in AI data centers

1. High energy consumption demand: AI data centers require a large amount of electricity to process large amounts of data, which has become an important key driving force for the development of AI.
2. High water consumption demand: According to Bluefield Research data, the total water consumption of global data centers increased by 6% annually from 2017 to 2022. The daily water consumption may reach 450 million gallons by 2030.

Water use similar to carbon emissions

1. Scope classification: Water usage in data centers can be divided into Scope 1, Scope 2, and Scope 3. Scope 1 refers to on-site server cooling water, Scope 2 refers to off-site water for power generation, and Scope 3 refers to server manufacturing supply chain water.
2. Indirect water use in Scope 2: The huge demand for water in the traditional power generation process of thermal power plants makes the water use in Scope 2 of the data center much greater than in Scope 1.

Future challenges

Water resource competition: Data centers in areas with scarce water resources may face intensified competition due to high water consumption demand, leading to tight water supply and even forcing data centers to shut down.