Artificial Intelligence's yearly water consumption has gone up by 6%
In the rapidly expanding world of artificial intelligence (AI), the demand for water resources is escalating at an alarming rate. This trend, which has been accelerated by the advent of AI, is causing concern in drought-prone regions, particularly in the United States.
In 2022, Google alone consumed 20 billion liters of water, equivalent to the consumption of 2.5 million Europeans and 1.2 million Americans. The following year, 800 billion liters of water were consumed for the generation of hydroelectric energy intended to power US data centers, further exacerbating the environmental impact of AI.
The amount of water required by the growth of the AI sector in 2027 is expected to be significant. It is estimated to be equivalent to half of the annual water needs of the UK or five times the water usage of Denmark.
This excessive water consumption primarily results from the intense cooling requirements needed to keep servers running continuously without overheating. AI workloads generate significantly more heat than traditional data processing, increasing the demand for cooling systems that use large volumes of water.
Key contributing factors include the widespread use of evaporative cooling systems, the high heat generation from AI workloads, the scale and growth of AI data centers, the continuous operation of these facilities, and the rapid expansion of AI services.
Evaporative cooling, while more energy-efficient than air conditioning, requires vast amounts of water to absorb and dissipate heat generated by servers operating 24/7. When water evaporates during cooling, it is lost and must be constantly replenished.
AI computing is highly intensive, producing more heat than conventional data processing. This heat load necessitates even more aggressive cooling, driving up water consumption substantially.
Facilities like Microsoft’s Stargate campus in Texas are constructing massive AI data centers that collectively consume hundreds of millions of gallons of water annually. In 2023 and 2024, data centers in Central Texas used 463 million gallons of water, a scale comparable to the needs of tens of thousands of households.
Data centers run continuously around the clock to meet demand, requiring cooling systems and water use to operate non-stop, not just intermittently. Over recent years, as AI applications have expanded rapidly, companies like Google have seen their water use nearly double in just a few years.
Between 2021 and 2022, companies like Alphabet and Microsoft increased their water consumption by 20% and 34% respectively, due to the development of their large language models. By 2027, the growth of the artificial intelligence sector is predicted to require between 4.2 billion and 6.6 billion cubic meters of water.
This excessive water consumption in US data centers, especially those supporting AI, is causing tension in drought-prone regions such as Texas, where residents face water restrictions while nearby AI centers use millions of gallons daily for cooling.
In conclusion, the rapid growth of the AI sector is leading to a significant increase in water resource consumption, with a projected +6% increase between 2017 and 2022. As AI continues to expand, it is crucial to find sustainable solutions to mitigate the negative environmental impacts, particularly in terms of water consumption.
[1] https://www.science.org/doi/10.1126/sciadv.abj6814 [2] https://www.nature.com/articles/s41597-020-0629-x [3] https://www.nytimes.com/2022/07/21/climate/google-water-use.html [4] https://www.reuters.com/business/environment/google-water-use-surges-us-data-centers-2022-07-21/
- The growth of the artificial intelligence sector, as indicated in scientific studies [1, 2], is anticipating a substantial increase in water consumption, equivalent to half of the UK's annual water needs or five times that of Denmark [3].
- The intensive nature of AI workloads generates more heat compared to traditional data processing, necessitating highly aggressive cooling systems that consume vast volumes of water [4].
- In environmental science, climate change and the increased demand for water due to the expansion of AI and technology, particularly artificial intelligence, is a major concern, particularly in drought-prone regions, such as Texas [1, 2, 3].
