AI operations are always resource intensive, especially when it comes to water and electricity. However, the recent AI boom has increased water consumption by 34% for Microsoft, 22% for Google, and 3% for Meta as of 2022.
In recent years, major technology companies such as Microsoft, Google, and Meta have significantly increased their use of water to cool their data centers, and as advances in generative artificial intelligence (AI) rapidly advance, they are becoming more and more concerned about their environmental impact. There are concerns. Financial Times report.
With millions of users relying on online services, Microsoft, Google, and Meta have seen water consumption skyrocket.
Uncovering and addressing the undisclosed water footprint of AI models, especially given worsening freshwater scarcity, prolonged droughts, and aging water infrastructure, according to a study from the University of California, Riverside published in the journal Nature. This is urgently needed.
This concern is exacerbated as major technology companies race to introduce generative AI-powered products with extensive language models capable of processing vast amounts of data. These models require significant computing power, which drives the construction of large server farms that use chilled water for their cooling systems. Some of it evaporates during the process, but some can be recycled.
Water is a key resource in a variety of energy production, including hydroelectric and thermal power plants, and there has been a notable spike in consumption by tech giants.
In 2022, Microsoft consumed approximately 22 million cubic meters of water. This equates to 22 billion liters of water. That’s enough to fill an Olympic-sized swimming pool more than 8,800 times.
Google, on the other hand, consumed about 6 million to 7 million cubic meters of water, or about 6 billion to 7 billion liters. Meth consumption was modest at 2 billion liters.
Consumption increased further in 2023. These companies have set goals to replenish water resources, aiming to return more water to aquifers and other systems than they consume by 2030 through initiatives such as strengthening irrigation infrastructure and restoring wetland systems. want to be.
Experts predict that demand for AI will drive water withdrawals to unprecedented levels, estimated at 4.2 billion to 6.6 billion cubic meters by 2027, equivalent to almost half of the UK’s annual water consumption. There is.
In a recent lawsuit, residents of West Des Moines, Iowa, raised concerns about a cluster of data centers consuming a significant portion of the district’s water supply. Shaolei Ren, an associate professor at UC Riverside, likens the water consumption of popular chatbots like ChatGPT to having him “drink” a 500 ml bottle every 10 to 50 interactions. We emphasized that AI models can consume large amounts of water.
AI companies are increasingly calling for greater transparency and data disclosure, demanding a detailed breakdown of water consumption across different computing services. Some companies, such as OpenAI, have expressed a commitment to improving efficiency, while others, such as Google, have declined to comment on the issue.
Experts emphasize the need for comprehensive reporting on the environmental impact of AI models, especially amid global concerns about climate change and dwindling water resources. Kate Crawford, research professor at USC Annenberg, emphasizes the importance of understanding the true environmental impact of generative AI tools amid the climate crisis.
