Increasing water consumption of AI for cooling data centers raises environmental concerns, prompting tech giants to seek sustainable solutions.
Artificial intelligence (AI) isn’t just a power-hungry technology; it’s also consuming vast amounts of water. Initially, public attention focused on the electrical power required to run AI algorithms. However, as these algorithms become increasingly complex, data centers require more water to cool the servers that make AI functionalities possible.
The Water Footprint of AI
When users interact with chatbots like ChatGPT, data centers require around 500 milliliters of water for every 5 to 50 user prompts. This is significant, especially considering that chatbots have become some of the fastest-growing websites globally. Major tech giants like Microsoft and Google admit they’re making efforts to mitigate the environmental impact. But the current rate of water consumption remains a red flag for environmentalists and industry experts alike.
In just a year, from 2021 to 2022, Microsoft saw a 34% increase in water consumption. Google wasn’t far behind with a 20% growth. This equates to an alarming extra 1.7 billion gallons of water consumed between the two companies. The water comes from sources near their data centers. For instance, Microsoft’s ChatGPT operates from data centers situated next to cornfields in western Des Moines, Iowa, utilizing water from local rivers like the Raccoon and Des Moines.
Local Impact and Unintended Consequences
Residents in areas where these data centers are located feel the impact, often without realizing the cause. For example, the water rates at Des Moines Water Works have soared almost 80% since 2007. The growing thirst of AI technology and data centers contribute to this surge, highlighting the technology’s unintended environmental impact.
Shaolei Ren, a researcher, emphasizes the need for awareness regarding the resources AI systems consume.
Without proper understanding and transparency, conserving these valuable resources becomes a far-reaching challenge. Tech companies are working to address the issues. Microsoft aims to switch to 100% carbon-free energy by 2030, focusing on sustainable practices outlined in their 2022 sustainability report. Similarly, Google aims to achieve net-zero emissions across its entire operations and value chain by the end of the decade.
Comparable Thirst in Cryptocurrency Mining
While not directly related to AI, the environmental toll of cryptocurrency mining serves as an eye-opening comparison. A study by Cambridge University in 2021 revealed that Bitcoin mining annually uses energy equivalent to Argentina’s entire electricity consumption. Tesla even halted its support for Bitcoin due to environmental concerns, setting off a series of discussions on the crypto industry’s sustainability worldwide.
While tech giants strive to enhance their data centers’ energy efficiency, the key to a sustainable future for AI lies in proactive management. Companies must invest in renewable energy sources, innovative cooling techniques, and transparent reporting on their resource usage. Google has already noted that due to efficient practices, the rise in AI’s energy demands is slower than initially predicted.
As AI technology continues to advance, its resource requirements, particularly water, cannot be ignored. To ensure the long-term sustainability of AI and its broader societal benefits, it’s crucial that both the industry and consumers understand the environmental costs involved and take steps to mitigate them.