The majority of the growth in water consumption is due to an increase in AI workloads. Data centers use massive cooling systems to prevent overheating of the servers powering AI computations, which require significant amounts of water. While the spike in water use is concerning, the bigger issue remains energy consumption. Both Microsoft and Google are working on ways to make their systems more energy-efficient and reduce their water footprints. Microsoft aims to become "water positive" by 2030, indicating a commitment to not only save water but also return more to the environment than what is taken out.
Key takeaways:
- Generative AI models like ChatGPT consume vast amounts of energy and water, with Microsoft and Google reporting a significant increase in water consumption at their data centers.
- Microsoft's data center water use increased by 34% from 2021 to 2022, consuming more than 1.7 billion gallons of water, while Google reported a 20% spike in its water consumption.
- Most of the growth in water consumption is attributed to an increase in AI workloads. For instance, OpenAI’s GPT-3 model consumed more than 85,000 gallons of water during its training.
- Microsoft and Google are working on ways to make their systems more energy-efficient and reduce their water footprints. Microsoft has also explored creative solutions like locating its data centers in the ocean to use seawater for cooling.