Big Tech's thirst for water is sparking a global resistance (restofworld.org)

The rise of AI technologies, particularly after the launch of ChatGPT, has revealed a significant environmental concern: the immense water usage associated with training and operating large language models and data centers. A study from the University of California highlighted that the training of the original ChatGPT required approximately 185,000 gallons of water, drawing attention to the hidden costs of AI development. Companies like Google and OpenAI downplayed these figures, suggesting that individual queries consume minimal water, while independent experts countered that significant water resources are consumed for cooling large data centers, which can use up to 5 million gallons per day, equivalent to major city usage. This escalating demand for water has ignited resistance from communities around the world, particularly in water-scarce regions like Chile and Uruguay. Local activists have successfully opposed large tech companies’ data center projects on the grounds that such facilities exacerbate existing water shortages. For instance, Google’s proposed data center in Cerrillos was projected to use water equivalent to the yearly consumption of 80,000 residents. Communities argue that these companies prioritize their resource needs at the expense of local populations, highlighting the tension between technological advancement and environmental sustainability—a crucial issue for the AI/ML community as it confronts the broader implications of its operational demands.