Tag
The article examines the water crisis driven by AI data centers, noting that cooling concentrates thermal load in single watersheds, and proposes distributed compute over idle hardware as a potential solution to spread water and energy demand.
An exploration of the water consumption associated with training and running AI models, often overlooked in discussions of AI's environmental footprint.