Comprehensive review reveals surging water and electricity consumption across AI hardware, highlighting an urgent need for direct liquid cooling and sustainable Green AI practices.
The rapid proliferation of generative artificial intelligence (AI) has transformed the global computing landscape, but this progress comes with a significant environmental price tag. As AI models grow in complexity, their demand for electricity and water primarily for cooling high-density hardware has reached levels that rival entire nations. This review provides a critical quantification of these footprints, highlighting the shift from energy-intensive training to high-volume inference. We demonstrate that current cooling strategies are reaching their physical limits, necessitating a transition to direct liquid cooling and 'Green AI' principles to ensure the long-term sustainability of the AI revolution.
No takes yet. Share an insight, caveat, or question.
Alobid et al. (2026) studied this question.
Synapse has enriched 4 closely related papers on similar clinical questions. Consider them for comparative context: