Yale Climate Connections published an analysis examining how bad AI is for the environment, covering carbon emissions, water consumption, and energy draw from data centers supporting AI workloads. The piece aggregates documented figures on AI's resource footprint rather than relying on projections. Specific metrics cited include water usage per query and per-training-run emissions estimates from major AI models.
Published environmental accounting of AI's resource costs from an academic-adjacent outlet adds to a growing body of evidence that regulators, investors, and utilities are citing in policy and procurement decisions. Documented figures on water and carbon give concrete benchmarks to communities and agencies evaluating new data center permits.
Yale Climate Connections' aggregation of specific environmental metrics for AI and data centers, covering emissions and water consumption, triggered selection. This is distinct from the previously published Yale heat output study and the pollution study, offering a broader multi-factor environmental accounting.