Getting started with measuring AI’s carbon footprint | Computer Weekly
Summary
AI workloads are driving exponential growth in compute, power and cooling needs, pushing datacentres toward higher-density racks, liquid cooling and alternative power options such as small modular reactors. The article urges integrating sustainability signals into developer platforms so teams make cost- and carbon-aware decisions at build and deploy time, arguing that GreenOps and FinOps work best as outcomes of product design rather than separate compliance exercises. It highlights practical inefficiencies such as duplicated datasets and forgotten cloud resources and recommends automation to track liveness, archive or delete dormant assets, and rightsize compute and storage. Gartner recommends a component-based measurement approach and full lifecycle accounting of datacentre equipment to accurately calculate AI's carbon footprint and guide mitigation efforts.
