Saralnama
Google has released a detailed analysis of the environmental impact of AI text queries, estimating that a typical text prompt consumes 0.24 watt-hours of energy, emits 0.03 grams of CO2 equivalent, and uses 0.26 milliliters of water—comparable to nine seconds of TV viewing. Despite the small per-request impact, the overall energy use is significant due to the high volume of AI-powered searches. The company attributes a 33-fold reduction in energy consumption per query over the past year to software optimizations, including a Mixture-of-Experts approach that activates only necessary model parts, compact model versions, efficient data center management, and custom AI accelerators. Google's analysis includes energy for CPUs, AI accelerators, memory, cooling, and hardware production but excludes networking, end-user device energy, and training costs. The report highlights that while AI-related electricity demand is rising, improvements in efficiency and increased renewable energy use have reduced carbon emissions per unit of energy. Google advocates for broader adoption of comprehensive measurement frameworks to improve AI environmental efficiency as capabilities advance. (Updated 23 Aug 2025, 08:00 IST; source: link)