Google says it dropped the energy cost of AI queries by 33x in one year
a day ago
- #Environmental Impact
- #AI
- #Energy Consumption
- US electricity use is up nearly 4% this year compared to last year, breaking decades of flat usage.
- The rise in electricity demand is linked to the expansion of data centers, many built to support AI growth.
- Increased coal use (up 20% in generation share) is meeting some of this rising demand, raising environmental concerns.
- Assessing AI's environmental impact is challenging due to lack of detailed data on hardware usage and AI query loads.
- Google's new analysis provides rare insights, showing a 33-fold drop in energy per search over the past year.
- Energy estimates for AI are a moving target, complicating environmental impact assessments.
- Analyses must consider not just processor energy but also memory, storage, cooling, manufacturing, and training energy costs.