United StatesTechnology and telecom2017 – ongoingAlphabet (Google)
Autonomous neural-network control of data-centre cooling
Google reported around 30% average cooling-energy savings after AI began controlling equipment directly
Problem
Google wanted to reduce cooling energy in already optimized data centres. Its earlier AI system recommended actions that operators had to review and implement.
Approach
DeepMind and Google put neural networks in direct control of cooling at multiple data centres. The system evaluated sensor readings every 5 minutes and selected actions within safety constraints. Local controls checked each action before execution, while operators retained supervision.
Results
Takeaway
Direct control retained local checks and operator oversight. The savings concerned cooling efficiency relative to historical performance, rather than total data-centre electricity use.
Categories
- Size
- Large enterprise
- Industry
- Technology and telecom
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and MLOptimization
- Adoption
- Built in-house
- Goal
- Cost savings
- Evidence
- Company's own claim
Sources
Source summaries
DeepMind described autonomous cooling control operating in multiple Google data centres. It reported average savings of around 30% and showed performance tracking beginning in 2017.
Full details are in the original sources.
