Google's AI cuts data centre power consumption
Google has found it can reduce the cost of cooling in its data centres by 40%, and decrease its data centres' overall energy consumption by 15%, using a machine learning algorithm developed by its artificial intelligence (AI) company, DeepMind.
The power needs of data centres fluctuate depending on multiple factors, such as the weather (more cooling is needed during warmer weather) and demand from users, and Google has succeeded in using machine learning to predict these patterns and make the necessary adjustments to achieve more energy efficiency.
Google says the system will be employed in all of its data centres by the end of 2016, and plans to share details of its system in a White Paper in six weeks' time, offering other companies the option of optimising their power consumption in the same way.
Some analysts estimate that data centres are responsible for 2% of greenhouse gas emissions due to the energy they consume.
Optimising their energy consumption is not the only way in which major tech giants are attempting to reduce their environmental impact. Google and Facebook have voiced commitments to incorporate more renewable energy usage into their data centre structures, while Apple states that all its data centres have been powered by 100% renewable energy sources since 2012.