Log In   |   Sign up

New User Registration

Article / Abstract Submission
Register here
Press Release Submission
Register here
coolingZONE Supplier
Register here

Existing User

            Forgot your password
John O | August 2018

Google tuns data center cooling over to AI to reduce energy usage by 40 percent

By Josh Perry, Editor


DeepMind Technologies, an artificial intelligence (AI) company based in London (U.K.) that was purchased by Google in 2014, recently announced that it is applying its machine learning algorithms to Google data centers to reduce the amount of energy dedicated to cooling by as much as 40 percent.


 DeepMind is using machine learning to reduce the energy consumption of Google data centers. (Wikimedia Commons)


According to the company’s blog, “Reducing energy usage has been a major focus for us over the past  10 years: we have built our own super-efficient servers at Google, invented more efficient ways to cool our data centres and invested heavily in green energy sources, with the goal of being powered 100 percent by renewable energy. Compared to five years ago, we now get around 3.5 times the computing power out of the same amount of energy, and we continue to make many improvements each year.”


DeepMind machine learning addresses several important features of cooling data centers, including the non-linear way that the equipment interacts with its environment, the inability of the systems to quickly adapt to external changes, and the unique nature of each data center that defies a universal solution.


Over the past two years, the company has applied machine learning to its data centers and has been working with Google closely over the past few months.


To make the machine learning intuitive enough to reduce cooling costs, DeepMind incorporated historical data about temperature, power, pumps, speed and more from thousands of sensors. This data was used to train deep neural networks.


“We trained the neural networks on the average future PUE (Power Usage Effectiveness), which is defined as the ratio of the total building energy usage to the IT energy usage,” the blog continued. “We then trained two additional ensembles of deep neural networks to predict the future temperature and pressure of the data center over the next hour. The purpose of these predictions is to simulate the recommended actions from the PUE model, to ensure that we do not go beyond any operating constraints.”


Experimenting with a working data center, DeepMind researchers saw a consistent 40 percent reduction in the amount of energy required for cooling, which is a 15 percent reduction in overall PUE overhead. According to DeepMind, this was the lowest PUE that the site had seen.


The MIT Technology Review added, “The project demonstrates the potential for artificial intelligence to manage infrastructure—and shows how advanced AI systems can work in collaboration with humans. Although the algorithm runs independently, a person manages it and can intervene if it seems to be doing something too risky.”

Choose category and click GO to search for thermal solutions


Subscribe to Qpedia

a subscription to qpedia monthly thermal magazine from the media partner advanced thermal solutions, inc. (ats)  will give you the most comprehensive and up-to-date source of information about the thermal management of electronics


Submit Article

if you have a technical article, and would like it to be published on coolingzone
please send your article in word format to articles@coolingzone.com or upload it here

Subscribe to coolingZONE

Submit Press Release

if you have a press release and would like it to be published on coolingzone please upload your pr  here

Member Login

Supplier's Directory

Search coolingZONE's Supplier Directory
become a coolingzone supplier

list your company in the coolingzone supplier directory

suppliers log in

Media Partner, Qpedia


Heat Transfer Calculators