"This will also help other companies who run on Google's cloud to improve their own energy efficiency," Google said in a blog about the achievement. "While Google is only one of many data center operators in the world, many are not powered by renewable energy as we are."
Google has set a goal to eventually power its data centers using 100% renewable energy. Today, the company claims, renewable energy is used for 35% of its power needs.
A graph displaying a typical day of testing using DeepMind's algorithm to recommend the most efficient power use effectiveness. The graph shows when the machine learning recommendations were turned on and off.
The company has also partnered with, or outright invested $1.5 billion, in 22 utility-scale wind or solar projects around the world, making it the largest corporate purchaser of renewable energy.
"When added up, these projects represent a total capacity of over 2.5GW, which is far more electricity than we use," Google said on its data center website. "To put this in context, this electricity is equivalent to that consumed by around 500,000 homes."
DeepMind, a London-based artificial intelligence company that Google acquired in 2014, is a neural network inspired by the human central nervous system that can actively learn about an environment in order to solve complex tasks.
Google's massive data center infrastructure supports Internet services such as Google Search, Gmail and YouTube, but its servers generate massive amounts of heat that "must be removed to keep the servers running."
"This cooling is typically accomplished via large industrial equipment such as pumps, chillers and cooling towers," Google said. "We began applying machine learning two years ago to operate our data centers more efficiently. And over the past few months, DeepMind researchers began working with Google's data center team to significantly improve the system's utility."
DeepMind used historical data -- such as temperatures, power and pump speeds -- that had already been collected by thousands of sensors in its data centers and used it to train the A.I.'s 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."
Additional neural networks were then used to predict the future temperature and pressure of data center in order to recommend actions.
Sign up for Computerworld eNewsletters.