Impressive stuff, that. But those AI demonstrations have yet to spell actual revenue. Until now — although the efforts are helping Google save money on its most expensive part.
DeepMind chief Demis Hassabis told Bloomberg that his unit recently began applying its advanced AI to Google’s data centers, finding ways to reduce the company’s sizable energy bill.
Google started using machine learning for its data centers two years ago, searching for ways to reduce costs for one of the company’s top expenses. A month ago, it aimed the more specialized AI tools from DeepMind at the problem of cooling these server farms. That cut the energy needed for cooling by 40 percent, the company said.
It didn’t offer a dollar figure for that, but it’s safe to assume that it means hundreds of millions in savings over the long haul.
In the past year, Google has invested in renewable energy for its data centers. It’s framing the DeepMind news in green terms, highlighting the potential energy savings for itself and — just as important for Google’s business — passed along to its cloud storage customers.
Here’s a portion of a blog post coming out later on Tuesday about the news:
The implications are significant for Google’s data centers, given its potential to greatly improve energy efficiency and reduce emissions overall. This will also help other companies who run on Google’s cloud to improve their own energy efficiency. While Google is only one of many data center operators in the world, many are not powered by renewable energy as we are. Every improvement in data center efficiency reduces total emissions into our environment and with technology like DeepMind’s, we can use machine learning to consume less energy and help address one of the biggest challenges of all -- climate change.
DeepMind technically sits outside of Google in Alphabet. (I’ve heard people describe it as in the “Alphaverse,” whatever that means.) But a rep said that Google was not paying DeepMind for its cost-cutting research here.
This article originally appeared on Recode.net.