Helping Google reduce its electricity bills

Demis Hassabis, co-founder of Google's DeepMind, said at an artificial intelligence conference in New York recently that Google is using technology from DeepMind to optimize the energy consumption of data centers, thereby saving a lot of energy costs.

In recent months, Google is using DeepMind's artificial intelligence system to control certain parts of the data center to optimize the energy consumption of servers and cooling equipment. Previously, DeepMind's technology was used to teach computers to play Atari games.

Hassabis said that this technology has reduced the energy consumption of Google's data centers by several percentage points. "From a cost perspective, this has brought huge savings and is also good for the environment."

Google said that such optimization has brought a 15% increase in energy efficiency. The energy efficiency index compares the power consumption of computers with the power consumption of infrastructure such as cooling systems.

Google said that in 2014, the power consumption was 4402836 megawatt-hours, equivalent to the average annual power consumption of 366903 American households. A large part of this power consumption comes from data centers, which are the basis of Google's network services and mobile applications.

Reducing power consumption by a few percentage points would be a huge financial help for Google. In the United States, electricity prices are typically $25 to $40 per megawatt-hour, according to the U.S. Energy Information Administration. So if data center energy consumption is reduced by 10%, Google can save hundreds of millions of dollars over a few years. Google acquired DeepMind in 2014 for £400 million, which was equivalent to more than $600 million at the time.

Google has previously applied machine learning technology to data centers. In 2014, Google said it was using neural network systems to predict changes in energy consumption over time so that equipment can be arranged more efficiently.

DeepMind's work goes a step further. The company's software can adjust the way equipment in data centers operates to improve energy efficiency. "This controls about 120 variables in the data center, including fans, cooling systems, windows and other equipment," said Hassabis.

He also said that this is just the beginning of the project. At present, DeepMind knows that this approach is effective, and it is also beginning to understand what information its artificial intelligence system is missing. Therefore, DeepMind may ask Google to deploy new sensors in the data center to help the software get more information.