Google Uses AI To Cool Data Centers

Google dramatically cut its energy usage for cooling its datacenters by up to 40%, with the help of its DeepMind artificial intelligence, the company announced Wednesday in a blog post.

Google now plans to direct DeepMind’s machine learning algorithm at other data center challenges, such as improving power plant conversion efficiency (getting more energy from the same unit of input); reducing semiconductor manufacturing energy and water usage; and helping manufacturing facilities increase throughput. DeepMind, which Google acquired in 2014, is using its AI engine to change how data center servers and cooling systems work to reduce power consumption.

Hassabis said Google was “astounded” by what his company’s technology could do, and has been surprised by how many “variables” in data centers it could address.

DeepMind’s system helped cut power usage in the data centers by several percentage points, “which is a huge saving in terms of cost but, also, great for the environment”, Hassabis said.

PUE is the ratio of the total building energy used by equipment, such as pumps, chillers and cooling towers, to the energy used by IT equipment such as Google’s servers.

Google said it used 4,402,836 MWh of electricity in 2014, equivalent to the average yearly consumption of about 366,903 US family homes.

Recent studies from Greenpeace estimate that the collective electricity consumption of our devices, data centres, and networks will jump from 7.4 percent of global electricity consumption in 2012 to between 7 percent and 12 percent by 2017. (Prices in different regions range from a few dollars to more than $100).

The 15% figure may not seem like much, but considering just how much energy the search giant uses per year, even a 15% improvement could mean savings of millions of dollars. In addition, each data center is unique, so Google can not apply the knowledge of how to operate one entirely to another.

Given that 4.4 million megawatt-hours of power was used in 2014 alone according to the company, power efficiency is a big deal. Human workers in the data centers are given the DeepMind recommendations, and they sometimes reject them. In all, DeepMind was able to make its data centres 15 percent more efficient. Now, it’s saving Google money in data centers.

However, Google said that operating these cooling systems in a complicated data center environment, while taking into account external factors such as the weather, is a highly complex task.