Google is looking at artificial intelligence (AI) technology to help it identify opportunities for data centre energy savings. The company is approaching the end of an initial 2-year trial of the machine learning tool, and hopes to see it applied across the entire data centre portfolio by the end of 2016.

According to the tech giant, which holds one of the world’s largest data centre footprints, the new AI software has already helped to cut energy use for cooling by 40%, and to improve overall data centre efficiency by 15%. The machine learning technology is being developed at Google’s DeepMind, which the firm bought in 2014 for an estimated £400 million (paywall).

Rich Evans, a research engineer at DeepMind, and Jim Gao, a Google data centre engineer, wrote in a blog post that the programme has been an enormous help in analysing data centre efficiency, from looking at energy used for cooling and air temperature to pressure and humidity. They added that this is an almost impossible task for a human manager, and argued that computer algorithms should be a critical component for effective data centre management.

Screen Shot 2016-07-20 at 15.29.05Over the coming months, the team hopes to expand the AI system to understand other complex infrastructure challenges, in the data centre and beyond. These include improving power plant conversion, reducing semiconductor manufacturing energy, water usage, and helping manufacturers increase throughput.

Data centres have a large environmental impact, accounting for around 2% of global greenhouse gas emissions. Google first revealed its data centre carbon footprint in 2011, since when it says that its efficiency has improved to attain 3.5 times the computing power for the same amount of energy. The machine learning trial hopes to boost that figure even further.

‘We are planning to roll out this system more broadly and will share how we did it… so that other data centre and industrial system operators – and ultimately the environment – can benefit from this major step forward,’ the team concluded in its blog.