Google cuts energy use in data centres using machine learning

When Google acquired British Artificial Intelligence (AI) company DeepMind for over 500 million in 2014, it was part of a wave of increasing hype about the potential of machine learning to transform societies and economies worldwide. Two years later and commercial opportunities for DeepMind and AI tech are beginning to be explored more. Beating humans at notoriously complex board games and within hospitals on healthcare projects has ensured that DeepMind remains in the headlines, but in terms of commercial value, the technology's full potential is still relatively unexplored. However, Google has announced that it has now unlocked one of those initial possibilities applying an AI system to control energy use in parts of its data centres. Power consumption is typically extremely high in data centres, where large electrical servers are prevented from overheating by a network of cooling units. During a testing phase in the first part of 2016, Google reportedly achieved a 40% reduction in energy use at its data centres through employing DeepMind to optimise consumption.