Searching for efficiency
How are data centers squeezing more work out of every electron, year after year? For Google, the answer comes down to a relentless quest to eliminate waste, at every level of our operations. We designed highly efficient Tensor Processing Units, (the AI chips behind our advances in machine learning), and outfitted all of our data centers with high-performance servers. Starting in 2014, we even began using machine learning to automatically optimize cooling in our data centers. At the same time, we’ve deployed smart temperature, lighting, and cooling controls to further reduce the energy used at our data centers.
Our efforts have yielded promising results: Today, on average, a Google data center is twice as energy efficient as a typical enterprise data center. And compared with five years ago, we now deliver around seven times as much computing power with the same amount of electrical power.
By directly controlling data center cooling, our AI-powered recommendation system is already delivering consistent energy savings of around 30 percent on average. And the average annual power usage effectiveness for our global fleet of data centers in 2019 hit a new record low of 1.10, compared with the industry average of 1.67—meaning that Google data centers use about six times less overhead energy for every unit of IT equipment energy.