|Fig. 1: Annual consumption of energy by data centers in the US.  (Source: N.Brackbill)|
Our technological society is quickly moving away from a local file system to a global file system, also known as "the cloud". Everything we want is accessible wherever and whenever we want it, from our photos on iCloud to our music on Spotify to our documents on Google Drive. Unfortunately, the cloud is not actually a magical gathering of data in the sky, but a distinctly unmagical gathering of servers in one of many, many data storage centers. We are not storing our data in some immaterial cloud; we are just storing it on someone else's computer.
For the consumer, there are innumerable benefits to this set-up, but there is one major downside for the companies: data storage. Holding on to all of our favorite cat photos (and backups of those cat photos) does not come cheap, even in digital form. Companies like Google run huge data centers across the country and world to store this data, plus of course the data we all use for common products like Google Maps. Data centers are expensive to build and maintain, but one of the most impressive costs is the electricity, both to run the servers themselves and to keep a giant room full of servers cool enough.
Data centers used 1.8% of the US's power in 2014, some 70 billion kWh, with Google alone using more energy than all of San Francisco. [1,2] While this is objectively a large amount of energy, the good news is that energy consumption is leveling off, even as cloud computing grows. In fact, the amount of energy used by data centers increased by 90% in the period 2000 - 2005, by 24% from 2005 - 2010, and only 4% from 2010 - 2014 (see Fig. 1). 
Energy consumption is slowing down due to technological improvements in server capability and organization, and further improvements and changes in management could actually lead to a decrease in the energy consumption. The best case scenario, which involves both upgrading hardware, improving utilization and management of the data centers, and a shift towards few large-scale data centers, could see an energy reduction of 45% by 2020. 
In addition, it is important to note that there are indirect energy effects of cloud computing. For example, a study suggested that downloading music took 40-80% less energy than buying a physical CD.  If this were the case for many products, it could be the case that cloud computing is more energy efficient than the local counterpart. However, in many cases there is not just a shift from a physical product to a digital product, but an entirely new paradigm of products and practices emerging, and the energy effects of this shift are hard to quantify.  What is certain is that with a focused effort on improving the energy efficiency of data centers around the US and the world, energy consumption can decrease even as the computational abilities of these centers soar.
© Nora Brackbill. The author grants permission to copy, distribute and display this work in unaltered form, with attribution to the author, for noncommercial purposes only. All other rights, including commercial rights, are reserved to the author.
 "United States Data Center Energy Usage Report," Lawrence Berkeley National Laboratory, LBNL-1005775, June 2016.
 Q. Hardy, "Google Says It Will Run Entirely on Renewable Energy in 2017," New York Times, 6 Dec 16.
 C. L. Weber, J. G. Koomey, and H. S. Matthews, "The Energy and Climate Change Implications of Different Music Delivery Methods," J. Ind. Ecol. 14, 754 (2010).