There are numerous organizations that are backing up data, and often times backing up the data to multiple locations (internet.org, amazon S3, dropbox, etc). Often times these sites taut site Geo-redundancy. But the data has to be physically stored somewhere.<p>A few facts I could glean (somewhat dated)<p>- In 2013 Amazon stores 2 trillion objects (http://gigaom.com/2013/04/18/amazon-s3-goes-exponential-now-stores-2-trillion-objects/). The only fact I could find about the size of those objects, is that each object has ~1-2k of metadata associated with it.<p>- In 2012 Dropbox has a billion files saved every day[2] (http://gigaom.com/2012/11/13/how-big-is-dropbox-hint-very-big/). Again size is not mentioned.<p>I wonder if anyone has actually done any research or have thoughts on the subject. The closest I could find was the work done by a redditor a thread about the Utah NSA facility discussion (http://www.reddit.com/r/restorethefourth/comments/1jf6cx/the_guardian_releases_another_leaked_document_nsa/cbedj6l?context=3). It leads me to believe that entire datacenters are built often just to keep up with the demand of the growth.