Oops!  Error.
Somewhere between bits and bytes, I made an error which I identify inline below.  The upshot is the capacity of the NSA data center (in terms of phone calls and time) is actually 10x larger than I'd previously calculated. It could probably hold the telephone audio of much of the world (America, Europe, and most of Asia) for 10 years.
           Jim Bell


From: jim bell <jdb10987@yahoo.com>
To: "cypherpunks@cpunks.org" <cypherpunks@cpunks.org>
Sent: Sunday, May 24, 2015 1:49 PM
Subject: Karl Rove's confusion about capacity of NSA's Utah data center.

On Fox News 5/22/2015, Karl Rove displayed his ignorance about the data capacity of the NSA's data center, which he said was intended to store telephone metadata, not the actual audio data from phone calls or (perhaps) the content of emails.  http://www.foxnews.com/transcript/2015/05/22/rove-critics-dont-understand-nsa-terror-tools/
 
Rove said:  "What it does is, it keeps a record of a phone number that is being called from to a phone number being called to and the date and time. So, unless the Tsarnaev brothers talked to somebody abroad to talked to the phone number of somebody who is identified as a possible terror suspect, they wouldn't -- they wouldn't have been caught by this program whatsoever.
And, again, I repeat, why is it that the opponents of Section 215 feel compelled to exaggerate by saying, oh, they're listening to our conversations?"
"Senator Paul on the floor of the Senate said, oh, they're listening in to thousands -- I mean, to millions of conversations. I mean, if we have to build a gigantic center in Utah to keep track of these phone numbers, how big a center would we need to have in order to keep the digital tapes of everybody who is having a telephone conversation? It simply does not happen."  [end of quote by Rove]
Rove is clearly assuming that the size of the data center in Utah is necessary to hold the metadata alone.  The article  http://en.wikipedia.org/wiki/Utah_Data_Center  states:
"An article by Forbes estimates the storage capacity as between 3 and 12 exabytes in the near term, based on analysis of unclassified blueprints, but mentions Moore's Law, meaning that advances in technology could be expected to increase the capacity by orders of magnitude in the coming years."   [3 and 12 exabytes is 3,000-12,000 petabytes, or 3 million-12 million terabytes.]
There are about 900 billion telephone calls made in America per year.  If it took 100 bytes to store the needed metadata per call, that would be 90 terabytes of information needed.  The capacity of the center is, therefore, 33,000x to 131,000x larger than would be needed to store that data for the US alone for one year.  The data stored in a single call could probably be compressed to 8,000 bits per second, so every second would require 80 times as much data to be stored as just the metadata alone.  [Correction:  The data stored in a single call could probably be compressed to 8,000 bits (1,000 bytes) per second, so every second would require 8 times as much data to be stored as just the metadata alone.}   If the average phone call is 120 seconds, that would require about 9600x  [Correction:  960x] the data size, well within even the lower estimate above of 33,000x.
So, the answer to Rove's question, "I mean, if we have to build a gigantic center in Utah to keep track of these phone numbers, how big a center would we need to have in order to keep the digital tapes of everybody who is having a telephone conversation? It simply does not happen." is simple:  "No, Karl, The center is just about the right size to store not only the metadata, but in fact the audio to all phone calls made in America in one year [correction:  10 years).  And in fact, a good deal larger, probably enough to store the audio of every phone call made in Europe as well, and most of Asia. [correction: for 10 years]"  
Would Rove be surprised to hear THAT?!?  Well, he'd probably claim to be surprised.  I suspect he'd defend himself by raising a newly-found fact that at the time that data center in Utah was proposed, hard drives were probably 1000x smaller in capacity than today.  In 2000, as I recall, a typical large hard drive was 2 gigabytes.  Today, the largest I've heard of is 8 terabytes, 4,000x larger.   http://techcrunch.com/2014/08/26/seagate-ships-an-8-terabyte-hard-drive-perfect-for-all-of-your-totally-legal-and-not-pirated-stuff/
         Jim Bell