Today AT&T has put up a blog post asserting both that “wireless data volume on our network continues to double annually” and that “over the past five years, AT&T’s wireless data traffic has grown 20,000%”. However, those statements can’t easily be reconciled with AT&T’s previous pronouncements in March 2011, when AT&T presented statistics claiming that traffic had grown by 8000% over the four years from 2007 to 2010 (because 20,000% growth in five years would imply ~150% growth in 2011).
However, even more importantly, AT&T’s blog post is somewhat at variance with their assertions to investors that traffic is currently growing at ~40% annually, as I noted in a story for GigaOm on Saturday.
So what’s the actual answer here? One factor appears to be that AT&T’s blog post is apparently obfuscating the issue by changing its definition from “mobile data” (in March 2011) to “wireless data” (in the current blog post). In other words, AT&T’s WiFi offloading (at Starbucks, Times Square, the Superbowl, etc.), which is helping to drastically reduce the growth of (on-network) “mobile data” traffic, is presumably now included in their statistics.
Another factor is that AT&T claims the 40% growth was only intended to represent the increased usage of existing smartphone users, and the smartphone penetration expanded from 42.7% to 56.8% during 2011 (a 33% increase, so if all subscribers, including new ones, increased their individual usage by 40% then this would equate to 86% overall usage growth).
The best way to reconcile AT&T’s apparently contradictory statements would be to conclude that on-net mobile data was 8000% (of the end 2006 number) at end 2010, and including WiFi offload it was 10000% of this level. In other words WiFi offload was 20% of total AT&T wireless data traffic. At the end of 2011 total traffic was 20000% of the end 2006 number, and during 2011 AT&T has stated that WiFi traffic tripled. Thus, it seems reasonable to assume that WiFi offload was ~6000% of the end 2006 number and on-net mobile data was ~14000% of the end 2006 number. In other words, on-net mobile data traffic volumes grew by ~75% (not 100%) in 2011 and ~30% of AT&T’s overall traffic was on WiFi.
As a cross-check, AT&T indicated that there were ~200M WiFi sessions on its network in Dec 2011, and if the above analysis is correct, there would have been about 9PB of WiFi data transmitted and received. That’s 45MB per session, which doesn’t sound too far off when “hotel locations account for approximately 45 percent of the total AT&T Wi-Fi network traffic”. However, unless and until we see some more specifics on data volumes for AT&T’s cellular network (separate from its WiFi network) it will be very hard to be certain just what the growth of on-network traffic really is.
Regardless, AT&T’s 40% traffic growth per user is far less than the 86% CAGR in data usage projected by Cisco today for US smartphone and tablet users between 2011 and 2016. For comparison, 40% CAGR for five years is about 5x growth in traffic, whereas 86% CAGR is equivalent to 22x growth in traffic per user.