|Cisco: U.S. Mobile Data Growth Falling 60-80%|
|Written by Dave Burstein|
|Tuesday, 29 March 2011 00:00|
92% and 120% are wild, unsustainable growth rates for mobile data. The U.S. will soon fall to a significant but more manageable 30-40%, about what wired networks have been handling since 2002 or so. That's according to the Cisco Visual Networking Index, universally agreed as the best predictor of bandwidth. Cisco measured traffic growth of 120% in 2010 from a very low base as smartphones took over the market. In a few years, nearly everyone will carry a smartphone and growth then should fall. They predict U.S. mobile growth will fall to 46% by 2015. The growth is clearly not exponential, and a simple projection of their data takes the rate closer to 30% for 2016-2017. Cisco's VNI is the gold standard in traffic forecasts. Cisco has relations with every major carrier in the world and a team of excellent professionals do the analysis. Five year forecasts always have uncertainty but Cisco has been generally on target since they began this project.
That's still pretty fast growth and will require good engineering and continuing capital investment to meet. On wireline, Moore's Law brings down equipment costs by 25-40%/year, which has allowed broadband networks to keep up with growing bandwidth demands while dramatically reducing capex. Landlines have become less congested despite 8 years of 30-50% growth because of technology change running just as fast. Wireless costs includes tower building and spectrum, where Moore's Law doesn't apply. The gear on the towers and in the network will keep up with the wireless growth rate, but how much capital the other costs will demand is a hard question.
One of the few willing to speculate on the required capex in a few years is AT&T Wireless CFO Pete Richter. He believes the current capital spending on wireless is a "spike" and that wireless capex as a percent of revenue will come down in a few years. A well-informed scholar points out we should treat that comment cautiously, because a CFO addressing investors has incentive to suggest capex will be low and cashflow high. My experience is that an old pro like Richter tries to be as accurate as practical to maintain credibility, but this is a hard one to call.
My guess is that having plenty of spectrum come available will reduce capex as % of revenue for a given level of bandwidth demand in 2015-2020 by a point or so from what it would have been. A Merrill Lynch analysis suggests many factors that could bring down wireless spending by mid decade.
When you're looking at $tens of billions of even a point or two difference is plenty of money, but far smaller when compared to margins or the $hundreds of billions of revenue that capex supports.
Every CFO and engineer has to plan carefully for the network upgrades needed, but the numbers certainly don't suggest a "crisis."