I woke up this morning to read New Street Research’s latest 3Q2015 Wireless Trends Review, subtitled “Competition Gets Ugly: The Big Guys Have A Problem” and while I agree that competitive pressures in the US wireless market are getting ugly, I’m afraid it is the analysts rather the “Big Guys” that have a problem. In particular, New Street’s “network capacity framework” that suggests Verizon “have half the capacity they will need by 2020 to maintain current network quality” is based on a completely flawed premise, just like most other pronouncements of an imminent spectrum crisis.
This network capacity framework calculates the number of “MHz-sites” that Verizon has for LTE in the top 25 markets and assumes that data demand will grow at 30% from 2015 to 2020 (reaching 10GB/LTE sub per month by then). Ignoring the fact that New St doesn’t know its Petabytes (1 million Gbytes) from its Exabytes (1 billion Gbytes), Verizon’s “aggregate LTE data demand” is projected to increase five fold between 2015 and 2020, while Verizon is expected to increase its deployed LTE spectrum from 68 to 138MHz and its number of LTE macrosites in these markets from 21K to 27.2K over the period, which supposedly will result in 2.5 times growth in network capacity. Thus New St concludes that Verizon will have only half the capacity needed to meet demand, without buying more spectrum from DISH.
This thesis is even more flawed than the Brattle Group study for CTIA back in June, because it completely fails to consider the potential for improvements in network efficiency over the period as LTE-Advanced technologies are deployed. Even Brattle Group acknowledges that “LTE+” will carry over 70% more bps per Hz than 4G LTE, based on a move from 2×2 MIMO to 4×4 MIMO, and the original source for these estimates (Rysavy Research’s August 2014 paper on “Beyond LTE: Enabling the Mobile Broadband Explosion“) acknowledges that many additional improvements are feasible.
Correcting for this error alone, and assuming capacity constrained locations are upgraded to LTE-Advanced by 2020, New Street’s supposed capacity shortfall is reduced from 50% to 14%, and so the number of additional macro sites that Verizon would need to build is reduced from 26.7K to only 4.5K (which based on the FCC’s October 2010 estimate of $550K per cellsite would cost only $2.5B in incremental capex). That hardly seems to justify Verizon spending tens of billions of dollars to buy DISH’s spectrum.
New Street also ignore other sources of incremental capacity (such as LTE-U) and reject small cells as infeasible because they have supposedly only 1/3 the capacity of macro sites (Rysavy in fact points out in Figure 68 of his report that through careful placement to meet hotspot demand, four picocells per macrocell could produce roughly a 10x increase in capacity).
Its hardly surprising that New Street conclude Verizon’s behavior is “perplexing” and suggest that rather than “discounting to hang onto subs” Verizon “should be taking up price to shed subs faster in congested markets to preserve superior network performance for their most valuable subs.” But perhaps Verizon do actually understand network engineering and the potential for network capacity enhancements better than analysts with a simplistic, erroneous spreadsheet model?
In these circumstances Verizon would also feel happy to say no to Ergen’s blackmail and refuse to pay more than their previous offer for DISH’s spectrum (which I guess is around $1/MHzPOP). In fact, it would seem likely that Verizon don’t want a deal at any price ahead of the incentive auction, if that would enable Ergen to bid up the price of spectrum once again. On the other hand, Ergen is undoubtedly keen to find a partner to endorse the value of his spectrum holdings and force Verizon back to the table. As part of that effort, I’m told he has been visiting Google again recently, just as he did back in 2012 when he was trying to pressure AT&T to do a spectrum deal.
One possibility for such a partnership would be to reinvigorate DISH’s rooftop small cell deployment plan, and meet Ergen’s AWS-4 buildout requirements, using the AWS-4 uplinks (2000-2020MHz) as downlinks in conjunction with his 1695-1710MHz uplink spectrum. Remember that Google has backed similar efforts in the past, when it funded Clearwire (rather than LightSquared) in 2008, in order to get a competitive 4G network built out quickly, and push Verizon and AT&T to do the same. However, Ergen has little more than a month left to get a deal done before the auction restrictions kick in, so time is not on his side.