Wednesday, October 31, 2018

How Much Share Will Fixed 5G Gain?

To mobile substitution we now have to look to fixed wireless substitution for networks built using cabling of some sort (hybrid fiber coax, fiber to home, fiber to cabinet, all copper).

As always, opinions vary about the potential amount of substitution, in urban or rural markets, with the greatest skepticism about impact on rural markets. In many cases, that is because of signal propagation characteristics of millimeter wave spectrum in such bands as 28 GHz and 39 GHz.

It is reasonable to suggest that many believe lower-frequency signals in the 3.5-GHz band might be more important in rural areas, as 3-GHz and 5-GHz frequencies already are used by fixed wireless providers in rural areas.

Analysts at CoBank are in the camp that believes millimeter wave spectrum will not have very much impact on rural internet access, and also believes fixed wireless using millimter spectrum will be less successful than Verizon estimates.

Still,  others might note that, in urban and suburban areas, even using skeptical estimates, Verizon alone might be able to take significant market share from other internet service providers using cabled approaches.

In the first five years or so, such gains might only represent share gains between 11 percent to 18 percent, in the areas where Verizon builds fixed wireless networks. In substantial part, such skepticism is founded on transmission distances that might range from 500 foot  to 1,000 foot cell radii.

To be sure, skepticism on the part of many observers, such as rural telcos who rely on cabled approaches, is to be expected. After all, fixed wireless is yet another alternative to existing telco or cable TV networks.

CoBank analysts believe the general absence of cable TV operators from the ranks of bidders for new millimeter spectrum indicate those firms are not worried about millimeter wave spectrum approaches. If they were threatened, CoBank analysts argue, cable operators would be bidding for spectrum, either to deny the use of such spectrum to would-be competitors, or to use the platform themselves.

Some of us would disagree with that logic. For institutional reasons, cable executives have disliked running services over any platforms they do not own, with a strong preference for HFC whenever possible. Already looking at use of leased facilities to support their early mobile efforts, cable executives might resist adding more one platform, especially if they conclude the revenue upside is limited.

Cable operators also seem to favor lower-frequency bands, such as 3.5 GHz, in part for reasons of cost, in part for reasons of signal propagation.

Nor would cable operators be able to simply “warehouse” spectrum very effectively, on a long term basis. “Use it or lose it” is the standard policy for new spectrum awards. And even if cable operators wanted to warehouse 28 GHz or 39 GHz spectrum, lots of other spectrum in the millimeter bands will be released for commercial use, and much of that spectrum will be available on an unlicensed basis (or licensed on a no-fee basis).


The other issue is the extent to which 5G mobile networks might emerge as similar substitutes for cabled network access.

In 2015, about 20 percent of U.S. households used mobile as the exclusive method for accessing the internet, according to the National Telecommunications and Information Administration.


Figure 1: Technologies Used to Go Online at Home,<br /><br />
 Percent of Households Using the Internet at Home, 2013-2015

As you would likely guess, lower-income households tend to do so at greater rates than higher-income households. But even in the highest-income households, 15 percent use mobile as the exclusive form of internet access.
Figure 2: Use of Mobile Internet Service Alone to Go Online at Home<br /><br />
 by Family Income, Percent of Households Using the Internet at Home, 2013-2015

No comments:

Many Winners and Losers from Generative AI

Perhaps there is no contradiction between low historical total factor annual productivity gains and high expected generative artificial inte...