Will Fixed Wireless Be a "Deploy at Scale" Choice for Telcos?

Up to this point, in U.S. and Canadian fixed network markets, cable TV providers have had a capex advantage over telcos. Their hybrid fiber coax networks have proven less costly to upgrade to gigabit speeds than telco networks, which often must break with copper-based access to match such speeds.

So, up to this point, the telco business decision has been whether there is a clear payback from ripping out copper access and replacing it with fiber to the home. Although fixed wireless and advanced forms of digital subscriber line are solutions in niche cases, the big choice has been to make the transition to FTTH or not.

The range of choices will change in the 5G era, as fixed wireless becomes a potential "deploy at scale" choice in instances where the FTTH business case is difficult.

The “politically correct” answer to the question of whether 5G fixed wireless competes with fiber to the home access is that “both have their place,” or are complementary. That has been the PC answer to the question of whether FTTH competes with digital subscriber line or cable modems as well.

Always, the answers have to be contextualized. The cost of each particular solution, for particular deployments, must be weighed against expected financial return, especially in competitive markets dominated by facilities-based providers.

One key element is cost per passing. The other key variables include take rates and revenue per account. At a high level, fiber to the homes least (per passing) in urban areas, most in rural areas, and somewhere in between in suburban areas.

source: ITU

The cost of a fixed wireless solution likewise varies by density: lower costs per location when connecting a high-rise apartment building or condominium; higher costs to connect individual homes. But nearly all observers would likely argue that fixed wireless should be less expensive than fiber to the home.

Cost estimates arguably are most developed for use of fixed wireless frequencies long in use, more speculative for 5G-based millimeter wave bands, as volume production and use of small cell base stations is just beginning.

But cost per passing is only one of the key business model drivers. In competitive markets, where there are two or more equally-talented and capitalized providers, the addressable market is never 100 percent of locations. Instead, two strong competitors essentially will split the market.

And that means the cost per customer is roughly double the cost per passing.

source: Siklu

In an era where one or multiple products might be sold, the average revenue per account also matters. And there, generally speaking, average revenue per account trends are clearly in the lower direction.

All of that--facilities-based competition; falling average revenue per account; cost per customer--means the cost of serving customers has to fall, even for next-generation networks that feature higher performance.

In that context, the question of “which platform” always includes a “cost to deploy” constraint. If the total customer base is limited because of competition, and average revenue per account is declining, then network cost (capex and opex) must drop.

That is why, all other things being equal, in competitive markets, the low-cost provider tends to win, when that provider has scale.

source: PwC

Some Consumer IoT Apps Will Generate $1 a Year in Revenue

Hopes for internet of things connectivity revenue face a daunting constraint, namely that per-year revenue can be quite low for some categories of consumer sensors such as home smoke alarms. Some existing retail tariffs work out to about $1.20 a year revenue per sensor.

So volume will be quite important. Moreover, the actual revenue from sensor connectivity alone will be relatively low, compared to the value of IoT platforms and apps.

The industrial sensor market predates our present notion of “internet of things.” In consumer markets, government mandates have driven much of the water meter or electricity meter demand. Industrial firms, on the other hand, long have used sensors to measure pressure, temperature and flow.

Those have been niche markets in the past. What is new is the expansion of sensors in many consumer-facing and consumer-used applications, as well as some new business markets such as sensors for parking, transportation, medical care and so forth.

source: Silicon Semicondutor

Newer apps, such as smart home apps or wearables, are expected growth areas in the consumer space.

source: IoT Analytics

A survey on ON World of nearly 100 low power wide area network operators found that unlicensed networks such as Sigfox and LoRa support 66 percent of the platforms in use today.

Sigfox and LoRa networks cover much of Europe and many parts of Asia Pacific with IoT services offered by telecom operators such as Arqiva, Bouygues, Orange, KPN, Proximus, Swisscom as well as a growing number of IoT independent operators such as Senet, Thinxtra and UnaBiz.

Licensed LPWA networks such as LTE-M and NB-IoT are growing even faster than unlicensed networks and make up nearly 33 percent of the LPWA operators surveyed.  NB-IoT network operator activity has accelerated over the past year and will grow 1800 percent in 20128.

One important observation is that IoT connections do not create much new connectivity revenue. Licensed IoT network operators such as Deutsche Telekom are selling €10 (USD $12) service plans  that support a single sensor for 10 years, using 500 MB of data. That obviously means incremental annual revenue is about $1.20.

Mobile operators such as Deutsche Telekom, China Telecom and Vodafone are aggressively rolling out their licensed NB-IoT networks in Europe, China and Australia.  In the U.S., all major operators including AT&T, Verizon, Sprint and T-Mobile are building licensed IoT networks and most of these are providing both LTE-M and NB-IoT services.  

By the end of 2018, most of the U.S., Europe and Asia Pacific will be covered with licensed IoT networks, ON World predicts.

U.S. Telcos Have Lost 87% of Voice Accounts in 18 Years

Mobile substitution is about to become a much-bigger commercial and policy issue, in large part because, for the first time, mobile substitutes for fixed network internet access are going to be commercial realities, sold by tier-one service providers with the marketing muscle to drive adoption.

History provides some idea of how much could change.

Once upon a time (in the 1970s and 1980s), it was believed the solution to the problem of universal communications had to be based on supplying connections to fixed networks. The great cost of building such networks in developing nations was therefore a great cause of concern.

Technology saved us. With the advent of mobile networks, we have nearly solved the problem of voice communications, globally. But there always are consequences for legacy products when new products displace them.

Over a span of less than two decades, voice services, the traditional telecom revenue driver, has virtually collapsed, in some markets. In the U.S. market, for example, line loss for telcos has been as much as 87 percent, between 2000 and 2018.

Ironically, the Telecommunications Act of 1996, the first major revision of U.S. telecommunications law since 1934, focused on enabling voice communications and ownership of voice switches. The emergence of the internet, and mobile substitution, seemingly had not occurred to lawmakers.

The point is that commercial business strategy and government regulation, no matter how thoughtfully considered in the moment, can fail to achieve its intended objectives because markets (both supply and demand) now change so rapidly.

source: US Telecom

Beginning with the 5G era, we are likely to find that technology once again solves an qually big challenge, namely supplying internet access to everyone. Nor is 5G the only important platform. We might well see that new constellations of low earth orbit satellites, and perhaps fleets of balloons or orbiting unmanned aerial vehicles, plus use of unlicensed and shared spectrum access platforms.

There are going to be winners and losers. It is not hard to predict that the business value of a fixed network will change. Such networks are going to drive far less revenue than in the past.

And that has implications for the amount of capital that can be invested in the business, where that capital is invested and how much innovation can be expected. Some of us would argue that enterprise revenue sources will become more important, consumer sources less important.

In the 5G era, it is possible that mobile networks and platforms will be able to match fixed network performance. On top of that, low earth orbit satellite constellations, or even Google’s Loon service, based on use of balloons, might be in place to provide rural internet access across the rural United States.

Even if some worry that competition in internet access is endangered, some of us would argue competition is destined to increase.

Can Independent ISPs Get 50% Market Share?

Can independent internet service providers (public or private) actually get 50-percent market share when competing against telcos and cable companies? Ting Internet believes so, but results from other firms suggest the level of competitor pricing really does matter. G

It always is difficult to quantify take rates for gigabit internet access services, as virtually no internet service provider ever releases such figures. That has been the case since faster internet access services, priced at a market premium, began appearing. The reason for the reporting reticence is that take rates for the fastest, most expensive tier of service tend to be minimal.

Still, ISPs do tout some figures. Mediacom, for example, claims that between 10 percent to 20 percent of its new accounts are buying  gigabit services costing between $125 a month to $140 a month. Again, it is hard to quantify what that means.

The actual number or percentage of account that change providers every year (churn) in the fixed network internet access business arguably varies between markets with strong offers (where fiber to home is sold, and where gigabit cable services also are sold).

Churn arguably is highest where a cable operator offering speeds in the hundreds of megabits per second competes with a telco only able to sell slower digital  subscriber line service. AT&T and Verizon, for example, tend to see low churn rates, where other independent telcos with less fiber to home tend to see higher churn rates.

Much obviously depends on pricing levels. In markets where a gigabit ISP sells 1,000-Mbps service at prices that match the current or legacy prices for slower service (perhaps 100 Mbps), take rates can climb dramattically.

What is harder to model are markets where a clear price premium exists. It will matter when a reasonably-fast standard offer costs $50 a month and the gigabit offer costs more than $125 a month.

Presumably, in such market, demand will be anchored by business demand, higher-income households and multi-user households.

Perhaps the most-optimistic provider to make public predications is Ting Internet, which tends to argue it will get adoption as high as 50 percent in any new market it launches, within five years or so. Initial take rates when first marketed appear to be about 20 percent.

In its fourth quarter 2017 report, Tucows reported Ting Internet take rates of 30 percent take rates in areas where it is able to market its gigabit service.

Ting prices its gigabit service at $90 a month. At that price, it is lower than Comcast and other cable companies charge, but higher than the $70 a month some other ISPs offer. The point is that Ting is pricing at a significant price premium to “standard” offers that offer less value (in terms of speed), but not pricing as high as cable companies or telco practice.

We are likely to see much more of this sort of independent ISP competition in the fixed market, not to mention 5G-based gigabit offers from mobile suppliers.

At least in principle, more than 100 Colorado communities could see some form of

municipal broadband network created, as voters in those communities have approved such moves.

Longmont, Colo. already has built out a portion of its planned gigabit internet access network, aided by that city’s ownership of a municipal power utility, meaning Longmont owns rights of way, distribution facilities, rolling stock and other assets helpful to creating a city-wide internet access network.

In Centennial, Colo., private internet service provider Ting Internet will piggyback on a new government network to be built by the city of Centennial itself.

Also, it sometimes is difficult to ascertain precisely what take rates are, since many independent ISPs challenging cable or telco suppliers seem to count “units sold” rather than “households served.”

That matters when an ISP sells two or more products, and then calculates adoption rates as “units sold divided by households passed.”

In other words, penetration is measured in terms of revenue-generating units, not “locations” or “households.” Each unit sold (voice, video or internet access) is counted against the base of locations. So a single location buying three services results counts as much as three other homes buying just one service.

source: Magellan Advisors

Customer “penetration” by household therefore is different from penetration measured as a function of units sold. The difference is that determining the magnitude of stranded assets hinges on how many locations passed generate revenue.

Assume that, on average, a typical household buys 66 percent of the total suite of services (two of three triple play services or  three of five services, for example).

The difference is significant. Measuring “penetration” by units sold, penetration appears to be as high as 76 percent to 87 percent. Measured as a function of homes generating revenue, penetration could be as low as nine percent, or as high as 44 percent, with a “typical” value being something between 20 percent to 25 percent of homes passed.

Penetration: Units Sold or Homes Buying Service?
	Morristown	Chattanooga	Bristol	Cedar Falls	Longmont
homes passed	14500	140000	16800	15000	4000
subscribers	5600	70000	12700	13000	500
units sold	39%	50%	76%	87%	13%
services sold	3	3	5	3	2
HH buys .66 =	2	2	3	2	1
Homes served	2828	35354	3848	6566	379
penetration	20%	25%	23%	44%	9%

It might be worth pointing out that all these communities (Morristown, Chattanooga, Bristol, Cedar Falls and Longmont) have municipally-owned utility companies, and might therefore represent a sort of best case for retail operations serving consumers.

That seems consistent with other evidence. In markets where a telco and a cable operator are competent, as is the attacking ISP (municipal or private), market share might take a structure of 40-40-20 or so, possibly 50-30-20 in areas where the telco does not have the ability to invest in faster broadband and the cable operator has the largest share.

Beyond the actual cost of the network, and the business role chosen by the municipality, details of revenue generation (homes that generate revenue as a percentage of total; number of services offered) are fundamental.

Beyond that are the other operating and marketing costs, overhead and need for repaying borrowed funds and making interest payments, on the part of the retail service provider.

One might argue that most other communities, without the advantages ownership of an electric utility provides, will often find the lower risk of a shared public-private approach more appealing.

Also, some ISPs might find the availability of some amount of wholesale or shared infrastructure makes a meaningful difference in a business model.

One might suggest there are a couple of potential practical implications. Efforts by incumbent ISPs to raise retail prices in the same way that video entertainment prices have grown (far higher than the rate of overall inflation) will increase the odds new competitors enter a market.

Higher prices, in fact, will increase the likelihood of new entrants entering a market, as the higher prices increase the attractiveness of doing so.

In at least some cases, the new competitors will be firms such as Verizon, which now has announced it will essentially overbuild an AT&T and Comcast markets in Sacramento, Calif.

Though it is not easy, more competitive ISPs are likely to enter more markets, as lower-cost access platforms evolve, helped in some cases by municipal facilities support.

Where that happens, it is conceivable that the incumbents will see a new limitation on their market share, dipping from possibly 50-percent share to a maximum of perhaps 40 percent each, on a long-term basis, assuming the new competitor is not eventually bought out by one of the incumbents.

If the Internet is Changing, Backward-Looking Regulation Maybe Not So Smart

It is a truism that regulators never can keep up with technology, which is a reasonable argument for caution where it comes to regulating just about anything central to the  internet, which keeps changing.

The logical implication might be that the historic light-touch treatment of anything related to computing likely should be the presumption, even if consumer protection or antitrust remedies might from time to time be deemed necessary.

The reason for such caution is that computing and the internet itself continue to change. And if there is anything history shows, it is that the leaders in any single era rarely--if ever--are the leaders in the succeeding era. So “punishing” a leader in one era, when an era is about to evolve, does not make too much long-term sense.

We might today believe it is nonsensical to consider the web browser a source of competitive advantage, as though the browser provides some sort of business moat. Nevertheless, not so long ago, the browser was considered a source of anti-competitive advantage.

Not so long ago, ability to offer phone services using Class 5 switches was considered to be a major pro-competitive step. Before that, long distance voice was the profit driver for the telecom industry. Things change, and change quite a lot.

Leadership in the mainframe computing era was not the same as leadership in the mini-computer or personal computing eras. And none of those leaders were key in the early internet age.

Someday, Google, Facebook and others might not be the leaders in some future evolution of the internet ecosystem. The point is that regulation is sometimes pointless when it is backward looking.

The internet has grown through multiple eras, any observer can note. Where the internet was a low-bandwidth method for researchers to communicate, it then became a consumer transaction and information platform, before becoming a social media vehicle, and now a content delivery mechanism. Next, the internet is on the cusp of possibly becoming a major platform for enterprise private networks.

With the coming internet of things evolution, autonomous vehicles and sensor networks of many types could represent the next big wave of development for the internet, as the internet has defined the most-recent eras of computing.

source: Chris Dixon

source: Keri Pearlson & Carol Saunders

source: Textio