Telcos Losing Internet Access Accounts at a Faster Rate to Cable

There is a reason much of the attention in the consumer internet access business now is focused on varieties of wireless access (balloons, low earth orbiting satellites, fixed wireless, bonding of licensed and unlicensed spectrum, 5G and millimeter wave).

Though the economics of fiber to the home have gotten better over the last few decades, the business model also has been under pressure from substitute products, the impact of competition on financial returns and the maturation of all legacy revenue models.

In the U.S. market, rival platforms already have had a huge effect, with cable companies using hybrid fiber coax platforms capturing all the net growth in the consumer segment of the internet access market.

The top cable companies netted 122 percent of the broadband additions in 2016, increasing gains from the 106 percent they gained  in 2015 and the 89 percent they got in 2014. To be sure, the leading telcos (AT&T, Verizon, CenturyLink) now have stepped up fiber-to-home deployments. But Verizon’s experience also suggests that that ubiquitous fiber deployment results in market share of about 40 percent to 45 percent.

The top telcos, in contrast,  lost 600,000 accounts in 2016, compared to a loss of about 185,000 subscribers in 2015. As the consumer internet access market is mostly a zero-sum gain, cable gains are at the expense of telcos.

In other words, under perhaps the best of circumstances (dense urban markets, lots of capital, marketing muscle), a big telco still loses leadership of the internet access market to cable competitors. To be sure, all-copper digital subscriber line services are where telcos mostly lose the battle to cable.

That scenario might remind you very much of Sprint, back in the days when it was bleeding former Nextel accounts, even as legacy Sprint accounts grew slightly, or lost just a bit of share. In other words, some segments of the customer base are more susceptible to predation.

Telcos face great difficulties in rural areas and less-dense suburban areas, where FTTH costs arguably would not produce a clear financially-positive result. Cable company networks always were built for low-cost access, always were broadband and now are easier to upgrade incrementally than telco networks.

All of that means there is a market opportunity, for all competitors to cable companies, in new platforms (mobile, Wi-Fi, LEO satellites, balloons, unmanned aerial vehicles, fixed wireless, 5G) now are getting serious attention to FTTH.

ISPs	Subscribers  4Q 2016	Net Adds
Cable Companies
Comcast	24,701,000	1,372,000
Charter	22,593,000	1,604,000
Altice*	3,907,000	122,000
Mediacom	1,156,000	71,000
WOW (WideOpenWest)**	718,900	20,600
Cable ONE	513,908	12,667
Other Major Private Company^	4,790,000	90,000
Total Top Cable	58,379,808	3,292,267
Phone Companies
AT&T	15,605,000	(173,000)
Verizon	7,038,000	(47,000)
CenturyLink	5,945,000	(103,000)
Frontier^^	4,271,000	(243,000)
Windstream	1,051,100	(44,000)
FairPoint	306,624	(4,506)
Cincinnati Bell	303,200	15,800
Total Top Telco	35,519,924	(598,706)
Total Top Broadband	92,899,732	2,693,561

source: Leichtman Research

5G Spectrum Will Feature Coverage, Capacity, Indoor Pattern

Up until the 4G era, mobile and Wi-Fi spectrum have been seen as having rather clear roles: lower frequency for coverage; mid-band for capacity and Wi-Fi for indoor distribution. In the 5G era, that tripartite pattern will be even more pronounced, with the key changes coming in the millimeter wave bands, which are seen as best suited for capacity and indoor coverage.

source: Ofcom

If ROIC is What Matters, So Does Market Share

Market share plays a large role in financial results. Looking at market share and return on invested capital (ROIC) for the three largest players in Thailand, China, and Indonesia since 2015, you can see that financial return and market share tend to be directly related.

My rule of thumb is that the leader has twice the share of number two, which in turn has twice the share of provider number three. In Thailand, China and Indonesia, the general pattern holds, In Thailand the pattern holds well. The leader has 53 percent share, number two has 31 percent and number three has 17 percent share.

The theoretical model I use would call for shares of 50/25/13, or any other ratio that maintains the same sort of relationships between market share among the top-three providers.

The Chinese market diverges: the leader has 69 percent share; number two 17 percent and number three 14 percent.

In Indonesia, the market leader has 65 percent share; number two has 18 percent and number three has 17 percent.

My approach also suggests that any market not having the 50/25/13 structure is unstable, with the big changes to be expected in share shifts from the number-one provider to number two. The third provider in all three of these markets actually is doing as well as a stable structure would suggest.

Those relationships also include varying rates of return on deployed capital, with one possible and important exception. Where the leading provider generally earns the highest ROIC, and in some cases can wring extraordinary profits out of market leadership, the relationship between market share and ROIC is less predictable for providers two and three, in any of these markets.

Still, the general pattern is important. Market share matters because it is related, more or less in a linear fashion, with profit and return on invested capital, although perhaps in a less certain way than market share structures tend to take, in stable markets.

Source: Reperio Capital

Source: Reperio Capital

ROIC issues for access services are one reason why the strategy of “moving up the stack” or “up the value chain” continues to be relevant and necessary for at least the larger telcos, cable companies and similar access providers. Mergers to gain scale will help, in many cases, but the fundamental problem--lack of organic growth for network services--will be a key constraint.

Market consolidation, in many cases, also will help, as market share and financial return tend to be correlated. So increasing share will tend to boost financial performance. In all three countries studies by Reperio Capital, market share and ROIC were related. In the clearest example, the number three provider in all three countries had single digit ROIC. Provider number one had high or even extraordinarily high ROIC.

ROIC Will be Key Concern in 5G Era

Some observers point out that the higher costs of 5G networks--driven by small cell architectures and the need for more backhaul and more radio sites--is going to mean 5G networks cost more than 4G networks. Logic suggests the merit of that view. Macrocell networks (4G and earlier generations) require fewer towers, sites, radios, frequency coordination and backhaul than microcell networks. 

And that means a current problem--return on invested capital--is going to become even more important in the 5G era. 

Return on invested capital, not earnings (EBITDA) or cash flow, now is the way major telcos have to measure business results, argue analysts at PwC. The fundamental reason is that “growth is gone and it’s not coming back,” say PwC analysts.  “Downward trends in return on invested capital (ROIC) are the result of a number of factors, from regulated price reductions to cannibalization of legacy revenues by OTTs, along with high capital intensity required to support demand for data,” say researchers at EY.

So if revenue growth is muted, what matters is how well access providers monetize invested capital. That is not going to be easy.

source: EY

Philippines telco Globe Telecom had a return on capital of about 7.7 percent in the third quarter of 2016. Airtel in India points out that, in 2016, some Indian telcos might have had a one-percent return on capital.

That 7.7 percent rate is about in line with global telecom firm financial returns from the 1990s until perhaps 2006. But there can be wide variation. In the U.S. market, in 2013, Verizon has ROIC of more than 13 percent, while CenturyLink has ROIC of minus 7.4 percent.

source: Bloomberg

source: Bloomberg

ROIC issues for access services are one reason why the strategy of “moving up the stack” or “up the value chain” continues to be relevant and necessary for at least the larger telcos, cable companies and similar access providers. Mergers to gain scale will help, in many cases, but the fundamental problem--lack of organic growth for network services--will be a key constraint.

Market consolidation, in many cases, also will help, as market share and financial return tend to be correlated.

TIA 5G Survey Suggests Potential Business Model Trouble

A new study issued by the TIA finds that 33 percent of respondents expect their companies will be offering commercial 5G by the end of 2020. However, there is a lack of consensus regarding the competitive advantage that might be gleaned from being first to market.

Some 48 percent of respondents said “being first to market” was unimportant or “only somewhat important” for their company’s competitive position. About 35.5 percent of respondents believe it is very important to lead deployment, though. Another 6.4 percent extremely important.

In general, operators in technologically advanced and mature markets place greater emphasis on being the first to launch 5G, while those that operate in less technologically advanced or less mature markets do not, the study, conducted by Tolaga Research, and sponsored by InterDigital, found.

source: TIA

Almost a third of respondents plan to launch pre-standard 5G products, one way of determining 5G deployment “high perceived value.”  Australia, China, Japan, Korea, and the United States are in that group.

That is probably a reflection either of market maturity (5G promises revenue upside--and significant upside) as well as expectations of higher initial demand for internet of things products and services that will be supported by 5G.

But the survey also suggests the business models for 5G are fluid and even unknown, to a large extent.

About 56 percent of respondents believed mobile broadband services would be “significantly transformed.”

Only 30 percent of respondents believed that machine-type communications will be significantly transformed by 5G, and 50 percent expected that ultra-reliable and ultra-low latency services will be not be significantly transformed by 5G.  

That will be a problem, if the views prove correct. Only 56 percent of respondents believe 5G will fail to significantly transform mobile broadband. If so, 5G will be an expensive way to possibly transform 4G mobile access.

Worse, should respondent views prove correct, is that 70 percent seem to believe M2M will not be a key benefit of 5G, while only half think low-latency new services will be transformed.

Ignoring for the moment the impact of that specific terminology (“transformed” rather than “enabled,” for example), most executives are signaling trouble for the business model.

If 5G promises big new revenues in IoT and M2M, and operator executives seem not to believe that will happen, there could be serious business model issues, of the “still a dumb pipe” variety. There is only so much incremental financial value to be wrung from 5G access speeds. Most of the revenue upside is going to come (observers currently believe) from IoT apps and services.

If 5G mostly is a “faster dumb pipe,” then the capital investment is likely to produce paltry returns.

AI Can Help with This Type of Problem

A report by the Mobile Marketing Association and RadiumOne finds 66 percent of marketers are not “fully confident” they know the “most critical” digital signals that they can correlate with behavior.

This sort of problem--lots of data and algorithms that are not able to interpret such data--seems the sort of problem artificial intelligence is ideally suited to solve.

Fully 58 percent were not fully confident their re-engagement efforts were effective against customer churn, while half called into question their ability to acquire new customers and re-engage those they'd lost.

The signals marketers considered the most important for mobile branding was that garnered from content sharing from apps (29 percent), mobile site visits (28 percent), and app installs (27 percent). For mobile direct response, its purchase data (38 percent), geolocation and bookmarked content (both 35 percent).

Is Mobile A Substitute for Wi-Fi?

For decades, observers have wondered and speculated about whether Wi-Fi could be a substitute for mobile data access.

Now we might have to ask the reverse question: under what conditions is mobile access a functional substitute for Wi-Fi?

The new question comes as all four leading U.S. mobile operators now offer unlimited usage plans. To be sure, none of the plans is “truly” unlimited. After a certain amount of usage, users will find themselves throttled, in terms of top speed. The other push might come with 5G.

Cisco, in fact, already believes 5G usage will reverse recent trends, and see mobile access growing at the expense of Wi-Fi. Tariffs likely will prove to be the driver. At suitable pricing levels, users simply will not have incentives to use Wi-Fi for offload.

Also, not every account needs, nor will every consumer benefit from, unlimited access. For many, though, the new unlimited feature might mean less value for Wi-Fi offloading. In India, for example, new 4G networks now mean mobile is faster than Wi-Fi.

Still, more Wi-Fi spectrum is coming. In the U.S. market, the Federal Communications Commission, for example,  is releasing about 7 GHz of new unlicensed spectrum  for use.

Ofcom, the United Kingdom communications regulator, has decided to authorize an additional 125 MHz of capacity for Wi-Fi  in the 5-GHz band (5725 GHz to 5850 GHz). That is the first step in what might be additional actions, including allowing Wi-Fi use in the 5850-MHz to 5925 MHz band now used to support mobile operations.

The additional spectrum will allow for wider channels, which should improve bandwidth efficiency and support higher speeds. The number of 80 MHz channels will increase from four to six. Three additional 40 MHz channels and six additional 20 MHz channels would also be available.

Ofcom also says itg could open up spectrum for Wi-Fi at 5350-MHz to 5470-MHz band as well.  

source: Ofcom

Ironically, moves to add more Wi-Fi spectrum in the millimeter wave bands in the U.S. market are happening at the same time that all four leading U.S. mobile service providers now are offering unlimited use plans. In principle, that means some customers might no longer have an incentive to switch access to Wi-Fi, since doing so does not save money or improve user experience (Wi-Fi tends to be slower than 4G).   

Telco Revenue Growth: GDP-Plus or GDP-Minus?

Nobody can yet tell how well big new revenue streams might develop on 5G platforms, from internet of things apps running over those networks or from telco movement into the applications space. But it is not incorrect to forecast dire financial repercussions for most access providers if such big new sources are not identified, then created and owned. Avoiding dumb pipe status rightly is the chief strategic problem for most telcos.  

As analysts at STL Partners (Telco 2.0 Research) note, between 2009 and 2016, though revenues grew at 68 public telcos, growth rates now are close to zero. Others, such as J.P. Morgan Chase analyst James Sullivan, say telco revenue growth now happens at a slower rate than gross domestic product growth, an insight that tracks the observed linear change rate quite well.

Also, 5G might well intensify the pressure, in large part because the network architecture will require lots of small cells (creating a denser network with more backhaul), though possibly less spending on spectrum (as unlicensed and more-affordable shared spectrum is used, and as huge blocks of lower-cost millimeter wave capacity is licensed).

The basic problem some see is that usage, revenue and infrastructure investment are non-linear, compared to the voice era. Back then, higher usage meant higher revenue, with incremental investment in network resources. In the internet era, and especially so in the coming gigabit era of the internet, usage skyrockets, while revenue grows at slower and non-linear rates, while network investment grows disproportionately as well.

Some might also note that supply will exceed demand in some cases, putting pressure on business models.

The fundamental problem, one might argue, is that no matter how fast, or how cheap internet access gets, revenues do not necessarily grow to match usage. That has been an issue for a couple of decades already, so there is little reason to believe matters will change.

Many argue that virtualized networks, able to provide bandwidth on demand, or network features on demand, will help. That might well be true. Still, some will argue, revenues beyond “access” must be created, to avoid a “death spiral” based on “skinny margin” or “no margin” access services.

68 major telecoms groups – aggregate revenue, 2009-2016

source: Telco 2.0 Research  

No, 5G Will Not the "Best" Network for Every Use Case

We should not be surprised when, inevitably, there are criticisms of 5G networks because 5G fails to be a single network that supports every application and service, equally well. In fact, no single network ever has done so. The telegraph network did not support voice. The voice network did not support video or data communications.

Even a hundred years ago, when the only application was analog voice, the fixed network did not work well in rural areas, for example.

By the 1960s and 1970s, satellite networks began to deliver long distance voice traffic, especially in isolated and remote areas. As additional applications created new markets, networks proliferated.

Cable TV, satellite TV, mobile networks, specialized business data networks (X.25, internet protocol, ISDN, SONET, ATM, frame relay) always have co-existed with the broad consumer voice networks.

That 5G “does not do everything,” for every geography, customer, application, device or business model should be a given. No network ever has done so.

Nor, by design, is 5G expected to do everything, everywhere, at lowest cost and highest efficiency. By design, 5G will embrace a network of networks approach, virtualizing access in many ways, for example.

But 5G also will use new frequencies (millimeter wave) that require small cells, which also means 5G will not--everywhere--be a replacement for 4G. Millimeter waves are the antithesis of low-band (frequencies at 800 MHz and lower).

Since 5G also is expected to underpin new machine-to-machine communications, we already see the rise of specialized narrowband networks designed to support low-bandwidth communications that are bursty and need to travel longish distances, without consuming much power.

Already, it seems, 5G is not the best way to support many M2M apps and devices (though 5G architects are working on those problems). So, no, 5G is not a panacea for every communications geography, customer or application. No, it will not be the single network to do everything. No network can do so.

On the other hand, 5G might be the “best network” approach for some important new apps and revenue sources, in some places. Some of those requirements apply to people, some to machines. Because 5G uses assets on a virtualized basis (access and core), it might be the “best” approach for any number of apps and use cases, even when it is not the “5G” access network doing the work, but some other wireless, mobile or fixed platform.

In 5G Era, Will Value of Mobile, Fixed Networks Change?

Almost nothing about 5G is completely certain; especially the new business models, revenue sources and cost structures that might emerge. For several decades, revenue and customer growth has been the province of mobile operators. But 5G creates new demands for radio backhaul and radio density (small cells), suggesting a boost in value for owners of fixed network infrastructure.

At the same time, more value will accrue to fixed networks that are efficient about backhaul: the lowest-cost provider wins, in other words. And, ironically, for the first time in a couple decades, owners of multiple or “integrated” platforms (mobile and fixed) might fare better than “mobile-only” or “fixed-only” contestants.

A simplistic phrase describing the shift is “the fundamental value of a fixed network is backhaul.” It’s a simplification, but a useful concept. Backhaul is the biggest, or among the biggest, operating costs for many mobile operators, somewhere in the “25 percent of total operating cost” range, typically.  In some cases, backhaul might represent 30 percent of operating costs, for mobile operators who do not own fixed network assets.

The move to small cells is going to increase the number of backhaul sites by an order of magnitude or more, so any retail provider with the ability to attack that cost will have an advantage. That is why cable TV operators are so optimistic about the value of their networks, which over the past few decades have shown an ability to deploy lots of bandwidth at lower cost than telcos have been able to accomplish.

If you think about Comcast’s “homespots,” they essentially function as small cell sites, able to support mobile connections and data offload to the fixed network. The “fiber deep” deployments required to support consumer internet access also will allow cable operators to leverage those assets to support small cell backhaul, either for “own use” or as a wholesale connectivity product for third parties.

So there will be tighter integration between fixed and mobile networks in the 5G and coming eras, as dense radio networks (wireless fixed and mobile) require huge numbers of high-capacity fixed links.

That sort of thinking is behind Verizon’s “One Fiber” strategy, which aims to leverage a single optical fiber transport and distribution network to support a range of services and customers, from residential broadband to business data services to small cell backhaul.

So one might predict that, in a growing number of cases, financial returns for “integrated” service providers will exceed returns for mobile-only or fixed-only operators, once the 5G era has begun.

We might already see some glimmers of that, as some analysts note that financial performance by integrated operators, compared to mobile-only operators, has oscillated.

From 1998 to 2011, mobile-only operators in emerging markets, especially, outperformed integrated operators in developed markets, according to J.P. Morgan equity analyst James Sullivan. From 2012 to about 2016, developed market integrated operators performed better. But Sullivan thinks the emerging market mobile segment is improving as consolidation trends take hold.

What remains to be seen are developments in developed markets, where the 5G era will value backhaul assets. Also, fixed network capabilities might improve with new fixed wireless business models.

Artificial Intelligence is Going Mainstream

Artificial intelligence used to be a “science project,” but AI increasingly is becoming a practical tool for home appliances and customer service apps, among others. That said, the industry selling AI software and services remains a small one.

Dave Schubmehl, research director at IDC, calculates that sales for all companies selling cognitive software platforms--excluding companies like Google and Facebook, which do research for their own use--added up to $1 billion in 2015.

He predicts that by 2020 that number will exceed $10 billion. USAA, the insurance provider, has been testing ways to use AI to fine-tune its detection of identity theft, for example. USAA also is looking at ways to use AI to improve customer service.

Using AI technology built by Saffron, a division of Intel, USAA has found it can match broad patterns of customer behavior to that of specific members, and 88 percent of the time it can correctly predict things like how certain people might next contact USAA and what products they will be looking for when they do. Without the AI, USAA’s systems were guessing right 50 percent of the time.

With ABI Research forecasting more than 120 million voice-enabled devices will ship annually by 2021, voice control, which combines speech recognition and natural language processing, is quickly becoming the key user interface within the smart home.

AI-driven customer service applications also are becoming common.

IoT Security, Not Market Structure, Likely to be Regulator Concern

Though policymakers now debate the best market structure for mobility suppliers (are two, three or four suppliers necessary to support robust investment, innovation and consumer benefit?), it seems unlikely they will have similar quandaries about coming special-purpose IoT networks.

For starters, there is no “consumer interest” to weigh, as IoT networks will be about machine-to-machine applications, not human communications. There are potential economic angles, but not the traditional “consumer welfare” concerns.

In other words, market structure and standards are likely to be “market driven,” not created by standards bodies, as will be the case for 5G. That is not to deny some big issues.

Cyber-security is likely to become a huge concern, though security likewise is not the sort of issue traditional regulation is best suited to handle. So to the extent regulation becomes an issue, it is unlikely to concern market structure, but rather security requirements for devices and services.

Will Dish Network's NB-IoT Network Actually be Built?

When a wave of consolidation has finished in the U.S. mobile market, it is unclear where some assets--Sprint, T-Mobile US, Dish Network--will reside. In the case of Dish Network, it is unclear whether the firm remains independent (probably not), and what use will be made of its mobile spectrum.

Though there has been speculation that its lower-band spectrum eventually would augment another carrier's 4G mobile operations, it is perhaps possible that the assets will provide an overlay IoT capability as well. Much hinges on what happens over the next several months to a year.

Without much doubt, whatever asset changes occur will happen long before the stated Dish plan is completed. So it is possible the spectrum assets wind up, as expected, as part of another carrier's 4G spectrum resources, and the NB-IoT network is not built, at all. On the other hand, it might take three years for Dish to build such a network, so it has to begin some amount of work soon, to ensure any potential buyer the spectrum assets are viable, if Dish--or its spectrum--are not acquired in 2017.

So even as Dish Network is raising $1 billion in a bond sale, to build a narrowband Long Term Evolution (NB-LTE) network, it is not clear that actually will happen.

Dish says it use its spectrum licenses in the AWS-4 Band and Lower 700 MHz E Block to build a “5G-capable network,” focused on internet of things (“IoT”) apps and services, and expects to have a network built, covering 70 percent of the U.S. population, by March 2020.

Such a network essentially is a 4G narrowband network in a class with other low power, wide area (LPWA) networks designed to support sensor communications, not human users. NB-LTE is a “5G” network in the sense that it aims to support machine-to-machine communications, not in the sense of being part of the formal 5G standard, as such.

That has cost and performance implications. NB-LTE is optimized for low bandwidth communications (250 kbps downlink, 20 kbps uplink).

That helps clarify a key part of Dish Network’s wireless strategy, which is to avoid losing rights to its spectrum because it has not made meaningful progress towards building an actual network. It also is reasonable to assume that building an LPWA network will cost far less than building a full 4G or 5G network, so Dish keeps its asset safe, while spending less money to do so.

According to Federal Communications Commission rules, Dish has to activate a network with 40 percent signal coverage, using the  700 MHz licenses it purchased in 2008 (a deadline it will miss), or, alternatively,  reach a 70 percent buildout by March 2020. That latter target is what Dish now will have to meet. Similar requirements are in place for other spectrum Dish has rights to use.

By focusing on NB-LTE, Dish can claim it is building a network that does not compete head to head with the LTE networks operated by the other big mobile service providers, as NB-LTE aims to support machine-to-machine devices, not human end users.

Much still remains to be established. Dish has no special competence in M2M services and industries, and no operational experience with mobile services. It likely would need a partner to both build and provide retail services. Beyond that, the market size for M2M services is unproven, if considered the next wave in mobile revenues and Dish’s ability to gain enough market share also is unknown, given competition from several would-be nationwide IoT networks and IoT efforts by the other four major U.S. mobile service providers.

As always is the case when new platforms emerge, existing platforms and capabilities, as well as forthcoming technologies and features, are labeled in a way that makes them part of the future. That is the case with NB-LTE, which formally is a 4G standard.

On the other hand, the network’s stated purpose and ultimate use will be for LPWA devices, a class of applications considered the hallmark of coming 5G business models.