Can New ISPs Sustain Themselves? Can Telcos?

Revenue, in addition to the cost of the network and other operating costs, is part of the complex  fiber to customer business case. Traditionally, one might have argued that fiber to the home made for tough payback models.

But we now are seeing lots of activity by gigabit Internet access providers that have decidedly different potential economics, driven mostly by the payback model.

For starters, traditional modeling has assumed a mandatory city-wide construction project, and in the U.S. market, at least, a likely ability to convert only about half of all potential sites to the status of “customers.”

Over time, service providers also have moved from a “single service” model to the “triple play,” a change that allows fixed network service providers to build a sustainable model even when only 40 percent to half of sites are “customers.”

The latest innovation is the “build by neighborhood” approach, essentially allowing service providers (ISPs, telcos, cable TV companies) to “cherry pick” their targets, instead of using the older “universal service” approach.

That approach earlier had been used by business-focused service providers (competitive local exchange carriers) to compete for business customers. The basic approach has been to target clusters of customers, using leased access to reach them, and then, over time, deploying backhaul facilities to lower cost.

The new evolution has firms building fiber-rich facilities to neighborhoods where the estimated demand is highest.

The question is long term sustainability, as the price-value relationship is redefined. We might assume investment costs per-location, and potentially, cost per customer, are lower than they were for Verizon Communications.

We might also assume costs are higher than for a new hybrid fiber coax network supporting DOCSIS 3.

That might put the cost per location somewhere between a high of $2,250 per site and $660 per location on the lower end, including activated drops, but not including any required customer premises equipment.

Since none of the new ISPs seem to be using HFC, we can assume their costs are closer to what any other telco would pay for fiber to home network elements. But current prices for telco style fiber to home networks might be far lower than in the past.

The key breakthrough would be if telco style FTTH costs were to approach cable TV HFC costs.

Assume a smaller operator could manage to connect a customer and activate service (including customer premises equipment) for about $300, where a telco connection might cost as much as $800.

Much depends on what services a service provider actually is offering. Internet access will have one cost profile, but video service will add costs.

If the network portion of cost is in the legacy $1450 per passing, then a new contestant might expect to invest $2900 per customer, at 50 percent take rates, plus a connection cost of $300, for a total of $3200.

Against that, assume $70 to $130 of revenue, in some cases. Over a three-year period, that is about $2520 to $4680 in revenue.

Some ISPs, such as Sonic.net, are retailing gigabit connections plus voice for $40 a month, though. That implies three-year revenue of just $1440 per subscriber.

All of that is before operating costs, franchise payments and marketing. Clearly, service providers are counting on longer customer life cycles, incrementally higher revenues and muted operating costs.

Perhaps prices now paid for network elements and cabling are lower than $1450 per passing, though, perhaps in the $900 range per passing. That helps, but does not make for an easy positive business case.

At 50 percent take rates, investment per subscriber at $900 per passing is $1800, plus $300 for connecting and activating a customer, or about $2100.

At 33 percent take rates, costs per customer are about $3000. As always, the business model is highly sensitive to adoption rates.

So a key expectation might be that take rates will be closer to 50 percent than 33 percent.

But the business case still is not easy.

On the other hand, market dynamics might be trending in a direction that makes the gamble less risky.

If one argues the telco business case keeps getting worse, while cable TV holds its own, in many markets, the competition might evolve towards a “cable TV versus independent ISP” model, with telcos becoming weaker competitors.

There is only so much a telco can do to reduce costs, given dividend payout obligations and union workforces, plus pension obligations.

Ultimately, there will be growing questions about sustainability, the possibility of bankruptcies or other restructurings, with the possibility that new sets of owners might be able to create a more sustainable business model. As crazy as that now seems, it cannot be excluded as an eventual outcome.

ISP "Cherry Picking" Will Have Financial Repercussions

“Cherry picking,” as helpful as it is for new entrants to communication markets, reduces the addressable market for the largest service providers that collectively serve most consumers.

That can have the unwanted effect of causing price increases for all remaining customers, as fixed costs are spread over a small number of paying customers.

That noted, nearly all gigabit network construction occurs on a cherry-picked basis, where ISPs build in areas where there is customer demand. Those neighborhoods tend to be higher-income areas, it goes without saying.

That clear tendency to focus on the lowest-cost, highest-return geographic areas makes clear financial sense--both for incumbents and attackers--but also tends to conflict with the notion of universal service.

That is an obligation typically only one provider has in any market: the incumbent telco. Eventually, that obligation might become onerous, not only for the provider of last resort, but also for customers, a declining number of which will have to pay for the fixed costs of the whole network.

Economically, that is the same problem faced by nations where the number of retirees exceeds the number of working people who pay the taxes to support all the retiree benefits. Similar problems increasingly will be faced by individual firms whose markets are not growing fast enough to generate the revenue required to support retiree benefits.

If the new trend where smaller ISPs launch gigabit networks continues, and more importantly, if most major U.S. markets start to feature three large ISPs (cable, telco, Google), the problem of stranded assets will become a big problem for telcos and cable.

The companion issue is whether the new entrants can create networks affordably enough, and maintain operating costs low enough, to compete in markets where addressable market share might be about 33 percent of homes and other locations.

Traditionally, that has proven quite difficult on a “one service” retail platform. The difference now is the additional revenue provided by triple play services, where an ISP might serve one house, but sell two to three services, boosting gross revenue past $130 a month, where it might otherwise be $40 to $50 on a single-service basis.

Still, there are clear business model issues. All three of the core triple play services are under pressure. Consumers are abandoning fixed network voice, even if that trend tends to be obscured by the very fact of triple play bundling, where consumers rationally might conclude they are better off buying all three services, even if demand is high only for two services.

With the U.S. Federal Communications Commission moving towards licensing rules that would allow over the top linear video providers access to the same programming, on the same terms, as now purchased by cable TV, satellite and telco providers, the likelihood of new OTT suppliers is growing.

And even high speed Internet access, arguably the mainstay and anchor service for any fixed network service provider, increasingly is challenged by new suppliers offering gigabit access for $40 to $80 a month, a price that often is paid for rival services operating between 15 Mbps and 105 Mbps.

The biggest strategic changes are new services provided by suppliers with lower capital or operating costs, or both. That often means a cable TV operator or independent ISP.

In the United Kingdom, for example, cable TV providers supply the overwhelming number of the fastest connections. And “Project Lightning,” the new Virgin Media investment program, will extend higher speed access to about four million additional premises over the next five years.

After the investments, Virgin Media will increase its footprint by about 33 percent, increasing its ability to sell the fastest Internet access services to perhaps 66 percent of the country.

Negative Scale Returns in Fixed Networks Business?

We might be in the early stages of another evolution of the U.S. fixed network business. The first stage had new providers, especially cable TV companies, emerge as key contestants in the fixed networks business, reducing incumbent telco market share by about half.

With Google Fiber and now a growing range of gigabit access networks by smaller firms, we might see the first of a wave of shifts to yet another wave of providers, namely independent and smaller providers.

Some might argue that if the trend continues, the smaller providers will simply be absorbed by larger providers, as has happened in the business in the past. That might not happen this time.

For starters, the advantages of scale might be diminishing, in certain respects, in the fixed networks business. In fact, some might argue the advantages of scale never have been so great in the fixed networks business, where as much as 80 percent of the capital investment has been in the local access network where costs are not shared, or not shared very much.

If that is the case, then scale might not be as big a value as has been believed. As we now are seeing many forms of decentralization and localized supply in the U.S. economy, perhaps we now are getting a glimpse of localism in fixed network supply, where many smaller and independent firms might emerge, in smaller city markets.

Though perhaps some consolidation could occur, the limits of scale might kick in fairly quickly, making large roll-ups uneconomical.

Dramatically lower prices will propel the localization moves. As retail prices dive, the advantages of larger provider scale become disadvantages, as there is less room in the business model for the overhead associated with tier one operations.

Some Internet service providers continue to push at dramatically-lower retail pricing levels for high speed Internet access, including gigabit prices. Sonic.net, for example, now is launching gigabit access plus voice for $40 a month, in Brentwood, Calif.

Sonic.net plans to bring its new service to about 8,000 homes within the next 15 months. Sonic.net also will provide a free 5 Mbps Internet access service (without phone service) for up to five years, after payment of an activation fee of $300 to $400.

To be sure, Sonic.net is a small ISP, and no matter what it does, it cannot affect many consumers in the U.S. market. On the other hand, the offer does suggest Sonic.net believes the economics work, for a bundle of gigabit Internet access and voice service, billed at $40 a month.

That does not mean a tier one service provider could do so, but a smaller firm, with lower overhead and operating costs, arguably can do so.

Paradoxically, as some claim there is too little competition in the U.S. Internet access business, competition is busting out all over. Existing municipal broadband providers have dropped prices for gigabit service from about $300 to about $70 to $80 a month.

Google Fiber has expanded to seven metro areas and another five metro areas are under evaluation. Brooklyn Fiber is firing up gigabit service for business customers in a part of the New York market, and some believe the firm eventually will start pitching service to consumers as well.

Even some incumbents, including CenturyLink, are offering gigabit service in Colorado markets, with plans to build in 16 metro areas. AT&T, for its part, has said it would build gigabit networks in neighborhoods in as many as 100 U.S. communities.    

As more “overbuilders” enter the Internet service provider business, providing more consumer choices, the exercise of that choice makes the business model harder for legacy providers (telcos, cable TV companies, satellite Internet providers, wireless broadband ISPs, municipal broadband providers).

It is easy enough to predict what likely will happen, longer term. Tier one service providers will start to invest capital elsewhere (international expansion, non-access assets, mobile or other technologies).

That will put tier one service providers that do so at a competitive disadvantage, but that also will allow more market share to be gained by smaller providers with lower operating and overhead costs.

One might note what has happened already in the tier one segment of the business. Verizon and AT&T now drive revenue volume and growth in the mobile segment, not the fixed network business.

Verizon has been divesting smaller landline markets to Frontier Communications, now perhaps the largest former-rural operator that has transitioned to a new business anchored by services provided to small business customers.

At the same time, Comcast and other cable TV companies, operating with lower costs, have begun taking serious market share from AT&T, Verizon and CenturyLink. It might not be unfair to characterize the change as cable becoming the dominant consumer services provider, while

AT&T and Verizon become mobile firms.

Price wars that destroy profit margins are virtually inevitable. In fact, some would argue, that is precisely what Google Fiber hoped would happen.  

Sonic.net , based in Santa Rosa, California, has been aggressively building its fiber to the home ISP business, using an approach similar to Google Fiber, selectively building in neighborhoods or communities where it believes it has a chance to disrupt market dynamics by offering more for less.

So far, Sonic.net has targeted smaller communities in northern California, including

Brentwood, Sebastopol, the Sunset neighborhood of San Francisco, Novato’s Hamilton Neighborhood, Healdsburg, Santa Rosa, Petaluma, San Francisco’s Bernal Heights neighborhood, Berkeley and San Francisco’s Castro neighborhood.

We might be seeing something new in the fixed network business: an age where scale economics apply in a lesser range of cases.

Does Spectrum Now Represent 80% of Dish Network Value?

The U.S. mobile business remains as unstable as ever in the wake of the recent AWS-3 spectrum auctions, which saw Dish Network acquire another $13 billion worth of spectrum assets.

In fact, some might attribute almost the entire value of Dish Network to its spectrum holdings. Conservatively, some might estimate the spectrum holdings at 56 percent of total value. Others might peg the spectrum as high as 80 percent of total value.

If the value of Dish spectrum now is about $72 billion, and the value of the video subscription business is $18 billion, then spectrum potentially represents 80 percent of total value.

Others might conservatively estimate the value of Dish Network spectrum at $36 billion, if it can be put to use in an active mobile network. Otherwise, Dish Network loses the right to use the spectrum, and its value dips close to zero, in the absence of any buyers.

Others might argue that the value of Dish Network spectrum potentially exceeds the present value of Sprint or T-Mobile US, even if Dish has no network and no customers.

But Dish Network faces perilous choices. If it cannot show 40 percent of its owned spectrum actually is active, supporting a retail network and service, by 2017, Dish loses the right to use the spectrum.

So what Dish Network does is among the biggest questions about the U.S. mobile market.

Dish Network at one point in 2013 was regarded as a legitimate bidder for Sprint, or at least significant Sprint spectrum, even if SoftBank eventually bought T-Mobile US.

Some have speculated that T-Mobile US might also have been SoftBank’s target, instead of Sprint, under some circumstances.

Illiad, the French telecom concern, actually did try and buy T-Mobile US in 2014.

Many believe Dish Network might make a bid for T-Mobile US, eventually. That is among the best options if Dish Network really believes its future survival depends on it becoming a mobile service provider.

Was a roll-up of assets--a combination of Sprint and T-Mobile US--always part of the strategic plan when SoftBank decided to buy Sprint, or was that an opportunistic development only after the strategy had been implemented?

That isn’t clear, from the outside looking in. What is clear is that the strategic terrain around Sprint and T-Mobile US continues to be volatile and unstable.

Maybe AT&T, Verizon, Sprint and T-Mobile US Don't Actually Make Money

When somebody says something “can’t  be done,” there normally are unstated conditions. Sometimes what is meant is that a thing cannot be done because physical conditions will not allow it (the “it violates the laws of physics” problem).

That’s the case for me beating Usain Bolt in the hundred yard dash.

Sometimes the unstated conditions refer to limitations of knowledge. In some cases, a specific set of people, in a specific discipline, might not know how a certain process can be made to work, but without claiming there is some actual physical impediment.

In many other cases there is only a business problem. A specific organization or entity might not be able to “accomplish something” with its present cost structure or capabilities, even if some other entity, with a different cost structure or skills, might be able to do so.

Verizon executives claimed, in 2011, that the new FiOS network would never be as profitable as its old copper network, a perhaps shocking statement.

To be sure, context matters. The old copper network largely supported operations that were conducted in a monopoly environment, with nearly-universal take rates and rate of return guarantees.

FiOS competes in a market where Verizon often has less than half its former market share, where mobile is the preferred voice network, its cable competitors are fierce rivals in high speed access and have the biggest share in video entertainment. And then FiOS competes also with DirecTV and Dish Network.

Stranded assets, beyond the cost of the FiOS network or the profit margins on new services such as video entertainment, are issues. Where it offers FiOS, Verizon might get only 30 to 40 percent video take rates, while cable gets a similar amount of share, and satellite gets the balance.

Verizon might get half of voice customer share, but overall demand for fixed network voice is declining, and is propped up by triple play bundles where a growing number of customers buy voice in large part because the overall cost of high speed access and video is more affordable.

So some might argue that firms such as Verizon might have lost money with their investments in fiber to the home platforms, for example.

The argument is that profits, over a 15-year period, will not recover capital investment.

Some might argue that is a problem for other telcos as well, which have a tough time earning a return on invested capital.

In fact, some might argue Verizon’s divestiture of lines to Frontier Communications illustrates the problem. The Verizon asset values arguably are less than Verizon spent to build FiOS.

The more shocking argument is that Verizon and AT&T might only be breaking even on mobile network investments, while Sprint and T-Mobile US might actually lose money on their mobile business.

Craig Moffett, Bernstein Research equity analyst, estimates that AT&T's financial return on its mobile network capital investment is about at  break-even, whileVerizon's return is only a bit  better. T-Mobile and Sprint are actually underwater, having invested more to create their networks than they are likely to earn.

Still, keep in mind the context. Even if Verizon or AT&T make much less money than you would suppose, while, Sprint and T-Mobile sometimes might not make money at all, that does not mean no other service providers are in the same situation.

Cable TV companies arguably have higher profit margins than Verizon and AT&T in the fixed network business. And it is not so clear that cable TV companies or others, when they enter the mobile business, might not fare better than AT&T or Verizon, as well.

The point is that as difficult as fiber to the home or mobile services might be for some providers, others might be able to do better. What “cannot be done” has to be taken in context. Arguing that “we cannot do so” does not mean others similarly cannot do so.

Mobile Device Replacement Cycles Slow

Mobile installment payment plans, replacing the device subsidy model long favored in the U.S. mobile market, have had some expected, and some unexpected impacts on smartphone adoption and replacement behavior.

Even if the new plans are largely constructed to be revenue neutral to the customer, there does appear to be a slowdown in replacement rates, according to Recon Analytics.

That was a concern when the new retail plans were created, and the concern appears to be justified, to an extent.

In 2014, roughly 143 million mobile phones were sold in the United States, and 90 percent of them were  smartphones. But that represents a drop of 25 million phones from 2013 when approximately 168 million phones were sold (only half of them were smartphones).

The decline in phone sales is caused by the rise of equipment financing plans, but also by slower new subscriber additions.

At the same time, consumers' phone purchase habits have changed significantly.

A growing number of American consumers appear to delay their phone upgrades to take advantage of the lower monthly service prices carriers offer to consumers who wait to upgrade phones at the end of their two-year contracts.

On the other hand consumers who are purchasing replacement phones are focusing on newer, higher priced devices, says Roger Entner, Recon Analytics principal.

While device sales declined by 15 percent year-over-year, device revenues increased by about five percent.

So, as expected, the shift to installment plans appears to be slowing the device replacement rate, a trend that arguably will slow innovation.

Entner expects device sales to fall by five perent to 136 million in 2015 and to fall again by four percent to 131 million in 2016.

In fact, says Entner,”handset replacement has abruptly slowed to the lowest rate since we began calculating the metric.”

Device replacement cycles are lengthening, reaching 26.5 months in 2014, an increase of 4.1 months compared to 2013.

In the past, U.S. consumers typically upgraded their phone at three points in time. Many upgraded at about one year,  when a new generation device was launched.

The next window was an upgrade every two years when the contract expired.

That now is changing.

The percentage of devices replaced at the traditional two year time point fell from 40 percent in 2013 to 16 percent in 2014, while replacement at three years grew from 15 percent in 2013 to 35 percent in 2014.

So nearly half of consumers upgrade every year, but more than a third keep their devices for three years.

Slower device upgrades might also have an effect on service provider efforts to provide more spectrum, as older devices often cannot get access to new spectrum that newer devices can support.

A six year old iPhone 3G will achieve download speeds of 2 Mbps on a 3G network, but a new iPhone 6 will be able to operate 25 times faster due to its new 4G LTE access.

To some extent, the delay in device upgrades also means it will take longer to migrate users off of congested legacy spectrum and onto new bands.

5G will be "4G on Steroids"

Despite the hype about future fifth generation network mobile networks, it is quite likely that 5G will simply be an evolution of concepts pioneered on fourth generation Long Term Evolution networks.

Among the more-important capabilities are access to multiple networks, built on today’s use of Wi-Fi and carrier networks; small cells; more complex antenna arrays; and a wider range of devices and applications.

As typically happens, new frequencies will be added, since carriers have to maintain existing networks while they add the next generation.

The degree to which millimeter wave spectrum (30 GHz to 300 GHz) will be a feature of 5G is not yet so clear, but is conceivable. The issue is not so much whether millimeter wave frequencies will be important for backhaul (they already are), but how effective they will be for distribution (direct access by mobile and other devices), using sophisticated new antenna arrays (multiple input, multiple output).

Fifth generation mobile networks, most supporters are quick to say, will not be about “faster speeds” or novel air interfaces.

Instead, at least for the moment, the emphasis is on making all access platforms available to mobile and untethered devices, seamlessly. And that will mean 5G is a development of 4G, not a sharp break from 4G.

In other words, all the fundamental concepts already exist, and all the fundamental underlying technologies already are being developed to support future 5G networks.

Those of you with long memories will see the pattern. All new next generation networks are preceded by a period where all marketing emphasizes “compatibility” with the new emerging standard. That is likely to be the 5G pattern as well.

On the other hand, 5G will bring new spectrum to bear, in particular in the millimeter wave area, where much work is being done to commercialize the use of frequencies that historically have been unworkable for communications applications.

Moore’s Law, though, allows cheap processing that enables networks to take advantage of formerly-forbidding millimeter wave frequencies (30 GHz up to perhaps 300 GHz).

In the United States, the 38.6 GHz to 40 GHz band already is used for licensed high-speed microwave data links and the 60 GHz band can be used for unlicensed short range (1.7 km) data links.

The 71 GHz to 76 GHz, 81 GHz to 86 GHz and 92 GHz to 95 GHz bands are also used for point-to-point high-bandwidth communication links.

The issue is whether sophisticated signal processing also will allow use of millimeter waves for local distribution (access) , in addition to trunking (point to point backhaul).

Observers will readily admit there are issues, ranging from propagation distances to atmospheric issues (rain fade) and even oxygen absorption. But at millimeter wave frequencies, extremely high bandwidths are possible, if distance is limited.

Still, 5G is highly likely to  be an extrapolation of trends we already see. Supporters will likely not appreciate the characterization of 5G as “4G on steroids,” but that is likely to be quite correct.