Wednesday, June 28, 2017

AT&T Wants to Virtualize the Access Network

AT&T will conduct a 10-Gbps XGS-PON field trial in late 2017, a key step towards . virtualizing access network within the last mile network. We tend to underplay the importance of such virtualization, as a key input into the business model.

By some estimates, operating costs represent the overwhelming portion of U.S. telecom service provider costs.

Nothing else comes close, as a driver of cost, or arguably, the overall business model, in the short term. That is true only if gross revenue remains at current levels. Any serious revenue declines would cause huge pressure for the business model.

And since the access network historically represents 80 percent of total capex or opex, anything AT&T and other service providers can take to reduce opex will be quite important for the business model.


The next-generation PON trial will aim to give consumer and business customers multi-gigabit per second (Gbps) internet speeds, with the business model being enhanced by sharing of much of the fiber investment needed to support 5G trunking infrastructure.

XGS-PON is a fixed wavelength symmetrical 10Gbps passive optic network technology. It also can coexist with the current GPON technology. It can provide 4X the existing downstream bandwidth. And it’s as cost-effective as GPON.

AT&T’s vision is to put some elements of XGS-PON in the cloud, and use open hardware and software designs. Those steps will save time needed to manage, deliver, monitor, troubleshoot and provide care services to customers.

5G is a Big, and Risky, Bet on the Future

It is easy enough to point out that we are far removed from the days when telecom was a safe, slow-growing, low-innovation industry with little competition. Telecom used to be considered a "utility" industry. It retains some elements of that heritage, but now is anything but a "guaranteed rate of return" industry.

Today, nothing is guaranteed: not revenue, not profits, not customers, not the business model, nor continued existence. So pointing out that 5G will be risky is only to note that the whole industry now is far more risky than ever before.

Big bets have to be routinely placed, and 5G is no different.

Among the potential 5G business model challenges is a mismatch between the expected value of use cases and hoped-for revenue opportunities. Put simply, the gigabit speeds 5G will enable for “enhanced mobile broadband” are not generally needed, so value will be low. But operators making investments will be looking for a financial return, so there will be pressure to price faster 5G services at higher levels than 4G. But 4G still will work for most consumer applications.

How important will 4K or 8K video, virtual reality and augmented reality really prove to be, early on? And how much more will consumers pay to get such support?

At the same time, the greatest hope for incremental revenue growth might come from enterprise apps generated by various internet of things applications, even if those take longer to develop.

At least, those are reasonable inferences from what industry executives currently believe. A survey by GSMA Intelligence found 69 percent of respondents believed enterprise apps (for IoT) would be the most-important drivers of new revenues. Just 23 percent believed consumer apps would be the most-important new revenue source produced by 5G.

In fact, three of four very-important new drivers of revenue are in the enterprise space, not the consumer space. In other words, it is easily possible to argue that 5G is the first mobile network generation of greatest value for enterprise applications, not direct consumer use cases.



Still, big risks now are simply part of the industry context. My own rule is that service providers are going to need to replace at least half of all current revenue every decade, from now on.

Whether that can be done, consistently and sustainably, only on the strength of connectivity services is an open question. Moving up the stack (getting into new lines of business in other parts of the ecosystem with higher revenue, margins and value) is imperative, not optional.



5G is a Big Gamble for an Industry That Has to Gamble Big

Among the potential 5G business model challenges is a mismatch between the expected value of use cases and hoped-for revenue opportunities. Put simply, the gigabit speeds 5G will enable for “enhanced mobile broadband” are not generally needed, so value will be low. But operators making investments will be looking for a financial return, so there will be pressure to price faster 5G services at higher levels than 4G. But 4G still will work for most consumer applications.

How important will 4K or 8K video, virtual reality and augmented reality really prove to be, early on? And how much more will consumers pay to get such support?

At the same time, the greatest hope for incremental revenue growth might come from enterprise apps generated by various internet of things applications, even if those take longer to develop.

At least, those are reasonable inferences from what industry executives currently believe. A survey by GSMA Intelligence found 69 percent of respondents believed enterprise apps (for IoT) would be the most-important drivers of new revenues. Just 23 percent believed consumer apps would be the most-important new revenue source produced by 5G.

In fact, three of four very-important new drivers of revenue are in the enterprise space, not the consumer space. In other words, it is easily possible to argue that 5G is the first mobile network generation of greatest value for enterprise applications, not direct consumer use cases.



The Internet Access Problem: Investments Not Matched by Revenue Growth

The big problem with mobile internet access or fixed network internet access is that exponential increases in supply--with investments to supply faster speeds and higher capacity--are not matched in a linear way with revenue increases that match the magnitude of investments.

In fact, some argue that most such investments now are bearing low to possibly negative financial returns. That, it is argued, is the case in the U.S. telecom market, where most service providers have negative cash flow.

Note the huge cost contribution “operating expense” represents. Nothing else comes close, as a driver of cost, or arguably, the overall business model, in the short term. That is true only if gross revenue remains at current levels. Any serious revenue declines would cause huge pressure for the business model.


EY consultants for example, argue that  “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.”

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



With one exception, you can see that average U.S. mobile internet access speeds have grown by an order of magnitude in about seven years, close to what you would expect from a Moore’s Law rate of increase (doubling about every two years).

“Looking at the maximum download speeds, it looks like there's a 2x jump every two years or so—from 50 to 60Mbps in 2014, to 120 Mbps in 2016, and now to 200 Mbps,” PCmag says.

The results were obtained by drive tests conducted by PCmag.


Average speeds, as you would expect, are lower, but still follow the “close to Moore’s Law” rate of improvement.

That is one of the surprises in the networks business. One would not expect speed or performance increases for physical networks (civil engineering projects) to improve as fast as chips. But that seems to be happening in fixed and mobile networks.

The difference is that cost is not declining as those performance increases happen, as has been the case for computing devices, for example.



Harvesting in the Linear Video Business

“Harvesting revenue” is a proven business strategy for declining markets. And it might turn out that linear video will prove to be among the most-successful examples of harvesting the communications industry ever has seen.

It will not be the first such example. Service providers long have harvested voice revenue,  access line revenue, text messaging and legacy business service revenue. Over the next several years, a harvesting strategy will happen in linear video entertainment.

And, for some time, it might be possible to grow revenues on a shrinking customer base.

Research firm SNL Kagan recently predicted that the U.S. linear TV industry (consisting of cable, satellite and telco service providers will lose 10.8 million customers by 2021. Total linear TV subscribers will be around 82.3 million at that time, which will be 20 percent less than the peak.

But some analysts believe market share in linear and OTT video still will lie with today’s providers of linear video subscription services, not the pure-play over the top providers.

In some clear ways, it already is possible to argue that the era of linear video subscriptions has passed its peak, as seen in net shrinkage of accounts. Revenue growth is another matter, as, so far, aggregate revenues are still growing, on a shrinking subscriber base.

Overall annual spending on subscription video and TV services in the U.S. market will peak at $130.3B in 2019, with revenues starting to decline to $125.7B by 2022, Strategy Analytics predicts.

Annual revenue growth for emerging players will fall to 4.4 percent by 2022, which you might argue represents “success” for a product in a declining phase of its life cycle.

Still, today’s linear video leaders, such as Comcast and AT&T, still will account for more than 80 percent of total market revenues in 2022. In part, that is because average revenue per linear account is far higher ($100 a month) than for an OTT service ($10 a month).

After 2022, it is possible OTT will begin to grow share.



5G Era: Opportunity and Peril

The 5G era will represent both opportunity and peril, and not all for reasons directly related to 5G itself. Industry dynamics are the reason.

  • Legacy revenue sources are mature or declining
  • Service providers will need to replace half of current revenue over a decade
  • Huge new revenue drivers must therefore be found
  • We are near the end of the “connectivity drives revenue” period
  • Scale always matters, but will matter more, in the 5G era

At a high level, it is easy enough to note that the industry has changed dramatically over the last four decades.

It also is clear enough to note that--in the most mature markets--all legacy revenue sources are declining. In developing markets, those trends often are not so clear, as the mobile markets are not yet mature, or at a peak, and subscriber growth continues to fuel revenue growth. Eventually, that stops.

To put matters starkly, revenue growth has stopped, in the mature markets.  

Most believe 5G is significant because it sets the stage for developing huge new revenue sources to replace declining legacy revenues. Even there, though, there are issues.

Will the new revenue sources be large enough to offset expected losses of legacy revenue? And will connectivity revenue continue to drive the business model?

Also, moving up the stack in the 5G era will require scale. Connectivity markets often are local. App businesses almost always involve huge scale. And that will favor those contestants with the ability to leverage capital and reach to assemble large market opportunities.

In other words, not every mobile operator will be able to “move up the stack,” and will have to be content with being a connectivity provider, operating with much-lower costs, since it will be very hard to grow the revenue top line.

In recent years, mobile internet access has driven revenue growth in developed markets, and is starting to assume that role in many developing markets as well, notes James Sullivan, Asia-Pacific equity analyst at J.P. Morgan.

That is why the search is on for new sources of revenue beyond voice, text messaging, linear video entertainment and even internet access.

The search will be difficult. It is fair enough to point out that telcos once were nearly exclusively in the “apps” business, first with voice, then with text messaging and more recently with video subscriptions.

The advent of internet access was a big move in the direction of “dumb pipe.” And it is hard to argue with the thesis that it is “dumb pipe” (internet access) that underpins all further generations of “telco” revenue models, even when not the exclusive revenue source.

Internet access is precisely a means to get access to apps and services, not an “app” in its own right.

So we might soon see other inklings that “connectivity revenue” ceases to be the single largest contributor to service provider revenue, even if it remains vitally important. As difficult as the effort will be, over time, growth has to come from app layer services and value, not simple connectivity (access).



Source: JP Morgan

Tuesday, June 27, 2017

App and Connectivity Firms Share Local Access Investments?

Will app providers and connectivity providers ever routinely share the cost of internet access facilities? Granted, it never has happened before. But the uncertain payback from fiber access facilities could lead to new thinking.

For fixed service providers, the growing amount of stranded assets is a key issue. Mobile service providers face different issues.

Unlimited mobile internet poses a threat to investments in capacity, in large part because there is little, if any, incremental revenue to be gained when usage is nearly unlimited and prices essentially are capped.

Eventually, that lack of return could prompt previously-unthinkable partnerships, including app providers funding optical access in the same way they often help fund (and own) subsea transport networks.

Partnerships between carriers and over-the-top players to fund deep fiber therefore emerges as a possibility, argue Deloitte consultants. But it is equally possible that more application providers will simply decide to build their own access networks.

Many will argue app providers cannot afford to do so, and will point to Google Fiber as evidence. Those objections are valid, but also must account for other possibilities in the future, as huge amounts of new millimeter spectrum are made available, including huge amounts of unlicensed spectrum.

Also, more powerful radio platforms and open source network elements will help the business case for such “overbuilds” as well.

So it is that over-the-top players might choose to fund fiber deployment by owning assets or forming partnerships with carriers.

Much as they view any other investment, service providers are likely to prefer ownership of assets that provide differentiation, and be more willing to spin off, partner or lease other parts of the infrastructure that do not provide clear business advantage, much as big carriers have spun off tower networks, data centers, network operations and technology development in general.

So shared infrastructure models could emerge for last mile fiber access. That would make access fiber a form of leased real estate.
But that is not a new, or terribly unusual problem. It has been clear for some time that there are limits to the amount of revenue growth possible from all connectivity services (voice, text messaging, internet access). In the consumer space, telco growth is increasingly reliant on video entertainment services.

In the business space, it likewise will be necessary to “move up the stack,” into the applications and platforms part of the ecosystem.Easier said, than done.

Internet of things also could provider new opportunities to move up the stack. Gartner predicts that affluent households will have up to 500 connected devices by 2022.

“In such cases, carriers could increase revenue by offering integration, network security and traffic management services,” Deloitte consultants argue.

“Another potential opportunity is working with, rather than against, over the top players,” Deloitte argues, as in the case of sharing access infrastructure costs. That has happened routinely in the undersea transport space, but would be a major change for the retail access networks business.

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...