Wednesday, November 15, 2017

Auto Industry Mirrors Telecom Industry Transition

Eventually, the auto industry will make vehicles as the foundation of a range of other revenue-generating activities, but not as the sole driver of revenue, much as the telecom industry is in transition to business models that are built on the need for connectivity, but not connectivity revenue itself.

It is easy to forget that whole industries, not just products created by any industry, have product life cycles. Now it is the auto industry as a whole that seems to have reached a peak of its life cycle, something we already have seen in the telecom industry regarding voice and messaging products, might be seeing in mobility and video entertainment as well, and will soon affect internet access.


With the caveat that global trends sometimes are not reflected in some particular markets, Moody’s Investors Service says the global auto industry outlook is negative, with “stagnant or falling demand for vehicles, a shift back to larger vehicles despite new energy efficient technologies, historically high levels of lease expirations and lengthening auto loan terms” in the U.S. market.

Eventually, as the legacy revenue model erodes, the industry will try--as others have--to create a new business model. For the auto industry, that might well include a shift to autonomous vehicle transportation or other transportation services, not the production, sale and maintenance of autos themselves.

If the telecom industry provides any useful model, the model will still include “making vehicles,” as telecom builds on its connectivity platform. But as connectivity itself does not drive revenue growth, making vehicles will become just a part of the revenue stream.

It is worth recalling that the only reason the early telecom industry built networks at all--and continues to do so--is that the application it wanted to sell required such networks. LIke Facebook, Google, Netflix or Amazon, which require the existence of internet access networks for their business models, so early telcos needed telephone networks to sell voice.

Eventually, if they are successful, automakers will continue to make vehicles, but only as an underpinning for their transportation services businesses.

Tuesday, November 14, 2017

AI in Telecom: Customer Service is an Early Use Case

As was the case for cloud computing, so artificial intelligence is going to appear in consumer-facing apps where the user is not always aware of its presence. Voice interfaces provide the best example.

That also seems to be the case in telecommunications as well. But AI also is expected to play a growing role in network operations as well.

The most-popular AI applications in use by a number of tier-one U.S. telcos include customer service apps such as chat bots. In those roles, AI-assisted apps automate customer service inquiries, route customers to the proper agent, and send prospects with buying intent directly to sales people, according to Tech Emergence.

Those use cases also are obvious in the area of speech and voice services for customers, allowing customers to explore or purchase media content by spoken word rather than some other method.

In the network, AI is starting to be used for predictive maintenance, allowing staffs to fix problems with telecom hardware (cell towers, power lines) before they happen.

Likewise, AI is used to support self-optimising networks (SON). It also is possible that AI will be used to create “deep neural networks” to support customer engagement tasks with those networks.

Software defined networks (SDN) and Network Function Virtualisation (NFV) also have use cases for AI, allowing customers to interact with services behind the network, for example.

At the customer service level, AT&T leverages AI to process all “online chat interactions”. predictive maintenance as a major AI initiative within the company.

Verizon has launched Exponent, a set of services offered to other global carriers. The suite of digital tools is designed to allow customers to apply their data to personalized marketing campaigns, laser-targeted advertising, and deep customer engagement.

Comcast uses AI to support its X1 voice remote interface.

The Charter Communicatins Ask Spectrum virtual assistant uses AI to help customers with troubleshooting, account information or general questions about Spectrum services. The AI-driven assistant named Angie was designed by Conversica.

DISH Network works with Amazon to support customer use of its digital video recorder, integrating voice response with Amazon’s Alexa.

None of those customer-facing apps are likely going to produce a “wow” reaction. But all are practical, every day implementations of artificial intelligence.

Monday, November 13, 2017

Will Autonomous Vehicles Increase or Decrease Traffic?

You might think significant use of autonomous vehicles would increase--or at least not affect--primary reliance on public transportation. You might also guess that use of autonomous vehicles would reduce use of traditional taxis.

A study conducted by Boston Consulting Group suggests the former would not happen, while the latter would. The study looked at existing and expected traffic patterns in downtown Boston.


The risk of unintended consequences arguably is substantial. If autonomous vehicles make transportation  cheaper and more convenient, traffic congestion could increase.

If people use autonomous vehicles more often and in an ad hoc manner, more congestion could result.

Greater congestion could also result from a rise in certain types of zero-occupancy trips, such as when empty autonomous vehicles cruise the streets to sautonomous vehiclee on the costs of parking.

The base case assumes that 56 percent of the trips start, end or occur entirely within the 0.45-square-kilometer study area involve public transit, 33 percent involve a traditional personal vehicle and 11 percent involve taxi or ride-hailing services.

Scenario A, the evolutionary scenario, assumed a substantial shift from traditional to autonomous privately owned cars and a steady increase in the use of shared modes of mobility.

Specifically, it assumed that 11 percent of trips would be by traditional private car, another 11 percent by privately owned autonomous vehicle, 50 percent by public transit, and 22 percent by ride-shared autonomous vehicle taxi.

Traditional taxis and ride-hailing account for the remaining six percent of trips in this scenario.

Scenario B postulated a revolutionary change from privately owned vehicles to the on-demand use of electric autonomous vehicle fleets. This scenario assumed that 34 percent of trips would be by public transit, 24 percent by single-passenger autonomous vehicle taxi, 14 percent by ride-shared autonomous vehicle taxi, and 28 percent  by autonomous vehicle shuttle bus.


source: Boston Consulting Group

Business Case for Much More Fiber Remains a Challenge

Verizon’s decision to deploy NG-PON2 networks, AT&T and Verizon fixed wireless plans and use of unlicensed spectrum, plus spectrum sharing, are among the ways various internet access providers are working to improve the economics of next generation fixed networks.

As always, competition and the business model are key constraints.

Policymakers and regulators always have to balance competition and investment incentives in the communications business. Too much perceived competition lowers profits, and inhibits investment; but lack of competition tends to inhibit consumer-facing innovation.

At the moment, incentives are a key problem where it comes to next generation fixed network investment.

If consultants at Deloitte are correct, and as much as $130 billion to $150 billion in additional optical fiber investment is needed in the U.S. market, under conditions where financial incentives do not exist, something will have to give.

As Deloitte sees matters, at the moment, U.S. service providers do not have incentives high enough to support that level of deployment of new facilities. For starters, some major suppliers--including both cable TV and telcos--have high stranded assets.

If you assume a typical cable operator has revenue generated by roughly 60 percent of its passed consumer locations, while at telco gets revenue from about 40 percent of residential locations, stranded assets range from about 40 percent to 60 percent of the invested distribution network.  

That obviously creates a return on invested capital problem. Unless something changes, in terms of revenue upside, many of the largest internet service providers will have a hard time vastly increasing their optical network investment.


NG-PON2 Explains Verizon's New Trunking and Access Architecture

Not since its decision to deploy fiber to the home for consumers, replacing its twisted-pair copper access media, has any single decision made by Verizon Communications likely been as consequential as its decision to deploy NG-PON2 networks using time and wavelength division multiplexing.

Bandwidth, by itself, is not the consequential implication, although NG-PON2 supports as much as 10 Gbps per account or location, and 40 Gbps per wavelength.

The big implication is the ability to logically separate wavelengths on a single optical fiber, and then dedicate wavelengths to end user accounts. Also, separate wavelengths can be used to deliver different services to different users (or support different applications) at a single location.

In principle, wavelengths also could be assigned to wholesale customers and can reuse existing optical fiber running GPON, which Verizon does, for the most part.

“Technologies such as NG-PON2 present exciting new opportunities for vendors, such as delivering residential and business services on multiple wavelengths over the same fiber,” said Vincent O’Byrne, Verizon director of technology.

In essence, NG-PON2 logically separates delivered bandwidth from physical media, to an extent. NG-PON2 also physically separates discrete wavelengths from each other, allowing discrete wavelengths to carry different services, with varying degrees of symmetry (upstream/downstream) and capacity.

In essence, what Verizon hopes to do is create a single physical trunking network that can support services for virtually any application or use case by using discrete wavelengths, and reusing much of the existing PON infrastructure.

What clearly is different now is a new set of applications beyond the fiber to home apps GPON was designed to support. The new use cases include fiber to the building for multiple-dwelling units, enterprise networking, mobile backhaul and fronthaul, as well as cloud-based radio access networks.

Supporters also believe operating costs will be as much as 30 percent lower than other alternatives, in part because expansion can be done incrementally, reusing passive investments to a large degree, and featuring relatively-modest upgrades of active components.

Those NG-PON2 features are behind Verizon’s new thinking on optical fiber deployment to support small cells, enterprise and consumer bandwidth requirements. One way of noting the change is to say Verizon hopes it no longer will build separate trunking networks for enterprise, mobile networks and residential and small business customers.

Instead, it hopes to deploy a single physical infrastructure, and then use separate wavelengths to deliver services (mobile backhaul, enterprise, small or medium business, consumer use cases).

NG-PON2 capabilities also are the reason Altice has decided to scrap the hybrid fiber coax access platform, alone among U.S. cable operators.

Researchers at CIR say $2 billion a year will be spend on 5G trunking infrastructure through 2022, with half the annual total spent in the United States.

Chinese service providers will spend more than $130 million on 5G backhaul in 2022. China will end up being the fastest-growing market for backhaul, CIR calculated, followed by South Korea.

“The technology that will dominate 5G backhaul will be NG-PON2,” CIR analysts predict. “By 2022, more than $890 million will be spent on this technology for 5G backhaul."



The time and wave division multiplexing allows for higher bandwidth (up to 10 Gbps for any user with a total of 40 Gbps, going to 80 Gbps later) and optimal flexibility relative to bandwidth per user, fiber management, service convergence and resource sharing.

But capital expense also is expected to be 30 percent lower, with less operational complexity than dense wave division multiplexing, as well.

TWDM-PON offers up to four wavelength pairs (eight in the future) that can each be configured at different bitrates (10G/10G, 10G/2.5G, 2.5G/2.5G) to best address the specific requirements of residential, business, or backhaul services. Providing up to 10-Gbps symmetrical speeds on each wavelength.

The bottom line is that Verizon, and likely Altice, will be making the most-important change in distribution and access network design in decades.

IoT Growing Fast in Transportation; Oil and Gas, Survey Finds

A new survey of manufacturing; oil and gas as well as transportation industry executives suggests internet of things adoption has been especially rapid in the transportation and oil and gas industries.

The internet of things is not actually new in the manufacturing industry. Industrial and transportation firms in 2017 already support more than three billion devices. About 87 percent of respondents to a survey sponsored by BSquare report already using the internet of things.

Of particular importance for mobile service providers are the transportation use cases. Some 93 percent of transportation industry executives reported having an IoT application in use. In substantial part, that is because industry regulations now require emissions and electronic logging reports.

A few use cases are widespread, including device connectivity and simple data forwarding, in use by 88 percent of respondents. Some 66 percent say real-time dashboards and monitored activity are in use. Advanced analytics, such as machine learning, cluster analysis, and artificial intelligence, from data scientists are used by half of respondents.

Fully 63 percent of respondents in the transportation industry say they implemented an IIoT application within the last year. Likewise, some 66 percent of oil and gas industry say they have implemented IIoT within the past year.

Only about 33 percent of manufacturing industry respondents said they had implemented within the past year.

Other business problems IoT is expected to assist with include vehicle performance, cited as an objective by 95 percent of respondents; logistics use cases, cited by 65 percent of respondents and lower operating costs, an objective for 23 percent of respondents.

“Gaining better visibility into and control over business-critical equipment” is a top objective for respondents. More than 90 percent of adopters say device health was a driver for industrial internet of things adoption, including real-time device information, better device management, and device optimization.

Some 67 percent said logistics was an objective, operating cost reduction (24 percent) and increased production volume and better compliance (18 percent) also being desired outcomes.

As you would guess, some 78 percent of respondents of decision-makers and influencers in the manufacturing, oil and gas, and transportation industries report using internet of things in the sense of devices connected and sending data to controllers and servers.

About 56 percent say they use IoT for real-time dashboards and monitoring. Some 46 percent use machine learning, cluster analysis, artificial intelligence or advanced analytics.

Connected devices in the industrial segment are predicted to grow to more than 7.5 billion by 2020.


Sunday, November 12, 2017

How Much Revenue Boost from Gigabit or Multi-Gigabit Internet Access in Consumer Market?

Singapore is among few internet access markets where a statement such as “the jump from 500 Mbps to 1 Gbps really is not that great” actually can be made. In most fixed internet access markets, and for most potential customers, the option is not yet available.


That noted, in markets where gigabit internet access is widely available, the pricing premium for a gigabit service--compared to a 500-Mbps service--ranges from between 16 percent to 40 percent, according to Ovum.


On the other hand, at the moment, it appears that a boost to 10 Gbps offers “much greater revenue growth potential,” according to Ovum analyst Kamalini Ganguly. Optimists might well argue this is the case as faster speeds have in the past provided a rationale for higher prices, and a doubling of speed from 500 Mbps to 1 Gbps will be dwarfed by a boost of an order of magnitude (10 times) from 1 Gbps to 10 Gbps.


It would be historically correct to note that, over time, the cost per delivered gigabit of data has declined. It also would be true to note that, over time, customers consume more data. So lower units costs are balanced by higher consumption.


That is largely the reason why average revenue per account for mobile data tends to grow, over time. The only issue is how wide the revenue delta will be as speeds continue to rise, and end user consumption continues to rise.


By one line of reasoning, end user willingness to pay cannot exceed growth of disposable income, household income or a certain percentage of total household income, absent other key changes in spending habits.


In other words, if typical spending is two percent to three percent for all communications services, mobile services or data services, most buyers are unlikely to change those patterns very much, over the long term. To change the consumption curve in a significant way, spending on other products has to decrease, or household income has to increase (the rising tide that lifts all boats, even if spending percentages do not change).




Also, much hinges on the degree of competition and adoption rates, one rightly might argue. That already has been the case. If past is prologue, the revenue boost from 10 Gbps might not be as great as some hope.


So far, gigabit services have boosted internet service provider revenue only up to 11 percent, says Ovum. “Reasons for that include low premiums in competitive markets and lack of take-up of gigabit services.” In other words, low adoption rates and competitive pricing have limited the ability to sell the higher-priced services.


Ovum bases its conclusions, apparently in part, on present experience of 10-Gbps services.


“The lowest 10-Gbps price premium we found was 47 percent,” said Ovum. That was the posted retail tariff for Singapore ISP M1's premium over its own 5-Gbps plan.


“The premium increases to 173 percent if compared with the 2-Gbps services. In other markets, the 10-Gbps premium over 1-Gbps ranges from 215 percent to 327 percent,” Ovum says.


Several caveats likely are in order. Historically, though price premiums for faster-speed services, as with mobile internet plans featuring higher usage, have cost more, such price premiums tend not to last.


Among the reasons: over time, the leading offer becomes a standard offer, with standard pricing; customers tend to upgrade, but not to the highest level of service; so over time, the standard offer is deemed “good enough” by most consumers.


That might not be the case for enterprise and business buyers, who logically are the lead buyers for the fastest services.


“The higher price points mean that 10-Gbps services are aimed more at enterprises (which have higher and symmetrical bandwidth requirements) and within the consumer segment at subsegments such as high-end gamers, who can afford the higher price points,” Ovum argues. “Other target groups include SMEs with large content requirements.”


The point is that there are some reasons to predict growing spending by accounts on internet access services, based on growing consumption and faster speeds. Just how much spending can grow will be bounded by growth of household income, shifts in consumption patterns, the degree of competition in each market and advances in technology.

But some of us will argue that it is the limitation of disposable income that will be key. There is only so much a typical household or person will spend on internet access. That is less the case for business accounts, but even there, a limit remains as to how much total communication spending will happen. That also tends to be a function of entity income (revenue and earnings).

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...