5G Will be Built on Innovations First Seen in 4G

It now is clear that 5G will be built on 4G, in several different ways. Consider only the issue of bandwidth, which will increase in the 5G era by an order of magnitude or two orders of magnitude over 4G.

No single technology element will drive all or most of the speed increases. Instead, a few different approaches, used together, will be key:

• New spectrum
• Unlicensed spectrum
• Spectrum sharing
• Wider channels
• New radios
• New modulation
• Small cells
• New devices and chips

All of those elements, together, will be pioneered first for higher-speed 4G networks, and then will be developed in a 5G context.

In the near term, small cells, new chips and devices, better radios and modulation and shared spectrum will have the most impact. Most of the new spectrum will come later. The reason is simple: most of the new spectrum lies in millimeter wave regions beyond 6 GHz, and have not yet been licensed for commercial use to support 5G.

But it is fair to note that all the other fundamental improvements will first see commercial application to support 4G networks. In the spectrum (physical capacity) area, several approaches--new, unlicensed and shared spectrum--will be key, and in many ways are inseparable.

The biggest innovations come in the use of shared spectrum, in several ways. First, new dynamic ways of enabling spectrum use will be commercialized, using methods such as the Citizens Broadband Radio Service (CBRS), which allows new commercial users access to licensed spectrum allocated to government users, without relocating those users, using new spectrum access systems.

source: Qualcomm

Also, 4G will be the first mobile network generation to make use of coordinated aggregation of licensed and unlicensed spectrum that essentially bonds those assets to support 4G access.

This is a step beyond “offload” of traffic from the mobile network to Wi-Fi, and actually integrates licensed and unlicensed resources so that, from a device or user perspective, both sets of spectrum assets are part of a unified access network.

In other words, spectrum sharing will happen within existing licensed bands, as well as across licensed and unlicensed bands.

Use of better radios (multiple input, multiple output) also will be seen first in 4G, and will boost throughput about four times, ultimately.

source: Qualcomm

Taken all together, the enhancements first seen in 4G will continue with 5G, creating something historically new in the retail parts of the mobile and telecom business: capacity abundance where all prior networks have been bound by scarcity.

That has huge business model implications. To the extent that all telecom business models and profit margins have been built on “scarcity,” the fundamental underpinnings of the business will change.

When there is abundance, economics teaches, unit prices fall. Scarce goods with known demand always get “cheaper” in terms of retail price when there is much more supply.

Like it or not, in the “access” part of the telecom revenue model, “dumb pipe” economics and business models are going to hold. That trend has been made clear enough in the whole internet era as more retail bandwidth has been accompanied by lower prices per bit.

In other words, over time, the retail cost of consuming internet bandwidth has dropped continually. That, in turn, has enabled (or forced) near-zero pricing levels, which is why telcos always fear becoming simple “dumb pipe” providers and why the search for ways to create roles “up the stack” also is considered so important.

That will be the key challenge for mobile operators in the 5G era: even as the network itself offers vastly more bandwidth, the retail price of units of bandwidth will keep falling, forcing ISPs to look elsewhere within the ecosystem for roles that have higher profit margins and different revenue streams.

"Fiber to the Home" is No Longer the Right Answer, or the Right Question

For the last 30 years, in many quarters, the generic answer to almost any retail telecom problem (bandwidth, new services, operating cost, future-proofing) was “fiber to the home.”

It is reasonable to ask, given the outsized role now played by mobile networks (and despite the acknowledged value, in some markets, for some apps and business models) globally, whether “how to do FTTH, when, where and why,”  is even the right question to be asking.

It is relevant to ask how best to supply gigabit-per-second internet access (eventually) to every potential user.  

source: Ericsson

Part of the issue is that fixed voice lines have been in decline since it peaked at 1.3 billion in 2008, with significant reductions in developed countries. That means a growing “stranded asset” problem, as access network investments must be paid for by a shrinking base of customers.

That effectively raises the “investment per customer” threshold.

In Europe, adoption rates (as percentage of population, not location) dropped from 45 percent to 26 percent in the last 10 years, according to Statista. That accentuates the problem, as homes and small businesses buy lines “per household” or “per location,” not “per person.” Enterprises buy trunks, not lines.

Europe: Percent of Population Using a Fixed Voice LIne

source: Statista

The global peak for fixed line subscriptions appears to have happened about 2006, with continued declines, going forward. In the United States, fixed lines peaked in 2000 or 2001.

source: FCC

Global Fixed Line Subscriptions

source: ITU

In the U.S. market, fixed line subscriptions (as a percentage of households) dropped from 95 percent to just over 50 percent over the decade starting at 2000.

Customers shifted purchasing of fixed network services from voice to internet access over the same period, partly offsetting voice line losses. Fixed broadband worldwide grew from 220 million to 794 million over a decade.

The point is that “fiber to the home” is not the answer to revenue, profit of growth problems, as we have passed the peak of fixed network subscriptions, as a percentage of total human network connections. We might still see incremental, but low growth, globally, in the fixed network subscription area, but growth will be modest.

The lesser problem of “how” to supply internet access therefore is better phrased as “how do you provide gigabit per second speeds to every location or device?” Formulated that way, mobility and fixed wireless often will be better answers than “fiber to the home.”

“Fiber to the home” is not the universal (or mostly universal) answer to the question of “how do we create a platform for growth?” FTTH is not even the answer to the question of “how do we provide gigabit per second speeds to consumers?”

What we care about is “gigabit access speeds.” Physical media is a lesser--and more tactical--problem, with more answers than once seemed possible.

Cable modems on hybrid fiber coax networks, satellites, fixed wireless and even standard mobile access are potential answers. More exotic solutions might yet also arise (balloons, unmanned aerial vehicles).

AI Can Boost Productivity 40%, Accenture Argues

Artificial intelligence has the potential to increase productivity 40 percent or more by 2035, according to Accenture and Frontier Economics.

AI could increase economic growth rates by an average of 1.7 percent across all industries through 2035, according to consultants at Accenture.

Information and communication (4.8 percent), manufacturing (4.4 percent) and financial services (4.3 percent) are the three sectors that will see the highest annual gross value added growth rates driven by AI in 2035.

The bottom line is that AI has the potential to boost profitability an average of 38 percent by 2035 and lead to an economic boost of $14T across 16 industries in 12 nations by 2035.

Artificial intelligence adds value by boosting the productivity of some other process or product. AI is not so much a “big new product” as a feature of a cloud computing service, analytics capability or an enabler of a product’s features (voice recognition for home computing appliances).

source: Accenture

source: Accenture

How Do You Measure the Size of the Artificial Intelligence Market?

It is very hard at the moment to define with much precision the “size of the artificial intelligence market,” for some of the same reasons that it is hard to quantify the size of “e-commerce” or mobile payments or banking.

Are we measuring the value of transactions, the incremental revenue generated by new marketplaces, the value of new software, hardware or services to create, install or support such transactions?

Depending on how we approach quantifying “the market,” it is possible to come up with big--but almost useless numbers.

With artificial intelligence, the cleanest way to measure might be the sales value of hardware and software that creates the AI capabilities. But that is a problem, as well. Most of AI value is not derived by “selling” the capability, but by using the capability to support something else that actually represents a current line of business activity.

So AI adds value by boosting the productivity of some other process or product. AI is not so much a “big new product” as a feature of a cloud computing service, analytics capability or an enabler of a product’s features (voice recognition for home computing appliances).

In other words, if AI, or machine learning, is a feature of a computing operation or process, most of the actual value will be embedded in some other product, and will remain difficult to measure. The direct, incremental and “new” markets are mostly software products that provide AI functions. And that is a difficult figure to create.

Indeed, the AI “outcomes” are mostly of an intangible nature: what is happening? What should I do?

source: R Ray Wang

At the moment, most of the actual AI activity is acquisitions of small software firms able to supply the AI functions.

source: CB Insights

IoT: Billions of New Connections by 2020

The total number of IoT connected devices (not including wearables) is expected to grow from 1.6B in 2014 to anywhere between 20B (conservative view) and 46B (disruptive view) by 2020, according to Bell Labs.

Of this total, mobile IoT devices will be between 1.6 billion and 4.6 billion in 2020. Despite this massive adoption and traffic growth of 50 to 70 times from 2014 levels, traffic load is not the issue. All traffic generated by IoT devices will only account for two percent of the total mobile traffic by 2020.

In contrast, the signaling requirements will be significant. A typical IoT device may need 2,500

transactions or connections to consume 1 MB of data.  In the disruptive view, daily network connections due to mobile IoT devices will grow by 16 times to 135-fold by 2020 and will represent three times the number of connections initiated by human-generated traffic.

source: Bell Labs Consulting

Does LPWAN Complement or Compete with 4G, 5G?

Nobody can yet say for sure how the various internet of things access platforms will fare in commercial markets. Some will argue that the specialized IoT access networks using unlicensed spectrum have (and might continue to have) cost advantages over networks based on use of licensed spectrum.

Others argue that device costs and retail prices eventually will become equivalent. It is clear that “wide area” or “long range” connectivity is where the IoT wireless platforms will compete. In many “local” cases, IoT devices and apps will use existing Wi-Fi or other “close by” connections, then connect to fixed networks.

The issue for connectivity service providers is that such connections do not necessarily increase revenue for access providers.

source: Sigfox

source: Sigfox

4 Competitive Dimensions: Connectivity is Just One of Them

By 2025--barely eight years from now--connectivity will be only one of four key competitive dimensions in the telecom industry. “Control, content and cloud” will be the other three key dimensions, according to researchers at Bell Labs.

You can call that an example of why “moving up the stack” now is an imperative, not a “nice to have” example of generating “ancillary” revenue. In other words, value and revenue are likely to hinge on analytics and insights, as well as marketplaces, services and apps that build on insight.

Moving up the stack typically requires both resources and scale beyond the capabilities of most smaller telcos. That is one reason why a huge wave of industry consolidation, in part to achieve customer mass and scale that leads to higher gross revenue and profit, is going to happen over the next decade or so.

Moving up the stack is going to require capital, human resources and geographic scale that only the very-largest firms will be able to attain. And most of the largest firms will get there by acquisitions, both horizontal and vertical.

Horizontal acquisitions will be for scale in the basic connectivity business. Vertical acquisition will be to acquire applications and insight assets that represent industry growth and value propositions.

source: Bell Labs