Is Google Mobile Edge Cloud Another Dumb Pipe Outcome?

Google Cloud says it is helping telecommunications companies monetize 5G as a business services platform, in part by prototyping and developing edge computing use cases in retail, manufacturing and transportation. 

The Global Mobile Edge Cloud program aims to create a portfolio and marketplace of 5G solutions built jointly with telecommunications companies; an open cloud platform for developing these network-centric applications; and a global distributed edge for optimally deploying these solutions, Google Cloud says.

A collaboration with AT&T is part of that effort, where AT&T supplies the network while Google Cloud supplies the computing functions. 

That might strike you as yet another instance of telcos supplying connectivity (dumb pipe) while somebody else supplies the applications. Realists might say that is about as good as it gets, in many cases.

A few tier-one service providers might hope to create platforms or services and applications for vertical markets. Already, one can see that in prior efforts to diversify into applications and platforms. Between 2017 and 2018, for example, AT&T significantly added non-telecom revenues. 

source: GSMA Intelligence

That is crucially important for connectivity providers in developed markets, as organic growth for connectivity services is low to negative. 

“For many operators, particularly those in developed markets, non telecoms services are the only source of growth,” GSMA Intelligence says. In other words, moving “up the stack” or “across the ecosystem” might well be the only driver of significant revenue growth. 

That is a somewhat contentious statement, as financial analysts virtually always prefer that connectivity providers “stick to their knitting” and perceived core competencies, and avoid venturing further afield into new lines of business outside the core. But in a slow-growth to no-growth environment, that could very well lead to organizational decline. 

“The biggest challenge will be finding the right balance between defending core operations and exploring opportunities beyond the core,” GSMA Intelligence argues. That is an ever-present challenge for virtually all organizations facing pressure on their core revenue models. The overwhelming tendency is to spend energy and resources defending the existing business, even if that means the future is neglected. 

The Google Cloud initiative might be a reasonable compromise. It allows mobile service providers to create new value for apps and use cases that require use of 5G. And to the extent that 5G is the natural successor to 4G, it defends the legacy revenue stream (mobile service and mobile internet access). 

Still, neglecting the “up the stack or across the ecosystem” initiatives could be challenging. In 2025, for example, GSMA Intelligence predicts that connectivity services will represent about five percent of total revenues for internet of things spending. Fully 67 percent will be generated by applications, platforms and services, while 28 percent will be created by professional services. 

source: GSMA Intelligence

The Difference 5 Years Makes: Google Goes from Threat to Partner

Once upon a time, many top telco executives ranked Google as a bigger threat than other telcos. A 2015 survey of 101 service providers sponsored by Openet Telecom, including respondents from every region, found “over the top” application providers were viewed as the most-significant competitors. 

In fact,  app providers were deemed bigger threats than other mobile operators, mobile virtual network operators, Wi-Fi first MVNOs, fixed network operators or free Wi-Fi providers. 

These days, many in the telco ecosystem are partnering with Google, in the form of Google Cloud. 

Now Google Cloud has its Global Mobile Edge Cloud (GMEC), which will deliver a portfolio and marketplace of 5G solutions built jointly with telecommunications companies; an open cloud platform for developing these network-centric applications; and a global distributed edge for optimally deploying these solutions. AT&T already is working with GMEC. 

In some indirect ways, suppliers of software infrastructure for telcos are using Google Cloud as the platform for services supplied to telco customers. Netcracker Technology, for example, deploys its entire digital BSS/OSS and orchestration stack on Google Cloud. 

In part, that is because Google Cloud is pitching its platform to the telco ecosystem. Anthos, for example, is Google Cloud’s open application platform for telecommunications firms, enabling them to deploy, manage, and optimize their applications, whether they are on-premise or in the cloud, to deliver its suite of products across multiple private and public clouds, on-premise environments, and at the network edge.

Amdocs, for its part, now uses Google Cloud to deliver Amdocs’ OSS and BSS systems. 

The Most-Important Math Function for Device, App, Network Businesses

The sigmoid function arguably is among the most-important mathematical expressions one ever encounters in the telecom, application and device businesses. It applies to business strategy overall, new product development, strategy for legacy businesses, customer adoption rates, marketing messages and  capital deployment, for example. 

The sigmoid function applies to startups as well as incumbents; software and hardware; products and services; new and legacy lines of business. Describing a specific relation between sets, the sigmoid function also is required whenever neural networks are created. 

source: Innospective

The concept has been applied to technology adoption in the notion of crossing the chasm of value any technology represents for different users. Mainstream users have different values than early adopters, so value propositions must be adjusted as any new technology product exhausts the market of early adopters. Early adopters can tolerate bugs, workarounds or incomplete on-boarding and support experiences. They tend to be price insensitive. 

Mainstream users typically require fully-developed customer support, costs that match value and a developed ecosystem (they do not want to write their own apps). Scale is not a huge issue early on, since the number of customers is limited. All that changes with adoption by the mass market, when support at scale is necessary. 

The S curve also is embedded into the concept of the product life cycle or new product development. Simply put, every product eventually exhausts its market. That further implies a constant need for new product development, which must necessarily begin before the succeeding product has reached its peak adoption level, and before that product begins its decline. 

source: Medium

Among the important takeaways is that technology or product adoption is logarithmic, not linear. What happens early in technology or product availability is quite different from what happens when any technology or product is demanded by the mass market. 

Product attributes and the ecosystem required are highly disparate for early adopters, compared to mass market customers in the growth phase, which is different from attributes required to attract late adopters. 

So how does this apply to 5G?

We’ve already started to hear stories about how consumers or enterprises are “disappointed” with 5G, even though 5G availability is still rolling out, even though there are different flavors of 5G with different strengths and weaknesses (because coverage and capacity, as always, are trade-offs). 

It is worth recalling that it took 10 years--in Europe--for 3G to reach adoption levels ranging from 30 percent to 60 percent. Take rates for 4G took a decade to reach 80 percent, and about five years to reach 50 percent adoption. 

If 5G is close to 4G in value, it will be about another five years before half of consumers actually buy the service. That is a good illustration of the S curve adoption model, something that applies to new services of all types, provided by incumbents or startups. 

As applied to 5G, it is easy to understand why early mixed reviews are understandable. The biggest performance boosts come with millimeter wave service that is going to take some time to supply, meaning few users actually have sustained use of that form of 5G. 

The other issue is the “obvious experience advantage” of 5G over 4G, compared to the difference in experience between 4G and 3G. Many do not recall, or did not experience the transition from 3G to 4G. Simply put, 4G brought immediate and obvious improvements in user experience of using the web from a mobile device, where 3G experience was painful.

That will not generally be the case for the transition from 4G to 5G. There are almost no use cases consumers will generally encounter or desire where 5G speed, capacity or latency advantages translate immediately into better experience. 

In other words, 4G service quality is quite good, compared to 3G experience when 4G launched. In fact, many users on low-band networks might not always even detect a significant difference in experience. 

As we already have encountered with fixed network performance, gigabit per second speeds--compared to services offering 100 Mbps to 200 Mbps--actually do not yield tangible experience benefits for any single user, though useful for multi-user households where simultaneous 4K streaming happens, lots of simultaneous gaming occurs or when multiple users are uploading lots of video content. 

It can be argued that 5G launches represent that same sort of situation: the capacious millimeter wave services cannot generally provide experience gains because 4G suffices (for the moment). 

That is bound to lead to some user disillusionment. 

The story can be quite different for a mobile service provider, deploying 5G in part for other reasons. As end user bandwidth demand continues to grow, there comes a point where 4G just runs out of room for improvement. 

That matters because cost per bit matters. Basically, mobile operators have to keep supplying more bandwidth to end users, but at about the same retail prices. There is some room for improvements at the margin, but the trend for decades has been that consumer prices have remained the same, or fallen, while the supply of bandwidth has increases, in some cases, at about the rate one would expect from Moore’s Law (doubling about every 18 months to 24 months). 

So end user experience “at the moment” is not the big issue. Supporting user experience in a few years, when the 4G network cannot do so at lower cost, is the big issue for a bandwidth supplier. 

Eventually, consumer benefits will be seen. But even so, lower cost per delivered bit would be reason enough for mobile operators to move to 5G now, before the next capacity crunch hits.

The ultimate creation of new services, apps and use cases, some of which will provide direct and indirect revenue, also are important. But the move to 5G is supported--one can argue--strictly by the need to deliver internet bandwidth at far lower cost. 

Latency and capacity improvements are nice, and have happened with each succeeding digital generation. And those improvements have lead directly to new use cases and value creation. Still, the bottom line is that mobile networks must drive down the costs of supplying internet access. 5G does that.

So Moore's Law Is Not "Dead"

Though it is logical to worry that Moore’s Law improvements cannot continue, the latest analysis of transistor counts by IC Insights suggests the law remains intact. Though growth rates in some product categories have slowed, doubling of transistors per chip every two years remains a guideline that the industry continues to follow,” IC Insights says.

That is important because Moore’s Law allows computing and storage costs to keep dropping, while performance keeps increasing. That enables any number of useful innovations, ranging from applied artificial intelligence to voice-controlled appliances of all sorts, more-precise e-commerce operations, automated and robotic machines and higher-resolution video. 

Moore’s Law that states there is a doubling of the number of transistors per chip every two years.  A corollary is that price remains the same or drops as well. 

source: IC Insights

Many argue Moore's Law is dead, which would have vast implications for all products and services using computing, storage or information processing, if alternative means of continually boosting device and application performance could not be found. But such alternative means already are recognized.

Wireless Has Transformed Network Architectures, More Coming

Wireless and the internet have radically transformed access network topologies. In the past, the architecture featured a long distance (core) network, distribution networks (metro trunking network) and access networks (local loop) plus premises wiring. 

Mobile networks--depending on how one wishes to describe it--substitute a wireless access network for cables or add a wireless tail to the cabled access network. These days, the indoor private network tends to use Wi-Fi rather than cabled Ethernet. 

source: Ericsson

Wireless local loop is a concept some service providers have employed for many decades, largely as a solution for customer access in lower-density scenarios, and initially seen as a voice solution.

 Wireless internet service providers and satellite providers are other connectivity providers that have for decades used either point-to-point or point-to-multipoint microwave networks to provide communications. But many thought DECT might become a standard for wireless local loop, in the mid-1990s. 

Satellite direct has been a delivery technology since the 1980s for video and other services, including both direct to consumer and master-antenna services aimed at residential complexes. Geostationary satellites provide internet access today, while new constellations of low earth orbit satellites are likely to do so in the future, with additional use of non-traditional platforms that might include free-floating balloons or other high-altitude platform systems (HAPS).  

source: Semantic Scholar

The point is that wireless “last mile” or “first mile” approaches have become more common since the early 1980s, reaching significant adoption in the 4G era in the form of mobile internet access, sometimes as a substitute for fixed access. Many believe 5G fixed wireless will be more attractive, as it will better match cabled network speeds and prices. 

Fixed wireless based on 5G will allow new entrants such as T-Mobile to attack the home broadband market that has, up to this point, been dominated by cable TV companies. That can be quite attractive for an attacker with zero present market share. 

Fixed wireless using 5G might also provide new opportunities for existing providers where other approaches, including fiber to the home, do not seem especially attractive. 

But network architectures are affected in other ways, as well. Small cell networks extend the metro optical fiber network much deeper towards end users, creating a “fiber to the small cell” network. 

At the service level, the 5G network core being fully virtualized, network slices through the core network, terminating at wireless and mobile endpoints, represent a new level of service assurance formerly possible only with virtual private networks.

Moore's Law Matters for Rural Internet Access

At some point, Moore’s Law matters for rural internet access supply and costs. The reason is that communication networks serving low-density areas are expensive, but Moore’s Law materially reduces such costs by constantly increasing the power of computing solutions and slashing the cost of such capabilities. 

That can be seen in the cost of a transistor over time, which allows the cost of computing to decline in half every 18 months to 24 months. 

The reason we are able to use millimeter waves commercially for consumer communications is Moore’s Law, which reduced the cost of applying sophisticated signal processing so much that the formerly-unusable millimeter wave spectrum now can be used even for consumer communications and access networks. 

The Tennessee Advisory Commission on Intergovernmental Relations estimated in a 2017 report that connecting 160,000 unserved homes in areas of the state ineligible for funding through the FCC’s Connect America Fund would cost $125 million to $799 million. That works out to about $78,125 per location. Other estimates place network costs at $33,000 per customer even at 70-percent take rates. Lower take rates raise the cost per customer. 

No consumer-reaching commercial network could ever hope to break even on a network with such high costs, as even networks costing $2,500 per location often face challenging economics. 

source: NTCA

But Moore’s Law, applied to space-based network launch costs, for example, has reduced costs for low earth orbit satellites by 20-fold over the last decade. 

Moore’s Law also allows us to do sophisticated spectrum sharing and spectrum aggregation, again improving our ability to supply connectivity services at lower costs than was possible in the past. 

Moore’s Law also powers the increasing bandwidth of optical fiber and hybrid fiber coax, enables the application of artificial intelligence to voice processing and pattern recognition, all of which mean e-commerce insights and customer service capabilities grow more powerful over time. 

Moore’s Law underpins our ability to create virtualized networks that cost less to build and operate, as well. 

That is not to underestimate the importance of financial subsidies or human agency,, as in many rural areas even the most-efficient platform might fail to generate sufficient revenues to sustain operations and service. 

Lower costs per bit for mobile bandwidth as well as fixed network  capacity have been the trend for decades. Slowly, those advantages will accrue in rural areas, even if the bandwidth gap between urban and rural areas does not completely close.

India Mobile Battle Over Revenue Recalls U.S. Battle over Common Carrier Regulation

Adjusted gross revenues taxes represent four percent to eight  percent of India mobile operator gross revenue. But there had been a dispute between mobile operators and the Indian government about the definition of gross revenue, resulting in a 14-year battle that ended in 2019 when the highest Indian court upheld the government’s definition. 

source: Bloomberg

At stake is possible sustainability of Vodafone-Idea, as the firm does not generate enough cash flow to pay the owed amounts. Bharti Airtel owes a sum representing perhaps three full quarters of cash flow. The owed sums also represent something like a fifth of Airtel and Vodafone Idea debt. 

The government said AGR included all revenue, from all sources. Operators sid it applied just to connectivity revenues earned from communications service supplied to customers, and not to any other sources, such as interest income, for example. 

The disagreement is reminiscent of the distinction between regulated, rate-controlled telecom service and non-regulated services supplied by U.S. telcos. That dispute also concerned the regulation--and hence taxes upon--new services, as distinct from legacy services.

That important difference grew in 1960 as then-monopolist AT&T began to offer data processing services, with a Federal Communications Commission decision referred to as Computer I. The decision established a principle that data processing would not be regulated under common carrier rules. 

The issue was further clarified in 1979, when a regulatory distinction was drawn between “basic” service (common carrier regulation) and “enhanced” service (computing or data services, for example, which are unregulated) by the Second Computer Inquiry, generally referred to as Computer II. 

If a service provider offered a pure transmission over a path that is transparent in terms of its interaction with customer supplied information, the FCC considered this to fall into the basic service category. 

Basic service includes processing the movement of information and computer processing, which includes protocol conversion, security, and memory storage. The category of basic service is everything from "voice telephone calls" to a phone company's lease of private line service. 

If a carrier offers services over common carrier transmission facilities that employ computer processing applications that act on the format, content, code, protocol or similar aspects of the subscriber's transmitted information; provide the subscriber additional, different, or restructured information; or involve subscriber interaction with stored information, the FCC considered that to fall into the enhanced services category, not governed by common carrier rules.