Service Provider R&D Spending is Low, for Reasons

Many will note that connectivity provider research and development spending is low, compared to spending by major app providers, for example. There are lots of reasons for that state of affairs.

In the U.S. market, for example, technology development in the monopoly era was conducted not by the operating units but by Western Electric, which after the Bell System breakup was separated out as part of AT&T, then privatized as Lucent, before sale to Alcatel and then to Nokia.

In the competitive era, most of the research and development has been done by industry suppliers, in part because that always was the way most such activity happened. That might be seen in R&D budgets for industry suppliers, as opposed to service providers.

Telco Research and Development Spending, $ Million
		2013	2014	2015	2016	2017
AT&T	R&D	1,488	1,730	1,693	1,649	1,503
	As % of sales	1.2%	1.3%	1.2%	1.0%	0.9%
BT	R&D	226	116	97	81	79
	As % of sales	0.9%	0.5%	0.4%	0.3%	0.2%
Deutsche Telekom	R&D	114	112	127	99	68
Orange	R&D	918	861	854	830	824
	As % of sales	1.9%	1.9%	1.8%	1.7%	1.7%
Telecom Italia	R&D	48	65	61	52	51
	As % of sales	0.2%	0.3%	0.3%	0.2%	0.2%
Telefónica	R&D	1,231	1,307	1,241	1,066	1,014
	As % of sales	1.8%	2.2%	1.9%	1.7%	1.7%
Ericsson	R&D	3,701	4,172	4,000	3,632	4,356
	As % of sales	14.1%	15.9%	14.1%	14.2%	18.8%
Huawei	R&D	4,632	6,168	9,001	11,536	13,544
	As % of sales	12.8%	14.2%	15.1%	14.6%	14.9%
Nokia	R&D	3,082	2,292	2,502	5,880	5,784
	As % of sales	20.6%	16.6%	17.0%	21.1%	21.2%
source: Light Reading

More recently, R&D investments have become less directly correlated with spending, however. One is the trend towards use of open source, which tends to mean development is less correlated with actual spending. Network functions virtualization also means that less capital has to be invested in hardware to achieve the benefits of software advances.

But some might note that connectivity providers simply have less profit from operations to invest in research or development. In a perfect world, with much-higher profits, service providers might have the ability to invest more. As a practical matter, they cannot.

The Big Difference Between Consumer and Business Adoption of New Products

SIP trunking is among communications markets that never has grown the way I would have predicted, namely with a rather-clear inflection point, as has been the case with consumer innovations such as the internet itself, PCs, mobility and smartphones.

In retrospect, I suspect that is a difference in adoption rates of connectivity products in business segments, compared to consumer markets. Simply, business adoption of products more often is linear; consumer product adoption non-linear.

With the caveat that most products and services do not become mass market successes, consumer adoption of most popular products is exponential and non-linear, with an S-curve adoption pattern.

source: Forbes  

source: Visual Capitalist

Few business products ever seem to have that sort of exponential growth.

source: Maximize Market Research

source: IDC

source: Grandview Research

source: IDC

source: Thorn Technologies, Statista  

The point is that business products might well not have a clear inflection point, or an S-shaped adoption curve, even if consumer products almost always do have such adoption curves.

5G Does Not Change Traditional "Capacity Versus Coverage" Issues

One way of making sense of claims about 5G performance is to recall that the traditional trade off between coverage and capacity will continue to hold, even as huge new blocks of millimeter wave spectrum are released for use.

Ultimately, all the noise about strategy and assets and advantage aside, our knowledge about existing use of mobile and other devices should provide a clue to network and spectrum choices.

Many note that perhaps 80 percent of data is consumed by mobile users who are stationary, in their homes, offices or other non-moving locations. That, in part, explains the high use of Wi-Fi by smartphone customers.

The obvious implication for 5G networks is that although coverage (with lower capacity) works for outdoor and mobile use, the places where millimeter wave spectrum really adds value are locations where users are stationary (and might otherwise use Wi-Fi).

So we might as well understand, early on, that peak 5G speeds (gigabit) will tend to be found only in urban and other high-density areas where people live and work. Though 5G will supply faster speeds in mobile scenarios, speeds are unlikely to reach those possible in fixed locations, simply because the physics does not support it.

Though it is easy to debate the merits of using spectrum below 6 GHz, or in the higher-frequency millimeter bands (24 GHz, 28 GHz, 39 GHz for example), the traditional trade offs between coverage versus capacity will still have to be made.

The only difference is that the range of frequencies will shift. Where 2 GHz once was seen as “high frequency” with better capacity performance, while 800 Mhz was better for coverage, now any spectrum below 6 GHz is seen as the “coverage” suppliers, while millimeter wave spectrum is seen as the capacity bands.

source: GSM History

Ultimately, we are likely to see all service providers rely on sub-6 GHz for coverage, even if that means sacrificing capacity (speed). To supply the gigabit per second speeds 5G makes possible will require the millimeter bands, but with coverage limitations.

To be sure, millimeter saves also offer greater spectral efficiency than lower-frequency signals. Simply, the higher the frequency, the higher the potential bandwidth.

How Big an Opportunity for New Content Aggregators?

A survey of consumers in the United States, United Kingdom and Brazil suggests they are receptive to any content provider that could supply most of the programming they want, and might be willing to spend more money to get such a service.

The good news is that there might be an opening in the market for an aggregator that could do so. The bad news is that it likely is impossible for any provider to do so. Content rights are so fragmented it is extremely doubtful any single provider could aggregate most of it.

On the other hand, the survey, by Vanson Bourne, and sponsored by Amdocs, also suggests there could be new opportunity for any supplier able to do a better job of bundling available content.

Perhaps the biggest unknown is whether consumers actually are willing to spend significantly more than they do at present to get “all” or “most” of their desired content. Most consumers likely would be unwilling to spend what it would take to aggregate “all” desired programming, under any circumstances.

The practical challenges is to assemble a package that is “good enough,” not “best,” as consumers will pay for “good enough,” and highly unlikely to pay for a service that provided nearly all desired content.

source: Amdocs

Even consumers who buy four or more discrete video services say there still is desired content they cannot watch.

Respondents suggested that, to view all of the content they want to watch on a regular basis, they would need to pay almost 50 percent more (about $30 a month) than they're currently spending, meaning a total of $126.59 per month or over $1,500 per year.

That probably should not be taken to mean they would spend that much; rather that this is what they would have to spend, at the moment, to get all of their preferred content.

The results might suggest to some that consumers are willing to spend significantly more than they presently do on video entertainment, even if cord cutting trends suggest consumers want to pay less than they already do, even if there are content limitations.

Others might suggest there is some additional demand, especially for providers that can package more of the desired content into a single service.

The study--not surprisingly--found that 68 percent of U.S. viewers aren't satisfied with the range of TV and video content they currently have, despite spending an average of $85.71 per month on TV, movie and video subscription services per household.

It also is worth noting that 58 percent of respondents with access to a subscription service say they also do not pay for at least one of those services.  

Some 70 percent of U.S. respondents stated that they would be prepared to pay for a single provider that could package all of their preferred content into a dedicated service bundle. Whether that is possible is the issue.

Some 69 percent of consumers said that they would be happy to switch content providers if such a single package were available.

The typical household spending about $72 a month on video subscriptions, the study suggests.

U.S. Household Spending on Communications is Quite Low

When assessing impact, it helps to focus on the few inputs that tend to drive most of the output. That is true for carbon footprint or household spending. The greatest portion of household spending in most countries comes in three categories: housing, food and transportation.

There are some outliers. Transportation is unusually high in Mexico. Healthcare spending is unusually high in the United States and food spending is above average in Russia.

Still, generally speaking, households can only make significant changes in those three areas, as those three areas might represent 65 percent of total spending.

source: Visual Capitalist

All household communications added together rarely represent more than a couple of percent of total spending. The point is that no matter how much a consumer seeks deals in that area, it does not affect total spending very much.

Entertainment spending--which includes what people spend on their pets--is probably twice to three times as high as communications spending, depending on the size of the household.

source: Bureau of Labor Statistics

The other obvious implication is that communications really does not cost too much. Prices are not “too high” when total communications spending represents such a low percentage of total spending.

In fact, communications spending is a small enough category that the U.S. government does not break it out when reporting household spending.  

Magnetic North is Drifting Faster

Magnetic north has been drifting slowly since 1900 and seems to have been drifting faster since 2000. The magnetic field is caused by the earth’s molten core. Navigation systems have to be adjusted to account for the changes,  as a result.

source: Nature.com

Pricing Models in Precision Agriculture (IoT-based Services)

Revenue models for precision agriculture (use of internet of things sensors, apps and services) might not be directly applicable in the connectivity business, but will have to be understood if service providers decide to explore roles in distribution (sales channels) of such services.

Retailers who offer precision agriculture programs often charge by the acre, per instance, percentage of yield or per unit sold.

That roughly corresponds to connectivity business pricing models based on flat rate, pay per use, revenue sharing or usage volume.

Pricing Model	Advantages	Disadvantages
Bundled Per Acre Flat per acre rate with all “precision services” offered by the retailer included in the plan.	Ease of billing Farmer knows cost up front Separates customers into two separate groups (participating in precision program or not). Allows for easier analysis of yield trends within each of the two groups. Allows group data analysis to be presented to entire group in the program to improve production within the group. Likely to create a closer relationship for all inputs with the grower (less shopping around). Easier to build in a call center/service center fee.	Large upfront commitment for producer Potential to be an expense that is cut during low profit margin years Each additional service outside of package requires surcharge Difficult to “dabble” with just a few acres. Farmer usually must commit to all or none. Machinery limitations of grower may not allow full implementation of program Accurate data analysis requires trustworthiness of grower and operators to properly adjust/calibrate equipment
Per pass or service Each pass across the field or individual service has a price associated with it.	Able to charge a higher total price vs. bundled model due to less “sticker shock” Allows producers to pick and choose their individual needs Allows company to analyze margins of each individual service Ease of entry/exit from use of specific services Grower and retailer work together to develop best plan for each operation	Bookkeeping becomes more challenging as each service needs to be tracked separately. Harder to prove benefits of the service as there is no “Large group” of producers doing same practices. Requires higher effort from salesforce as each individual product or service must be “sold” Easy for grower to say “not this year” on certain services
% of yield bump/gain (example: a ten bushel/acre yield bump was added to a grower’s average and they share four of those bushels with their service provider)	Places accountability with the service provider to deliver results.  “The saying put your money where your mouth is” applies here. Grower is more likely to commit as it is obvious that they only pay if they receive a yield gain Potential for higher adaptation by grower and producers Keeps service provider in closer contact with grower throughout the year as their paycheck depends on the grower’s success! This is a unique approach, sure to make conversation as farmers talk to farmers, making for self-advertisement.	Delayed collections as retailer must wait for harvest to be completed Determining baseline yield establishments Weather variability Measurement of yield (yield monitor, grain cart, certified scale tickets?), Is a yield monitor calibrated by the grower accurate enough? Determining a market price for the grain Risk of negligible or no yield bump equates to no payment to service provider May “discourage” growers to shoot for higher yields No consideration for reduced inputs. Model may not be sustainable long term
Per unit of product sold Service is bundled with each unit of fertilizer, chemical, seed, etc.	Ease of bookkeeping Services buried within the price of a retailer’s commodities Spreads the overhead cost of precision ag professionals and technologies over all customers	Encourages growers to cut back on inputs Easier for competition to undercut a retailer’s price if grower doesn’t see value of buried service charges

source: PrecisionAg

How do You Know a Process is Using Machine Learning?

Most of the time, the typical consumer or worker encountering any form of artificial intelligence is interacting with some process enhanced by machine learning. ML is based on algorithms and statistics to find patterns in massive sets of data.

These algorithms use statistics to find patterns in massive amounts of data. They then use those patterns to make predictions.

source: MIT Technology Review

AI is a Feature, Not a Product

Channel partner organizations in both communications and information technology industries already are trying to figure out what they can sell to business customers in the artificial intelligence area.

The problem is that, for most products, AI is an attribute or function, not a stand-alone product. As applied to information technology operations of various types, that means AI is a feature of a product, not the actual product that is viewed as the solution to a business problem.

source: Tractica

IS Cost of Internet Access a Problem? Yes and No

People sometimes complain about high prices for U.S. internet access. As a percentage of household income that is not true, in the United States or any other developed nation, but people believe the “high price” charge anyway.

According to the International Telecommunications Union, prices adjusted to reflect purchasing power parity across nations suggests mobile broadband cost perhaps $15.90, in 2015, in developed nations, compared to a world average of $26.70 and far lower than the $30.80 paid in developing countries or $39.90 in lesser developed countries.

Looking at fixed broadband prices, customers in developed countries paid about $27.80 in 2015, compared to the world average of $56.30, the developing country cost of $67.30 and the lesser developed country average of $134.

LIkewise for mobile broadband, which in developed countries costs one percent of gross national income per person, compared to the world average of five percent, the developing country average of perhaps 7.5 percent and the lesser developed country average of about 16.5 percent of GNI per person.

source: ITU

Population density and country size do play a role in the cost of providing service to customers. Large countries and countries with large rural areas will find high costs to serve customers in rural areas. In Canada, 90 percent of people can be connected using facilities that cover just 3.3 percent of the land mass.

In Australia, 90 percent of people can be connected using facilities that cover just 4.32 percent of the land mass. In the United States, connecting 90 percent of people can be connected using facilities covering about 31 percent of the land mass.

The point is that costs to connect rural customers will be quite high in countries with huge areas of sparse population.

source: Deloitte