Cable Industry's Own Numbers Show General Decline In Investment Over Past Seven Years

from the net-neutrality? dept

So, yesterday we wrote about the cable industry lobbying group NCTA blatantly misleading with statistics about how much broadband investment is coming from the cable side. The story it was pushing was that investment in broadband has continued to go up and up (and thus, “changing” the rules would somehow lead to less investment).

Beyond changing the number of years before showing the totals, this chart is exceptionally misleading in that it is showing cumulative investment, rather than the actual rate of investment. Matthew Yglesias called NCTA out on this, leading NCTA to more or less admit Yglesias was right, though denying that he was. Still, NCTA finally released some actual year-by-year numbers. Of course, as people quickly noted, those year-by-year numbers seemed to support Yglesias’ point. Furthermore, they weren’t adjusted for inflation.

Thankfully, one of our readers using the name cincinnatus (nice one), posted the inflation adjusted numbers, and we’ve taken the liberty of making a new chart with those numbers:

From there, the story of cable continually increasing investment looks… very different. There was that big initial build-out of broadband in the late 90s/early 2000s, but then expenditures dropped, and other than minor ups and downs it’s been pretty flat. In fact, since 2007, the trend is pretty clearly downward on investment. Yes, there’s a tiny bump in the last two years, but that’s at best flat spending, not any real increase.

Now, it’s possible to claim — as some cable industry supporters have been — that because of greater efficiency and better technology, spending the same amount is fine because they get more value out of fewer dollars. But, if that’s the case, then that’s the argument the cable industry should be making, rather than the wholly misleading argument that investment in infrastructure has continued to rise over the last decade. It’s quite clear from the chart that this is simply not true.

Filed Under: , , , ,
Companies: ncta

Rate this comment as insightful
Rate this comment as funny
You have rated this comment as insightful
You have rated this comment as funny
Flag this comment as abusive/trolling/spam
You have flagged this comment
The first word has already been claimed
The last word has already been claimed
Insightful Lightbulb icon Funny Laughing icon Abusive/trolling/spam Flag icon Insightful badge Lightbulb icon Funny badge Laughing icon Comments icon

Comments on “Cable Industry's Own Numbers Show General Decline In Investment Over Past Seven Years”

Subscribe: RSS Leave a comment
Anonymous Coward says:

Investing vs Investment

It seems the misleading word here is “Investing” – saying you’re investing more in something every year implies that the amount you invest each year increases from the last year… whereas claiming your investment increases every year speaks to the cumulative investment you’ve made year over year.

I’m sure this is how they’re spinning it.

Anonymous Coward says:

Re: Investing vs Investment

Speaking in terms of “investment” also doesn’t speak anything towards investment in anything that benefits consumers.

For all we know, it could be investment in privacy-invasive tech, throttling tech, relationships with other telco’s, investment in CS size as an alternative to investment in fixing problems, investment in lobbying, investment in PR, etc…

Anonymous Anonymous Coward says:


I always differentiated between maintenance and repair in accounting for those types of expenses in the businesses that I have run. Sometime the difference was substantial, and a low number might mean something got deferred. In this case, I bet they are including such expenses, and that is not investment.

In normal accounting maintenance and repair go into operating expenses or the P&L statement. Investment goes on the Balance Sheet with the depreciation going into the operating statement, unless they are trying to screw with the IRS.

Investment would be expansion and/or improvement in some part of their system. We know by various means that this is not the case.

We want details, or it is just a lie.

Cincinnatus says:

I actually didn’t know Snowden used Cincinnatus… I just thought of him as a great general and statesman. Glad we like the same dude.

Two things about these data. 1. It’s possible the data were already inflation corrected by the ncta, but they never replied to my request for clarification ( This doesn’t change the basic trend, but makes the peak around 2001 less massively large. 2. Their original graph not only has the problems noted, but also had a nonlinear (non logarithmic) y axis. Edward Tufte was spinning in his grave–despite still being alive–over that graph.

Whatever says:

not misleading

It’s not really misleading at all, it’s an indication of the value of their networks as a result of their investments. It’s hard to deny that they have put 200 billion on the table.

Now, even with what you seem to considering damning numbers, they are putting a ton of money of the table every year, and have done so in a constant fashion. The only peak is in late 1990s, where they would have being doing the initial building of their networks for the internet. Otherwise, they have been pretty much spending 15 billion a year. Considering that much of the networking equipment is actually cheaper these days, that means that they continue to make significant investment and upgrades. Why beat them up for doing the right thing and constantly putting money on the table?

Whatever says:

Re: Re: not misleading

Do you think it’s any more honest to claim a general decline, which is not supported by the numbers? Do you think it’s any more honest to (as was done in the original post on this yesterday or the day before) to group investment into three year groups to try to make it look like there was a sudden big drop, by putting a couple of “up” years into three year groups?

The total investment number is honest. Representing it as an annual investment is not honest. Claiming a general decline is equally dishonest.

JM says:

Re: not misleading

Interesting that you’d mention $200 billion. That’s about the amount that the baby Bells and cable companies got in the form of tax breaks and direct subsidies from the Telecommunications Act of 1996.

I wonder how much of their own money the cable companies actually put up, after all of this is said and done.

They didn’t “[do] the right thing”; they did their best to earn as much money as possible. For a while they did that through investing in their infrastructure. For the last several years they’ve been coasting and investing at a lower rate THAN ANY TIME IN THE LAST 25 YEARS.

I believe the phrase is, “don’t piss on my leg and tell me it’s raining.”

Anonymous Coward says:

Re: not misleading

For all I know their accounting/reporting practices could be as good as Hollywood accounting. How would anyone know? Why should I believe any numbers they give me. It’s not like the government will audit them and catch them in a lie and actually punish them. With the whole revolving door mess it’s clear the regulators are in on the scam.

Until I see independent auditors audit them I won’t believe any numbers they give me. and even then the only real way to keep them honest is to allow competitors to enter the market. If not then the government should directly set prices so that they are only allowed to make a normal profit (which is enough to ensure development and maintenance) and not an economic profit, and the government should regulate them very closely (and basically run the operation) with independent auditors. Otherwise the U.S. will continue to just keep falling behind in terms of broadband services while ISP’s keep telling us baseless claims and lies that no one believes including the bought politicians and shills around here (including yourself) that only pretend to believe them in public because they have no conscience.

Government established natural monopolies should not be for profit entities.

Anonymous Coward says:

Re: Re: not misleading

These numbers basically exist in a vacuum, so anyone trying to use them to prove anything is a bit wacky. What’s the number of dollars invested per user? Per bit of traffic carried? Just what the hell is “investment in broadband networks?” Is it just pipes/routers/switches, or does it include the phone system at Comcast’s complaint center? How much of the capital expenditure is replacing worn out equipment with similar equipment, versus adding new and better equipment? How does unregulated US capital expenditure compare to those in place with utility regulation?

The biggest argument for disbelieving the conclusions reached by the broadband providers is that they are the ones who provided the information. As said by Cicero (following the Roman theme!): “Who benefits?” But the actual data is basically meaningless.

Allen (profile) says:

With all of the chapter 11 restructuring and distressed asset sales that followed, you can argue that the y 2000 peak was a case of over investment in which case subsequent years might be better described as flat. It doesn’t excuse them trying to claim that investment has been increasing or distorting the truth with graphic distortion shenanigans, but telecom investors lost billions in the dot com bubble and that does not indicate investment levels were normal or reasonable.

Whatever says:

Re: what about investment per account?

Then again, if you looked that the per subscriber cost of doing upgrades, you would likely notice it’s dropping as well. It’s hard to compare the initial buildout period around 2000 to upgrades in the current term and try to figure it out. The initial cost per user was very high, it just isn’t the same anymore.

It’s one of the reasons that the price of internet access is stable or dropping, and not increasing.

Josh in CharlotteNC (profile) says:

Re: Re: what about investment per account?

So why do my bills keep going up with no increase in speed (even the advertised rate that no one can reasonably expect to get)? Why does the actual speed of my connection get worse, have higher latency, and more dropped packets? Why do only other countries reap the corresponding savings of lower cost of equipment and lower costs per subscriber?

Anonymous Coward says:

Re: Re: Re: what about investment per account?

Not a huge fan of broadband providers here, but could the cost increase just be inflation (in other words, have consumer broadband costs increased more than ~2%/year)?

Of course, inflation doesn’t explain higher latency and dropped packets. But is that a general trend (except in specific cases like Netflix)?

Andrew D. Todd (user link) says:

Re: Re: Re:3 I Did Do The Math (Mechanical Engineers Are Trained To Always Do The Math) [to: John Fenderson, #31]

Some years ago, I built an abstract telecommunications model, by laying out wire paths for a typical suburban block, repeating the process for the subdivision, and so upwards. The conclusions I came to were:

10 feet of cable is one’s pro-rata share of getting from the Local Exchange (say, a couple of miles away) to the rest of the world.

10 feet of cable is one’s pro-rata share of getting from a U-Verse Cabinet (or the equivalent), 2000 feet from one, to the Local Exchange.

40 feet of cable is one’s pro rata share of getting from the Primary Plugbox at the street corner to the U-Verse Cabinet

500 feet of copper wire is one’s private task of getting to the Primary Plugbox at the corner.

A telecommunications network looks sort of like a “fractal,” and virtually all its length is in the branches. You can see that the cost is overwhelmingly in the last chunk of wire leading to the house, because that is not shared.

Bona fide investment works out to laying out new subscriber loops in a new subdivision, at the same time that the streets are being paved. This would not precisely correlate with housing construction, but the long-term relationship would be close. This investment has nothing much to do with improvement of service for existing customers. Investment in telecommunications is fairly durable. Ducts, as installed in the ground, are much more expensive than the actual cables inside them. Digging a ditch, and putting one or more ducts inside it, costs something like ten dollars a foot, or fifty thousand dollars a mile. You are paying for such incidentals as the eventuality of hitting a gas line. Cables, in turn, are much more expensive than the switches which tie them together. As for switching components, which do become obsolete, a defensible cost for the primary switching machine is about twenty dollars per subscriber loop, and assuming that this is amortized over two years, that is only about a dollar a month. The primary switching machine merges many individual subscriber loops into a packet stream at gigabit rates, and upstream switches enjoy much greater economies of scale. The cost of a switch is primarily determined by how many ports it has, not how much data it moves. Again, the cost of building telecommunications plant comes down to the local ditch and the local duct, and often depends on whether it is feasible to use a Ditch-Witch or not.

There will be some differences for cable television plant, as distinct from telephone plant, but the differences will not be major. The main point is that a cable network cannot be more geometrically linear than the streets it runs over.

Josh in CharlotteNC (profile) says:

Re: Re: Re:2 what about investment per account?

It’s not inflation. The cost goes up 10% or more every 2-3 years for the same (or decreasing) level of service.

The latency and packet loss is at the edges of the networks. Without going into too much detail, the links between the ISP network (Comcast, Time Warner, Verizon, ATT) and the backbone/tier 1/other internet providers’ network (Level3, Cogent) are at or above capacity. Historically, when these links get close to capacity, they are upgraded, and in most cases, this is extremely cheap to do (sometimes as cheap as running an extra cable or two through a co-location data center from one rack to another). Consumer ISPs that have an effective monopoly have been increasingly not upgrading these links as internet traffic has increased. It’s not just Netflix, but Youtube, other video, streaming music providers, bittorrent, and just general web page traffic that has been increasing, and as such all is being effected. While it is all being effected, it is only most noticeable in situations where latency and data integrity is key – streaming video/music, VoIP, online gaming, and such. If Techdirt takes an extra second or two to load, no big deal, but when your video stream buffers or pauses, or when your command to dodge out of the way of that bullet/sword doesn’t reach the server, you notice.

Andrew D. Todd (user link) says:

Re: Enjoying the geeking out here

Here you are, then. This is rather messy in form, and hasn’t been cleaned up. A long time ago, I was taught, in doing statistical work, to always scale the problem two different ways, and see if I got the same answer. That also applies to simulation-modeling. So in this case, I laid the system out hierarchical in one case, and extremely decentralized (mesh net) in the other, and the answers came out approximately the same. I’m sure there are sundry errors, but the conclusions are in factors of ten, and I don’t believe the errors are that big.
Subscriber Loop To Primary Plugbox

Upgrade: DSL terminal, 1Mbps
Single wire (True Subscriber Loop), 500 ft from Subscriber to
Primary Plugbox

Existing: copper wire.

Upgrade: copper wire, recertified for DSL, 1 Mbps
Primary Plugbox to Local Exchange, Centralized Version
Primary Plugbox

Assuming rectilinear layout, following streets and respecting
property lines, the plugbox commands an area of 1000 feet by 1000
feet, 1/25 square mile, or about four city blocks. Taking a
notional small-town/suburban population density of 3000 per sq.
mi., 3 per household, this works out to 40 subscribers, each
house being about 160 ft from its neighbors.

Distance from subscriber: 500 ft.

Existing: copper-to-copper physical connection

Upgrade: 40 Mbps, Local, remote-controlled optical-to-DSL
switching computer.

Primary Ganged cable, 1600 ft

pro-rata share per subscriber line of the ganged cable: 40 ft.
per line.

Existing: ganged copper wires

Upgrade: 40 Mbps, one optical fiber (grossly underutilized).
Secondary Plugbox

Assuming rectilinear layout, this plugbox command a 3 by 4 array
of primary plugboxes, and areas as described above. The 12
Primary Ganged Cables converge at a plugbox serving about 480
lines, 1/2 sq. mi. (3000 by 4000 ft), and about 1500 people.

Distance from subscriber: 2100 ft.

Existing: copper-to-copper physical connection

Upgrade: 480 Mbps, Local, remote-controlled optical switching
Secondary Ganged Cable, 4,500 ft

pro-rata share per subscriber line of the ganged cable: 9.4 ft.
per line.

Existing: copper-to-copper physical connection

Upgrade: 480 Mbps, one optical fiber (underutilized).
Tertiary Plugbox

Still assuming rectilinear layout, this plugbox command a 2 by 3
array of secondary plugboxes, and areas as described above. The 6
Secondary Ganged Cables converge at a plugbox serving about
2880lines, 3 sq. mi.(9000 by 8000 ft), and about 9000 people.

Distance from subscriber: 6600 ft.

Existing: In process of being furnished up with a DSL switch.

Upgrade: 2880 Mbps, Local, remote-controlled optical switching
Tertiary Ganged Cable, 8000 ft.

pro-rata share per subscriber line of the ganged cable: 2.78 ft.
per line.

Upgrade: 2880 Mbps, one optical fiber (still underutilized on one
Primary Local Exchange

Assuming an only partially rectilinear layout, let us say that a
primary local exchange commands a town of 20160 households and
60,000 people spread over 21 square miles. The Primary Local
Exchange is fed by 7 Tertiary Ganged Cables. One Tertiary Plugbox
is centered on the exchange, the other six are within a radius
of 10,000 feet.

Distance from subscriber: 14,600 ft.

Existing, optical-copper switch, in progress of being refurbished
to a mixture of DSL and optical connections to substations
overlaid on tertiary plugboxes.

Upgrade: 20 Gbps optical switching computer.
Primary Plugbox to Local Exchange, Decentralized Version
Assume that plugboxes are 1000 feet apart, and have connections
to their neighboring plugboxes to the north, east, south, and
west, and shared cables running to them, and that all traffic is

handed over from plugbox router to plugbox router, internet
fashion. Each plugbox therefore owns 2000 feet of single-fiber
optical cable, shared among 40 subscribers for a pro-rata share
of 50 ft. Assume that the router in the plugbox is capable of
handling four multiplex channels in each direction (40 Gbps) and
any point in the network is therefore capable of doing 160 Gbps,
eight times the town’s total outside connection requirement.

Continental Trunk Switching Network, Centralized Version.
Primary Local Trunk Cable
20 mile switched trunk cable from Local Exchange to County

pro-rata share of the trunk cable: 5.25 ft. per line.

Existing: one optical fiber with obsolete transmitter and

Upgrade: 20 Gbps, same optical fiber, say, refurbished to support
two multiplexing frequencies.
County Exchange

At this point, we switch from suburban population density to
overall quasi-rural density. Assume one central town, and six
surrounding towns at a distance of twenty miles. Since the cable
is crossing the spaces between towns, it runs straight, with
minimal circuity. Making some allowance for rural population,
this might be 180 persons per square mile. A county has a
population of 420,000 persons and 140,000 lines in an area of
about two thousand square miles.

Existing: obsolete switches.

Upgrade: 140 Gbps optical switching computer.
County to Area code Trunk Cable, 300 mi

pro-rata share of the trunk cable: 0.95 ft. per line.

Upgrade: Single Optical fiber, multiplexed.
Area Code Exchange
The Area Code consists of seven counties, laid out as before, at
a distance of thirty miles. Its population is three million with
a million lines, in an area of about 30,000 square miles. for a
population density of 100 persons/square mile.

Upgrade: 1 Tbps switching computer.

Intermediate Distance Trunk Cable, 300 mi.

pro-rata share: about 1.43 ft. per line

Upgrade: 1 Tbps, one fiber with wavelength division multiplexing,
finally fully utilized.
Regional Exchange:

10 area codes in a region of 300,000 sq. mi., and a population of
30 millions and 10 million lines.

Upgrade: 10 Tbps switching computer.
Long Distance Trunk Cable, 2000 mi.

pro-rata share: about 0.95 ft. per line

Upgrade: 10 Tbps, ganged trunk cable with 10 optical cavities.
This does not mean manufacturing ten fibers. Instead, it might
mean starting with a glass blank preform with a number of areas
of different refractive index and then stretching it out to get a
multi-cavity optical fiber. This would bring optical-fiber making
within the scope of Moore’s Law.
Continental Exchange:

10 regions in a continental area of 3 million square miles, with
a population upwards of 300 million, having 100 million lines.
Continental Trunk Switching Network, Alternative Decentralized
Assume that every town has neighboring towns, north, east, south,
and west, at a distance of twenty miles, and shared cables
running to them, and that all traffic is handed over from town to
town, internet fashion. This works out to 10.5 ft. per line. Let
us assume that the cables are 10-ganged or 100 ganged (10 fibers
times ten optical cavities per fiber) as a matter of course. It’s
the same sort of routine overkill which puts a traditionally
defined supercomputer on every desktop. In this case, The
aggregate east-west and north-south bandwidth across a continent
is somewhere in the hundreds of Tbps or more.
Intercontinental Connection:
At this point, we drop the abstraction and recognize that the
continent is North America, and assume that we have half shares
in 50 Tbps cables to Europe and Asia. Our share works out to 8000

pro-rata share: about 0.42 ft of cable per line.


0.42 ft of oceanic cable per line is one’s pro-rata share of
getting from one’s own continent to the rest of the continent.

8.58 feet of cable (0.95 feet of it 10-ganged) is one’s pro-rata
share of getting from the Local Exchange to the rest of the
[alternatively, 10.5 ft, decentralized model]

52 feet of single optical fiber cable is one’s pro rata share of
getting from the Primary Plugbox at the corner to the Local
[alternatively, 50 ft., decentralized model]

500 feet of copper wire is one’s private task of getting to the
Primary Plugbox at the corner.


The point to be drawn from this model is that the true copper subscriber loop, in the strictest possible sense of the word,
overwhelms all other components of the telecommunications system.

Long distance works out to about ten feet per subscriber, whereas the houses are a hundred and sixty feet apart. Since the copper subscriber loop does not admit of being shared, per-byte or per-minute pricing is irrational. In fact, if there is one irrational element in telephone pricing, it is the fact that the subscriber loop can be rented on a month-to-month basis, rather than on a long lease.

Andrew D. Todd (user link) says:

Re: Re: Re: Enjoying the geeking out here

Incidentally, here is a book you might find interesting:

Jeff Hecht, Understanding Fiber Optics (1987). It has sections of questions after each chapter, and was probably intended for use in a community college course for retraining telephone linemen, or something like that. One curious point about it, from hindsight, is that the author knew about Packet Switching and Wavelength-Division Multiplexing, but he didn’t take them particularly seriously, so he laid out an enormously elaborate practice of cable installation aimed at giving each customer their own fiber leading all the way to the central office. In hindsight, that is massive overkill, but it offers an insight into the level of installation involved in building a global network.

Another worthwhile book is David macaulay, Underground (1976). This is a book of illustrative drawings setting forth the sheer complexity of what’s down there: building foundations, manholes, pipes, cables, subways, etc.

Brian says:

The writing is on the wall

Very soon there will be so many people losing their houses back to the banks that there will be hundreds of thousands of folks that won’t “need” broadband, and even more that will have to cut it from their household budgets. Simple business 101, if broadband providers could plan on making a profit from increasing their services, they would increase their services. Likewise, if they project a loss of customers, why bother doing more than “busy work” to keep the game going a little while longer.

Pragmatic says:

Re: The writing is on the wall

Laissez-faire at work… except that the person calling me a Socialist (I’m not*) told me we’re doin’ it wrong.

“Less regulation!” she said.

About that…

* Socialists are not keen on the middle class, or “bourgeoisie,” and I believe we need a strong middle class. Look up “Socialism” and “bourgeois,” and see for yourself what they really believe. That’s why I’m not and never will be a Socialist.

Add Your Comment

Your email address will not be published.

Have a Techdirt Account? Sign in now. Want one? Register here

Comment Options:

Make this the or (get credits or sign in to see balance) what's this?

What's this?

Techdirt community members with Techdirt Credits can spotlight a comment as either the "First Word" or "Last Word" on a particular comment thread. Credits can be purchased at the Techdirt Insider Shop »

Follow Techdirt

Techdirt Daily Newsletter

Techdirt Deals
Techdirt Insider Discord
The latest chatter on the Techdirt Insider Discord channel...