Our Aging Fiber Infrastructure

One thing that I rarely hear talked about is how many of our long-haul fiber networks are aging. The fiber routes that connect our largest cities were mostly built in the 1990s in a very different bandwidth environment. I have a number of clients that rely on long-haul fiber routes and the stories they tell me scare me about our future ability to move bandwidth where it’s needed.

In order to understand the problems of the long-haul networks it’s important to look back at how these fiber routes were built. Many were built by the big telcos. I can remember the ads from AT&T thirty years ago bragging how they had built the first coast-to-coast fiber network. A lot of other fiber networks were built by competitive fiber providers like MCI and Qwest, which saw an opportunity for competing against the pricing of the big telco monopolies.

A lot of the original fibers built on intercity routes were small by today’s standards. The original networks were built to carry voice and much smaller volumes of data than today and many of the fibers contain only 48 pairs of fiber.

To a large degree the big intercity fiber routes follow the same physical paths, either following interstate highways, but to an even greater extent following the railroad tracks that go between markets. Most companies that move big amounts of data want route diversity to protect against fiber cuts or disasters, yet a significant percentage of the routes between many cities are located next to fibers of rival carriers.

It’s also important to understand how the money works in these routes. The owners of the large fibers have found it to be lucrative to lease pairs of fiber to other carriers on long-term leases called IRUs (indefeasible rights to use). It’s not unusual to be able to shop for a broadband connection between primary and secondary markets, say Philadelphia and Harrisburg, and find a half-dozen different carriers. But deeper examination often shows they all share leased pairs in the same fiber sheath.

Our long-haul fiber network infrastructure is physically aging and I’ve seen a lot of evidence of network failures. There are a number of reasons for these failures. First, the quality of fiber glass today has improved by several magnitudes over glass that was made in the 1980s and 1990s. Some fiber routes are starting to show signs of cloudiness from age which kills a given fiber pair. Probably even more significant is the fact that fiber installation techniques have improved over the years. We’ve learned that if a fiber cable is stretched or stressed during installation that microscopic cracks can be formed that slowly spread over time until a fiber becomes unusable. And finally, we are seeing the expected wear and tear on networks. Poles get knocked down by weather or accidents. Contractors occasionally cut buried fibers. Every time a long-haul fiber is cut it loses a little efficiency, and over time splices can add up to become problems.

Probably the parts of the network that are in the worst shape are the electronics. It’s an expensive proposition to upgrade the bandwidth on a long-haul fiber network because that means not only changing lasers at the end points of a fiber, but at all of the repeater huts along a fiber route. Unless a fiber route is completely utilized the companies operating these routes don’t want to spend the capital dollars needed to improve bandwidth. And so they keep operating old electronics that are often many years past their expected functional lives.

Construction of new long-haul fiber networks is incredibly expensive and it’s rare to hear of any major initiative to build fiber on the big established intercity routes. Interestingly, the fiber to smaller markets is in much better shape than the fiber between NFL cities. These secondary fiber routes were often built by groups like consortiums of independent telephone companies. There were also some significant new fiber routes built using the stimulus funding in 2008.

Today a big percentage of the old intercity fiber network is owned by AT&T, Verizon and CenturyLink. They built a lot of the original network but over the years have also gobbled up many of the other companies that built fiber – and are still doing so, like with Verizon’s purchase last year of XO and CenturyLink’s purchase of Level3. I know a lot of my clients worry every time one of these mergers happens because it removes another of a small handful of actual fiber owners from the market. They are fearful that we are going to go back to the old days of monopoly pricing and poor response to service issues – the two issues that prompted most of the construction of competitive fiber routes in the first place.

A lot of the infrastructure of all types in this country is aging. Sadly, I think we need to put a lot of our long-haul fiber backbone network into the aging category.

Economic Lives of Fiber Assets

outdoor-indoor-cable-161One thing that anybody who builds a fiber network needs to deal with at some point is depreciation expense. Fiber networks are expensive and depreciation expense is a key component for measuring profitability and success. This is even true for non-taxable entities if you still create financial reports that include depreciation.

Companies differ in their approach to depreciating their assets. If the owner is a taxable corporation or cooperative they may prefer higher depreciation expense in the early years to shield the business from income taxes. But other owners care a lot about what their financial reports say and I know some fiber network owners that like lower depreciation. Accountants all understand that depreciation is a non-cash expense, but this is a nuance that is often lost on the public or the non-sophisticated reader of financial statements.

My job allows me see the books of a lot of different kinds of telecom entities and I see that depreciation rates used for telecom assets vary widely. There was a time when the FCC and state commissions set depreciation rates for big companies, and the rest of industry usually followed. But today a fiber provider is free to set depreciation lives within a surprisingly wide range.

There is only one authoritative source for depreciation lives which is from a bulletin published by the IRS in 2015.  That bulletin establishes a baseline for depreciation for tax purposes for fiber networks that assumes a conservative and short life for fiber assets. For example, the IRS life for fiber cable is 24 years. At the other end of the scale, I have clients who are using a 40-year life on fiber. From an accounting perspective this wide range is like night and day.

In my experience the economic lives suggested by the IRS are ridiculously short. There was a time in the 1980s when a 20 to 25-year life for fiber was probably reasonable. The early generations of fiber cable had manufacturing flaws that allowed small cracks to develop over time that eventually cause the fiber to become opaque and lose usefulness. Most of the fiber built in those days has deteriorated over the years and has been retired or is of limited use today.

But the manufacturing process for fiber cables has improved drastically in each succeeding decade. I’ve talked to engineers at the fiber manufacturers who estimate that today’s fiber cables might easily last for 50 to 75 years as long as it’s installed properly and not unduly stressed. And there is speculation that fiber might last even longer – we’ll just have to wait and see.

The same thing is true for fiber electronics. Thirty years ago electronics in general were not as well made or as robust as today. There were large clunky circuit cards that expanded and shrunk in outside use and then eventually went bad. And these cards were full of individual components that could fail. But today a lot of the brains of electronics is embedded in chips that can last for a long time.

I can remember back in the 1990s when the engineering mantra was that you designed electronics to last from 7 to 10 years. Within that time frame the equipment would either start having operational issues or else the manufacturer would stop supporting it. But today’s electronics are much hardier and more reliable. I have several clients that still operate the first generation BPON fiber network networks. The electronics on these networks were made 12 – 15 years ago and are still going strong, and during that time they have had almost no failures. But most companies used depreciation lives for the BPON electronics of between 7 to 10 years. The same thing is true with the electronics used to power backbone networks. I have clients still operating networks built 15 years ago at twice the expected economic life.

So my advice to clients is that if they they are not stuck with whatever deprecation rates they are using. If they have reasons to might want shorter or longer depreciation lives there might be a justification for changing the depreciation rates. When I first got into the industry everybody used rates within a narrow range, but today there is a huge amount of flexibility in settling depreciation rates.

If your financial statements are audited then your auditor might want a professional opion of why it’s okay to change rates. But there is a huge amount of empirical data to support using longer lives for both fiber and electronics. And if you want to shorten lives it’s fairly easy to point to the IRS rules.

Accounting is not supposed to be this flexible and one would expect the industry to have a more consistent range of depreciation practices. But once the regulators stepped out of the business of regulating depreciation lives it’s been the wild west from an accounting perspective. So if you don’t like what depreciation expense is doing for you, contact me and I can help you find a better answer.

What’s the Real Cost of Providing the Internet?

British-Union-Jack-FlagThere is an interesting conversation happening in England about the true cost of operating the Internet. As an island nation, all of the costs of operating the network must be borne by the whole country, and so every part of the Internet cost chain is being recognized and counted as a cost. That’s very different than the way we do it here.

There are two issues that are concerning British officials – power costs and network capacity. Reports are that operating the data centers and the electronics hubs needed to operate the Internet now consume 8% of all of the power produced in the country. And it’s growing rapidly. At the current rate of growth of Internet consumption it’s estimated that the power requirements needed for the Internet are doubling every four years.

Here in the US we don’t have as much of the same concern about power costs. First, we have hundreds of different power companies scattered across the country and we don’t produce electricity in the same places that we use the Internet. But second, in this country the large data centers are operated by the large billion-dollar companies like Amazon, Google, and Facebook who can afford to pay the electric bills, mostly due to advertising revenues. But in a country like England, that sort of drain on electricity capacity must be borne by all electric rate payers when the whole grid hits capacity and must somehow be upgraded.

And it’s going to get a lot worse. If the pace of power consumption needed for broadband doesn’t somehow slow down, then by 2035 the Internet will be using all of the power produced in the British Isles today. It’s not likely that the power needs will grow quite that fast. For example, there are far more power-efficient routers and switches being made for data centers that are going to knock the power demand curve down a notch, but there is no reason to think that the demand for Internet usage is going to stop growing anytime soon.

In Britain they are also worried about the cost of maintaining the network. They say that the bulk of their electronics need to be upgraded in the next few years. In the industry we always talk about fiber being a really long-term investment, and the fiber is so good today that we really don’t know how long it’s going to last – 50 years, 75 years, longer? But that is not true for the electronics. Those electronics have to be replaced every 7 to 10 years and that can be expensive.

In this country all of the companies and cities that were early adopters of FTTP technology used BPON – the first Fiber-to-the-premise technology. This technology was the best thing at the time and was far faster than cable modems – but that is no longer the case. BPON is limited in two major ways. First, as happens with many technologies, the manufacturers all stopped supporting BPON. That means it’s hard to buy replacement parts and a BPON network is at major risk of failure if one of the larger core components of the network dies.

BPON is also different enough from newer technologies that the new replacements, like GPON, are not backwards compatible. This means that in order to upgrade to a newer version of fiber technology every electronic component in the network from the core to the ONTs on customer premises must be replaced, making upgrades very costly. Even the way BPON is strung to homes is different, meaning that there is fiber field work needed to upgrade it. We have hopefully gotten smarter lately; a lot of fiber electronics today are being designed to still work with later generations of equipment.

This is what happened in England. The country’s telecoms were early adopters of fiber and so the electronics throughout the country are already aged and running out of capacity. I saw a British article where the author was worried that the networks were getting ‘full’ and that more fiber would have to be built. The author didn’t recognize that upgrading electronics instead can use existing fiber to deliver a lot more data.

England is one of the wealthier nations on the global scale and one has to be concerned about how the poorer parts of the world are going to deal with these issues. As we introduce the Internet into Africa and other poorer nations one has to ask how a poor country that already has trouble generating enough electricity is going to be able to handle the demand caused by the Internet? And how will poorer nations keep up with the constant upgrades needed to keep the networks operating?

Perhaps I am worrying about nothing and maybe we will finally see the cheap fusion reactors that have been just over the horizon since I was a teenager. But when a country like England talks about the possible need to ration Internet usage, or to somehow meter it so that big users pay a lot more, one has to be concerned. In our country the big ISPs always complain about profits, but they are wildly profitable. The US and a few other nations are very spoiled and we can take the continued growth of the Internet for granted. Much of the rest of the world, however, is going to have a terrible time keeping up, and that is not good for mankind as a whole.