Appendix G: Introduction to Digital Subscriber Line Landscape
The DSL network relies on traditional telephone line and is often categorized as ‘legacy’ assets. It repurposes the telephone copper pairs to run internet data traffic. The number of customers relying on these legacy assets are shrinking at mid-single digits annually. It is being cannibalized by newer technologies such as cable, fiber, and wireless.
Attempting to track what a wireline categorizes as legacy is often misleading. The definition of what is deemed legacy has changed several times. One way to think of legacy is circuits that are incapable of speeds greater than 10Mbps download. This minimum speed threshold is what changes, which is why defining it is difficult.
It is important to recognize that the switch between technologies (DSL/cable/fiber) occurs slowly over many decades. As a parallel, one can look at why mainframe computers and dial up connections are still being used today. It is unlikely that in the near future DSL will go away. Rather, much of DSL services that presently exists will likely not go away. Cable companies have likely already explored the opportunity to expand into these regions housing traditional telephone lines, but have purposely decided against it because the economics are not justified. As a result, many regions will be bound to DSL technologies for the foreseeable future. This should be factored accordingly when modelling future DSL sales.
Additionally, there are technologies like fiber-to-the-node which extends the lifespan of older DSL technologies. It does so by mixing DSL technology with fiber technology (creating a DSL/fiber hybrid network). Performance of DSL improves by an order of magnitude as a result. These investments are high return on investment as you’re replacing only a few centralized equipment running on better optics and algorithms, while getting speed benefits across the board. We touch more on this topic in the section on Fiber Supply vs. Demand.
Finally, much of the share in DSL has already been eroded from faster cable technology, especially in the residential environment. This is why most telcos’ DSL technology serves the enterprise segment, while cable takes the lion share of the consumer market.
Appendix H: Introduction to Fiber Landscape
Fiber optics offers much higher throughput (download/upload speed) and more importantly, better latency than coaxial cable (cable) or copper pair (DSL). The benefits in throughput are slightly more obvious but the benefits in latency will appear more obvious in the years to come. Applications such as virtual reality and virtual web conferencing will catalyze its relevance.
Fiber providers have been around for a long time but not well-known to consumers because their target customers are predominately service providers and enterprises. Some of the pure play fiber providers include Level 3 Communications (expected to be acquired by CenturyLink), Zayo, Cogent, and GTT.
These pure fiber providers offer two types of products. The first is known as IP Transit. If an individual wanted to connect to google.com, he/she would purchase an internet connection from the local provider. The local internet provider receives this request to connect to google.com but needs internet services itself in order to reach this website. The local internet provider purchases what is called IP Transit to connect to the rest of the internet.
The second product sold by pure fiber providers are very high speed internet circuits to enterprises. This is similar to regular internet connections for consumers but at higher speeds, security, and reliability. Some of the names of these business-grade internet connections are MPLS, VPLS, EPL, and EVPL.
These fiber providers compete with the local telcos and cablecos to provide fiber to enterprise customers. Fiber providers, telcos, and cablecos target enterprises in multi-unit dwelling or locations where there are many potential customers. This is due to the high capital requirement of the initial build out. The potential return laying a single fiber to a building of a hundred residents is much higher than a fiber to a single resident.
The major distinction between the pure fiber providers (think Cogent and Zayo) and the telcos/cablecos is their geographical focus and target customers. Telcos/cablecos focus on a local region and target both enterprises and consumers. On the other hand, pure fiber providers choose to target service providers and enterprises, not consumers. They have strategically chosen to build a highly-distributed network spanning across states, as oppose to building regional density. Interestingly, this approach after a decade appears to have been highly successful in breaking into the wireline market.
Appendix I: Fiber Economics: Bandwidth Demand vs. Pricing Pressure
For over a decade, we have seen dramatic decreases in the cost of IP transit (bandwidth). Rates have been decreasing around 20% annually ($/Mbps) for the past five years, while the prior five years have ranged from 50-90% decrease year-over-year. One might speculate that at a certain point it would be uneconomical to sell IP transit. However, that would be incorrect.
Cogent’s IP Transit business (previously disclosed as their ‘Net Centric’ segment) has been able to grown at mid-high single digits for the past five years. This occurred even as Cogent decreased their rates approximately 20% annually. If one were to aggregate Cogent’s total pipe size, which represents how much traffic could be pushed through their total network at any given moment, this pipe has been growing at about 30% annually for the past 4 years.
Regardless of what the future holds for pure play fiber providers like Cogent, one could say they have already succeeded. For one, they have correctly identified the business model to enter the wireline market which has a strong defendable moat (see Competitive Advantage). They hold assets which allow for a regionally-concentrated peer to expand to have national presence.
Driving the cost of bandwidth down is the continued improvements in algorithms, optics, spectrum optimization and more. On the demand side, we have devices like smartphones, servers, and PCs. It is however, the demand for content-richness and number of devices that have been causing demand to outpace supply. Much of the long-term value of fiber rests on this equilibrium. We will explore this supply and demand balance in the next section.
Appendix J: Fiber Supply vs. Demand
Will we have too much supply – hindering the long-term value of fiber? Is the supply created by wireless providers a substitute for the supply of fiber wireline? Will wireless replace wireline?
First let’s look at the supply side. It is important to first recognize much of the wireless network itself is fiber. Between the cellular towers is fiber built and managed by wireline providers. Wireline providers add supply by installing more fiber optic cables in conjunction with utilizing higher performing network equipment. Wireless providers add supply by relying on wireless network gear and spectrum. Supply grows linearly as more fiber is laid, and cell towers are built. However, exponential growth comes from innovations in optics and algorithms. The critical question is therefore considering the potential for improvements in algorithms and technology. We think about this in two ways,
- Moore’s Law – defining the improvement in computing speed (# of transistors on a chip will double every 18 months)
- Shannon’s Law – defining the maximum network speed (theoretical speed data can transfer given power and noise)
More significantly, technology improvements are not realized and implemented as quickly as they are discovered. There are delays in when the industry collectively decides to upgrade to the newer technology. An estimate lifecycle to refresh to a newer technology is about 5 years. Cellular speeds, for example, require chip manufacturers, policy-makers, regional wireless carriers, and more parties, to collectively work together to build and roll out the technology. Networks will lag behind devices because of the challenges relating to planning and implementation. Looking out 10 years from now, we should see 2-4x increase in supply created by these algorithmic/technological improvements.
Another opportunity for increasing supply is through spectrum optimization. Current communication devices mainly crowd around the 500MHz – 3GHz bands. Ten years from now, the spectrum availability should double, contributing to a 2x improvement in supply. Summarizing the supply side, we should see 4x, 6x or 8x improvement in the next 10 years due to improvements in such things like algorithms and spectrum.
On the demand-side, demand is primarily driven by [i] content richness and [ii] number of broadband devices. There has been a constant evolution of devices: phone, mp3 player, tablets, laptops, IP-based TV, HD TV, and Ultra HD TV. The bandwidth demand for each of these newer devices are growing at exponential rates. Nielsen’s law estimates bandwidth demand doubles to residential homes every 24 months. The demand for content richness in combination with the number of devices approximates to a 100x increase in the demand over the next decade.
To summarize, wireless is not a substitute for wireline, rather, connected technologies. The situation in which wireless could be a substitute for wireline is if bandwidth requirements remained low – ie. devices aren’t demanding more bandwidth. Ten years from now, we will see an increase in the supply. However, supply will grow in the ball park of 10x (rounding upwards). Demand side looks to increase by a 100x, greatly outpacing the available supply. The solution to the supply-demand imbalance will unlikely be a slowdown in data consumption or number of devices. Rather, wireline and wireless providers will be working together to add more supply. At the core, the network will be all fiber. At the edges, there will be a vast array of wireless technologies such as: macro cells (miles), micro cells (300m), pico cells (100m), WiFi (30m). In the near term, wireline providers will hint at this supply/demand imbalance through usage-based billing. In conjunction, wireless providers will implement rate plans (50GB, 100GB, 500GB).