Liberty Global provides residential broadband Internet access and video services. The company enjoys technical, regulatory and economic advantages relative to other providers. These advantages are persistent. They explain why, of the 45 million homes passed by Liberty Global’s network, 22 million (49%) subscribe to the company's services.
An analysis of the cost structure of the latest (5g) wireless technology shows that wireless broadband technology will not displace Liberty Global as the low-cost provider of residential Internet access.
For decades, people have bought into the idea that a wireless data network should be superior to any wired alternative. Sticking a radio on a mast must be cheaper and more convenient than digging up sidewalks and front lawns to run cable. It is worth exploring this idea to see how it has worked out so far. We use Liberty Global's home market in the UK for reference but the results apply to other markets.
The monthly cost of a mobile data subscription is roughly $25. At that price, consumers get an allowance of perhaps 10 gigabytes of data. That's an hour of HD video. In South Korea, a country that has invested heavily in the latest mobile network infrastructure, subscribers are limited to 100 gigabytes.
The monthly cost of a fixed Internet connection is roughly $50. It comes in the form of a coaxial cable, twisted copper wires (DSL, UTP) or sometimes a fiber-optic cable. The connection is typically terminated by a WiFi radio. These connections are used by households as well as restaurants, gyms, schools and other small businesses. Subscribers are allowed more than 3 000 gigabytes of data. That's 300 hours of HD video. At that price, subscribers with a cable connection typically get a large number of TV channels. In the UK, Liberty Global also throws in content from Netflix and Amazon.
In sum, the per-gigabyte cost of data over a wired network is at least an order of magnitude lower. This remains the case in countries that have recently upgraded their mobile networks.
Why is the per-gigabyte cost of a wireless network so much higher?
1) Propagating a radio signal through the air requires energy. It requires even more to drive it through a wall. A highly efficient 5g site with three 64T64R massive MIMO cells going full blast, might serve out 50 Gbit/s to users within a 1 000 meter radius. That site needs 10 kilowatts of power. In the UK, the electricty bill works out to $1 000 per month. That's $20 per gigabit of bandwidth.
In contrast, a Docsis 3.1 line card pushing 100 Gbit/s to a similar group of users over coaxial cable requires 200 watts. That's a $0.20 monthly cost per gigabit of bandwidth. Line cards are mounted in a 19-inch CMTS chassis. A CMTS is the device at the upstream end of a residential coax connection.
The CMTS performs the same function as a g5 radio. It converts the signal from a fiber-optic backhaul connection to an RF radio signal. The CMTS transmits over coax while the g5 antenna transmits over the air. Software within the CMTS allocates the bandwidth for each subscriber on the shared coax "tree". This network topology is called a data bus.
2) It is difficult to get zoning permission for cell sites. There is a lot of pressure on local authorities to minimize the number of cell sites in a given neighborhood. As a result, mobile network operators pay rent to co-locate their antennas at existing sites. For a typical site in the UK, an operator will pay roughly $3 000 a month to mount three antennas. Assuming these are state-of-the-art 64T64R 5g antennas capable of serving out a total of 50 Gbit/s, that works out to a monthly cost of $60 per gigabit of bandwidth.
3) Antennas need a license to transmit. These licenses are auctioned off at a national level. The previously discussed hypothetical 5g site would require 200 megahertz of spectrum to operate. At current prices in the UK, that's $500 million for a twenty-year license. For an operator like Vodafone with 25 000 sites, that works out to a per-site monthly cost of roughly $5 per gigabit of bandwidth.
Adding capacity means adding more cell sites and that is not economical. It is simply more profitable to maximize the per-site number of subscribers and so mobile network operators create price structures that attract many subscribers but discourage heavy use. As a result, everyone has a mobile subscripition but switches to WiFi to watch video or download software updates.
Industry experts expect that the next generation of mobile networks will use a large number of smaller cells to augment the existing macro sites. A Martian spying on planet Earth through a telescope would observe that this is exactly what Liberty Global already has. The company has millions of small radios attached to its network. They are usually indoors where people spend most of their time and local regulators have no control. Integrating this dense network of small cells with the existing network of mobile operators would instantly create the network architecture of the future.
What about DSL?
A DSL connection is created by transmitting a signal over a pair of copper wires. Unlike coax, the wires are not individually shielded from electromagnetic interference. They are bundled together in large numbers and then the trunk is shielded. This works for voice but at higher frequencies, the wires act like antennas. A signal transmitted over one wire is picked up by nearby wires. The phenomenon is called crosstalk and it degrades the signal. It is the reason why DSL has been unable to provide bandwidth at the same level as coax cable. It is physically impossible to serve many subscribers with a good high-frequency signal over a trunk of twisted pairs.
What about fiber?
CATV networks with their coaxial cable do not compete with fiber-optic networks. They are fiber-optic networks. Modern CATV networks run fiber to the curb and use cheap hybrid-fiber coaxial (HFC) signal converters to switch to coax for the last meters. When Liberty Global runs a fiber-optic line to a local school or business, the whole neighborhood benefits. This has important effects.
Running a fiber trunk brings at least 30 Gbit/s of bandwidth within meters of all subscribers along the route. Fiber requires even less energy than the coax trunk it replaces. The CMTS still controls the bandwidth for each subscriber but the capacity is available to any subscriber at the flick of a virtual switch. DSL providers used to claim that their dedicated lines were superior. It works both ways though. Where there is excess capacity, all subscribers benefit.
So coax networks do not compete with fiber networks. They are themselves dense fiber networks serving half of all households passed. No other data network can lay claim to this level of penetration.
Is it sustainable?
The laws of physics are fairly persistent. They dictate that it requires less energy to transmit data over a wire than through the air. While each new generation of radios is more efficient, the latest 5g radios still require more energy by two orders of magnitude for the same amount of data.
There will continue to be pressure on local authorities to minimize the number of cell sites in a given neighborhood. The decentralised nature of the decision-making process creates high costs and significant uncertainty for anyone devising a strategy to create a nation-wide wireless network.
The measure of Liberty Global's moat.
The per-gigabyte cost of data for Liberty Global's network is at least an order of magnitude lower than the cost over a 5g network. The root causes of this cost advantage are persistent.
First published 9th June 2019 by batbeer.