The Bell Tolls For Thee
A few weeks ago at&t announced a shift in the pricing of their wireless data plans. Previously you could get all of the wireless 3G data that you could eat for $30 per month. Now you can get 200 MB of data per month for $15 with further bundles of 200 MB available for $15 each, or you can get 2 GB of data per month for $25 with further bundles of 1GB costing $10 each.
The explanation for this is that most users don’t use too much data but that a very small number use lots of data and have a significant impact on the user experience of others and the cost to operate the network. That makes sense. Users are incentivized to use less data and those who use more will pay for it. The network will be under less strain and in the places where it is maxed out at&t will have financial motivation to improve it. Thus every one gets their email faster.
This is the very plausible story that at&t has spun and the tech press has either bought; or treated as true, but a bad deal for the consumer. The reality is that this narrative is back haul centric rather than air link centric and ignores the largest part of the problem for wireless data providers and thus obscures many obvious inferences about the motivation driving many current and future plans of carriers.
I know that you have an iPhone, but the math that follows is based on EVDO rather than WCDMA because I am handier with the relevant numbers of that technology. So we are going to assume that you have a Droid on Verizon and that Verizon has followed at&t’s pricing plans exactly. Neither of these things are true but they will be illustrative of the principle and allow me to use honest numbers.
The network components that connect your Droid smart phone to the internet can be broken down into two categories; back haul and air link. The air link is the radio connection between your phone and the cell tower, the last mile if you will, and the back haul is the fiber optic, microwave, or whatever network connection that connects the cell tower to the internet.
Back haul is a traditional network resource and from it’s point of view Johnny who streams 10 GB of HD video from Hulu to his phone each month uses ten times the amount of resources as Jimmy who streams 1 GB of music from Pandora each month. Back haul is infinitely upgradeable. If you have the money, you can always lay more fiber. So as long as Verizon gets more money for carrying more bits it is in their interest to lay more back haul, increase throughput, and sell more data. Sounds good. This is why they will go to a metered pricing model, to monetize the amount of data so that they can afford to carry more data to you, more quickly and make more money.
Wrong. You see, while the back haul is infinitely expandable, the air link is not. For a given wireless technology, a given amount of radio bandwidth can carry a set maximum amount of data. Thus as data usage increases exponentially, and users pay per bit, wireless data spectrum increases in value exponentially. Actually, at any time, it has an infinite value that has an upward bound of whatever wireless operators can spend on it without bankrupting themselves.
This next part is what is actually driving everything though. The air link has a variable throughput. The more bars of signal you have, the more data the cell tower can send to you in a given amount of time. So if everyone is making use of a cell at it’s edge and it is at it’s lowest throughput, that cell can transmit 80 times less data than if every one is standing right under the tower.
If your phone is downloading data for an entire month at the highest data rate possible, it could suck up 800 GB if you are under the tower as opposed to a maximum of 10 GB in a month of use at the lowest data rate which you would get at the edge of the cell. The important thing to know is that both of these cases use the same amount of air link resources. So if every one in a cell is an edge user and is on the lowest data plan, the carrier can make $750 off of that air link in a month. If everyone is a core user and on the same cheap plan, the carrier can make $60,000 of of that same air link in the same month. Big difference.
The carriers have a strong motive to make their data users high throughput rather than low throughput. The carriers only want to sell wireless data to people in high density locations with good signal strength. A cell tower in the middle of a busy shopping center will easily earn that $60k per month, per directional sector. Lots of people, using data, in a predictable location that can be provided with a strong signal. The same tower in a suburban residential neighborhood will struggle to make a couple of thousand dollars per month because enough users will be at the edge of the cell that aggregate data throughput will be low.
So what does it all mean? Getting people in their homes to not use wireless data is the biggest priority for carriers. at&t was forced to be the first to roll out metered pricing by the 3G iPad. It’s ideal use case is in the home. Watching movies over a 3G wireless data connection in your apartment where you only get one bar of signal strength costs at&t more than you are worth to them as a customer. Metered pricing motivates you to push the switch that uses wifi instead of 3G when you are at home in order to save money. Free wifi in Starbucks gets you used to pushing that button to save money, for at&t.
So what is the future? Even higher data rate cell towers, that transmit shorter distances unless absolutely necessary, coupled with a next generation wifi mesh network. Your car’s nav unit will update traffic data by sending a message to the car in front of it and so on and so forth until one of the cars right under the tower connects to the back haul and then passes your data back to you. Once the technology is widely implemented in such an automobile centric use model, personal devices will be easy to add into the mix because it will be the fastest and cheapest connection that people can get.
Such a business model, which I think is very likely, will have such strong network effects that carrier consolidation will be inevitable and wireless data providers will have to be regulated as utilites. However, since the benefits of the network effects of this model are the same whether the carrier has a nation wide monopoly or just a local monopoly, I expect at&t to get broken up once again within the next 20 years. That would be approximately 50 years after the first AT&T breakup and 25 years after the baby bell re-consolidation under Cingular. I love the symmetry.
