Google Fiber and Huntsville Alabama just announced an interesting public private partnership. This is something that’s new for Google. In this partnership Huntsville is going to build and own the network and Google will lease connections on the network. Other ISPs will also be able to get on the network making this an open access network.
The details of the arrangement were not announced but there have already been a couple of interested parties that have made public records requests about the deal, so we ought to know more soon about how it will work.
There are a number of different ways to operate an open access network. For instance, a city can only own the fiber network and leave it up to ISPs to install the needed fiber drops and the customer electronics. At the other extreme a city could pay for everything. Since it’s been widely reported that Google uses some proprietary electronics my guess is that Google will be responsible for the electronics and the city for the rest. But we’ll have to wait a bit to see those details.
If Google does utilize a custom set of electronics it will be interesting to see how the city proposes to handle adding other ISPs to the networks. There are a lot of networks that would have a hard time handling different kinds of electronics mixed everywhere throughout the network.
The real question that everybody is going to want to know is if the city can make enough revenue from this arrangement to pay for the network. I’ve modeled open access networks many times and about the only way I can see for the network owner to break even with open access is if there are a lot of customers using the network.
And that is the biggest dilemma for owning an open access work. The big open access networks in Europe have a very high overall penetration rate because there are literally a dozen quality ISPs that compete on each network – basically multiple Googles. But if customer penetration rates fall below 50% it gets harder to see a path towards profitability for the network owner.
Fairly simple math can be used to demonstrate the dilemma for open access. If the network has a high penetration rate, say 70% or higher like happens in Europe, then the network owner can charge a relatively small fee per connection and can still break even. But should that same network have a small penetration, say 30% or 40%, the network owner would have to charge twice as much per connection to recover their costs.
The dilemma for network owners is that charging a high connection rate naturally leads the ISPs to cherry pick – that is, not sign up customers with low revenues that don’t create a good enough margin over and above the cost of the network connection. To give an example of this, if a network has a connection charge of $15 per customer, then some ISP in the market is probably going to be willing to use that connection to sell relatively low-price broadband, perhaps at $35 to $40 per month. But if the connection charge is instead $30 per customer, then no ISP is likely to chase those same $40 customer revenue opportunities and will only pursue customers willing to pay more.
This puts network owners in an economic bind. If they charge a low rate but don’t get a lot of customers they don’t make enough revenue to recover their costs. But if they raise the connection charge they force the ISPs to cherry pick and only sell more expensive products, and the network owner still might not sell enough connections to break even. The higher the connection charge, the fewer the potential connection that can be sold. It’s an interesting economic dynamic and one that puts all of the risk on the network owner.
I’m sure the deal is good for Google or they wouldn’t have signed it. It certainly relieves Google of a huge capital outlay. What others cities are going to be most interested in is if this a good deal for Huntsville. Most of the open access networks in the country have not done well for the network owner and it will be interesting to see if having a premiere tenant like Google will make a difference in the open access dynamic.