Some of that will be eaten up by fixed costs of providing the service (for stuff like bandwidth, Google can probably get better prices than almost anyone else in the world, due to extensive peering), and for the provided electronics (which Google can order in huge quantities and pay comparatively little for, or even build themselves, like they do with their own internal servers and networking equipment). Whats left over is still a very large chunk of change, especially when you consider the fact that Google wont roll out fiber till a decent percentage of a neighborhood is willing to sign on (their calculated numbers are almost certainly designed to cover roll out costs)
I think this will easily pay for their roll out costs. Remember, they don't need to build out a national network to serve tons of places, they already have one. YouTube itself already uses something like 40% of the internet traffic to residential connections.
The important thing to consider here, is as far as Google is concerned, they only need to break even, they arn't in the same game the other ISPs are where it is their only profit center.
There is a large upfront free that customers have to pay for the installation. If the economics have been worked out correctly by Google, the customers themselves will pay almost all of the roll out costs.
Edit: I should clarify, a large upfront fee, or a reasonably large monthly fee.