That's what an article in Forbes says this morning, quoting a financial model put together by analysts critical of the effort. It seems crazy to think that Google has any intention of spending that much money on this demonstration, but the fact is that we still don't really know how much skin they're really putting into the game.
The $28B number comes a model that sees Google eventually connecting an eventual 15 million households at $850-1,250 each plus overhead and operational costs. Of course, one must note that Kansas City and the surrounding metro area only has a total population of about 2 million, and probably half that many households. So this model certainly assumes that Google will expand its effort to dozens of cities across the country.
They're obviously not going to do that if the economics are as bad as the Forbes article suggests, so there's actually no chance that Google is going to blow $28B on the project. If the experiment doesn't work, it surely won't even leave Kansas City let alone hook up everyone there.
But it's still an open question just how Google plans to make the economics more favorable on a large scale. Right now Google is even waiving its $300 construction fee for Gigabit service, which would have helped diffray the buildout costs a bit. Clearly they must have a different per household cost number planned, I just which they'd tell us what it is.
If this is a trial designed to show the rest of the country how to make FTTH into reality, then hopefully they will share with us the real numbers. We could use some real buildout cost numbers to play with, Google!