The limited deployments at the moment are BECAUSE they want it self-sustaining/profitable to continue to justify roll-outs.
Fiber providers live and die by their uptake rate. Most of the investment is in passing a house, but only about a third of houses you pass will subscribe. The cost of passing all the other houses has to be recovered from those that do subscribe. Historically, a condition of entering a city has been wiring up the whole city, without regard to customer interest in different neighborhoods. Google refuses to play that game. It only builds to Fiberhoods with demonstrated interest. That changes the economics entirely. But it's highly unlikely that approach will work across the country.
When their sole source of income is selling broadband. Part of the profit for google fiber includes the ads served up. If you think they don't track ad serving to their own customers based on the IP addresses they're accessing google from, you're crazy.
While I don't think ANYONE outside of Google knows their end-game - it's entirely possible that the fiberhoods are just proving the business model out before further investment.
The ad income is just peanuts compared to the monthly cost of Google Fiber. We are talking about a few bucks per month for ads versus $70 for fiber. Ads are just not going to move the needle in any meaningful way.
It was a (poor) business decision not to go for public cloud earlier, but not driven by datacenter capacity.
I'm quite confident they have the capacity. They have more than pretty much everyone besides maybe the government.
https://fiber.google.com/cities/sanfrancisco/ Google Fiber is coming to parts of SF using existing fiber infrastructure.