> But it is clearly not avoidable. If one of the largest tech companies in the world (with enormous profits and cash reserves) cannot afford to deploy fiber in major cities, it is clear that the private sector cannot reliably and affordably provide this service on the scale required.
First of all, the fact that one company tries something and fails doesn't mean that the entire problem is intractable. It means that that particular attempt didn't work. Other variants might. Pretty much every single successful startup you read about is doing something that was previously attempted unsuccessfully or with less success.
And to that point, Google's model wasn't fundamentally different from the model used by existing ISPs[0] - they were doing almost the exact same thing (deploying {fiber, cable} from Tier 1 networks to the last mile) and structured their pricing in the same way. Their main goal was to force competition, not to dramatically change the fundamental business model of ISPs.
For example, one proposal that has been thrown around is the unbundling of the last mile, in which the lines are deployed and maintained by one entity but leased out to multiple private ISPs who actually service the customers[1]. There are other possibilities as well that haven't been tested yet either. Google failed at being a "better traditional ISP", but that doesn't mean that:
1) anyone who tries to be a "better traditional ISP" is doomed to fail, or
2) Google (or someone else) can't be successful as an ISP by using a different, non-traditional model altogether
[0] with one exception: they sought (and received) carve-out agreements
[1] Incidentally, this is what's already done with wireless networks (MVNOs) and in some places (like NYC) for electricity supply (supply only, not the wires).