By shifting the competition from infrastructure to service. Infrastructure can't have competition. It costs too much per customer,
That's pure fiction and fabrication.
Citation needed. In rural areas, from what I've read, the average cost of fiber to the curb is somewhere in the neighborhood of $2,500 per customer. Assume that the ISP charges about the U.S. average fee—$35 per customer. Part of that goes to the upstream ISP for providing the bandwidth, and part of it goes to the cost of interest on that fiber cost, so let's say that maybe $25 of that actually goes towards paying down the principal. You're now talking about 8.3 years just to break even.
Ah, but it gets worse. The cost of buildout skyrockets if you're just doing it for one customer, so you basically have to build the capacity for the expected number of customers up front. This means that if you only get half the people to buy service, that number doubles. Get a third of the people to buy service, and it triples. Get four competitors in a market, and you're talking about a whopping 33 years before you break even. Care to place bets on whether some new technology will have superseded that fiber you're running before you pay it off at that rate? And that's before you factor in the cost of line maintenance, replacing signal amplifiers, upgrading hardware at the head end, replacing customer premises equipment, recovering the costs associated with nonpaying customers, handling technical support, or any of the other myriad costs associated with running an ISP.
So no, it's not pure fiction and fabrication. There's no way in you-know-where that anyone can feasibly create true competition at the infrastructure level unless you're in a big city or unless you manage to trick somebody into footing the bill who has insane amounts of money to burn. I've seen rural areas try to get competition, and it always fails. Why? Because the incumbent provider always has a tremendous financial advantage caused by having already mostly paid off their infrastructure costs. Get rid of that advantage by starting with multiple companies from the outset, and you might end up with two competitors if you're really lucky. Unfortunately, to be perfectly frank, a stable duopoly isn't much better than a monopoly in terms of how badly the customers get bent over a barrel, in my experience. Heck, the cell phone market is barely competitive, and that has four major players.
What you ideally want, assuming you truly want enough competition to actually do some good, is somewhere closer to the dozen competitors that we had back in the DSL era (where the phone company was required to lease lines to any ISP that paid them). Using the earlier math, if twelve companies all built separate fiber infrastructure, that would mean a 100-year break-even point, on average. Even in a big city, where the fiber cost is more like $750 per customer, that's still a 30-year payoff even without factoring in other costs. No bank in the world is going to underwrite that loan. The only way you can make it work is to take the infrastructure cost—the high barrier to entry into the market—out of the equation entirely.
without regulations that mandate full coverage, nobody will serve anyone except in higher-density areas.
So you're saying that regulations currently force people to subsidize a suburban and low density lifestyle and we should have even more of those subsidies. In different words, progressives and liberals who keep whining and complaining about the suburban lifestyle, the lack of public transportation, environmental destruction, lack of high speed Internet, are, in fact, subsidizing the very things they criticize, and the very problems they use to then justify even more subsidies. Thank you for pointing that out so clearly.
Nowhere did I even suggest that urban areas should subsidize rural areas. Maybe that was necessary when the first phone network was rolled out, but at the time, there was one phone company across almost the entire country, so by definition any subsidization had to be nationwide. That's not the case now, and it is unlikely that such a nationwide subsidy would be necessary.
With that said, within a given geographical region (whether that region is primarily rural or primarily urban), some subsidization is absolutely necessary. Without that, you'll get a network that is biased against the poor and thus (statistically speaking) biased against minorities. Almost without fail, when allowed to choose who they serve, companies have upgraded their services in rich neighborhoods long before they upgraded their services in poor neighborhoods, if they upgraded the latter at all. So without mandatory universal service requirements, you'll have fiber in Cupertino, but not parts of Sunnyvale. You'll have fiber in San Francisco and Berkeley, but not in Oakland. You'll have fiber in San Jose except for East San Jose. You'll have fiber in Palo Alto, but not East Palo Alto. And so on.
We're not talking about urban versus rural here. We're talking about unserved chunks right in the middle of major metropolitan areas, just a couple of miles away from areas that get service. And that's why you simply cannot give the wire providers a choice about who to serve. It must be all or nothing within a geographical region (e.g. a county). This is, of course, unrelated to the issue of competition; it's simply the right thing to do.