They don't compete. In 80% of the United States, people have only one real choice for low-latency, modern broadband.
Cherry-picking. What's a "real choice"? What's "low-latency"? What's "modern"?
It's not cherry picking. "Real choice" means a choice that meets the legal definition of broadband, and has low enough latency that normal use of the Internet isn't painful.
The most critical word was "broadband", which has a strict legal definition. Going forward, that definition is 10 Mbps down, 2.9 Mbps up. AFAIK, only one satellite service can provide that (just barely), and DSL can only provide that (just barely) in a few rare places that support ADSL+ with Annex M, and even then, only within about 1,000 feet of a central office or specially configured remote terminal, if memory serves. So the vast majority of DSL and satellite service no longer qualifies as broadband.
Low latency typically means "not satellite". Satellite adds approximately half a second of round-trip latency. At such high latencies, the Internet does not work very well:
- Web browsers can only open so many connections at a time, and most browsers limit you to three per hostname. So if you have a hundred tiny resources on a page, the page would take a minimum of 15 seconds to load over a satellite link, regardless of the speed of that link. And that's best-case performance; the actual bandwidth limits compound this. Ad blocking is pretty much mandatory, and even then, you're going to be miserable.
- VoIP calls are painful with half a second latency. In addition to being challenging to avoid stepping on each other's sentences, such long latency wreaks havoc on the technology used to prevent feedback and remove echoes.
- Using ssh over such a connection? Fugghedaboudit.
Most consumers have only one usable choice that qualifies as broadband. In a few areas, folks have two. Even that isn't enough competition to provide real choice, because duopolies tend not to compete more than absolutely necessary unless one of them is a newcomer, and even then, only for a short time (after which the entrenched monopoly usually runs them out of business, but if they don't, then competition still invariably settles down).
Now if you'll think back to high school economics, with supply and demand (a free market), if the supply gets too low and the demand stays high, the price goes up, and once it gets to a certain point, it becomes profitable for another player to enter the market and compete, and supply increases, and the price drops back. However, that can only happen when the barriers to entry are low enough to allow other players to feasibly enter the market.
When you have a per-household cost of $2,500–$5,500 for fiber service, even if you're a monopoly, it is going to take you well over a decade to pay off the infrastructure costs, assuming typical service prices. A business considering jumping into such a market has to ask themselves, "Can I steal 50% of the customers in this market, and then hold out against an entrenched monopoly for 20 years without them undercutting me so much that they bury me?" I think you'll find the answer is always "no".
For this reason, you'll never get the steady stream of market disruption required for supply and demand to function properly as long as each ISP has to provide its own physical infrastructure. It is simply a non-starter.