Average population density is borderline meaningless, it's the distribution that counts. A single cable between long stretches of nothing won't be the deciding cost factor, it's the net that has to be weaved afterwards. (Although - without having hard data - I'd guess that US-style sprawling suburbia is quite expensive to provide for..)

These prices are not valid outside of the higher density neighborhoods (by German measurements). You won't get that Internet outside of a settlement. Yes, theoretically the cost of connecting smaller villages/town that are rather isolated to the next fiber hub is much more noticeable, but as long as you don't service a location where you can't get at least 1k homes from a single fiber hub with 'normal' 10km transceivers over single mode cable, bill the cable to the customer as a setup fee (1$/m should be feasible, and you can split the cost if you can share part of the distance between multiple customers, by splicing the customer lines to a trunk cable), and buy reasonably priced peering. Maybe bill at a like 80% or 90% peak billing (traffic get's averaged over 5 min time slots, the 20%-10% highest ones get ignored, and the maximum speed left after that gets billed at x$/(Mbit/s), each month. This makes it clear that it's only expensive to provide the Internet speed if it's not sufficiently averaged out over time.

But regardless, serving anything with a population density high enough for car-price-competitive bus service of once per hour, walking distance max. 10min, between 7am and 9pm, should not be more expensive than Germany. At least if you can get fiber deployed without superfluous fees enacted by NIMBYs. E.g., along other occasions where the street get's dug up.

Look at a US population density map some time. Cable services are provided where people are, not evenly distributed across the country.