"Significantly less than 200 million Americans catching it" is still on the table.
"Herd immunity" is just the point at which the reproduction rate of the disease falls below 1 because the average infected person does not encounter anyone else to infect (because everyone else has already been infected) before they recover. It's not a binary switch, though -- as the percent of the population that has recovered grows, the reproduction rate shrinks steadily.
This means, for instance, that if we had X infected individuals before the lock downs, and then -- because we can't stay completely locked down forever -- we at some point have X infected individuals again after the lock downs end, we're actually in a much better place, because the reproduction rate of the disease is lower and that X will become 2X much slower than it would have before the lock downs.
This also synergizes with stay-at-home orders, which also reduce the reproduction rate of the disease. Right now we're relying entirely on stay-at-home to reduce the reproduction rate to as close to (or below) 1 as possible, but in a month or two, once a portion of the population has recovered, we'll be able to rely on a mix of "herd immunity" and stay-at-home to achieve the same thing; it's entirely reasonable that we can drag it out for years, and that less than 50 or 100 million Americans will catch it before a vaccine is developed.
It isn't hopeless.
> Around 0.5% will die, overwhelmingly those who had a low remaining life expectancy to begin with. This is a much larger body count than most people seem able to accept, but also much less dangerous to the average American than many believe.
We're currently losing 12-17 years of life per death, based on the statistics out of New York. It's older folks, sure, but it's hardly just taking people off their death beds.