A driver, a ticket inspector, and someone to run the café? Three people? Shared amongst a few hundred passengers? Is that really a significant cost?
Of course, there are also 44 hour routes. Or less popular routes without 300 passengers that are run for political reasons, or because they feed passengers into the more popular routes. Or the popular routes when run off-peak, so passengers can still use the route even if they may occasionally need some flexibility in when they travel. The video suggests amtrack employs 1 person per 4 riders, not per 100! For a route half the distance (DC to New York - 3.5-5 hours) they attribute $25/ticket to labor costs.
You've got stories like this where Japan is keeping open a train station for a single passenger - do you think that route has 300 people on it there? And this is in a country where trains are popular and some routes are packed to the brim in the middle of rush hour: https://www.citylab.com/transportation/2016/01/japan-keeps-t...
I recently watched https://www.youtube.com/watch?v=fwjwePe-HmA which covered much of the cost breakdowns - especially for American passenger rail. TL;DR:
* Labor is expensive. That 10 hour train ride is 10 hours you're paying every employee on the train. The faster speed of an equivalent plane ride means you're not paying pilots for as long, or allows them to make more trips.
* Rail is expensive. You have to buy the land between point A and point B. You have to build the rails, bridges, crossing guards, with lots of labor to build and maintain it all, with plenty of regulatory burdens to work through (ecological, safety, eminent domain, ...). The sky between airports, on the other hand, is free.
I've seen some suggestions that the sweet spot for trains is in medium length routes ("too far to drive, too near to fly"), with short haul routes being dominated by road vehicles (cars, taxis, buses) and long haul routes being dominated by airplanes (fast, don't need to buy/lease/maintain rails across thousands of miles)