Once autonomous driving works well, I see no reason why cars can't scream along I-5 at twice the current speed, too. So five hours could become three and a bit, door to door.
They could even autonomously draft behind each other for maximum efficiency. Cars could be designed with high speed drafting aerodynamics in mind. With that, a line of cars might actually be more energy efficient than a high speed train running at typical occupancy levels, or at least competitive.
I'd be surprised if a "car train" running on rubber tires, even if as aerodynamically efficient somehow as a high-speed train, could ever match the same efficiency per occupant-mile. Ignoring everything else, steel wheels on steel rails have a massive efficiency advantage from lower rolling resistance alone - and a train carrying 400 people is going to have far fewer wheels per passenger than a "car train" carrying 400 people.
Yep, rolling resistance is the big factor in favour of the train.
The big factors for the car-train are much smaller cross sectional area and the fact that it's always exactly as long as it needs to be, unlike the train which needs to run even when it's mostly empty.
Once you have steel wheels, air resistance is the main problem. For trains it largely only depends on the nose of the train, then the length doesn't really matter. Since your "car train" isn't smooth, it'll have much more overall air resistance.
They could even autonomously draft behind each other for maximum efficiency. Cars could be designed with high speed drafting aerodynamics in mind. With that, a line of cars might actually be more energy efficient than a high speed train running at typical occupancy levels, or at least competitive.