(And I think you're probably right that they've miscalculated how far they can stretch the patience of investors and how quickly they can achieve reliable self-driving tech)

Well firstly, there are autonomous trains. The DLR in London is autonomous, there are plenty of train systems that are autonomous. But here's a key to why autonomous trains are not important and autonomous cars are: A traditional train requires 1 driver. An autonomous train requires 0 drivers and 1 guard (generally). In both cases trains have a capacity of around 1000 people.

Autonomous cars won't need guards, and have a capacity of 1 driver per 4 passengers. So the saving per person is 250x higher for autonomous cars than autonomous trains.

That's not even taking into account that the technology for autonomous cars is almost identical to autonomous HGVs - who have huge problems due to the limits of how long they can travel before the driver can rest.

The Skytrain does in Vancouver.