He approached Tesla the same way he approached Paypal, with an assumption that technology would do the "Moore's Law thing" so he didn't really have to worry about that. But in fact, we are approaching the theoretical densities of what we can store in the kind of batteries suitable for cars. That makes Tesla more or less a curiosity rather than a generally practical vehicle for the future. That's not to say he hasn't had an impact. He has. But I see Tesla as the finest product that can be produced on that dead-end road.
And likewise for rocket power, I doubt he can come up with a way to make rocket propulsion twice as efficient as all the rocket scientists that have gone before him have been able to do. And that's what you have to do to do a powered return.
Maybe he will prove me wrong. But let's see him use that approach when traveling to -- and returning from -- satellite altitudes.
To be fair, it doesn't take as much fuel on the return trip because the return weight is a small fraction of the launch weight, so maybe the numbers will work out for him.