Elon thinking is that on long distance flight you increase moved cargo per vehicle per unit of time.

He was saying that flying to the other end of the world will be 20x faster vs. airplane.

So if you move 20x more cargo per unit of time, then cost of cargo per vehicle + static infrastructure are proportionally lower.

But we need to take into account that number of cycles of a rocket engine may not be as good as jet engines, ground / water infrastructure seems a lot more complicated, fuel costs are higher etc. etc.

I doubt the math will close on it taking everything into account, but maybe there will be some limited use cases. USA army seems to be thinking about it. Extremely rapid deployment capability all around the world may be worthwhile for them.