So why is Elon Musk so focused on improving transportation with his companies in the internet age?
Musk is paying attention to the world as it is, where location is still important. For example, London and San Francisco are still terribly expensive; if location was unimportant they would not be more expensive than other cities or towns.
The reality has largely been precisely the opposite: as transportation capabilities have increased, location matters more, and megacities rise in both absolute size and porportionate importance.
In a world in which it is difficult to transport physical goods more than a few miles, most production, consumption, and commerce is local. Where centres of production do emerge, it is precisely the insubstantiality of their output that leads to their concentration. Political, commercial, and educational centres were the first to emerge.
With industrialisation, we've seen centres of manufacturing production (usually dictated by transportation or availability of inputs), of banking, publishing, entertainment, cinema, and audio. "Washington, DC", "Wall Street", "Hollywood", "Silicon Valley", "Milan" (fashion), and London (politics, banking, fashion, publishing, art, ...) are all topnyms for products which are themselves largely ephemeral, but whose production becomes highly localised.
Why? becomes a very interesting question.
My sense is that it's analogue to Amdahl's law. In computing the limit of parallelisable processing is set by the nonparallisable component of a task. If half of operations cannot be parallelised, then the maximum possible speedup is 50%. If it's a quarter, the maximum speed-up is four times, if 10%, 10 times, etc. (The maximum is obtained only when there is infinite parallelisation, it represents a limit.)
For human activities, say, software production, the end result is something that can move around the world via digital communications largely seamlessly. But production requires computers, servers, training, support personnel, and the like. Yes, some of that can be distributed (as work-from-home has illustrated), but there remains some element of centralisation required. For the start-up community, that's often face-time with Sand Hill Road VC --- an artificial constraint, but a highly persistent one.
In banking, at least with high-speed trading, microseconds count. That's about 300 metres. The clock-tick time of an exchange literally defines the trading radius about it. (See the classic "500 mile email" story.)
It's highly counterintuitive, but friction decreases the significance of locations. Improved transportation increases it.
But slightly more seriously, not everyone can or wants to do everything through the internet. Sure, some white-collar jobs will be less location dependent, but that's hardly everyone. Some people will still want to see other people, and places. Stuff will still need to get from one place to another.
The internet and remote work is the best solution we currently have. However, in an ideal world we would all travel to work with each other, _if_ time and monetary costs of travel were low enough.
Travel is the best education you can get.