At a 2 seconds clearance between vehicles, the Tesla system at peak capacity of 4 passengers will move 2 people per second. So it's equivalent with a subway train at crush capacity running every 1000 seconds or 16 minutes. That kind of frequency is typical for some longer haul train links, airport trains with underground sections like the RER etc.

Of course, there are other factors that preclude running Loop at 2s frequency, like station capacity and ensuing bottlenecks on ramps etc., but bear in mind there is no comparison between a train filled at crush capacity, with most people standing and smelling someone else's armpit, that you have to wait for a quarter of an hour, vs a leather seated car that takes you point to point to your destination without stopping at any intermediary station.

The point is, they are very different systems that don't compete against each other. If you need a subway and have the money for a subway, by all means, build a subway. Loop competes with light rail, and it says it can do better than most of the light rail systems in operation, at least in the US. Both its claimed pphpd and per mile cost are sufficient to be financially competitive, to the point that it can attract private capital and generate profits.

And the pphpd they demonstrated, of nearly 5000, was achieved with Tesla cars. If they manage to introduce 8-12 seaters and cracks self driving in the tunnels (a much simpler problem then full on road self driving) than it could compete with many of the less congested subway networks.

It seems very irrational not to let them demonstrate what they claim, as long as they are using their own money. It's transport capacity that does not compete with other transit options and does not capture significant urban real estate.