(1) Cost. Building roads and highways inflicts both huge upfront capital costs and maintenance costs. A typical highway line has a throughput of 2,000 cars per hour. (2) Real Estate. Building the level of roads necessary to meet induced demand+ would take up an extremely large amount of land that can only be used for vehicles. (3) Local Roads. Local roads can only absorb so much capacity and they have functional throughput limits. If you wanted to make them faster, homeowners couldn't have driveways.
If you really wanted to, you could possibly make a city where driving is king and there is effectively no congestion. It would just be absurdly expensive and hostile to everyone who doesn't have access to car.
This is exactly what has been done in most of North America, except for the "no congestion" part.
I'm not sure which one would be quicker: having a really long access ramp (long because of the tunnel's depth), or having to wait in line to board an elevator that moves quite slowly.
This reality is also frequently ignored in the housing conversation when people advocate the solution to high prices is to simply add more supply, as if it's a given that supply could be created fast enough to overtake demand. I find that no one really ever does the work, taking into account the friction of real life on supply creation, to figure out if that's likely to be the case.
I suppose the limit might be quite high, but I've never seen it discussed at all.