Scaling costs for cars is linear (no congestion) to polynomial (congestion), whereas transit has scaling costs that are a step function on the level of an individual vehicle, and somewhat resembling a logarithmic scale in aggregate. That in itself isn't interesting until you understand the implications of subsidy policies.
For example, if you subsidize auto travel, you may pull people away from walking, cycling, or transit, and put them onto a congested freeway. Every marginal person that moves towards car use increases the costs of car use for every other car user. Transit, on the other hand, gets cheaper, on a per rider basis, with each new rider (although this can be muted if transit ridership is heavily biased in one direction, which can be the case for cities with strong central cores and no outer job centers, as half the buses may be riding at capacity while the other half are riding empty).
I essentially agree with you, but I'm of the opinion that if we would stop subsidizing cars, some small percentage of drivers would move back to transit, and that small percentage could be enough to make transit self-sustaining.