uber's unit economics will easily work out if they stop expanding (which is where the expenses are).
Their backend services have a "fixed cost", if you assume they've designed it to be scalable, such that the marginal cost of a new user doesn't add more cost to hosting and compute. Then fire most engineers, and keep some skeleton crew maintaining the services.
The other cost is obviously the payment to drivers. I believe the unit economics will work here, since uber is not making capital investments into equipment, and is paid per-ride. Competition would drive the margins down, but thin margin is still a positive unit economics. Right now, the cost is subsidized by uber, but only as a marketing tactic to obtain marketshare and drive out competitors (unsuccessfully i might add). Uber can choose to stop the subsidizing, which can then make the unit economics positive.