In theory if they can lower the costs substantially through innovation over the current taxi model they can grow the market out of private car usage. This would mean all parties (except car manufacturers) will benefit.
* You can't reduce DS expenses. You can offload them to drivers, but that will still be reflected in prices
* You can reduce wage cost by eliminating administrative load by having tools (apps, backends) do the job. Considering the size of Uber they are just transfering wages from operators to engineers.
* You can't reduce wehicle depreciation. You can reduce maintenance costs by having own workshop. Taxi companies might have that, for Uber that would be next to impossible due to practical reasons. You can reduce operating costs by buying fuel/insurance/etc in bulk. Agin, easy for cabs, difficult for Uber (not impossible though).
Personally I think a lot of Uber business model is hidden cost shifting onto the driver and society. Drivers are underestimating the true cost of being a Uber driver and society is losing out via inadequate insurance and tax avoidance. The drivers are not taking out the correct insurance and a lot of them are not declaring their income or collecting sales tax. I guess as long as it is all done on an app then it must be cool and OK.
What matters is passenger miles driven, not total number of unique car sales.
If you open up a new pocket of driving demand because it's easier to get a lift, then the total passenger miles driven goes up. Even more so if it is a pickup/dropoff model with 'empty miles'.
More miles driven->Cars worn out faster->more sales. More sales of likely generic cars less obsessed with expensive annual styling tweaks. More sales -> fewer models -> longer model runs = more profit.
I'd say the car companies are happy.