Earlier quoted context omitted.
So far SpaceX has used approximately a third of a billion dollars in launch hardware in reusability testing, almost entirely subsidized through commercial launches. It's a very savvy way to do R&D. Note that it takes a lot of foresight to design a rocket which can serve as both a good expendable booster and also be sensibly made reusable, and most of the optimizations for expendability would push the design away from…
It takes some smarts and will to realize that fully optimizing for reusability isn't the right move, too. It's a constant refrain in these comment threads to ask, why don't they use parachutes? Why don't they use wings? Why not do all sorts of things that would make it easier to reuse, instead of trying a crazy vertical landing suicide burn onto a boat? The paradox is that making reusability easier would make it hard…
This is a good example of both "worse is better" and the MVP concept. The Falcon 9 is not the best design for a reusable launcher that one could put together, I could list easily half a dozen optimizations that would lead to a vastly superior reusable vehicle, for example. But by going the expendable mode first route, and building a dead simple vehicle (2 stage LOX/Kerosene, a 60+ year old design) they've been able to iterate, do full-scale testing, and earn income. Income is the life blood of any operation, so being able to bring in profits early made it easier for SpaceX to invest in rocket R&D. And by using the expendable -> reusable route they're able to gain a lot of confidence in the behavior of their rocket as well as get multi-million dollar test vehicles to play with, all financed by their customers.