>Raptor engines are still expensive.
They stated a while ago the cost of a Raptor was already under $1 million and that they're aiming for <$250k each with 2.0. That'd mean those three cost, at most and prototype status aside, a few million. That's just really not expensive in rocketry, or big machine early development in general. Even ignoring all other investment they can get, they're probably making at least $10-20m+ net income per commercial launch of the F9 at this point. So every single launch could cover multiple Starship tests. It's really wild how frugal it is compared to what everyone has become used to in rocketry, and in turn the kinds of things they can try.
Edit: just for context at the polar opposite end of the scale, "NASA will pay $146 million for each SLS rocket engine" (they need four per launch, and they're all thrown away each time).
https://arstechnica.com/science/2020/05/nasa-will-pay-a-stag...