And they’re cheap and don’t reinvent an already solved wheel. Tesla’s approach to windscreen wiping is as dumb as investing tens of millions trying to design a better moustetrap using cameras and AI to detect whether or not a mouse is present, when a perfectly good solution that costs $1.50 has existed since the 19th century.
Exhibit B: Hyperloop One is finally closing down for good, years after anyone with a science degree in physics could have (and did) tell Elon it could never work at an economically viable cost.
There are some problems for which it makes sense to invest millions to see if you can do it better. Then there are are other problems for which it makes no sense at all (e.g. mousetrap), because even if you can solve it a different way, some basic calculations will tell you whether it could ever yield a better solution at lower cost or more margin, even under the most optimistic of assumptions.
Elon seems incapable of discriminating between these two scenarios, and thinks all problems should be solved a different way regardless.