Fair enough. It seems to me that if an application has high performance demands and is expected to be in service for years, it should eventually be rewritten in a faster language. I find myself designing/developing programs in Python because development is so fast, then rewriting the result in a faster language after the design phase is over.
The rewriting phase is basically a translation, not a redesign. In most cases I know exactly how the program should behave, so there's no design taking place.
The system serves a massive number of users in real time, so I don't think that the speed is necessarily an issue. Of course, it can be different in other cases...
But all in all, it's more the safety you get from static typing I'm interested in than a pure speed boost. Though pylint is not too bad, it's still not the same.