The second company was with a team that I joined who was specifically looking for Clojure developers. It ended up being 4 developers, only 2 of which really knew any Clojure previously. The other 2 were Ruby devs. We paired over the entire life of the project and it was a very effective way of transferring skill sets - for all involved. We all learned more and more quickly than I've experienced during any other equivalent time on a project.
At my current engagement, I'm the head of tech and was able to choose the technology. I have the permission of the rest of the executive team and given my activity with the local technology community I've had no difficulty finding good developers who are enthusiastic about using Clojure. This may change as we grow, but I know I've still got a pool to pull from.
Choosing a non-mainstream technology is often additional work for whomever is leading the decision in any organization. For me it's been worth it: we've gotten passionate people who are more engaged because we're using something they care about. The fallback is to just go with a more mainstream JVM language (Java) - hasn't been and doesn't seem like it'll be ann issue.