It seems to be extremely difficult to create enough value in the programming language space to justify a startup. Most of the new languages we're seeing today were bootstrapped by volunteer or hobbyist efforts. Even after these languages gain a "critical mass" in the hobbyist community, the language can still be unproven for some time (maybe I'm wrong, but I don't think Clojure [for example] has much "in the wild" use yet).

Programming language development is valuable because it exposes programmers to new ways of thinking about problems, but it's tough to translate that into monetary value in a world full of stable, time-tested alternatives.