How about writing an implementation of AMQP with Termite? Success in this sector would definitely be a nice example of Scheme in the industry (with all your money relying on it ^^)
Why try competing with an existing product, written in a language that greatly eases its implementation? Unless you can convince me that Termite could ease the implementation of really distinctive features that would be hard (if not impossible) to implement in Erlang, I'm not sure that's such a nice idea.
I think Scheme, as a language, is still in need of a niche market where it could really stand out and provide a competitive advantage. But don't get me wrong. I'm not talking about programmer productivity here. On this front, Scheme/Lisp is a definite killer. But you cannot take a large market share in any industry solely based on programmer productivity (*). That's not how business works, unfortunately for us... You need a kiiler app that can't be easily reproducible in most other languages (the barrier to entry).
(*) Maybe for businesses in the SaaS (Software as a Service) space, where the implementation language is totally irrelevant to the end-user, Scheme could help reduce development costs, provide higher margins and faster time to market (read: profitability!). It can even enable features difficult to replicate in other languages, thus providing a competitive advantage. Much like Lisp did for Viaweb in the 90's. But it must remain a secret weapon. I know at least one large company in the speech recognition industry that used Lisp before being acquired by Microsoft. But they were not advertising their use of Lisp at all.
Dominique