In 6 months you can only do something mundane. For instance, adapt SBCL so it flawlessly interacts with CPAN and remove all the legacy lisp stuff. Then you'd have a clean lisp with a great library environment. People could actually -use- it. Still not a great business idea, but it has a small chance to rock the world. Especially since Ruby and Python are slowly becoming popular the ahead-of-the-curve guys are getting edgy. They want to move on to the next big language, and they want that language to be Lisp.
Your mundane suggestion has quite a bit of merit. However the Lisp OS would rock the world in a much bigger way. What I described would actually be capable of running Linux apps, but would also open the door to advanced kernel hacking to ordinary application programmers. It would be capable of competing against a commercial version of Singularity. It would further enable innovation with regards to security and programming models. It would be in a position to become the Linux/BSD of the next generation of F/OSS operating systems.
It would hardly be a waste.