JB, I just wanted to thank you for responding openly to my earlier query. I just had a chance to read an digest it.
I would like to give the matter additional thought. My current opinion on the matter is that there is a free platform available that will run on 99% of commodity hardware, and nobody is currently leveraging it. Installation of this platform is so incredibly easy that it isn't even required (LiveCD/USB). If I were in AMDs position, I would put together a bare-bones distro (Kernel, X, Mesa, SDL, a steam-like-client, and the demos with the hot cg chick) and leverage it to showcase AMD's technology. The result would be that 'every' AMD powered PC becomes a console.
I realize that this is a lot like what sony did with FreeBSD (cellOS) and GLES2 (libgcm/PSGL), the difference would be that you already have a huge installation base. Existing customers that have a positive experience on the platform continue to purchase AMD tech, customers of "the other tech and platform vendors" will look over the shoulder of your customers and say "Hey what's that? How can I get that?".
I'll think more about what you said tonight, but as far as I can tell, a million bucks and two years could quickly turn into market domination for a tech vendor. You don't need Sony or Microsoft.
While GSoC is useful,unfortunately there seems to be a bunch of unfinished projects/tasks or unimplemented features happening with it.
I believe that the same can be said of any org that funds software development. Not all pound puppies find homes. It's sad. I wonder if GSOC is statistically better or worse than other similar initiatives (and why).