I don't think being an internal product/prototyping team changes the game at all. I have seen many new product innovation projects inside large organizations just like you describe, but operating like a startup, and executed by tight teams of 3 or 4 (maybe with some support from marketing, legal, etc.). And they operate far more effectively than such a top-heavy distribution described above (which I have also seen). Adobe, GE, Inuit, and Toyota have all publicized running projects like these in the last two years, and their teams were nearly always much smaller (I saw two of those cases first hand).
It's the methodology of how you execute product development that matters, not the team-size. The key element is speed, speed of learning, speed of shipping. The main drag on speed in product development is communication lag, and cost of delay from slowing down decision making. Coordinating 6 senior people in a modern corporation to try and ship something would take 3x as long as making that same decision (given mandate and budget) by a tight, tactical cross-functional product team.
I'd strongly question the assumption that organization size or protocol should dictate team size or make-up in new product development.