Porting to a dead platform is not a way forward, only the most blinded zealot would think that such a move would be anything other then stupifyingly shortsighted.
Ok then, please point us at the current alternatives on the PPC-market? Are there any other proven, relatively cheap and powerful PPC-platforms that are available in numbers and that have a future path? No? The only alternative would be to go x86, but before doing that, you would first need to assess what you are really aiming at with your OS.
If you want to go the commercial route and want to target at the desktop users, then there might be no other way than going x86. Hyperions problem is exactly that: I guess they originally planned to make some money with their OS, but how realistic is that on a license-restricted "only-PPC" and "we need custom HW" niche market?
The targets of the MOS-team might be different, for most of us they might still appear rather fuzzy, so I just wouldn't risk to tell them which route they should go.
