Brian - there are indeed a number of "sub-goals" on the path to photonics nirvana, and many are laid out in detailed pre-competitive R&D plans by Plat4M, DARPA, etc. Nothing makes sense until you can reliably manufacture an embedded waveguide that keeps the light inside, so industry needs to decide what materials and mfg process technologies work just for the photonics side. After that, then there are choices for what form the larger integration takes - I think it likely that waveguides and coupling rings may be on their own silicon (say, 90nm SOI), with the complex electronics on standard bulk CMOS die. In that scenario, you only need to deal with the laser diodes in CMOS... but then you need 2.5D or 3D stacking on an SOI interposer where hi-bandwidth chip-chip communications could take place. Later, more might migrate alongside transistors on the same die or in a true 3D stack. Then again, I'm not an expert here -- Si2 is learning rapidly as we engage to enable their needs in Si2 standards (hmm... not a bad reason for folks to join the SP TAB!).
Steve, we rarely get to the end goal in one step so it would seem reasonable that there are sub goals that could be defined with photonics. If this is true, what is the first way in which we could actually see them being used - chip to chip communications perhaps?