I think what is important to remember is that POETs approach since the purchase of DL and BB has been to focus on the quickest ROI which is 100G and 400G (industry seems to be passing over 200G). These are QSFP form factor transceivers outside the box that are in the form of a plug which is what the entire industry is using. Ayar Labs has been working with GF for I think 8 years with a very different effort that of course is very much associated with GF's business which is silicon. Finding a way to merge optical devices within silicon without sacrificing too much electronic efficiency of those silicon chips. The light source of course remains external to the chip as no one has figured out how to implant those devices efficiently but the germanium implants for detection have been available for some time and of course the passive capabilities for silicon are there. 

So what POET is doing is producing the best possible photonic devices in the material set that is optimised for that use. It allows for the best spacing and best node as well as best confinement and evanescent coupling while maintaining the lowest possible cross talk and of course the separation of thermal profile.

In other words to combine everything possible in silicon is a compromise which can in my opinion only result in something less than the best electronics and something less than the best photonics within the same chip. Clearly it does not have to be the best of both to have a market and clearly this is right there in Global Foundries wheelhouse as this is where the start-up began as a group of academics.

Maybe in the future we can talk about the best electronics and the best photonics combined in a monolithic fabrication flow associated with the GaAs POET platform.