The thing is no one raised an eyebrow at the mention of POET. No one said pardon me but what is a POET. The whole discussion revolves around the abilities to serve high volume communications between disaggregated memory and process. That is not going to happen with Ayer Labs. It is the opposite. The attempt to bring everything together except for the laser which remains remote. The whole issue of thermal control and photonics together on the same die does not make sense for massive scale interconnected processes. It can't be the future but it can be something very significant on a smaller scale. The microrings have to be heated and tuned at high temperature to match the high temperature environment of the processor which is integrated monolithically on the same chip. It is not something that can scale. They are talking about very high powered silicon and that is not what Ayer Labs does from anything I have seen. 

It is interesting to listen to the guy from UCSB talk about using VCSEL based communications for short distances since multimode fiber coupling/insertion is easier to achieve than single mode. That makes sense since mode matching is much closer between the two. As such he talks about how difficult it is for single mode fiber coupling into PICs. Not with POET as we all know.

This is all my opinion with a lot of spit balling but I believe it is sound logic.