It is complicated. The ability of the Optical Interposer platform to integrate different devices allows POET the ability to optimize the package for the specific application in ways that have not been available before. 

For example the DMLs (directly modulated lasers) are very energy efficient but are difficult to work with in high speed applications. At OFC Suresh took great pride when he declared that no one else had a 100G DML application (Lumentums laser). Since the DML is switched on and off by varying the supply voltage to produce a signal, the energy used is significantly less than a continuous wave laser which produces photons continuously where the signal is created using an electro absorption modulator (unutilized energy). Thus we can regard POETs ability to flip chip and passively butt couple the DML as a breakthrough to new levels of efficiency and POET saw this as an accomplishment that would elevate POETs stature in the community. Was this considered a position of strength? 

The result is POET now has two paths: the DML energy efficient path for lower lane speed applications and the EML for very high speed applications. Prior to EMLs becoming available at 200G and potentially beyond 200G at higher modulation rates greater than PAM4, POET was waiting on the development of an Si modulator for use with continuous wave lasers. Those Si modulators failed to materialize so EMLs now take the lead.

Marvel, Acacia and others will provide silicon photonics solutions - they will not be as competitive as an integrated solution with EML will be.

That being said, I think providing an integration platform for EML lasers would make the platform accessible to a lot of potential players that have already committed their system design down an EML path.  And simplify adoption in this regard. 

One could say it was a mistake to spend so much time and energy on 100G - and that we should have focused on 400G.  But that is all hindsight at this point - and today we know the market is at 800G and we need to take all of our cumulative learning and target it squarely where the demand is.  We were skating to where the puck was -- but we were also learning how to skate.  Now we are skating to where the puck will be.