Was the decision to which you refer made after the initial projection of some sales in the last part of 2017, and therefore a possible reason for the delay in sales? 

My understanding which is entirely based on second hand information which has been articulated by a few people here on the forum is that POET was working on a deal to sell the waveguide as a product. The waveguide is that part of the optical engine which simplifies the packaging as stated in the news release. POET has shared the cost of producing that product and it is very cheap as it does not require the expensive optics which need to be actively aligned. Thus a $1 cost to build a quad filter versus $20-$80 using existing technologies.

http://www.poet-technologies.com/docs/agm2017/POET_AGM_July%202017_FINAL-web.pdf slide 19

Using POET’s dielectric is a complete change to how optical engines are produced today and would require some very key technology transfer into the fabrication process to build the optical engines that go into the 100G transceiver. So I believe two overriding considerations to do or not do the deal were at play. Firstly just because it is a very low cost solution does not mean POET should settle for a low cost deal. Secondly why would POET sell the waveguide to someone else when they build the required detector and laser components to make a complete optical engine?

So in answering your question I do not believe that the sales were delayed but taken off the table to preserve that market for the optical engines that POET is building and ensure greater value for the dielectric.  

And remember that POET appears to have fine tuned what it is they will offer the market.They are focusing on the optical engines not the waveguide and not the full blown transceiver.