I think I understand Suresh's statement in simple terms to mean: development cost once, use developed AOC many times for many wavelegths.

But for Silicon Photonics how is it different and why do you have to redevelop for every AOC?

I appologize if this bores some of you, but I have found it is simple details like this that can make a huge difference in my understanding of the intended strategies. As Rainer pointed out, sometimes there needs to be a deliberate push towards a common language.