Butt couple the dielectric waveguide to Rockley's large silicon waveguides at the edge.

Many companies have so much invested in silicon photonics as the path forward and they can build everything they need in silicon and make it work except for a light source so they have to find a way to introduce a material (III-V) which can provide the required optical gain. It is a complex problem. The dielectric provides a path for efficient coupling and an efficient heat sink which remains isolated to from the silicon waveguides which would otherwise be exposed to the heat generated by the lasers especially if they were flipped directly onto the waveguide (perpendicular = Rockley).

Hybrid lasers which use molecular bonding are right there on top of the waveguide with no efficient way to heat sink except through the small area of bonds which are high stress points for taking up lattice mismatch which also becomes exacerbated across the range of thermal expansion co-efficient. And of course waveguide filtering become messed up by the heat.

It just seems to me to be so much easier and at the same time so much more efficient to use the architecture enabled by the POET dielectric interposer.