I don’t have time to talk about this right now but in short:

There are very big issues with any heteroepitaxy approach that Juniper, Intel and others have been trying to develop over many years. And they all limit the lifetime of these lasers. Efforts to reduce the lattice mismatch which is really at the root of the problem are still being actively pursued. But like most solutions to get around the problems new problems are created and sacrifices are required which degrade performance. POET is ready for prime time and the benefits are significant