I agree with Fj. One of the major attractions of POET is its ready ability for manufacture in existing fabs. 40 nm is probably as fast as available Silicon processors at 14 nm (I say this advisedly, on the basis that definitive figures for performance are not available though the theortetical advantages are well known). However, to make POET processors smaller should not present major difficulties as the nodal drop from 40nm to 28nm has already been achieved in Silicon. The techniques should be transferable to POET processors and at low cost. I cannot say how fast this will occur but, because the node reduction is well understood, it should be significantly shorter than current Silicon experience with a reduction say from 14nm to 10nm. Experience with recent node changes in Silicon reveals this can take 2 years and there are substantial signs of further delay and gigantic costs. It is possible that such a reduction is unachievable in Silicon (Intel, the largest leader, is already muttering about the use of group 111 - V elements which is a real sign of Silicon distress)

It is reasonable to assume that a POET 40 to 28 transition will be a much shorter period than 2 years with further nodal drops to come already achieved in Silicon. It is also likely that any fab with experience with 28nm manufacture could be adapted to produce POET at this nodal size. You may see mothballed fabs being reused and new life in existing fans.

David