Putting the ethical dilemmas aside, I'd like to know how such bioprocessors could possibly have:

>million times greater power efficiency when compared to digital processors

if bioprocessors have to support("run" the metabolism) all their organelles, including parts that are not at all involved in the signal processing, which is I suppose >99% of the cell, compared to digital processors that we literally built with the sole purpose of performing such operations and have logical gates close in size to the single layer of atoms already? What did we miss in the design?