another potential benefit
Global interconnects take multiple clock cycles due to propagation delay. My vague recollection is that the propagation speed is ~0.3c (due to the RC coefficients). If the conversion to and from photons is fast enough, this kind of thing could unblock one of the major limits on processor scaling, allowing bigger/badder cores with better single-thread performance.
Anyway, always cool to see this kind of fundamental work going on.