AFAIK the main issue with silicon electronics is not the propagation of electric signals in wire but the switching time of transistors, in such that wait time of computation is 99% inside the transistor and not on the wires between them.
Will photonics become competitive on a MIPS per watt basis in the long term?
Strange question? Photonics suffers from logic density issues, certainly not frequency or power usage compared to relatively inefficient transistors.
Maybe we'll see a return to building sized computers like the vacuum tube monoliths of yore
The near term use of silicon photonics is to move data between stacked dies and later on dies themselves. Not to actually perform computations
Is there currently an analogue to FETs in photonics?
MIPS stands for Microprocessor without Interlocked Pipeline Stages. It is a RISC (Reduced Instruction Set Computing) architecture developed by MIPS Technologies. MIPS processors are known for their simplicity, efficiency, and the wide use in embedded systems, consumer electronics, and as microcontrollers.