I don't think that the problem is with the lack of designers. There are companies making specialized chips all the time, so if there were a way to create an ef

I don't think that the problem is with the lack of designers. There are companies making specialized chips all the time, so if there were a way to create an efficient ternary circuits, then someone would have made one. For example, AI/neural networks do not care about base (they just need math), and they only need a few types of cells to be useful. But I have not heard of any applications of ternary there. Intuitively, I assume we'd still want to drive transistors to saturation, so we are not dependent on process parameters. Which means that each transistor still only handles two states, and we need two transistors for each trit. But if we have two transistors, we might as well use them for 2 bits and get even more state out of them.