I would say there is nothing in principle stopping us from moving away from transistors or including other compute substrates in our computers. Two places I think this is being explored is quantum computers and biological computers, both of which, iiuc, have been demonstrated, but are still impractical.
I think you could count signal processing as a form of computation, in which case I would suggest that audio, wifi, and bluetooth include computation components relying on resistors, inductors and capacitors rather than transistors. This may be more valid in audio filters used by musicians. Of course these are still electrical.
Some transistor memory does use light to reset the transistors, and old hard drives used magnetism to store data, however this is memory storage, not computation.
Some things keeping us focused on transistors has to be manufacturing ease, economics of scale, and leveraging knowledge to work with transistors. We understand how to build electronic and digital circuits very very well, and there needs to be a very strong justification to try moving outside of that.
And, as Donald Hobson pointed out, analogue processes are often quite imprecise without great expense. This includes the actuation and re-digitization if the signal is going to interface with digital computers.
So my point of view is that currently economic incentives, lack of technical capabilities, and lack of value proposition are the reasons computation is done in various kinds of digital electronic components, but as those change in the future, it is possible that non-digital-electronic compute components may become common.
I would say there is nothing in principle stopping us from moving away from transistors or including other compute substrates in our computers. Two places I think this is being explored is quantum computers and biological computers, both of which, iiuc, have been demonstrated, but are still impractical.
I think you could count signal processing as a form of computation, in which case I would suggest that audio, wifi, and bluetooth include computation components relying on resistors, inductors and capacitors rather than transistors. This may be more valid in audio filters used by musicians. Of course these are still electrical.
Some transistor memory does use light to reset the transistors, and old hard drives used magnetism to store data, however this is memory storage, not computation.
Some things keeping us focused on transistors has to be manufacturing ease, economics of scale, and leveraging knowledge to work with transistors. We understand how to build electronic and digital circuits very very well, and there needs to be a very strong justification to try moving outside of that.
And, as Donald Hobson pointed out, analogue processes are often quite imprecise without great expense. This includes the actuation and re-digitization if the signal is going to interface with digital computers.
So my point of view is that currently economic incentives, lack of technical capabilities, and lack of value proposition are the reasons computation is done in various kinds of digital electronic components, but as those change in the future, it is possible that non-digital-electronic compute components may become common.