The real-world comes to us as analog signals, things that vary over time continuously. The digital world converts the real-world into data to simulate what's going on. Digital is precise and provides powerful analytic capabilities but is very power hungry compared to an analog processor. It seems counter-productive to think about going back to analog as a processing medium basis. What's caused engineering to consider doing just that?

Digital processing is leveraged by being easily and extensively programmable. Analog processes are usually fuzzy. The outputs are approximately true with offsets, things vary with temperature and other physical influencers. What's going on now with companies like Aspinity is these offsets are nulled and the interface with the analog world is now programmable. This also allows analog computing to not be application specific.

Consider this: the use of energy to convert analog signals into digital versions is rapidly becoming greater than what is being generated by convensional means like the power grid because of the massive number of devices being used at any given moment world-wide. Network centers and cloud storge are power hungry enterprises. Many devices sit and wait to be activated, i.e., your TV waiting for the remote command, monitoring devices of any kind consuming power being mostly idle.

Now we will stragetize the use of digital approaches with analog to optimize the results and conserve the power to do it. But beyond that, there are other considerations. There is an effeciency gain by letting analog processor evaluate the environment and if required, wake up the digital system to do it's thing. There is a value gain as well. For example, an analog circuit could be listening to a bearing for ware. When it is determined that something is waring thin, the digital system is awoken to alert a network of the situation. Neuromorphic AI processors are being constructed to push the processing into the memory space it's self to avoid bus negociations and data transfers and replace that with AI arranged vector math that works directly on the incoming data. Currently, Mythic has the first commercial analog AI processor on the market. You can check out this amazing new world order for your self. These new chip familes do not depend on small nanometer fabrication so the cost and size are very low in comparison to the bleeding-edge digital processors designed to do the same job.

Well, then IBM has a light speed processor. This digital computer is energy effecient. It also has analog features build in like light amplification. This technology has migrated into a quantum computer. The analog concepts that match the real world and quantum digital techniques will ultimately displace the crude silicon processors of the past...ah present.