Is it really that insane? There is already a plethora of companies working with IOTA. No one knows much about the status of JINN due to prospective NDAs that would prevent any sort of chip production deals from becoming public. Companies are actively seeking out the foundation to see how they can become involved, not vice versa.

An IOTA efficient IoT device would simply have a tiny JINN ASIC to perform the PoW necessary to attach a transaction to the tangle.

I will add that I did mis-speak in that sentence. The hardware is not necessarily required, but will make the PoW for IoT devices negligible in terms of both transaction attachment times and energy usage. The tangle is expected to be running smoothly without the coordinator before any adoption of JINN hardware takes place.

Yes. If you expect commercial iot products to use unusual custom hardware you are out of your mind.

Commercial IoT products will be built to work in the world of IoT. If IOTA becomes the standard method of data and value transfer among IoT devices, manufacturers will implement them. Remember, device manufacturers will only include tiny JINN chips in their devices, nothing more.

You would have probably said the same about bluetooth chips when they first came out.

Trinary logic hardware will perform better for the PoW in low energy environments, which is extremely important for IoT. It's either ambitious and visionary, or else insane. Personally I think it's the former, cutting edge technology always seems insane until it doesn't. Time will be the judge.

They've never actually shown that trinary logic hardware will work better in low energy environments. CMOS transistors can be very efficient at switching, but waste power like hell if used at intermediate voltages. You REALLY don't want to be running your MOSFETs in the triode region. Which leaves the option of making trinary logic out of bi-state FETs, which requires extra die area and wastes power, or inventing an entirely new process/transistor type that is more efficient than modern CMOS and happens to be trinary... Good luck with that.

Look into radix economies https://en.wikipedia.org/wiki/Radix_economy

I'm aware of that. I don't really see the relevance to modern semiconductor engineering for cryptographic applications though. Differences in radix economy are typically only significant for very small numbers, far smaller than those used in cryptography. The decrease in number of required transistors is rapidly overwhelmed by the increased voltages required to discriminate between states and the resistive (heat) losses incurred by having a mid-valued state. If IOTA were a cryptocurrency for vacuum tubes where the power use is dominated by the cathode heater it might make sense to use trinary. But the last time I checked pretty much every processor around used CMOS FETs...

Edit: Idea! Hipstercoin - Trinary cryptocurrency for tube amplifiers! Play your vinyl and mine for hipstercoins at the same time!

No one knows.