The value proposition for using Akida is that you can do inferencing on the cheap, if you are doing it in a million smart-phones or cars or routers etc.
Otherwise you can do it on FPGA chips - slow and expensive in large numbers or on GPU's - even more expensive and power hungry because you're doing lots more work to get the same answer.
Akida lets manufacturers put their inferencing into tiny ASICS - which are cheap in large numbers and super efficient. This is because Akida uses SNN's not CNN's.
Manufacturers who use CNN's can't put their logic into ASICS as the model they use to do the inferencing will change for each new application.
That is our value proposition.
Our problem is very banal: our customers need to be so convinced by Akida, that they will take the risk and expense of taping out and fabricating those million chips. ie there is no easy entry.
Of course it could be possible for a number of potential clients to band together and do the tape-out and fab, but then everyone has NDA's these days.
So please add to your list
Unavailability of ready to use chips off the shelf. Cost of tape-out Cost of re-analyzing sensor data Fear of going out on a limb
BRN Price at posting:
16.0¢ Sentiment: Buy Disclosure: Held
(20min delay)