What does M3 tell us, page-7

I maintain that simply moving to a new medium - software to hardware - it not revolutionary, merely evolutionary. It does not provide a new capability, it merely does it faster, significantly faster, I agree.

revolutionary / evolutionary is probably merely semantics, but my concern is that I am yet to be convinced that speed alone is the key to deployment of new applications. Google car and many other driverless car implementations seem to function perfectly well with current hardware / software combinations.

The challenge for Brainchip is to find its killer app, and to build a development infrastructure that is as familiar and as well proven as current methodologies based on hardware / software.

The cochlear and the visual recognition are good examples of where Brainchip may offer advantages - such time dependent input signals play to its strengths. But once an object is categorised / recognised the logic of what to do with that information is a whole new ball game and in many cases is best handled with software.

The bottom line is that Brainchip could have a place, but that place is not as a generalised one size fits all solution. I am still trying to position it, but as yet, insufficient technical information is available to make an accurate determination. I do not trust broad marketing claims.