Mushy Brain

Disclosure: I am not an investor. I am an FPGA engineer by trade - I work with the latest Xilinx and Altera FPGAs daily.

The investor presentation smells just a tiny bit fishy.... The technology is sound, but the claims are way overstated - designed to get the uninformed all excited and giddy.

It talks up the SNAP as being orders of magnitude faster (presumably against a pure software implementation on a general purpose CPU), and being 1/1000th of the power consumption. But this is nothing special at all - being able to do specific computing tasks (especially parallel tasks...like implementing neural networks), much faster, and much lower power has always been the domain of FPGA / ASIC implementations (i.e. custom digital hardware design).

And yet the claim that their implementation is a "revolutionary digital hardware only design, no software". Nothing revolutionary here...

By comparing their SNAP implementation to a traditional general purpose CPU, they make it look much better than it really is to the current state of the art.

If Brainchip were any good, they would've brought on investment from the valley (or get acquired by Qualcomm/Google/Facebook), not merge with some mining company in Australia.

Yes, the future market and potential for neuromorphic processors would be HUGE...but brainchip...I think...is just playing it up too much for now...

BUT HEY. THEY'RE A STARTUP. THEY'RE ALLOWED TO OVERSTATE CLAIMS