Comment by g_airborne

So this might be an enormous opportunity for low-cost and more performant AI if someone was able to build an FPGA of some sort that could handle these types of computations as efficiently right?