Comment by xt00

Inference / Agentic AI implies "running models performantly using CPU cores" most likely (maybe with some optimizations / special AVX512 stuff) -- so essentially "welp, no sense in trying to build GPU's, we are too far behind nvidia to catch up".