Comment by ArtTimeInvestor

If we only look at the hardware (not the software) for inference, does Nvidia have any advantage over Intel, AMD and Google?

They all build GPUs/TPUs that can run LLMs, correct?