You can run LLMs in your own machine. Do you think a super computer would have issues? CUDA has optimizations, but you don't necessarily need it to do inference

You can run LLMs in your own machine. Do you think a super computer would have issues? CUDA has optimizations, but you don't necessarily need it to do inference at all. Those super computers are extremely powerful, it might not be as energy efficient as H100s, but it does the job.