← Back to context

Comment by electroly

7 hours ago

I can't answer for NVIDIA but AWS has its own training and inference chips, and word on the street is the inference chips are too weak, so some companies are running inference on the training chips.