The GB200 NVL72 delivers up to 30x faster LLM inference than the H100 while cutt

The GB200 NVL72 delivers up to 30x faster LLM inference than the H100 while cutting cost and energy use by up to 25x.