NVIDIA's official blog heralds the extensive deployment of the GB200 NVL72 system by CoreWeave, a pioneering AI cloud computing service provider. Joining an elite group that includes Cohere, IBM, and Mistral AI, CoreWeave stands as one of the first to adopt this advanced technology. Recent MLPerf benchmark tests reveal that the GB200 NVL72 system boasts a performance that is two to three times superior to that of the previous-generation H100 chip, markedly accelerating the training and inference speeds for large-scale models.
