Microsoft has introduced the new ND H100v5 virtual machine series, equipped with the latest NVIDIA H100 Tensor Core GPU and NVIDIA Quantum-2 InfiniBand network. This virtual machine features AI supercomputing GPUs, next-generation CPUs, low-latency networks, optimized host-to-GPU performance, and massive memory and memory bandwidth, offering unparalleled computational capabilities. When performing matrix multiplication operations using the new 8-bit FP8 floating-point data type, speeds are six times faster than the previous generation's FP16. The ND H100v5 VM helps large language models (such as the BLOOM175B end-to-end model) achieve up to double the inference speed, demonstrating their potential for further optimizing AI applications. The ND H100v5 VM series boasts the following features: AI supercomputing GPUs, featuring eight NVIDIA H100 Tensor Core GPUs; next-generation computer processing units (CPUs), powered by the 4th Gen Intel Xeon Scalable processors; low-latency networks, utilizing NVIDIA Quantum-2 ConnectX-7 InfiniBand; optimized host-to-GPU performance, with PCIe Gen5 providing 64GB/s bandwidth per GPU; and massive memory and memory bandwidth, using DDR5 memory.