Microsoft and NVIDIA have jointly announced the launch of the world’s first GB300 NVL72 supercomputer on Microsoft Azure, setting a new benchmark for large-scale AI infrastructure.
Each rack houses 4,600+ Blackwell GPUs and 36 Grace CPUs, delivering up to 5× higher performance per GPU, 37 TB of unified memory, and an astounding 1.44 exaFLOPS of total compute power.
Why It Matters
This partnership represents a massive leap in cloud-based AI computing. The NVL72 architecture merges NVIDIA’s cutting-edge Blackwell platform with Microsoft’s Azure infrastructure, offering unprecedented scalability for AI training and inference workloads.
It’s a clear signal that Microsoft’s focus on Azure AI infrastructure is paying off — positioning it as one of the most powerful compute providers in the world.
