NVIDIA and Microsoft are collaborating on a new cloud-based AI-focused supercomputer, which they claim will be “one of the most powerful in the world” when complete.
The new machine will use Microsoft Azure’s supercomputing infrastructure (opens in new tab) combined with NVIDIA GPUs (opens in new tab), networking and AI software. It is set to contain ND and NC series virtual machines specially designed for AI distributed training and inference.
The companies claim that the project is the first public cloud to include NVIDIA’s full AI stack and connect tens of thousands of NVIDIA A100 and H100 GPUs, NVIDIA Quantum-2 400Gb/s InfiniBand networks, and the NVIDIA AI Enterprise software suite to its platform. will add.
How will it be used?
The companies said the new machine will be used to help enterprises train, deploy and scale AI, including large models.
NVIDIA will also use Azure scalable virtual machine instances to research and advance generative AI.
This is an emerging area of AI (opens in new tab) in which fundamental models such as Megatron Turing NLG 530B form the basis for unattended, self-learning algorithms to create new text, code, digital images, video or audio.
The companies will also collaborate to optimize Microsoft’s DeepSpeed deep optimization software and NVIDIA’s full stack of AI workflows and software development kits, optimized for Azure, will be made available to Azure enterprise customers.
“Advances in AI technology and industry adoption are accelerating. The breakthrough of basic models has sparked a flood of research, spurred new start-ups and enabled new business applications,” said Manuvir Das, vice president of enterprise computing at NVIDIA.
It’s not just Microsoft looking to Nvidia to power its latest AI innovations.
Oracle and Nvidia announced a partnership at Oracle Cloud World 2022. Tens of thousands of Nvidia GPUs, such as the A100 and the upcoming H100, will support Oracle Cloud Infrastructure (OCI). (opens in new tab)