Microsoft and Nvidia collaborate to build new Azure-based AI supercomputer
Nvidia says it will conduct research and accelerate the development of generative AI using Azure's scalable virtual machine instances
Microsoft and Nvidia have announced a 'multi-year cooperation' to develop an artificial intelligence (AI) supercomputer that will be available to enterprises for the purpose of training, deploying and scaling AI models.
The AI supercomputer will make use of computing, networking and storage capabilities provided by Microsoft Azure, as well as Nvidia's GPUs such as Nvidia A100 and H100, its network kit and software stack.
Enterprises wishing to train and use sophisticated AI models at scale will be the target market.
To develop the system, the duo will use the Azure cloud platform's NC-series and ND-series virtual machines, which are GPU-based instances.
According to Nvidia, Azure will be the first public cloud to use its advanced AI software stack, adding tens of thousands of Nvidia A100 and H100 GPUs, Nvidia Quantum-2 400 Gb/s InfiniBand networking, and the Nvidia AI Enterprise software suite.
As part of the collaboration, Nvidia will conduct research and accelerate the development of generative AI using Azure's scalable virtual machine instances.
Nvidia describes generative AI as a rapidly developing field of artificial intelligence where unsupervised, self-learning algorithms are used to create new text, code, audio, digital images and video using fundamental models like the Megatron Turing NLG 530B.
Meanwhile, Microsoft will optimise its DeepSpeed library for new Nvidia hardware in an effort to lower computing power and memory usage during AI training workloads.
It will also work with Nvidia to make the company's stack of AI workflows and software development kits available to Azure enterprise customers.
"AI is fuelling the next wave of automation across enterprises and industrial computing, enabling organisations to do more with less as they navigate economic uncertainties," said Scott Guthrie, executive vice president of Microsoft's cloud and AI group.
"Our collaboration with Nvidia unlocks the world's most scalable supercomputer platform, which delivers state-of-the-art AI capabilities for every enterprise on Microsoft Azure."
Manuvir Das, VP of enterprise computing at Nvidia, noted: "AI technology advances as well as industry adoption are accelerating. The breakthrough of foundation models has triggered a tidal wave of research, fostered new startups and enabled new enterprise applications."
Nvidia is currently engaged in a number of supercomputer-related projects. Nvidia's GPUs are used by many of the most powerful supercomputers in the world, including HPE Perlmutter, IBM's Summit and Sierra.
The company is also working with Meta to develop an AI supercomputer.
Earlier this year, Meta said its research team had built one of the world's fastest supercomputers, using over 6,000 Nvidia GPUs.