Nvidia and Microsoft team up to deliver an AI supercomputer in Azure

Nvidia and Microsoft collaborate on supercomputers. On November 16, 2022, the companies announced a multi-year collaboration to build one of the world’s most powerful artificial intelligence supercomputers accessible from the cloud. “Our collaboration with Nvidia will produce the world’s most scalable supercomputing platform, bringing cutting-edge AI capabilities to every enterprise on Microsoft Azure”says Scott Guthrie, executive vice president of the Cloud + AI Group at Microsoft.

Tens of thousands of Nvidia GPUs

This supercomputer is based on the Microsoft Azure infrastructure and consists of Nvidia GPUs. Specifically, Microsoft will add tens of thousands of Nvidia A100 and H100 GPUs, Nvidia Quantum-2 InfiniBand, and the Nvidia AI Enterprise software suite to its Microsoft Azure platform. This platform is the first public cloud to integrate Nvidia’s AI technology bricks. Customers will be able to take advantage of this computing power to train and train the artificial intelligence models they want, even large models for language, to create complex recommender systems, and to work on generative AI.

Nvidia intends to use the virtual machines available on Azure to accelerate progress in generative artificial intelligence, which consists of algorithms that can create new content from existing content used to train it. A thriving field (some models are the basis for unsupervised self-learning algorithms that generate new text, image, video, audio, etc.). In addition to AI Enterprise, the partners also want to improve Microsoft’s DeepSpeed ​​​​​​​​deep learning optimization software.

Deploy AI solutions at scale

“The Breakthrough of Models [d’IA] has sparked a tidal wave of research, fostered new start-ups and enabled new enterprise applications, explains Manuvir Das, vice president of enterprise computing at Nnvidia. Our collaboration with Microsoft will provide researchers and businesses with the latest AI infrastructure and software to harness the transformative power of AI.”

All tools and software development kits optimized for Azure will be made available to customers. The goal is to offer this solution to manufacturers, researchers, and others interested in training, developing, and deploying artificial intelligence solutions at scale and quickly.

This is not the first collaboration between the two companies. Last year they created the world’s major model for language processing: the Megatron-Turing Natural Language Generation (MT-NLG). It would make it possible to achieve unmatched results for automatic natural language processing applications, and especially for text understanding functions. The latter can be used to train generative AI models, but this requires significant computational capacity. Hence this partnership.

Chose for you

Leave a Comment