A “multi-year cooperation” between Nvidia and Microsoft is underway to create “one of the most powerful AI supercomputers in the world,” which will be able to manage the enormous processing workloads required to train and scale AI. As part of the partnership, Nvidia will make use of Microsoft’s scalable virtual machine instances to quicken the development of generative AI models like DALL-E.
The AI supercomputer will make use of a number of Nvidia’s potent H100 and A100 data centre GPUs as well as its Quantum-2 InfiniBand networking architecture, which is based on Microsoft’s Azure cloud infrastructure. Nvidia asserts that the integration of Microsoft’s Azure cloud platform and Nvidia’s GPUs, networking, and complete AI suite would enable more businesses to train, deploy, and scale AI, including huge, cutting-edge models. Additionally, the two organisations will develop DeepSpeed, Microsoft’s deep learning optimization engine.
Nvidia said in a statement that the supercomputer could be used for the purpose of “research and can further accelerate advances in generative AI,” a relatively new category of large language models like DALL-E and Stable Diffusion that employ self-learning algorithms to produce a variety of content, including text, code, digital images, video, and audio. Due to the recent rapid expansion of these AI models, there has been a considerable increase in the need for robust computer infrastructure that can grow along with them.
Manuvir Das, vice president of corporate computing at Nvidia, stated that “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.” In order to capitalize on the revolutionary potential of AI, Das continued, “Our collaboration with Microsoft will offer researchers and businesses cutting-edge AI infrastructure and software.”