Nvidia and Microsoft introduced Wednesday a multi-year collaboration to construct an AI supercomputer within the cloud, including tens of 1000’s of Nvidia GPUs to Microsoft Azure. The brand new settlement makes Azure the primary public cloud to include Nvidia’s full AI stack – its GPUs, networking and AI software program.
By beefing up Azure’s infrastructure with Nvidia’s full AI suite, extra enterprises will be capable to prepare, deploy and scale AI – together with giant, state-of-the-art fashions.
“AI expertise advances in addition to trade adoption are accelerating,” Manuvir Das, Nvidia’s VP of enterprise computing, mentioned in an announcement. “The breakthrough of basis fashions has triggered a tidal wave of analysis, fostered new startups and enabled new enterprise purposes.”
Azure cases already function Nvidia A100 GPUs paired with Nvidia Quantum 200Gb/s InfiniBand networking. Future cases might be built-in with Nvidia H100 GPUs and Nvidia Quantum-2 400Gb/s InfiniBand networking. In the meantime, Nvidia’s software program suite, Nvidia AI Enterprise, is licensed and supported on Microsoft Azure cases with A100 GPUs. Assist for Azure cases with H100 GPUs might be added in a future software program launch.
Azure enterprise prospects may even get entry to Nvidia’s full stack of AI workflows and software program improvement kits, optimized for Azure. The businesses may even collaborate to optimize Microsoft’s DeepSpeed deep studying optimization software program.
The brand new deal additionally offers Nvidia the possibility to make use of Azure VMs to advance its analysis in generative AI.