NVIDIA and Microsoft Expand AI Infrastructure Through Azure and Windows PC Deployments

Microsoft and NVIDIA are expanding their collaboration to deploy next-generation infrastructure across Azure cloud services and Windows PCs. The partnership includes the deployment of NVIDIA Vera Rubin NVL72 systems in Azure data centers to support large-scale model training, inference, and agentic applications. The companies are also working to extend confidential computing to bare-metal and rack-scale AI infrastructure to protect sensitive data during processing.

In recent performance benchmarks, Azure demonstrated significant scaling capabilities. According to MLPerf Training v6.1 results, Azure scaled to 10,240 NVIDIA GPUs using GB200 NVL72 systems on the DeepSeek-V3 model. Additionally, in MLPerf Inference v6.1, Azure’s NVIDIA GB300 NVL72 systems led cloud submissions for DeepSeek-R1 interactive throughput by more than 10%.

The collaboration extends to enterprise customization through Microsoft Foundry, which offers NVIDIA Nemotron open models. EY utilized Programmed Training in Microsoft Foundry to adapt Nemotron models for classifying corporate purchasing records, a process that reportedly improved accuracy on evaluation datasets by 55% and reduced costs by 15x compared to proprietary frontier models. The expanded Nemotron 3.5 lineup includes Nemotron 3.5 Super for long-running workflows and Lightning for fast-response applications.

On-premises and edge deployment options are also expanding. Azure Local will support NVIDIA RTX PRO 6000 Blackwell GPUs, as well as NVIDIA Jetson Thor and Jetson Orin edge AI platforms for industrial applications like robotics and machine vision. NVIDIA physical AI models, including Cosmos, are also being integrated into Foundry Local.

For desktop users, NVIDIA and Microsoft will showcase local large language model experiences on Windows PCs powered by NVIDIA RTX Spark. Furthermore, the companies are introducing Direct Stream capabilities to GPU-accelerated Windows 365 Cloud PCs, allowing for up to two 4K streams at 60 frames per second. A limited preview for these Cloud PC capabilities is currently open, with a public preview scheduled for early 2027.

FI Journal
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.