NVIDIA said on November 19, 2026, at Microsoft Ignite, that it and Microsoft are widening their joint offering from Azure data centers to enterprise customization tools and Windows PCs. The announcement, written by Dave Salvator, covers new infrastructure, more open models on Microsoft Foundry, and more options for running AI close to enterprise data.

On infrastructure, NVIDIA says Vera Rubin NVL72 systems are running in Azure data centers, and the companies are expanding access to Azure virtual machines built on them, described as an Azure preview. NVIDIA and Microsoft are also working to extend confidential computing to bare-metal and rack-scale AI infrastructure, so customers can verify the computing environment before releasing sensitive models, data or keys, and they are collaborating on power efficiency using NVIDIA MaxLPS. NVIDIA points to benchmark results: in MLPerf Training v6.1 Azure scaled to 10,240 GB200 NVL72 GPUs on the DeepSeek-V3 model, the largest scale in that round, and in MLPerf Inference v6.1 Azure's GB300 NVL72 systems led cloud submissions on DeepSeek-R1 interactive throughput by more than 10%, according to Signal65.

For model customization, Microsoft's serverless Programmed Training service gives developers control over post-training while Foundry manages the infrastructure, and Foundry Managed Compute now supports deploying custom model weights. NVIDIA says EY used Programmed Training to adapt Nemotron open models for classifying corporate purchasing records, improving accuracy on its evaluation dataset by 55% and cutting cost 15 times compared with proprietary frontier models. That result is NVIDIA's account of a customer project and has not been independently checked. The Nemotron 3.5 lineup on Foundry Managed Compute includes Super for long-running agents and Lightning for fast responses and tool use, plus Embed, Rerank, Parse and Content Safety models, with Bolt and Transcribe listed as coming soon.

Beyond the cloud, Nemotron 3.5 support is expanding in Foundry Local for Azure Local environments. Azure Local will support NVIDIA RTX PRO 6000 Blackwell GPUs from hardware partners and is expanding to Jetson Thor and Jetson Orin edge platforms for machine vision, robotics and autonomous operations, with Cosmos physical AI models coming to Foundry Local. On Windows PCs with NVIDIA RTX Spark, the companies will demonstrate local large language model experiences at Ignite. For Cloud PCs, Direct Stream capabilities are coming to GPU-accelerated Windows 365, supporting up to two 4K streams at 60 frames per second, with a limited preview open for sign-ups and a public preview planned for early 2027.

In science, the BioNeMo Agent Toolkit and ALCHEMI Toolkit are joining the Microsoft Discovery catalog, with an initial integration using the RFD4-Proteina model developed with the University of Washington's Institute for Protein Design. Microsoft AI also uses NVIDIA Dynamo for reinforcement learning rollout inference during post-training and RTX PRO 6000 Blackwell GPUs for Bing search models.

The post is a vendor announcement, so performance, accuracy and cost figures are NVIDIA's claims and the preview and coming-soon items carry no firm dates beyond early 2027.

References