Nvidia and Microsoft are widening their artificial intelligence partnership, putting Vera Rubin systems into Azure data centres and extending their joint work from the cloud to factory floors and Windows PCs. The expansion, announced at Microsoft Ignite, covers open models, edge hardware and scientific research tools.
Rubin runs on Azure
Nvidia Vera Rubin NVL72 systems are now running in Azure data centres, providing infrastructure built for demanding reasoning, agentic applications and large-scale training and inference. Access is expanding through Azure’s Vera Rubin preview, although Nvidia and Microsoft gave no pricing, virtual machine series name, region list or general-availability date.
The joint engineering shows in the latest MLPerf results. Azure scaled to 10,240 Nvidia GPUs on DeepSeek-V3 in MLPerf Training v6.1, the largest scale in the round. In MLPerf Inference v6.1, Azure’s GB300 NVL72 systems led cloud submissions on DeepSeek-R1 throughput by more than ten percent, per Signal65.
Open models on Foundry
Microsoft Foundry is gaining more of Nvidia’s open models and customisation tools. Programmed Training lets developers control post-training workflows while Foundry manages the infrastructure, and Foundry Managed Compute adds support for deploying custom model weights. The Nemotron 3.5 lineup covers long-running agents, fast responses, retrieval, document parsing and content safety.
AI closer to the data
The partnership reaches beyond the cloud. Nemotron 3.5 support in Foundry Local brings reasoning, coding and tool-using agents to Azure Local, and Azure Local will support Nvidia RTX PRO 6000 Blackwell cards. Azure Local is expanding to Jetson Thor and Jetson Orin for machine vision, robotics and autonomous operations.
At Ignite, RTX Spark Windows PCs will show local large language model experiences, as developers run AI applications on devices. Direct Stream is coming to GPU-accelerated Windows 365 Cloud PCs, with two 4K streams at 60 frames per second, and a limited preview open now, public preview in early 2027.
Science and Microsoft’s own stack
For scientific research, Nvidia’s BioNeMo Agent Toolkit and ALCHEMI Toolkit are expanding into the Microsoft Discovery catalogue, bringing life science, drug discovery and materials models into Microsoft’s agentic research platform. An initial integration uses the RFD4-Proteina model to generate candidate protein structures and amino acid sequences for laboratory validation.
Microsoft is also using Nvidia technology inside its own products. Microsoft AI uses Nvidia Dynamo inference software for reinforcement learning rollout during post-training, and Nvidia RTX PRO 6000 Blackwell cards now power Bing search models. The two companies are building the partnership into both customer offerings and Microsoft’s own services.
Our Read
This is a breadth play, not one product launch. Rubin on Azure is the headline, but the announcement stitches cloud, local and edge into one stack, with Foundry as the glue. Nvidia gets its hardware and open models into the second-biggest cloud; Microsoft gets a platform to sell against rivals.
The timing matters. The expansion lands days after a day-long Microsoft 365 outage and its chief executive warned that critical infrastructure is largely unpatchable. A partnership built on scale, confidential computing and power efficiency answers doubts, though the preview carries no dates, prices or regions.
Why It Matters
Rubin is Nvidia’s architecture, and Azure hosts a share of enterprise AI work, so the pairing sets the terms for demanding agents and reasoning workloads. The open-model and edge pieces matter too: they let businesses deploy AI on their own infrastructure instead of renting frontier models by the token.
Background
Nvidia and Microsoft have deepened their partnership, from the agentic AI stack shown at Build to this Ignite expansion. Nvidia reported record revenue as Rubin ships, and it has been locking in chip substrate supply to keep the ramp going. Azure is the first place Rubin runs at cloud scale.
Source: NVIDIA blog


0 Comments