Skip to header Skip to main navigation Skip to main content Skip to footer
Computernik Networks
Remain at the cutting edge of technology

Main navigation

  • Home

NVIDIA Vera Rubin NVL72 Systems Are Running in Azure as Microsoft Expands Preview Access

NVIDIA headquarters in Santa Clara, California, photographed from across the street with the company sign in the foreground.

NVIDIA Vera Rubin NVL72 systems are now running in Microsoft Azure data centers, and the two companies are expanding access to Azure virtual machines powered by them, NVIDIA announced on November 19 at Microsoft Ignite. NVIDIA describes the step as Azure's Vera Rubin preview, aimed at demanding reasoning, agentic applications, and large-scale model training and inference.

The announcement is a preview expansion, not a general launch with published terms. Neither company announced pricing, a virtual machine series name, a region list, or a general-availability date in the Ignite post. Microsoft's own Ignite preview adds useful context: in March it became the first cloud to bring up a Vera Rubin NVL72 system for validation, and in August it delivered the first production Vera Rubin systems in its data centers.

Security and power are central to the work. Microsoft and NVIDIA are collaborating to extend confidential computing to bare-metal and rack-scale AI infrastructure, using NVIDIA Confidential Computing and attestation so customers can verify the computing environment before releasing sensitive models, data, or keys. The companies are also collaborating on power efficiency using NVIDIA MaxLPS, as operators try to get more performance from the power already available to a data center.

NVIDIA pointed to Blackwell results, not new Rubin benchmarks, as evidence of the co-engineering behind the preview. In MLPerf Training v6.1, Azure scaled to 10,240 NVIDIA GPUs with GB200 NVL72 systems on the DeepSeek-V3 model, which NVIDIA called the largest scale in the round. In MLPerf Inference v6.1, Azure's NVIDIA GB300 NVL72 systems led cloud submissions on DeepSeek-R1 interactive throughput by over 10 percent, according to Signal65 figures cited by NVIDIA.

The cloud hardware was only part of the Ignite package. Through Microsoft Foundry, enterprises gain wider access to NVIDIA open models and customization tools. Microsoft's serverless Programmed Training gives developers control over post-training workflows while Foundry manages the infrastructure, and Foundry Managed Compute adds support for deploying custom model weights. NVIDIA cited EY, which used Programmed Training to adapt Nemotron open models for classifying corporate purchasing records, improving accuracy on its evaluation dataset by 55 percent and lowering cost by 15 times compared with proprietary frontier models.

The expanded Nemotron 3.5 lineup on Foundry Managed Compute includes Super for long-running agents working across coding, research, and business workflows, and Lightning for applications that need fast responses and tool use, alongside Embed and Rerank, Parse, and Content Safety models. Bolt and Transcribe are coming soon, NVIDIA said.

Beyond Azure itself, Foundry Local brings Nemotron 3.5 to Azure Local environments, which will also support NVIDIA RTX PRO 6000 Blackwell GPUs from hardware partners and expand to NVIDIA Jetson Thor and Jetson Orin edge platforms for machine vision, robotics, and autonomous operations. On Windows PCs powered by NVIDIA RTX Spark, the companies showcased local large language model experiences at Ignite, and GPU-accelerated Windows 365 Cloud PCs gain Direct Stream capabilities 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.

Microsoft is also a customer of the stack it sells. Microsoft AI uses NVIDIA Dynamo to support reinforcement learning rollout inference during post-training, and Microsoft uses NVIDIA RTX PRO 6000 Blackwell GPUs for Bing search models. For enterprise buyers, the practical takeaway is narrower than the headline: Rubin capacity is physically running in Azure and preview access is widening, but purchasing decisions still wait on the pricing, regions, and availability terms neither company has yet published.

Chips & Semiconductors
Cloud Computing
Data Centers
Artificial Intelligence
Nvidia
GPUs
Servers

Visit Red River New Mexico

Visit Our Vacation Website, Red River New Mexico

Copyright © 2026 Rocky Mountain Madman LLC - All rights reserved

Developed and Designed by Rocky Mountain Madman LLC