NVIDIA and Microsoft at Ignite: Vera Rubin NVL72 in Preview on Azure, Nemotron 3.5 in Foundry, Windows 365 Direct Stream in Limited Preview
NVIDIA says it has expanded AI collaboration with Microsoft from the Azure cloud to the edge and Windows, including a Vera Rubin NVL72 preview on Azure, Nemotron 3.5 models in Foundry and a limited preview of Windows 365 Direct Stream.
News date: About 5 min read
Contents
In its official blog, NVIDIA says it has expanded its collaboration with Microsoft at Microsoft Ignite. The expansion covers Vera Rubin NVL72 virtual machines on Azure, NVIDIA open models and customization in Microsoft Foundry, and AI deployment at the edge and on Windows. Vera Rubin NVL72 is NVIDIA's rack-sized system, which NVIDIA says is built for demanding reasoning, AI agents, and large-scale model training and inference. Here, "edge" means the place where a business generates and uses its data, such as a factory or its own server room, rather than a remote cloud. According to NVIDIA, Vera Rubin on Azure is in preview, Direct Stream for Windows 365 is in limited preview, and some Nemotron models are marked as coming soon. Enterprises get more options, from cloud training and model customization to running AI locally and at the edge, but each item is at a different stage.
What changes were announced?
- Azure: NVIDIA says Vera Rubin NVL72 systems are running in Azure data centers. The two companies are also working to extend confidential computing to bare-metal and rack-scale AI infrastructure. Confidential computing protects models and data while they are being processed. Bare metal means using physical servers directly, and rack-scale means systems built as whole server racks.
- Foundry: Microsoft adds Programmed Training, a serverless post-training service, which means Foundry runs the servers for you. Post-training means adjusting an existing model's behavior using a company's own data. Managed Compute adds support for deploying custom model weights, which are the learned parameters that make up a customized model.
- Edge: Foundry Local expands support for Nemotron 3.5 so it can be used in Azure Local, a company's own on-premises environment. Azure Local will support RTX PRO 6000 Blackwell GPUs from hardware partners. It will also extend to Jetson Thor and Jetson Orin for industrial uses such as machine vision and robotics.
- Scientific research: BioNeMo Agent Toolkit and ALCHEMI Toolkit are expanding into the Microsoft Discovery catalog.
What stage is each offering at?
| Item | Status or specification as described by NVIDIA |
|---|---|
| Azure Vera Rubin NVL72 | Systems are running in Azure data centers; the post describes Vera Rubin on Azure as a preview |
| Nemotron 3.5 Super, Lightning | Listed in the expanded Foundry Managed Compute lineup |
| Nemotron 3.5 Bolt, Transcribe | Coming soon |
| Windows 365 Direct Stream | Limited preview open for sign-ups; public preview planned for early 2027 |
| Direct Stream specification | Up to two 4K streams at 60 frames per second |
| Azure Local | Will support RTX PRO 6000 Blackwell; extends to Jetson Thor and Orin |
What does this mean for enterprise AI workflows?
NVIDIA says Programmed Training gives developers control over the post-training process, while Foundry manages the underlying infrastructure. NVIDIA cites EY as an example. EY used Programmed Training to adapt Nemotron open models to classify corporate purchasing records. On its evaluation dataset, accuracy improved by 55%, and cost was 15 times lower than with proprietary frontier models, meaning the most advanced closed commercial models. This is a vendor-provided case study, and the post does not say whether the results apply to other scenarios.
NVIDIA also says Microsoft AI uses NVIDIA Dynamo to support reinforcement-learning rollout inference during post-training. It adds that Bing search models use RTX PRO 6000 Blackwell GPUs. On performance, NVIDIA cites MLPerf Training v6.1, an industry benchmark round. In that round, Azure scaled to 10,240 GPUs with GB200 NVL72 systems on the DeepSeek-V3 model, the largest scale in the round.
Frequently asked questions
Is Vera Rubin NVL72 on Azure generally available yet?
NVIDIA says the systems are running in Azure data centers and describes Vera Rubin on Azure as a preview. The post gives no general availability date.
When can I use Windows 365 Direct Stream?
According to NVIDIA, a limited preview is open for sign-ups, and a public preview is planned for early 2027. Windows 365 Cloud PCs are Windows desktops that run in Microsoft's cloud and are streamed to your device.
Are Nemotron 3.5 Bolt and Transcribe available?
NVIDIA says both are coming soon and does not list them as available. It describes Bolt as built for high-throughput, low-latency reasoning and Transcribe as built for speech recognition.
What does confidential computing provide?
NVIDIA says confidential computing and attestation can help customers verify the computing environment before releasing sensitive models, data or keys. The two companies are still working to extend this protection.
Browse the latest news in this topic
Found this guide helpful? Share it with someone who needs it
Google Quantum AI Makes the Cover of Science: Quantum Contextuality Holds at Up to 105 Qubits, With a 98.3% Win Rate in a Coordination Game
Google announced that its quantum processor showed quantum contextuality in tasks spanning up to 105 qubits, with a 98.3% win rate in a coordination game versus an 88.9% classical limit. This is a research result, not a product launch.
Google: ARMORY CASTLE Exploited VMware vCenter Zero-Day CVE-2026-123890, Affecting vCenter 7.x and 8.x; Broadcom Has Released a Patch
Cloudflare: Election-Day Traffic to 2026 U.S. Midterm Election Sites Surged 363%, While Attackers Shifted from DDoS to Exploiting Vulnerabilities
Project Suncatcher: Google Targets Q1 2029 for First Heavy-Lift Launch on Blue Origin's New Glenn
Sources
Compiled from the references and verification dates listed above; fares, schedules, product features, and terms can change, so please confirm the latest details on the relevant official website before travelling, booking, or purchasing.