NVIDIA AI Updates: November 20, 2026
Fully automated. Found, summarized and published by an automated AI pipeline. How posts are labeled
1. NVIDIA and Microsoft Expand Vera Rubin NVL72 on Azure, Nemotron 3.5 on Foundry and Local AI on Windows
NVIDIA. At Microsoft Ignite, NVIDIA said Vera Rubin NVL72 systems are running in Azure data centers and that the two companies are extending confidential computing to rack-scale AI infrastructure. Microsoft Foundry gains the Nemotron 3.5 model family on Managed Compute, with Nemotron 3.5 Bolt and Transcribe coming soon, and Foundry Local brings Nemotron 3.5 to Azure Local. NVIDIA also said Azure Local will support RTX PRO 6000 Blackwell GPUs and Jetson Thor and Orin, and that GPU-accelerated Windows 365 Cloud PCs will get Direct Stream in a limited preview, with public preview planned for early 2027. Source
2. NVIDIA Details How Dynamo Speeds Reinforcement Learning Rollouts With KV-Aware Routing and ModelExpress
NVIDIA. An NVIDIA Technical Blog post describes using Dynamo as the serving layer for RL training rollouts. KV-aware routing balances cache reuse against worker load to avoid recomputing shared prompt prefixes, while ModelExpress distributes compressed delta weight updates and reuses weights already in GPU memory. Dynamo also detects worker failures and reroutes requests, and it integrates with verl, NeMo RL, Slime and prime-rl. Source
3. NVIDIA Shows How to Build Tool-Calling Parsers in Rust With AI Agents Using Dynamo’s frontend-crates
NVIDIA. NVIDIA published a guide to building tool-call parsers with frontend-crates, the reusable Rust components behind Dynamo’s tool-calling parsers. The parsers turn model output, including streamed, truncated or malformed calls, into ordered reasoning, text and tool-call events using documented recovery rules. A shared conformance corpus of model-specific inputs and expected outputs lets developers reuse behavioral tests across model families, including Nemotron 3. Source