NVIDIA and Microsoft are expanding a shared stack for agentic Artificial Intelligence deployment across Windows devices, Azure cloud services and local environments. The collaboration spans new Windows systems for running agents, NVIDIA-accelerated Microsoft Fabric, open models on Microsoft Foundry, OpenShell integration with GitHub Copilot, physical Artificial Intelligence tooling and Azure data center infrastructure for large-scale inference.
NVIDIA and Microsoft are reworking Windows PCs for agent-based computing with RTX Spark laptops and small desktops, and DGX Station for Windows deskside systems. RTX Spark is positioned as the first Windows PC platform built for personal agents, with 1 petaflop of AI performance, up to 128GB of unified memory, all-day battery life, and full AI and graphics performance unplugged. Systems arrive this fall from Microsoft Surface, ASUS, Dell, HP, Lenovo and MSI. DGX Station for Windows is aimed at enterprise workflows and applications, powered by the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip with up to 748GB of coherent memory and 20 petaflops of FP4 performance, and it runs frontier models of up to 1 trillion parameters. Systems are expected from ASUS, Dell, GIGABYTE, HP, MSI and Supermicro in Q4. Both products run NVIDIA OpenShell.
On Microsoft Foundry and Azure, the partnership adds NVIDIA, Anthropic and OpenAI models, along with Hermes special agents, to hosted agents in Foundry Agent Service. Anthropic’s Claude models now run natively on NVIDIA GB300 Blackwell Ultra systems on Azure, with customer availability in the weeks ahead. NVIDIA Nemotron 3 Ultra is available this month on Foundry managed compute, alongside Nemotron 3.5 ASR and Nemotron 3.5 Content Safety. NVIDIA says its open model portfolio on Foundry now covers agentic, physical and scientific Artificial Intelligence, including Cosmos 3 and Earth-2 weather models. NVIDIA Agent Toolkit, NemoClaw and CUDA-X libraries including cuDF, cuOpt, AI-Q and NeMo are also being made accessible for building production agents.
Microsoft Fabric Data Warehouse now includes NVIDIA accelerated computing. Microsoft’s internal benchmarking delivering SQL execution up to 6x faster than the CPU-powered baseline and up to 7x faster than three other leading cloud data warehouse providers for high-concurrency workloads. Microsoft is also integrating NVIDIA’s open source physical Artificial Intelligence tools with Azure and its Physical AI Toolchain to support robots, autonomous vehicles and industrial systems.
For local and hybrid deployments, Microsoft is bringing Foundry Local on Azure Local to the NVIDIA RTX PRO 6000 Blackwell Server Edition platform, paired with the NVIDIA Nemotron model family. Foundry Local on Azure Local now supports multinode deployments and the vLLM runtime for latency-sensitive and sovereign scenarios. In software security, NVIDIA OpenShell is now integrated into GitHub Copilot, isolating each agent in a sandboxed container and applying code-defined policies before outbound calls can reach files, networks or credentials.
At the infrastructure level, Microsoft’s Fairwater Wisconsin Artificial Intelligence factory is now live, ahead of schedule, running hundreds of thousands of NVIDIA Grace Blackwell systems as a single AI factory and linked with a similar facility in Georgia. Microsoft has also validated the NVIDIA Vera Rubin platform for deployment across Azure data centers. Vera Rubin slots in alongside Blackwell with no retrofits, delivering up to 10x inference throughput per megawatt and reducing cost per agentic token by an order of magnitude. NVIDIA says Dynamo and Grove extend those gains in software for AKS and Kubernetes-native distributed inference orchestration.
