NVIDIA and Microsoft announced an expanded partnership at Microsoft Build, covering a unified stack for agentic AI deployment across Windows devices, Azure cloud, and local/on-premises environments. Key announcements include RTX Spark laptops and DGX Station for Windows (purpose-built AI hardware for agents), NVIDIA open models (Nemotron 3 Ultra, Cosmos 3) on Microsoft Foundry, GPU-accelerated Microsoft Fabric Data Warehouse delivering up to 6x faster SQL execution, NVIDIA OpenShell secure agent runtime integrated into GitHub Copilot, and the NVIDIA Vera Rubin platform validated for Azure data centers offering up to 10x inference throughput per megawatt. The partnership also extends to physical AI with Cosmos 3 integration into Azure's Physical AI Toolchain.
