Anthropic announced that its Claude AI models are now available in Microsoft Foundry on Azure, marking the first deployment on Nvidia hardware. The Claude models operate on Nvidia’s GB300 Blackwell Ultra GPU systems, enabling Azure customers to develop autonomous and domain-specific AI agents.
The deployment utilizes Nvidia GB300 NVL72 systems with Quantum-X800 InfiniBand networking. This setup is intended to enhance inference performance and efficiency while lowering total ownership costs for enterprise AI workloads.
This announcement follows a statement from Nvidia and Microsoft in June, indicating that Claude models would soon “run natively” on Azure’s Nvidia GB300 systems. The deployment aligns with growing enterprise interest in agentic AI applications.
The launch builds on a strategic partnership established in November 2025, where Anthropic committed to purchase $30 billion in Azure compute capacity. Nvidia and Microsoft agreed to invest $10 billion and $5 billion in Anthropic, respectively.
Initially, Anthropic’s compute commitment was set to reach up to one gigawatt of capacity using Nvidia’s Grace Blackwell and Vera Rubin systems. This partnership allowed for Claude’s integration across Microsoft’s Copilot family, which includes GitHub Copilot and Microsoft 365 Copilot.
In October 2025, Microsoft deployed a large-scale Azure cluster featuring over 4,600 Nvidia GB300 Blackwell Ultra GPUs, initially designed for OpenAI’s inference workloads. The cluster’s advanced capabilities position Azure as a multi-model platform for enterprise customers, with Anthropic’s Claude now the second major AI model family running on this infrastructure.
