Anthropic has launched its Claude AI models on Microsoft Azure, harnessing the power of NVIDIA’s latest Blackwell Ultra GB300 GPU platform. This collaboration aims to accelerate the deployment of intelligent, agent-driven AI applications in enterprise environments.
The integration brings two Claude models—Opus 4.8 and Haiku 4—into general availability. Running on NVIDIA GB300 NVL72 systems with Quantum-X800 InfiniBand networking, these models target advanced agentic workloads, including autonomous and specialized sub-agents capable of handling complex tasks across varied business domains.
Microsoft Azure’s cloud infrastructure supports this rollout with billing, authentication, and governance controls, ensuring enterprises can securely manage AI applications. Additionally, the NVIDIA Secure Agent Workspace Reference Design offers a structured framework for operating autonomous agents within governed environments, controlling identity, network access, credentials, and runtime policies at the infrastructure level.
This partnership deepens the integration of NVIDIA’s accelerated computing technologies into Anthropic’s stack, enabling enterprises to inject domain-specific skills into Claude agents. By using NVIDIA-verified agent capabilities, organizations can embed AI agents more thoroughly into their operations, effectively turning these models into foundational components of the enterprise software ecosystem.
The deployment builds upon a strategic alliance announced previously by Microsoft, NVIDIA, and Anthropic, focused on expanding enterprise access to powerful AI models running on NVIDIA hardware within Azure’s cloud environment. Anthropic plans to continue expanding its AI offerings on Azure to meet evolving business demands.