Azure AI Infrastructure Expands With NVIDIA at GTC

Introduction

Microsoft used NVIDIA GTC to outline a broader Azure AI strategy focused on production-ready agents, inference-optimized infrastructure, and Physical AI. For IT leaders and platform teams, the announcement signals faster paths from AI experimentation to governed enterprise deployment.

What’s new in Microsoft Foundry

Microsoft expanded Microsoft Foundry as its enterprise AI platform for building, deploying, and operating agents at scale.

Key updates include:

• Foundry Agent Service is now generally available for building AI agents that can reason, plan, and act across tools, data, and workflows.
• Observability in the Foundry Control Plane is also generally available, giving teams better visibility into agent behavior and operations.
• Voice Live API integration for Foundry Agent Service is now in public preview, enabling voice-first and multimodal real-time agent experiences.
• A refreshed Microsoft Foundry portal is now generally available.
• NVIDIA Nemotron models are now available in Microsoft Foundry, expanding the model catalog for enterprise AI workloads.

Microsoft also highlighted additional integrations with security and governance partners such as Palo Alto Networks Prisma AIRS and Zenity.

Azure AI infrastructure gets a major boost

Microsoft is also expanding Azure infrastructure to support inference-heavy, reasoning-based AI workloads.

Highlights include:

• Azure is the first hyperscale cloud to power on NVIDIA Vera Rubin NVL72 systems in Microsoft labs.
• Vera Rubin NVL72 is expected to roll out to Azure datacenters over the coming months.
• Microsoft says it has already deployed hundreds of thousands of liquid-cooled Grace Blackwell GPUs globally in under a year.
• Azure Local now has initial support for the NVIDIA Vera Rubin platform, helping customers in sovereign and regulated environments run advanced AI closer to where data resides.

Physical AI and digital twins on Azure

Microsoft and NVIDIA are also deepening work around Physical AI.

New capabilities include:

• A public Azure Physical AI Toolchain GitHub repository integrated with NVIDIA Physical AI Data Factory and core Azure services.
• Deeper integration between Microsoft Fabric and NVIDIA Omniverse libraries.
• Support for workflows that connect live operational data, digital twins, simulation, and AI-driven actions.

This is especially relevant for manufacturing, energy, and industrial operations where organizations want AI to move beyond dashboards into real-time decision support and automation.

Why this matters for IT admins

For Azure administrators and enterprise architects, these announcements point to three priorities:

• Better tooling for moving AI agents from pilot to production
• More infrastructure options for high-performance inference and reasoning workloads
• Stronger support for regulated, sovereign, and edge scenarios with Azure-consistent governance

Next steps

Organizations evaluating Azure AI should review Microsoft Foundry GA features, assess Nemotron model support, and watch for Azure availability of Vera Rubin NVL72. Teams in industrial sectors should also explore the new Physical AI Toolchain and Fabric-Omniverse integration for digital twin and simulation use cases.