Microsoft & NVIDIA: Scaling AI Infrastructure with Foundry & New Accelerators | GTC 2024

Microsoft and NVIDIA are deepening their collaboration to accelerate the development and deployment of artificial intelligence, with a significant focus on bringing AI capabilities beyond traditional digital environments and into the physical world. Recent announcements at NVIDIA GTC 2026 detail expanded capabilities for Microsoft’s Foundry platform, recent Azure AI infrastructure optimized for demanding workloads, and tighter integration between Microsoft and NVIDIA’s software and hardware ecosystems. This builds on years of partnership, integrating hardware, software, and infrastructure to power advancements in AI.

Expanding Microsoft Foundry for Production-Ready AI

At the core of Microsoft’s strategy is Foundry, positioned as an operating system for building, deploying, and managing AI at scale. Foundry leverages Azure services to combine models, tools, data, and observability into a unified system designed for production AI agents. Key updates include the general availability of the next-generation Foundry Agent Service and Observability in Foundry Control Plane. These tools are designed to streamline the development and operation of AI agents capable of reasoning, planning, and executing tasks across various tools, data sources, and workflows. The Control Plane provides developers with end-to-end visibility into agent behavior, aiming to improve both productivity and trust in AI systems. Corvus Energy is already utilizing Foundry to automate inspection workflows across its global fleet, demonstrating a practical application of the platform.

Further simplifying the path from prototype to deployment, Microsoft is previewing Voice Live API integration with Foundry Agent Service, enabling developers to create voice-first, multimodal, real-time agentic experiences. Alongside this, a refreshed Microsoft Foundry portal and expanded integrations with security platforms like Palo Alto Networks’ Prisma AIRS and Zenity are now available, enhancing both the developer experience and runtime security throughout the agent lifecycle.

Bolstering Model Choice with NVIDIA Nemotron

Microsoft is similarly expanding the range of models accessible through Foundry, now including NVIDIA Nemotron models. This addition complements the existing selection of models available on Azure, encompassing reasoning, frontier, and open-weight options. This move follows a recent partnership bringing Fireworks AI to Microsoft Foundry, allowing customers to fine-tune open-weight models like NVIDIA Nemotron for low-latency performance, even at the edge. The ability to fine-tune and deploy models closer to the data source can reduce latency and improve responsiveness for certain applications.

Scaling AI Infrastructure with NVIDIA Vera Rubin

Microsoft recognizes that inference-heavy AI workloads—those focused on applying trained models to new data—are driving new demands on infrastructure. To address this, the company is deploying next-generation NVIDIA systems within its Azure datacenters, designed for efficient power, cooling, and rapid upgrades. Notably, Microsoft is the first hyperscale cloud provider to power on NVIDIA’s Vera Rubin NVL72 systems in its labs, with plans to roll them out into liquid-cooled Azure datacenters in the coming months. Hundreds of thousands of liquid-cooled Grace Blackwell GPUs have already been deployed across Microsoft’s global datacenter footprint.

This infrastructure investment extends to sovereign and regulated environments, offering customers greater control over where their AI workloads run and how they evolve. Foundry Local now supports modern infrastructure and large AI models, and initial support for the NVIDIA Vera Rubin platform on Azure Local is available, bringing accelerated AI capabilities to customer-controlled environments. This approach aims to balance performance with the governance and security requirements of regulated industries, leveraging Azure Arc and Foundry Local for consistent operations.

The Rise of Physical AI

Microsoft and NVIDIA are also collaborating on “Physical AI,” extending AI’s reach beyond digital experiences into the physical world. This operate centers on the NVIDIA Physical AI Data Factory Blueprint, with Microsoft Foundry serving as the platform for hosting and operating Physical AI systems on Azure. By integrating this blueprint with Azure services, developers can build, train, and operate workflows that connect physical assets, simulation environments, and cloud-based training pipelines. A public Azure Physical AI Toolchain GitHub repository has been released, integrated with the Nvidia Physical AI Data Factory and core Azure services, to support this effort.

Furthering this integration, Microsoft and NVIDIA are deepening the connection between Microsoft Fabric and NVIDIA Omniverse libraries. This allows for the integration of live operational data with physically accurate digital twins and simulations, enabling organizations to monitor physical systems in real-time and use AI to inform decision-making. Customers in manufacturing and operations are using this approach to move beyond simple monitoring and alerts to coordinated, AI-driven actions across machines, facilities, and workflows.

Looking Ahead: From Experimentation to Impact

Microsoft’s strategy focuses on delivering reliable, production-scale AI by combining its global infrastructure, platforms, and real-world systems with NVIDIA’s latest innovations. The goal is to enable customers to continuously operate intelligence, running inference-heavy, reasoning-based, and physical AI workloads with the necessary performance, security, and governance. The company emphasizes that AI transformation isn’t a solitary endeavor, but rather a process that thrives within interconnected ecosystems. The next steps involve continued refinement of these tools and platforms, broader deployment of next-generation infrastructure, and expansion of the partner ecosystem to address specific industry challenges. The focus remains on moving customers from initial experimentation to measurable business value through the practical application of AI.