NVIDIA and Microsoft announced a deepened collaboration at Microsoft Ignite 2026 to expand accelerated computing from cloud data centers to edge and personal computing devices. Central to their partnership is the deployment of NVIDIA Vera Rubin NVL72 GPU systems within Azure data centers, designed to handle complex AI workloads such as large-scale training and agentic reasoning applications.
These GPU systems, co-designed and optimized by both companies, underpin Azure's AI infrastructure delivering enhanced performance demonstrated in MLPerf benchmarks, with Azure scaling beyond 10,000 GPUs for demanding models. To address enterprise security needs, NVIDIA and Microsoft are extending confidential computing capabilities to protect sensitive AI models and data during processing, including attestation protocols to verify computing environments.
Beyond cloud infrastructure, the collaboration introduces expanded offerings through Microsoft Foundry, allowing enterprises to customize and deploy open-source AI models like NVIDIA Nemotron 3.5. Programmed Training enables customers to adapt models to specific business datasets and workflows, demonstrated by EY’s successful adaptation of Nemotron models for corporate record classification, yielding substantial accuracy improvements and cost reductions.
Deployment options now extend closer to enterprise data via Azure Local environments equipped with NVIDIA RTX PRO 6000 Blackwell GPUs and support for NVIDIA Jetson edge platforms, enabling organizations to run AI models directly on-premises or at the edge for applications such as robotics and industrial automation.
On the consumer and workstation front, NVIDIA RTX Spark devices introduce local large language model capabilities on Windows PCs. Additionally, GPU-accelerated Windows 365 Cloud PCs are gaining Direct Stream features to support high-performance graphics streaming, targeting professional and enterprise workloads with low latency even under challenging network conditions.
For scientific research, NVIDIA’s BioNeMo and ALCHEMI toolkits are integrated with Microsoft Discovery, enabling natural language-driven computational biology workflows and drug discovery processes powered by AI models including those developed with the University of Washington.
Microsoft is also internally adopting NVIDIA technologies such as Dynamo inference software to boost the efficiency of its AI model training and service delivery, illustrating the pervasive application of NVIDIA acceleration across Microsoft’s AI ecosystem.
While the collaboration significantly advances AI infrastructure and deployment versatility, limitations remain regarding integration complexity for enterprises, the need for specialized hardware to access full capabilities, and ongoing challenges in securing AI models at scale across diverse environments.
Source & further reading
NVIDIA and Microsoft Expand Accelerated Computing From Azure to Windows PCs ↗Source published 2026-11-19. Claims and limitations should be checked against the original announcement or research.