NVIDIA and Microsoft announced new integrations during Microsoft Ignite 2025. Microsoft’s Fairwater AI superfactory will use next-generation NVIDIA Spectrum-X Ethernet switches and NVIDIA Blackwell GPUs for large scale training. Microsoft will also deploy more than 100,000 Blackwell Ultra GPUs in NVIDIA GB300 NVL72 systems for global inference workloads.
Azure introduced NCv6 Series virtual machines powered by NVIDIA RTX PRO 6000 Blackwell GPUs. These support multimodal AI, simulation, industrial digitalization and visual computing. The same technology can run on Azure Local which allows low latency workloads at the edge.
Both companies are enhancing software performance through continuous full stack optimization. This helps accelerate models used by Microsoft AI teams and reduces operating costs. NVIDIA tools like TensorRT-LLM and DGX Cloud Benchmarking support these improvements.
The partnership also brings AI directly to enterprise data through SQL Server 2025 with NVIDIA Nemotron and NIM microservices. The NeMo Agent Toolkit now connects with Microsoft Agent 365 which helps developers build workplace AI agents inside Outlook, Teams and other Microsoft apps. Further work includes cybersecurity research, robotics development with NVIDIA Isaac Sim, industrial workflows through Omniverse libraries and 3D content interoperability on Azure.
Leave a comment