Timed with the Microsoft Ignite convention working this week, NVIDIA is increasing its collaboration with Microsoft, together with by the adoption of next-generation NVIDIA Spectrum-X Ethernet switches for the brand new Microsoft Fairwater AI superfactory, powered by the NVIDIA Blackwell platform.
The collaboration brings new integrations throughout Microsoft 365 Copilot, in addition to the general public preview of next-generation Azure NC Sequence VMs powered by NVIDIA RTX PRO 6000 Blackwell Server Version GPUs, NVIDIA Nemotron integrations to speed up AI for Microsoft SQL Server 2025, capabilities for onboarding AI brokers in Microsoft 365 and optimizations for high-performance inference, cybersecurity and bodily AI.
Microsoft’s AI Superfactory connects the landmark Fairwater knowledge middle in Wisconsin with a brand new, state-of-the-art facility in Atlanta, Georgia. This massive-scale infrastructure will combine a whole bunch of hundreds of NVIDIA Blackwell GPUs for large-scale coaching. As well as, Microsoft is deploying greater than 100,000 Blackwell Extremely GPUs in NVIDIA GB300 NVL72 methods being deployed globally for inference.
“Our collaboration with NVIDIA is constructed on driving innovation throughout all the system and full stack, from silicon to providers,” stated Nidhi Chappell, company vice chairman of product administration at Microsoft. “By coupling Microsoft Azure’s unmatched knowledge middle scale with NVIDIA’s accelerated computing, we’re maximizing AI knowledge middle efficiency and effectivity, which is of paramount significance for our clients main the brand new AI period.”
Probably the most demanding workloads for OpenAI, the Microsoft AI Superintelligence Workforce, Microsoft 365 Copilot and Microsoft Foundry providers will probably be powered by this infrastructure. Prospects like Black Forest Labs are additionally utilizing NVIDIA GB200 NVL72 methods to coach next-generation multimodal FLUX fashions that energy visible intelligence.
To attach this large infrastructure, Microsoft is deploying next-generation NVIDIA Spectrum-X Ethernet switches in its Fairwater AI knowledge middle — the biggest and most subtle AI factories ever constructed — delivering the efficiency, scale and effectivity required for OpenAI to run large-scale AI fashions and purposes.
New Azure NCv6 Sequence VMs with NVIDIA RTX PRO 6000 Blackwell GPUs at the moment are in public preview on Azure, increasing the Blackwell platform to offer right-sized acceleration for a number of workloads together with multimodal agentic AI, industrial digitalization with NVIDIA Omniverse libraries, scientific simulation and visible computing. This flexibility extends from the cloud to the sting with Azure Native, enabling highly effective sovereign AI options whereas bringing low-latency, real-time AI to wherever knowledge must reside.
This permits enterprises to seamlessly develop, deploy and handle AI-powered digital twins and generative AI purposes with NVIDIA RTX PRO 6000 Blackwell GPUs from the Azure cloud on to their manufacturing facility flooring, on-premises knowledge facilities or safe edge places.
Software program Optimizations Ship a Fungible AI Fleet
The NVIDIA platform on Azure, spanning NVIDIA Blackwell and Hopper GPUs, accelerates the most recent fashions from the Microsoft AI Superintelligence Workforce, together with textual content (MAI-1-preview), real-time voice (MAI-Voice-1) and high-fidelity picture era (MAI-Picture-1) — bringing new multimodal experiences throughout Bing Picture Creator and Microsoft Copilot.
Central to NVIDIA’s collaboration with Microsoft is constructing a fungible fleet — a versatile, constantly modernized infrastructure that may speed up any workload with most effectivity. That is achieved by steady, full-stack software program optimizations that ship compounding efficiency good points and maximize throughput throughout all the AI lifecycle and throughout a number of NVIDIA architectures on Azure. The good points additionally prolong to workloads past generative AI, together with knowledge processing, vector search, databases, digital twins, scientific computing and 3D design.
This co-engineering saves important prices for purchasers, making AI tasks that have been as soon as theoretical now economically viable. For instance, the continual full-stack optimization work has instantly contributed to an over 90% drop within the value of widespread GPT fashions for finish customers on Azure in two years.
Ongoing optimization work now extends to Microsoft Foundry, the place the NVIDIA TensorRT-LLM library helps increase throughput, scale back latency and decrease prices for a variety of widespread open fashions.
NVIDIA and Microsoft have additionally partnered to optimize their fleet for AI workload efficiency by the NVIDIA DGX Cloud Benchmarking suite. Engineering groups from each firms labored intently collectively to establish bottlenecks and implement infrastructure tuning, driving efficiency good points. By attaining 95% of the efficiency doable utilizing the NVIDIA reference structure, Microsoft was named an Exemplar Cloud for H100 coaching.
From Clever Knowledge to AI Brokers
NVIDIA and Microsoft are integrating AI into the core of the enterprise, unlocking many years of proprietary knowledge saved in one of many world’s most trusted databases.
NVIDIA is accelerating AI within the new Microsoft SQL Server 2025 by integrating it with NVIDIA Nemotron open fashions and NVIDIA NIM microservices. This resolution delivers GPU-optimized, safe and scalable retrieval-augmented era instantly the place enterprise knowledge lives, within the cloud or on premises.
Plus, the collaboration extends to the brand new frontier of agentic AI within the office. The NVIDIA NeMo Agent Toolkit now connects with Microsoft Agent 365, enabling builders to construct, deploy and onboard compliant, enterprise-ready AI brokers instantly into the Microsoft 365 app ecosystem, together with Outlook, Groups, Phrase and SharePoint.
To energy these new enterprise brokers, Microsoft Foundry now gives NVIDIA Nemotron fashions for digital AI and NVIDIA Cosmos fashions for bodily AI as safe NIM microservices. Builders can use them to construct enterprise-grade agentic AI for an unlimited vary of purposes that profit from multimodal intelligence, multilingual reasoning, math, coding and bodily AI capabilities.
The collaboration can be tackling cyber threats for enterprises. Microsoft and NVIDIA are collaborating on analysis for new adversarial studying fashions, constructed on the NVIDIA Dynamo-Triton framework and the NVIDIA TensorRT suite of instruments, that may assist enterprises defend towards real-time cybersecurity threats with a 160x efficiency speedup in contrast with CPU strategies.
Bodily AI and Industrial Digitalization
NVIDIA and Microsoft are constructing the way forward for bodily AI. With NVIDIA Omniverse libraries accessible on Microsoft Azure, NVIDIA is unlocking end-to-end reindustrialization within the cloud by its developer ecosystem. Builders are remodeling industrial workflows, from computer-aided engineering with Synopsys to manufacturing facility operations with Sight Machine and SymphonyAI.
Robotics builders can faucet into the NVIDIA Isaac Sim open-source robotics simulation framework to unlock important workflows, from artificial knowledge era to software-in-the-loop testing for all sorts of robotic embodiments. Hexagon is constructing its AEON humanoid robotic primarily utilizing NVIDIA’s full robotics stack on Azure. Equally, the robotics platform, Wandelbots NOVA, working on Azure integrates Isaac Sim and Isaac Lab to simplify and velocity up simulation to real-world deployment.
As well as, NVIDIA and Microsoft are utilizing a standardized method for digital engineering to allow seamless OpenUSD interoperability throughout 3D workflows, making simulation and digital content material creation accessible within the cloud.
This expanded collaboration comes on the heels of a partnership introduced with Anthropic and Microsoft earlier immediately. NVIDIA and Anthropic will collaborate on design and engineering to optimize Anthropic fashions for efficiency, effectivity and whole value of possession, in addition to optimize future NVIDIA architectures for Anthropic workloads.
Be taught extra about NVIDIA and Microsoft’s collaboration and classes at Microsoft Ignite.