NVIDIA and Microsoft Propel Agentic AI Innovation: Transforming Industries from Cloud to PC
Agentic AI is revolutionizing industries by redefining scientific discovery and unlocking unprecedented research breakthroughs. Through a deepened collaboration, NVIDIA and Microsoft are accelerating the development and deployment of agentic AI-powered applications, spanning from cloud-based solutions to Windows 11 PCs. This partnership empowers researchers, developers, and enterprises to innovate faster, achieve greater efficiency, and deliver transformative outcomes.
Microsoft Discovery: Empowering Scientific Breakthroughs with Agentic AI
At Microsoft Build, Microsoft unveiled Microsoft Discovery, an extensible platform designed to transform the entire scientific discovery process using agentic AI. By integrating advanced AI tools, Microsoft Discovery accelerates the time-to-market for new products and streamlines end-to-end research workflows for scientists across industries.
A key highlight of this platform is its integration with NVIDIA ALCHEMI NIM microservices, which optimize AI inference for chemical simulations. These microservices enable rapid property prediction and candidate recommendation, significantly speeding up materials science research. Additionally, NVIDIA BioNeMo NIM microservices provide pre-trained AI workflows for drug discovery, empowering researchers to develop AI models more efficiently.
In one groundbreaking example, Microsoft researchers used Microsoft Discovery to identify a novel coolant prototype with promising properties for immersion cooling in data centers—all within 200 hours, compared to months or years with traditional methods. This showcases how agentic AI can drastically reduce the time and resources required for complex research.
Scaling Agentic AI with NVIDIA GB200 Systems on Azure
To support the growing demands of generative and agentic AI workloads, Microsoft is deploying tens of thousands of NVIDIA GB200 NVL72 rack-scale systems across its Azure data centers. These systems feature up to 72 NVIDIA Blackwell GPUs per rack, advanced liquid cooling, and high-speed interconnects like NVIDIA NVLink and Quantum InfiniBand networking.
The result? Azure’s ND GB200 v6 virtual machines deliver up to 35x more inference throughput than previous-generation ND H100 v5 VMs. This leap in performance sets a new benchmark for AI workloads, enabling seamless scaling to tens of thousands of GPUs for even the most demanding applications.
During the event, Microsoft Chairman and CEO Satya Nadella and NVIDIA Founder and CEO Jensen Huang emphasized how continuous software optimizations across NVIDIA architectures on Azure maximize developer productivity, lower total cost of ownership, and accelerate all workloads—from AI and data processing to generative tasks. These advancements also drive greater efficiency per dollar and per watt, benefiting customers across industries.
Expanding AI Capabilities with NVIDIA Microservices on Azure AI Foundry
Building on their collaboration announced at NVIDIA GTC, Microsoft and NVIDIA are expanding Azure AI Foundry with cutting-edge microservices. Developers now have access to the NVIDIA Llama Nemotron family of open reasoning models and BioNeMo NIM microservices, which deliver enterprise-grade, containerized inferencing for complex decision-making and domain-specific AI workloads.
Key highlights include:
- NVIDIA Llama Nemotron Super and Nano Models: These advanced models offer multistep reasoning, coding, and agentic capabilities, delivering up to 20% higher accuracy and 5x faster inference than previous iterations.
- Healthcare-Focused BioNeMo NIM Microservices: Tools like ProteinMPNN, RFDiffusion, and OpenFold2 address critical applications in digital biology, drug discovery, and medical imaging. Researchers and clinicians can leverage these tools to accelerate protein science, molecular modeling, and genomic analysis, ultimately improving patient care and driving faster scientific innovation.
This expanded integration simplifies scaling and ensures robust reliability, enabling organizations to deploy high-performance AI agents tailored to their specific needs.
Accelerating Generative AI on Windows 11 with RTX AI PCs
Generative AI is transforming PC software, creating entirely new experiences—from intelligent writing assistants to creative tools and digital humans. NVIDIA RTX AI PCs are at the forefront of this revolution, making it easier than ever to experiment with generative AI and unlock superior performance on Windows 11.
At Microsoft Build, NVIDIA and Microsoft introduced an advanced AI inferencing stack to streamline development and boost performance for Windows 11 PCs. Central to this effort is TensorRT for RTX, which has been reimagined specifically for RTX AI PCs. Combining industry-leading TensorRT performance with just-in-time, on-device engine building, this solution reduces package size by 8x, ensuring seamless AI deployment across more than 100 million RTX AI PCs.
Available natively through Windows ML, TensorRT for RTX delivers state-of-the-art performance while maintaining broad hardware compatibility. Developers can access TensorRT for RTX via the Windows ML preview starting today, with a standalone SDK set to launch on NVIDIA Developer in June.
Empowering Developers and Enthusiasts with NIM Microservices and AI Blueprints
For developers and enthusiasts eager to dive into AI, NVIDIA offers NIM microservices and AI Blueprints, which simplify the creation of advanced workflows. Recent updates include:
- FLUX.1-schnell NIM microservice: An image generation model optimized for speed.
- Updated FLUX.1-dev NIM microservice: Now compatible with a wide range of GeForce RTX 50 and 40 Series GPUs.
These tools enable faster performance with TensorRT and quantized models, running over twice as fast on NVIDIA Blackwell GPUs compared to native execution.
Additionally, Project G-Assist—an experimental AI assistant integrated into the NVIDIA app—allows users to control their GeForce RTX systems with natural language commands. Developers can create custom plug-ins using the Project G-Assist Plug-in Builder, leveraging no-code or low-code development with JSON definitions and Python logic. New plug-ins include integrations with Spotify, Twitch, and IFTTT, showcasing diverse ways AI can enhance PC and gaming workflows.
Join the Agentic AI Revolution
Through their collaboration, NVIDIA and Microsoft are driving unparalleled advancements in agentic AI, empowering researchers, developers, and enterprises to push the boundaries of innovation. Whether you’re leveraging Microsoft Discovery, deploying AI agents on Azure, or experimenting with generative AI on RTX AI PCs, the tools are here to bring your ideas to life.
Stay connected with the RTX AI Garage blog series to explore community-driven projects, tutorials, and inspiration. Join the NVIDIA Developer Discord channel to collaborate, share creations, and gain support from a growing community of innovators.
As agentic AI continues to evolve, NVIDIA and Microsoft remain committed to delivering cutting-edge solutions that accelerate discovery, enhance productivity, and transform industries. Dive in today and be part of the future of AI-powered computing.
