During the GTC event, Microsoft and NVIDIA unveiled significant advancements in AI infrastructure and collaborative efforts, centered on the NVIDIA Blackwell Platform. Azure will be among the first to deploy the NVIDIA GB200 Grace Blackwell Superchip, optimized for large-scale generative AI tasks and high-performance computing. This superchip promises up to 16 TB/s of memory bandwidth and offers up to 30 times faster real-time inference for trillion-parameter models compared to previous GPU generations. Microsoft and NVIDIA are also integrating the GB200 into Azure AI infrastructure to enhance performance and scalability for complex AI models.
Moreover, their collaboration extends to deploying the NVIDIA Quantum-X800 InfiniBand networking platform, which supports advanced AI techniques and expands GPU capabilities across massive scales through its in-network computing capabilities. Together, these innovations are tailored to accelerate generative AI workloads such as natural language processing, computer vision, and speech recognition.
Microsoft also announced Azure NC H100 v5 virtual machines powered by NVIDIA H100 NVL Tensor Core GPUs, designed for AI inference and mid-range training workloads, as well as complex simulations like computational fluid dynamics and molecular dynamics. This new VM series leverages NVIDIA’s advanced GPU memory and NVLink v4 technology for enhanced bi-directional communication speeds, aiming to boost performance and scalability in AI and HPC applications.
Additionally, Microsoft and NVIDIA are enhancing healthcare and life sciences innovation by integrating Azure with NVIDIA DGX Cloud and the NVIDIA Clara suite. This partnership aims to accelerate clinical research, drug discovery, and medical image-based diagnostics, thereby improving efficiency and effectiveness in healthcare.
In the domain of AI deployments, NVIDIA GPUs and Triton Inference Servers will support AI inference predictions in Microsoft Copilot for Microsoft 365, enhancing productivity and creativity with real-time contextual intelligence. Furthermore, NVIDIA NIM inference microservices are being integrated into Azure AI, providing cloud-native microservices optimized for AI inference on foundational models.
Lastly, Microsoft is adopting NVIDIA Omniverse Cloud APIs to empower developers in industrial sectors to integrate digital twin technologies into their design and automation workflows. These APIs enable the creation and operation of physically accurate digital twins, facilitating advancements in autonomous machines and simulation workflows.
Overall, the collaboration between Microsoft and NVIDIA at GTC showcases their commitment to advancing AI and accelerating innovation across various industries through cutting-edge technologies and integrative solutions.
