At the NVIDIA GTC AI Conference, Microsoft and NVIDIA made several announcements across the entire technology stack, targeting developers, startups, and organizations.
Microsoft announced today that Azure AI Foundry now offers NVIDIA NIM microservices. NIM microservices are pre-built containers powered by NVIDIA inference software. They are designed to optimize inference for more than two dozen popular foundation models, enabling developers to easily deploy Gen AI applications. Microsoft also revealed plans to integrate the NVIDIA Llama Nemotron Reason open reasoning model.
Additionally, Microsoft is working with NVIDIA to optimize inference performance for open-source language models on Azure AI Foundry. For example, they worked with NVIDIA on performance optimization for Meta Llama models using TensorRT-LLM. Developers will be able to experience improved performance when using these models without any additional steps.
Microsoft today announced the general availability of the following services:
- Azure Container Apps serverless GPUs with support for NVIDIA NIM that will allow anyone to run AI workloads on-demand with automatic scaling, optimized cold start, and per-second billing with scale down to zero when not in use.
- Azure ND GB200 V6 virtual machine (VM) series accelerated by NVIDIA GB200 NVL72 and NVIDIA Quantum InfiniBand networking.
Microsoft also confirmed that the upcoming NVIDIA Blackwell Ultra GPUs and the NVIDIA RTX PRO 6000 Blackwell Server Edition will be available on Azure. The NVIDIA Blackwell Ultra GPU-based VMs will be available later in 2025.
Omar Khan, Vice President, Azure Infrastructure Marketing, wrote the following regarding the new announcements at GTC 2025:
Today’s announcements at the NVIDIA GTC AI Conference underscore Azure’s commitment to pushing the boundaries of AI innovations. With state-of-the-art products, deep collaboration, and seamless integrations, we continue to deliver the technology that supports and empowers developers and customers in designing, customizing, and deploying their AI solutions efficiently.
Finally, Microsoft and NVIDIA have launched preconfigured NVIDIA Omniverse and NVIDIA Isaac Sim virtual desktop workstations, and Omniverse Kit App Streaming, on the Azure marketplace. These VMs will enable developers to quickly start developing and deploying digital twin and robotics simulation applications and services.
0 Comments - Add comment