Azure Updates: AI Foundry; NVIDIA; VMs

MSDW Insight: As the month of March draws to a close, Microsoft placed AI development, AI Foundry, and its partnership with NVIDIA center stage. 

Microsoft VP for Azure Infrastructure Marketing, Omar Khan, looked at the ways that Microsoft and NVIDIA are working together to accelerate AI development. Building on a longstanding partnership, the two companies announced that NVIDIA’s Blackwell platform is being integrated with Azure AI services infrastructure. Simultaneously, the two companies incorporating NVIDIA NIM microservices in AI Foundry, and working with organizations like BMW, Dentsu, Harvey, and OriGen on their innovation initiatives.

“Microsoft is also working closely with NVIDIA to optimize inference performance for popular, open-source language models and ensure they are available on Azure AI Foundry so customers can take full advantage of the performance and efficiency benefits from foundation models. The newest addition of this collaboration is the performance optimization for Meta Llama models using TensorRT-LLM,” he wrote.

Aksha Sharma, Microsoft’s CVP for AI Platform also touched on AI Foundry and NVIDIA NIM. By integrating NVIDIA NIM into AI Foundry, Microsoft has endeavored to offer zero-configuration deployments, support seamless Azure integration, and provide “enterprise-grade” reliability. NVIDIA’s AgentIQ is also part of the picture, providing an open-source toolkit to connect and optimize groupings of AI agents. It is intended to support profiling, optimization, dynamic inference enhancement, and be integrated with Semantic Kernel.