Azure Storage for AI workloads | OD870

Saurabh Sensharma and Vishnu Charan TJ cover how Azure Storage can be used to improve performance and cost efficiency for AI inference workloads, including agent-based scenarios.

Overview

The session explains how Azure Storage fits into the AI stack to:

• Securely connect enterprise data to AI models.
• Accelerate inference by improving data access patterns and storage throughput.
• Reduce GPU idle time by speeding up model loading and distribution.
• Integrate with Microsoft and open-source AI frameworks to support scalable, agent-based applications.

Topics called out in the session description and chapter list

Storage for AI and AI for Storage

• High-level framing of how storage enables AI workloads, and how AI can be applied to storage scenarios.

Azure Storage integration across the AI stack

• How Azure Storage integrates across infrastructure and AI frameworks.

Clients and tools for AI workloads

• Azure Storage clients and tooling relevant to AI workload data access.

Deployment paths for AI workloads

• Running AI workloads with storage across:
  • Azure AI Foundry
  • Azure Kubernetes Service (AKS)
  • IaaS-based deployments

Storage requirements for agentic inference

• Storage roles and requirements when building agent-based (agentic) inference systems.

Inference optimization with caching

• Prompt caching as an optimization technique.
• Explicit caching using Azure Blob Storage.
• NIXL integration demo (as listed in chapters).

Faster model loading and distribution

• Approaches to reduce model load time and improve distribution.
• Run:AI Streamer and distributed cache (as listed in chapters).

Bringing enterprise data to AI

• Azure integrations for connecting enterprise data to AI systems.
• Foundry IQ (as listed in chapters).

Storage Center

• Introduction to Storage Center and recap (as listed in chapters).

Session metadata

• Event: Microsoft Build 2026
• Session: OD870
• Language: English (US)
• Track: Cloud platform & data