Azure Storage product managers walk through how Azure Blob Storage powers AI inference at scale across three pillars: prompt/KV caching, fast model loading, and enterprise data ingestion. Key demos include the NIXL Azure Blob backend enabling 6x faster KV cache cold starts (52s to 8s for 100k token prompts), the RunAI model streamer achieving near line-rate model weight loading (86 Gbps), and a new AKS distributed cache using local NVMe that delivers 2.5x faster model loading at cluster scale. On the data side, Azure Files is now a native data source for Azure AI Search, and Foundry IQ provides a unified governed knowledge layer for agents backed by Blob storage. A new Storage Center hub is announced in limited preview as a unified entry point for storage-to-AI workflows.