Microsoft is aggressively scaling its data and AI ecosystem, reporting that its unified platform, Microsoft Fabric, now serves 35,000 customers. This growth represents a significant jump from the 25,000 clients reported in September 2025. The platform, which integrates services like Power BI, Data Factory, and OneLake, currently generates $2 billion in annual recurring revenue for the Redmond-based firm—a 60% year-over-year increase.
How is Microsoft Fabric accelerating data performance?
To handle the heavy lifting of modern data analytics, Microsoft is introducing GPU-based query acceleration for its Fabric data warehouses. According to the company, this setup can process 100 GB of data for 64 concurrent users seven times faster than standard configurations. Unlike many competitors that rely on Nvidia’s RAPIDS library, Microsoft has built a proprietary framework using NVLINK and Infiniband.
This architecture includes an abstraction layer designed to distribute query plans across various coprocessors, including GPUs, ASICs, and FPGAs. Industry observers expect this foundation to eventually support Microsoft’s own Cobalt 200 CPUs and Maia 200 AI chips. Additionally, Fabric now supports Spark 4.0 and Delta 4.0, with a “custom live pool” feature that slashes notebook startup times from several minutes down to three to five seconds.
What role does Rayfin play in agentic AI development?
Microsoft is positioning its new open-source SDK, Rayfin, as a “back-end on demand” for developers building agentic AI applications. Currently in public preview, the SDK allows developers to define data models using TypeScript decorators. The framework then handles the heavy lifting of database management, authentication, and API hosting through Azure services.
Analysts at Futurum Group note that Rayfin changes the typical development lifecycle by treating governance as the foundation of the application rather than a final checklist. By keeping the entire lifecycle within the Fabric ecosystem, Microsoft aims to ensure that AI agents operate within established security guardrails. However, current documentation indicates that these apps are best suited for prototypes and dashboards, as they do not yet support complex multi-step transactions or stored procedures outside of Fabric’s single sign-on environment.
Is Fabric prepared for enterprise-grade AI deployment?
While Microsoft continues to expand its offerings, some analysts point to gaps in the current architecture. William McKnight, an independent analyst, notes that Microsoft has not yet fully integrated Fabric with its Purview governance and security layer. Furthermore, the absence of specialized LLMOps functions remains a hurdle for organizations looking to move AI agents into production.
In contrast, competitors like Snowflake have already introduced dedicated governance features, and Databricks is expected to announce similar advancements soon. For now, enterprises are often forced to build their own “AI harnesses” to manage the interchangeability of LLMs, tools, and servers, as no single platform currently offers a seamless, cross-platform governance solution.
Frequently Asked Questions
What is Fabric IQ?
Fabric IQ is a semantic layer introduced by Microsoft to help build knowledge graphs for AI agents. It uses ontologies—glossaries of business concepts—that are now accessible via Microsoft Foundry, Copilot Studio, and Agent 365.
Can Fabric connect to third-party databases?
Yes. Microsoft has introduced general availability for connectors to SAP Datasphere, SharePoint, OneDrive, and Oracle DB. Public previews for “shortcuts” to Azure DB for MySQL and Dremio are also available, allowing users to surface data directly in OneLake.
What are “Operations Agents”?
Operations Agents are now generally available in Fabric. They allow businesses to automate tasks by configuring instructions and triggers in Fabric Real-Time Intelligence, which then executes actions through Power Automate.
Are you integrating agentic AI into your data workflows? Share your experience with Fabric or other platforms in the comments below, or subscribe to our newsletter for the latest updates on enterprise data trends.
