Key points
- Microsoft unveiled the Database Hub in Fabric, a unified management console for diverse database services.
- Fabric IQ provides a semantic layer to give AI agents contextual business understanding.
- New migration assistants will help customers move Azure services like Data Factory to Fabric.
At FabCon and SQLCon 2026 in Atlanta, Microsoft announced a significant convergence of its database and data platform offerings. The company is integrating its SQL Server and Microsoft Fabric data analytics platform into a single, unified architecture. This move aims to simplify management across transactional, operational, and analytical workloads, particularly as customer demand shifts toward AI-powered applications.
The centerpiece of this integration is the new Database Hub in Microsoft Fabric, now in early access. Microsoft says this hub provides a unified database management experience, bringing services like Azure SQL, Azure Cosmos DB, and SQL Server (via Azure Arc) into a single view. The goal is to allow teams to explore, govern, and optimize their entire database estate from one place without redeploying existing services.
Database Hub introduces an agent-assisted management model using Microsoft Copilot. The system can surface changes, explain their impact, and suggest actions, aiming for a more proactive and resilient operational model. Microsoft emphasizes that humans remain in control of goals and trust boundaries. A savings plan for databases, promising up to 35% lower costs versus pay-as-you-go on select services, was also announced.
Beyond unifying the database layer, Microsoft is enhancing Fabric’s core data unification with Microsoft OneLake. OneLake acts as a single logical data lake, connecting data across clouds, on-premises, and third-party platforms. The company is expanding its Mirroring capability to more sources, including SharePoint lists and Oracle, with extended features like Change Data Feed now in preview as a paid option.
Security for OneLake is also being strengthened. OneLake security, now generally available, allows data owners to define roles and enforce row and column-level controls through a single unified model that follows the data. Interoperability remains a focus, with native reading from OneLake via Azure Databricks Unity Catalog in public preview and Snowflake interoperability already generally available.
Preparing data for AI requires more than unification; it demands harmonization and semantic meaning. Fabric’s analytics engines, including Spark and T-SQL, are being updated. A new Runtime 2.0 (preview) focuses on large-scale computation with Apache Spark 4.x. Materialized lake views are now generally available, simplifying medallion architecture implementation.
For real-time scenarios, Maps in Fabric reached general availability, adding geospatial context to agents. The next critical layer is semantic understanding, handled by Fabric IQ. Unlike point solutions, Microsoft says Fabric IQ creates a shared business context from productivity signals, institutional knowledge, and live business data. This context is housed in ontologies—frameworks of entities, relationships, and rules—that agents and users can navigate instead of raw tables.
Fabric IQ ontologies will soon be accessible via an MCP server in preview. A major new addition is planning in Fabric IQ, enabling budgets and forecasts directly on semantic models. This creates a single source for historical, real-time, and forward-looking data. Power BI’s semantic model technology and Fabric’s graph database underpin Fabric IQ. Direct Lake on OneLake is now generally available, and graph in Fabric, for querying complex relationships, is nearing general availability.
To act on this context, Microsoft is building specialized AI agents. Fabric data agents (now generally available) act as virtual analysts for specific domains. Operations agents monitor real-time data and take proactive action. These agents can be used within Fabric or as knowledge sources in tools like Microsoft Foundry and Copilot Studio.
For developers, the Fabric Model Context Protocol (MCP) is advancing. Fabric local MCP is generally available, connecting AI coding assistants to Fabric. Remote MCP (preview) provides a secure cloud execution engine. Git integration now supports selective branching, and new open-source projects like Agent Skills for Fabric and Fabric Jumpstart aim to accelerate development. The Fabric Extensibility Toolkit (FET) is now generally available with full CI/CD support.
Finally, Microsoft is simplifying migration from other Azure services. New migration assistants for Azure Data Factory, Synapse Analytics, and Azure SQL are in public preview. These tools help move pipelines, notebooks, and database schemas into Fabric with AI-assisted compatibility checks. The announcements reflect Microsoft’s strategy to position Fabric as a comprehensive, AI-ready data platform that unifies previously separate tools and services. The company highlights customer adoption, noting Fabric serves over 31,000 customers and SQL Server 2025 is growing rapidly.