Microsoft held the third annual Fabric Community Conference (FabCon) in Atlanta from March 16–20, 2026, co-located for the first time with SQLCon, the Microsoft SQL Community Conference. The combined event drew more than 8,000 attendees across nearly 300 sessions at the Georgia World Congress Center.
The colocation was deliberate. Microsoft used the event to position Fabric not as an analytics platform but as the control plane for the entire enterprise data estate — explicitly collapsing the historical separation between transactional database workloads and analytical workloads. SQL Server 2025, Azure SQL, and SQL in Fabric were framed as a single, consistent SQL engine spanning on-premises, PaaS, and SaaS environments, with Fabric as the unification layer rather than a replacement.
Two framing numbers anchored the keynote: Microsoft Fabric now serves more than 31,000 customers two and a half years after general availability, and SQL Server 2025 is growing more than twice as fast as the prior version.
The Strategic Narrative
Three themes ran through the keynote and most partner recaps:
- Convergence of operational and analytical data. Databases are being repositioned as participants in the AI and analytics workflow rather than upstream systems that feed it. Database Hub, OneLake, and expanded mirroring are the technical mechanisms behind this shift.
- Semantic context as a first-class capability. Fabric IQ introduces ontologies, graph, and semantic models as the grounding layer for AI agents. The recurring message: agents are only as reliable as the semantic context they operate over.
- Action, not just insight. Operations agents, planning in Fabric IQ, and translytical task flows shift Fabric toward a platform that takes action on data rather than only reporting on it.
Major Announcements
Fabric IQ
The headline announcement. Fabric IQ is an umbrella semantic intelligence layer with five components:
- Semantic Models — extended beyond Power BI reporting into AI grounding.
- Graph in Fabric — a scalable graph database for modeling relationships across customers, partners, and supply chains. Going generally available in the weeks following the conference.
- Data Agents — now generally available. Function as virtual analysts that answer business questions grounded in governed data.
- Operations Agents — new. Built on Real-Time Intelligence to monitor data and trigger workflows or remediation autonomously.
- Ontologies — the only entirely new component. Define business entities, relationships, properties, rules, and actions as an agent-readable domain model. Available in preview.
Fabric IQ ontologies will become accessible through an MCP server in preview, enabling AI agents to discover and act on the semantic layer programmatically. Fabric IQ sits alongside Work IQ (productivity signals from Microsoft 365) and Foundry IQ (AI Foundry institutional knowledge) within Microsoft’s broader IQ framework.
Microsoft also announced planning in Fabric IQ — a native enterprise planning capability that supports budgets, forecasts, and scenario models built directly on top of governed semantic models, eliminating the data movement and reconciliation typically required between planning tools and analytics platforms.
Database Hub in Fabric (Early Access)
The other headline announcement and the centerpiece of the SQLCon message. Database Hub provides a unified management experience for heterogeneous database estates, bringing Azure SQL Database, Azure Cosmos DB, Azure Database for PostgreSQL, Azure Database for MySQL, SQL Server (via Azure Arc), and Fabric Databases into a single view.
Database Hub does not replace existing tools or force consolidation — it centralizes observability, governance, and optimization across environments. It introduces an agent-assisted, human-in-the-loop model with Copilot-powered insights that surface what changed, why it matters, and recommended remediation.
OneLake
- OneLake security going GA in the weeks following the conference. Role-based, row-level, and column-level controls under a single unified permissions model that follows the data.
- Native read from OneLake through Azure Databricks Unity Catalog — now in public preview. Reduces friction for organizations operating both Fabric and Databricks.
- Snowflake interoperability GA. Bidirectional reading of Iceberg data managed by Snowflake or Fabric, with the ability to natively store Snowflake-managed Iceberg tables in OneLake.
- Microsoft Foundry integration. OneLake catalog accessible directly within Foundry, enabling one-click connection of OneLake items as knowledge sources for AI solutions.
- OneLake file explorer for Windows going GA, bringing browse, upload, download, and edit capabilities for OneLake data into the standard Windows file system experience.
Mirroring Expansion
- New sources: SharePoint, Oracle, SAP Datasphere, Dremio, and Azure Monitor.
- Change Data Feed support on mirrored data, enabling incremental and CDC-aware downstream consumption.
- Views on mirrored data plus shortcut transformations to simplify access and preparation.
- Snowflake mirroring support for views in preview.
Engineering and Compute
- Runtime 2.0 (preview) — described by Microsoft as the most significant compute upgrade since Fabric reached general availability. Purpose-built for large-scale data computation.
- Spark 4.0 shipped in Experimental Public Preview (a lower maturity tier than standard preview, not recommended for production workloads).
- Materialized Lake Views introduced as part of the broader compute refresh.
- Workload management features for the Fabric admin portal, including a Manage Workloads tab with centralized workload visibility and tenant assignment controls.
- Branched workspaces and selective branch-out for Git integration, improving the developer experience for feature-based workflows.
Power BI and The Semantic Layer
- Direct Lake on OneLake general availability, with native security enforcement, cross-item modeling, and import-class performance without data movement or refresh.
- Translytical Task Flows GA — enables write-back actions from Power BI reports.
- Maps GA, adding native geospatial analytics.
- Semantic model AI capabilities expanding the role of semantic models as agent grounding.
SQL Server and SQLCon-specific Announcements
The SQL Server message was deliberately additive — modernization without disruption.
- SQL Server 2025 “Expose to Fabric” — zero-copy exposure of SQL Server data to OneLake.
- DiskANN public preview in SQL Server 2025 for vector search.
- SSMS 25 with GitHub CI/CD integration.
- SQL MCP server for AI agent access to SQL data.
- Azure SQL Database savings plans offering up to 35% savings versus pay-as-you-go pricing, depending on commitment.
Migration Tooling
- Fabric migration assistant for Azure Data Factory and Synapse Analytics moves existing pipelines, Spark pools, and notebooks into Fabric, supporting incremental modernization.
- Migration assistant for SQL databases handles schema import via DACPAC, AI-assisted compatibility issue resolution, and guided assessment workflows.
Strategic Partnerships
- NVIDIA partnership. Real-Time Intelligence and Fabric IQ are being integrated with NVIDIA Omniverse libraries to enable Physical AI scenarios — intelligent digital twins, predictive maintenance, autonomous logistics, and energy optimization.
Practical Implications
For organizations running Microsoft data platforms, the key planning considerations coming out of the event:
- Semantic foundations matter more than ever. Fabric IQ ontologies are powerful, but they require consistent underlying schemas to be useful. Organizations with fragmented semantic models will need to invest in foundational data modeling work before realizing the full benefit of agent-grounded AI.
- The mirroring expansion changes ingestion economics. Direct mirroring with CDC support from SAP, Oracle, and other enterprise sources reduces the case for custom ETL pipelines for many bronze-layer scenarios.
- Database Hub does not require consolidation. Organizations with mixed database estates can adopt the unified management plane without migrating off existing platforms.
- Migration paths are real, but timelines are typically 6–12 months out. Partner observations from the event consistently noted that organizations are planning rather than executing immediate moves, particularly from on-premises SQL Server.
- Operations agents introduce a new class of automation. They are distinct from data agents in that they take autonomous action on real-time signals. Governance and approval workflows around agent-initiated actions warrant early planning.
What’s next
The next FabCon + SQLCon Europe event takes place September 28 – October 1, 2026 in Barcelona, Spain, with a dedicated SQL track. The next North American FabCon + SQLCon is scheduled for Spring 2027.