FabCon Europe 2026: the Microsoft Fabric announcements that matter for data teams

Microsoft's Fabric Community Conference Europe, held together with SQLCon in Barcelona at the end of September, drew about 5,000 people and roughly 100 announcements. The message was simple: the semantic models and lakehouses you build in Fabric now feed Copilot and AI agents directly. That is exciting, and it also means messy models will be seen by far more people. Here is the shortlist that matters for Power BI developers, data engineers and capacity admins.

Your semantic models now power Copilot

The headline is Fabric IQ, now generally available inside Microsoft Copilot Chat and Cowork. Business users can ask a question in Copilot and get an answer grounded in the Power BI reports and semantic models they already have access to, with no extra AI token cost. Alongside it, the Fabric IQ MCP endpoint and Fabric data agents in Copilot Studio (now with interactive visuals) are GA, so other agents and tools can use the same models.

Also in this area: Power BI's agentic experiences, the local Power BI Authoring MCP server and PBIR as the default report format are GA, and table visuals now export up to 500,000 rows to Excel. Ontologies (a richer layer of business definitions), semantic views in OneLake and Microsoft 365 Copilot inside Power BI are in preview, the last one switched off by default.

Capacity: overage instead of throttling

For capacity admins, this is the change to act on first. Capacity overage is GA: when a capacity runs out of compute, Fabric can buy extra automatically instead of throttling users, billed at three times the pay-as-you-go rate on eligible excess capacity unit hours, with a configurable 24-hour limit.

For very large estates, F4096 and F8192 SKUs are now available, and the Capacity Metrics app gained new views.

Data engineering and the warehouse

Getting data in: mirroring and Data Factory

Mirroring keeps a live, analytics-ready copy of a source in OneLake without building pipelines. Two new sources are GA:

• Google BigQuery, useful for teams that run Google Analytics or marketing data there but report in Power BI.
• SharePoint Lists, finally ending the "export the list to Excel every Monday" routine.

Business Central and Salesforce Data 360 sharing are in preview. In Data Factory, dbt jobs are now GA, so you can run your dbt projects as native Fabric jobs, and Copy jobs gained change data capture, event triggers, smart retry and SCD Type 2 handling. A Modern Power Query Editor and a Dataflow Gen1 to Gen2 upgrade wizard are in preview; note the upgrade is in place and cannot be undone, so test on a copy.

Git, governance and developer tools

Git integration got the features teams kept asking for: compare and commit, selective branching and bulk item-definition APIs (GA), with deployment plans and file-level commits in preview. The Fabric CLI, Terraform provider, and a remote and local Fabric MCP server all improved. Fabric REST APIs now document clear limits: 500 calls a minute per identity for the platform APIs, 200 for the job scheduler and 500 for long-running operations.

SQL and databases

From the SQLCon side: DiskANN vector indexes are GA in Azure SQL Hyperscale, SQL Managed Instance and SQL database in Fabric, SSMS has an Agent Mode, and SQL Server runs on Azure Local in connected and disconnected modes. A new certification, Microsoft Certified: SQL AI Developer Associate (exam DP-800), was also announced.

What to do first

For the latest Power BI features, read the Power BI October 2026 update. For the bigger picture, see the data stack in 2026 and how Databricks and Snowflake are approaching the same agent era.

Sources: James Serra's summary of the FabCon Europe 2026 announcements (8 October 2026) and Bitmetric's FabCon Europe 2026 recap. Performance figures are vendor-reported. Preview features and dates can change.

Alessandro Ecclesie AgazziWritten by Alessandro Ecclesie Agazzi, analytics engineer in London. Updated 17 October 2026. About me