Microsoft Fabric & Power BI — February 2026 Update

The February update has landed, and while it's not the headline-grabbing release we saw around FabCon last year, there's some useful features - particularly if your team lives in Power BI day-to-day or you're building out a data platform on Fabric.

The Power BI reporting improvements are mostly about usability. The Input Slicer is now GA — it was called the Text Slicer in preview, but it's been renamed and it's production-ready. If your reports deal with long lists of codes, categories, or names where a dropdown slicer falls apart, this is a great addition to slicer functionality. You type to filter, choose the match mode (contains, starts with, exact, etc.), and the report responds. You can also now paste a list of values directly into any standard slicer — copy a column from Excel, paste it in, done. Small feature, but anyone who's manually scrolled through a 200-item slicer list will appreciate it!

The Card visual now defaults to 10 callouts instead of 5, and clicking a category within the card cross-filters the rest of the page. For KPI dashboards this means more data on screen and more interactivity without touching the report design.

Copilot gets two practical updates. In Power BI Apps, you no longer need to pre-select a report before asking a question — Copilot works out which data to pull from and asks a clarifying question if it needs to. You can also ask it to summarise what an app contains, which is useful for onboarding new users. The prompt character limit has also jumped from 500 to 10,000, so you can give Copilot meaningful context rather than cramming everything into a tweet :)

On the data platform side, the OneLake Catalog now covers all Fabric item types including Workspace Apps, and opening a semantic model from anywhere in Fabric gives you a consistent detail page with the full schema visible. It's a governance and discoverability improvement that matters as environments grow. Worth noting for admins: from 8 February the Fabric Copilot capacity tenant setting is enabled by default across all tenants. It doesn't change any existing configurations, but it's worth checking in the admin portal if your organisation has a preference.

The Data Engineering updates are worth a look if you're managing data movement pipelines. Incremental copy from a Fabric Lakehouse now supports both CDF (Delta Change Data Feed) and watermark-based methods in a Copy job. CDF remains the recommended approach as it captures inserts, updates, and deletes cleanly, but watermark-based is now available for tables where CDF isn't enabled — useful when you don't control the source table configuration. Dataflow Gen2 gets Relative References for Fabric connectors, which is a meaningful CI/CD improvement. Previously, connecting to a Lakehouse or Warehouse embedded hardcoded workspace and item GUIDs in the script — meaning every deployment needed manual updates. Relative References lets you select "(Current Workspace)" and the script uses the item name instead, so the same dataflow works cleanly across dev, test, and production without modification. Finally, Data Factory can now read large CSV files in parallel. It determines how to split the file based on your multiline configuration and processes the parts simultaneously — a solid throughput improvement for anyone ingesting large flat file exports.

A decent update. Nothing that fundamentally changes the game, but a lot of things that make existing workflows faster and less painful — which is exactly what a mature platform should be doing. For all details check the official blog