dbt on Microsoft Fabric.
dbt does not move or own your data. On Fabric it compiles your models to T-SQL that runs inside your Fabric Warehouse, so the same dbt project you would run anywhere targets the lakehouse you already pay for.
What running dbt on Fabric looks like.
The platform is the target, not the project. dbt-fabric points dbt at a Fabric Warehouse, and your models compile to the T-SQL the Warehouse runs. Incremental models compile to MERGE, slowly changing dimensions use dbt snapshots, and your staging and marts map straight onto the bronze, silver and gold medallion layers. Sources, tests, freshness and documentation behave exactly as they do anywhere else, so a Fabric build is a configuration choice, not a different project.
One project, the Fabric Warehouse as the target.
Author once. dbt generates the T-SQL Fabric expects, and the platform-specific patterns sit at the edges.
- MERGE for incremental models
- Snapshots for slowly changing dimensions
- Bronze, silver and gold medallion layers
- One OneLake estate, governed in Fabric
No lock in: dbt generates the SQL and Fabric runs it. The same project moves to another platform without changing how your team works.
What you get on Fabric.
T-SQL on the Warehouse
dbt-fabric compiles your models to T-SQL that runs in the Fabric Warehouse, inside one governed OneLake estate.
Efficient runs and history
Incremental models use MERGE so runs touch only what changed, and dbt snapshots record slowly changing dimensions.
Medallion by design
Staging cleans the bronze data and marts build the silver and gold tables, so the layering lives in your dbt project.
No licence on top
dbt Core and the dbt-fabric adapter are free to run. You pay for the Fabric capacity you already use, nothing more.
dbt on Fabric, in short.
How does dbt run on Microsoft Fabric?
dbt uses the dbt-fabric adapter to target a Fabric Warehouse. Your models compile to T-SQL that the Warehouse executes, so the data never leaves your Fabric estate. The project structure, tests, sources and documentation are exactly the same as on any other platform.
How do incremental models and history work on Fabric?
Incremental models compile to MERGE statements so each run touches only the new and changed rows rather than rebuilding the whole table. Slowly changing dimensions use dbt snapshots, which record history as it changes. Both compile to native Fabric Warehouse SQL.
Does dbt fit the lakehouse and medallion pattern?
Yes. dbt models map cleanly onto bronze, silver and gold layers: staging models clean and standardise the raw bronze data, and marts build the governed silver and gold tables the business reads. The layering lives in your dbt project and runs inside one OneLake estate.
Is there a licence cost to run dbt on Fabric?
No. dbt Core is open-source and free to run, and so is the dbt-fabric adapter. You pay for the Fabric capacity you already run; dbt adds no licence on top of it.
Want this on your Fabric estate?
Tell us how you build on Fabric today, and we will show you what a dbt project looks like on it.