Why this lesson matters
You already build reports, publish semantic models and manage workspaces in the Power BI service. When your organisation says "we're moving to Fabric", it is easy to assume you need a new tool and must start over. You don't. Microsoft Fabric is the platform that Power BI now sits inside. The portal you sign in to every day is the Fabric portal. Microsoft's documentation says so directly: Power BI is part of Fabric, and app.powerbi.com and app.fabric.microsoft.com lead to the same service.
So what changes is not the place you work. It is what you can build there. Up to now, your data arrived from somewhere upstream: a SQL Server, a SharePoint folder, an IT-owned warehouse. With Fabric, the tools that build that upstream layer sit in the same workspace as your reports.
Facts and feature lists in this lesson come from Microsoft Learn, read on 20 October 2026. Fabric changes every month, so check the docs again before you rely on anything marked preview.
The analogy: Power BI moved into a bigger house
Think of Power BI as a well-run flat. It has your reports, semantic models, dashboards and apps. Fabric is the building that flat now sits in. The other floors hold data engineering, data warehousing, data science and real-time analytics. The whole building shares one water supply, and that supply is OneLake.
OneLake is a single logical data lake for your whole tenant, built on Azure Data Lake Storage Gen2. You don't provision it, you don't need an Azure subscription, and every Fabric workload reads and writes to it. Tables are stored in the open Delta Lake format (Parquet files plus a transaction log). This is the key idea: a table that a data engineer writes in one tool can be read by your semantic model without being copied again.
The workloads, in plain terms
As of October 2026, Microsoft's overview lists these workloads:
Workload What it does Who usually drives it
------------------------ --------------------------------------------- ---------------------
Power BI Reports, semantic models, dashboards, apps You
Data Factory Dataflows Gen2 (Power Query) + pipelines You, increasingly
Data Engineering Lakehouses, Spark notebooks, Spark jobs Data engineers
Data Warehouse T-SQL warehouse, stored as Delta in OneLake SQL developers
Data Science ML models and experiments Data scientists
Real-Time Intelligence Streaming data, Eventhouse, KQL queries Ops / IoT teams
Databases SQL database in Fabric, mirroring App developers
Industry Solutions Industry-specific data solutions Varies
Fabric IQ (preview) Shared business semantics, ontology, agents Still emerging
Gotcha: Fabric IQ is labelled preview. Preview features can change or disappear, so don't build production work on them without a fallback.
Two of these will feel familiar quickly. Dataflows Gen2 are Power Query, the same M language and the same editor, except the output can go to a lakehouse or warehouse table rather than living inside one model. Power BI itself is unchanged.
Old items versus new items
In a Fabric-enabled workspace, the New item menu offers far more than before. Here is how to sort what you see:
FAMILIAR POWER BI ITEMS NEW FABRIC ITEMS
- Report - Lakehouse (files + Delta tables, Spark and SQL)
- Semantic model - Warehouse (T-SQL tables, full read/write SQL)
- Dashboard - Notebook (Python / Spark SQL code)
- Dataflow (Gen1) - Pipeline (orchestration and scheduling)
- Paginated report - Dataflow Gen2 (Power Query that writes to OneLake)
- Eventhouse (real-time / KQL data)
One detail is worth knowing from day one. When you create a lakehouse, Fabric automatically creates a SQL analytics endpoint for it. It is a read-only T-SQL view over the lakehouse's Delta tables. Only Delta tables appear there, so CSV files in the lakehouse's Files area don't. Also, since 5 September 2025, Fabric no longer creates a default semantic model automatically when you create a lakehouse. Older tutorials that say "open the default semantic model" are out of date. You will build your own semantic model.
Trade-off: These new items are not free extras. They run on a Fabric capacity, and Direct Lake semantic models require one too. Import and DirectQuery models still work with any Power BI licence. Level 2 covers capacities, SKUs and the trial.
What actually changes for you
- Where you shape data. Heavy transformation moves upstream into lakehouse or warehouse tables that many models can reuse. You stop copying the same Power Query logic into five .pbix files.
- A new storage mode. Direct Lake reads Delta tables in OneLake straight into the same VertiPaq engine that Import uses. It doesn't copy all the data at refresh time. A Direct Lake refresh (called framing) updates only metadata and usually takes seconds.
- Import and DirectQuery still matter. Microsoft's own guidance says Import often suits a self-service analyst who needs to move quickly without waiting on IT. Fabric gives you more options. It doesn't make the old modes wrong.
The course project: sales files to a Direct Lake report
Every lesson builds one piece of a single solution in a Fabric trial workspace. Here is the plan:
[1] Raw files sales_2026.csv, products.csv, stores.csv
| (land in the lakehouse Files area)
v
[2] Ingest + shape Dataflow Gen2 (Power Query you already know)
|
v
[3] Delta tables lakehouse Tables: fact_sales, dim_product, dim_store
| (readable via the SQL analytics endpoint)
v
[4] Orchestrate pipeline runs the dataflow on a schedule
|
v
[5] Semantic model Direct Lake on the Delta tables
|
v
[6] Report the Power BI report you already know how to build
A sample of the raw input we'll use:
OrderID,OrderDate,StoreID,ProductID,Quantity,UnitPrice
100001,2026-01-03,S01,P104,2,19.99
100002,2026-01-03,S02,P221,1,149.00
100003,2026-01-04,S01,P104,5,19.99
Notice that only steps 5 and 6 are classic Power BI. Steps 1 to 4 are the new territory, and the rest of the course teaches them one at a time.
Check: confirm your starting point
Before Level 2, check what you can see today:
1. Browse to https://app.fabric.microsoft.com and sign in.
2. Open any workspace you own and select + New item.
3. Look for Lakehouse, Warehouse, Notebook and Pipeline.
- Visible -> Fabric items are available to you.
- Only Power BI -> no Fabric capacity or trial yet, or the admin
setting that lets users create Fabric items is off.
Level 2 sorts this out.
4. Write one sentence per box in the diagram above saying
which Fabric item it uses. Keep it: we'll check against it later.
You pass if you can put every item in the diagram in the right workload (for example, notebook = Data Engineering, dataflow = Data Factory, report = Power BI).
Key takeaways
- Fabric is a SaaS platform that contains Power BI. Your portal, reports and semantic models carry on as before.
- Its main workloads are Power BI, Data Factory, Data Engineering, Data Warehouse, Data Science, Real-Time Intelligence and Databases, plus Industry Solutions and Fabric IQ (preview).
- OneLake is one lake per tenant, built on ADLS Gen2. Tables are stored in Delta format, so different tools can share them without copies.
- New items such as the lakehouse, warehouse, notebook and pipeline need a Fabric capacity. A lakehouse gets a read-only SQL analytics endpoint automatically.
- Direct Lake runs on VertiPaq like Import, but refreshes only metadata. Import and DirectQuery are still valid choices.
- The course project goes: sales files, Dataflow Gen2, Delta tables, pipeline, Direct Lake model, report.