Microsoft Fabric Foundations Update: What Report Builders Should Change Now

The Problem We Are Solving

If you followed the Microsoft Fabric Foundations session recorded in 2024 and the workspace now looks different, it is hard to tell whether your report is out of date or Microsoft has simply changed the route around it.

The awkward part is that “Fabric has changed” can mean two very different things. One change may affect a semantic model your report already uses. Another may only matter to the data engineering team building lakehouses, notebooks and pipelines behind the scenes.

A semantic model is the layer that gives report data its business meaning: tables, relationships, measures and names that Power BI reports use. Microsoft previously created a default semantic model automatically for some Fabric items. That behaviour has changed.

The useful question is not “How do I rebuild the whole Fabric platform?” It is:

Does this change affect a report I own, and what is the smallest safe action I should take?

The Short Answer

Most report builders do not need to rebuild their reporting workflow because Fabric added Task Flows, changed its default semantic-model behaviour or expanded Direct Lake.

Start with the report you already maintain:

• If it uses Excel, SharePoint, SQL or another source through a normal Power BI Import model, the retirement of Fabric default semantic models does not by itself require a rebuild.
• If it uses a semantic model that was created automatically from a Fabric lakehouse, warehouse or mirrored database, check which reports depend on that model before anyone renames or deletes it.
• If you are creating a new model from one of those Fabric items, expect to create the semantic model deliberately rather than receiving a default one automatically.
• If someone is proposing Direct Lake, treat that as an architecture decision that needs a Fabric capacity, suitable data in OneLake and a small proof of concept. It is not the new default for every self-service report.
• If you see a Task Flow above the workspace item list, treat it as a visual organiser. It does not force you to rebuild the items underneath it.

What Changed, and What Should You Do?

Run This Five-Minute Report Check

Choose one important report. Do not start with the whole estate.

1. Name the report and its owner. If nobody owns it, record that before changing anything.
2. Identify the semantic model. Power BI previously called this a dataset. Note the model name and workspace.
3. Check whether the model comes from a Fabric lakehouse, warehouse or mirrored database. If it does not, the default-model change is probably not your immediate issue.
4. Look for dependent reports. Use the semantic model details and related-content information available in Fabric or ask the workspace admin to confirm. Do this before deleting an apparently unused model.
5. Record the storage mode. If it is Import, there is no general requirement to move it to Direct Lake. If it is Direct Lake, confirm who owns the upstream data preparation and capacity.
6. Confirm capacity and licences. Ask the admin for the actual workspace capacity SKU and the licences held by authors and viewers.
7. Write one decision: keep as is, tidy ownership, create a new independent semantic model, or run a Direct Lake prototype.

If you cannot complete steps 2, 3 or 6, that is not a failure. It means the next move belongs with the workspace or Fabric administrator.

The Licensing Check That Prevents Surprises

Licensing depends on both the person and the capacity behind the workspace.

Do not use this table as a purchase quote. Capacity assignments, workspace roles and tenant settings can change what a particular user can do. Ask the administrator to confirm the live setup before committing to a design.

A Message You Can Send to Your Admin

Recommended Move

Audit one real report before planning a Fabric rebuild.

If the report does not depend on a Fabric-created semantic model, stop there unless you have another reason to migrate it.

If it does depend on one, keep the model in place while you confirm ownership and dependent reports. For a new Fabric item, create a semantic model deliberately. For a proposed Direct Lake migration, run a small prototype against the current report requirements, capacity and support model.

That gives you a defensible answer without turning a reporting check into a data-engineering project.

Evidence Notes

Use Microsoft Learn to trust the product behaviour. Do not use it as proof that a particular tenant, report or migration is safe.

• Microsoft documents that default semantic models stopped being created automatically on 5 September 2025 and that existing default models were disconnected from their Fabric item by 30 November 2025. This proves the platform change. It does not prove that an existing model is unused or safe to delete.
• Microsoft describes Task Flows as a visual representation used to organise and navigate workspace work. A Task Flow can be hidden. It does not show the flow of data and does not require the underlying items to be rebuilt.
• Microsoft describes Direct Lake mainly for IT-driven, lake-centric analytics with large volumes of data in OneLake. The same documentation says Import mode can still suit a self-service analyst and recommends a prototype before choosing Direct Lake.
• Microsoft licensing documentation supports the F64 viewing threshold, the Pro authoring requirement and the boundary that Premium Per User does not provide Fabric capacity. It does not confirm the assignments, roles or tenant settings in your organisation.

This briefing supports an initial reporting decision. It does not validate report performance, data quality, security, row-level security, capacity sizing or migration effort in your tenant.