Why Fabric sunsetting default semantic models is a win
Microsoft has announced the sunsetting of default semantic models in Fabric — a move many data teams have wanted for a long time. It acknowledges what most of us saw in practice: the automatic models cluttered workspaces and added little real value.
Why it matters
- Workspaces were overcrowded. Fabric created a default semantic model every time you spun up a lakehouse or warehouse, and most of them went unused and confused people.
- Modeling was limited. You couldn’t fully customize relationships, date tables or calculations, so most teams ended up building custom models anyway.
What teams get
This shift shows Microsoft is listening. Teams can now:
- Build clean, lean semantic models designed around business logic.
- Avoid workspace clutter and simplify ALM.
- Spend their time on high-value work instead of boilerplate.
My perspective
I’m glad Microsoft took its users’ feedback into account. I’ve watched default models drag down workspace quality and confuse report builders, and this is a big step.
Update, October 2026: the second phase has happened. By November 30, 2025, Fabric had decoupled every remaining default semantic model into a standalone model. Our step-by-step cleanup guide covers what to do with them.
What to do next
- Audit your workspaces. Retire or archive the default models.
- Teach your team to build new semantic models with Direct Lake or Import.
- Work through the cleanup guide for the models that are left.
Goodbye clutter. Hello clarity.