Default semantic model on DirectLake

[jaryszek]

Hello,

I am trying to use a service principal on a remote semantic model on DirectLake. 

But there is a specific case:

https://learn.microsoft.com/en-us/fabric/fundamentals/direct-lake-overview

Power BI Direct Lake semantic models can be created and queried by Service Principals and Viewer role membership with Service Principals is supported, but the default Direct Lake semantic models on lakehouse/warehouse don't support this scenario.

Can anybody explain what is a diff between service principal semantic model on DirectLake vs default DirectLake semantic model?

Best,
Jacek

[v-pnaroju-msft]

Hi jaryszek,

Thank you for the update.

In addition to the points already mentioned by @ cengizhanarslan, please consider the following, which may help resolve the issue:

1. The default DirectLake semantic model inherits the Lakehouse or Warehouse security model and does not have its own independent permissions. It is not treated as a standalone Power BI dataset and behaves more like an embedded or system managed model.
2. Custom semantic models support enterprise governance features such as deployment pipelines and broader governance scenarios. Custom semantic models are user managed, whereas the default semantic model is system managed and optimised for simplicity.

Microsoft has announced sunsetting or strategic shift away from default semantic models in favour of explicitly created semantic models.

Additionally, please refer to the following links:
Power BI Semantic Models - Microsoft Fabric | Microsoft Learn
Sunsetting Default Semantic Models – Microsoft Fabric | Microsoft Fabric Blog | Microsoft Fabric
Decoupling Default Semantic Models for Existing Items in Microsoft Fabric | Microsoft Fabric Blog | ...

We hope this information helps to resolve the issue. Should you have any further queries, please feel free to contact the Microsoft Fabric community.

Thank you.

[v-pnaroju-msft]

Hi jaryszek,

We are following up to see if what we shared solved your issue. If you need more support, please reach out to the Microsoft Fabric community.

Thank you.

[jaryszek]

Hello, 

no. Problem is not solved. Discussion is still there. 

[v-pnaroju-msft]

Thankyou, @cengizhanarslan for your response.

Hi jaryszek,

We appreciate your inquiry submitted via the Microsoft Fabric Community Forum.

Thank you for confirming. Your understanding is correct, and the behaviour you are observing reflects the current design of DirectLake semantic model ownership and security.

We would like to inquire whether the solution provided by @cengizhanarslan resolved the issue. We hope the information supplied resolves your query. Should you have any further questions, please feel free to contact the Microsoft Fabric community.

Thank you.

[cengizhanarslan]

In Fabric there are two different kinds of DirectLake semantic models, and that’s what the doc is referring to.

When you create a semantic model explicitly (for example: New semantic model → choose tables → save it as its own object), that model is a first-class semantic model. It has its own permissions, supports service principals, Viewer role assignment, XMLA, REST APIs, etc. Service principals can query it just like users.

The default DirectLake semantic model is different. This is the auto-generated model that Fabric creates for every Lakehouse or Warehouse so you can immediately build reports. It’s meant for quick exploration and interactive user access only. It does not fully participate in the security and API surface, which is why service principals are not supported for it.

So the difference is not DirectLake itself, but how the model is created and owned:

• Default model = system-generated, lightweight, user-only

• Explicit semantic model = fully managed, securable, SP-supported

If you need to use a service principal, the fix is simple:
Do not use the default semantic model. Create a new semantic model explicitly on top of the Lakehouse/Warehouse tables and grant the service principal access to that model.

[jaryszek]

Thanks. 

I created semantic model like that: 
New semantic model → choose tables → save it as its own object

And i can use SP on it. without any issues. 

"This is the auto-generated model that Fabric creates for every Lakehouse or Warehouse so you can immediately build reports." 
about default. You still need to create it using power bi desktop or fabric. You need to create on "Create Semantic Model" button...

Best,
Jacek