Does importing a PBI Dataflow into a semantic model duplicate the source data?

[LenFi]

Let's take an example where you use import mode to get source data into a dataflow. You then combine multiple dataflows where your output is your dataflow entity. Once you import your dataflow entity into a semantic model(dataset), does that mean that your source data is duplicated? 

This is the response I got from Chat GPT, but I cannot find any documentation regarding this: 
"Imported Semantic Model: When you import data from a dataflow into a Power BI dataset, you're essentially creating a semantic model based on the dataflow's entities. However, this process doesn't duplicate the underlying data. Instead, it establishes metadata references to the data stored in the dataflow. The imported semantic model provides a structured representation of the data for analysis and visualization purposes.

Therefore, while you may have multiple artifacts (dataflows, datasets) referencing the same underlying data, there's no duplication of the actual data. This architecture promotes data consistency, reusability, and efficient management of data assets within the Power BI ecosystem."

Has anyone looked into this and can perhaps point to more information? Thank you!!

[lbendlin · Accepted solution]

Once you import your dataflow entity into a semantic model(dataset), does that mean that your source data is duplicated? 

yes, yes it does.   Only use dataflows if they provide value. Avoid using them if the Semantic Model can get the same data from the original source reliably and with good performance.  Or consider using Direct Query where appropriate.