you could try two things: 1. Use a pipeline with the Copy Activity and Connection REST (GET). Source is the URL for every file (might need some for pipeline activities to iterate the urls). Destination is the Folder in your Lakehouse. Then you can access it via notebooks. 2. Use a python notebook (no heavy spark needed) and use the temporary file system of the notebook session. Most standard python packages with filesystem functions won't support abfs paths directly. So you can use notebookutils.fs.cp to copy files out of the sessions filesystem to a Lakehouse files folder: # store to session filesystem with normal pandas tmp_path = "/tmp/myfile.xlsx" sales_df.to_excel(tmp_path) # copy from session to lakehouse workspace_id = notebookutils.runtime.context["currentWorkspaceId"] lh = notebookutils.lakehouse.get("YourLakehouse", workspaceId=workspace_id) lh_folder = f"{lh.properties['abfsPath']}/Files/My/Folder" notebookutils.fs.mkdirs(lh_folder) lh_path = f"{lh_folder}/myfile.xlsx" notebookutils.fs.cp(f"file:{tmp_path}", lh_path) While this might not be the best approach for data that could easily be saved as delta table, as described by Anthony above, its helpful if you need to use python packages that output other data that needs to be a file but the package filesystem methods don't support abfs. In my case it was with geopandas and topojson and I needed the plain json output. if you want to write to a table (delta), using a lightweight python notebook is usually a better way than spinning up the heavy spark cannon: %pip install pandas deltalake pyarrow import pandas as pd from deltalake.writer import write_deltalake your_df = pd.from_excel("...") write_deltalake(f"{lh.properties['abfsPath']}/Tables/my_table", your_df, mode="overwrite")