Large city LA has 1 million address, estimate to have 4000 x 100 = 400,000 commonplace or POI
Don't collect all 400k POI at once, break them down into small grid.
Example 1: Microsoft commonplace POI
This is national grid USNG with 100meter x 100 meter. Each grid has grid number which is USNG number.
This grid number : 11S LT 832 673
I collect 17 poi.
Example 2: Google commonplace POI
This grid number : 11S LT 850 675
I collect 28 poi.
You can choose different size of grid, for example small box is 100 meter, bigger box is 1000 meter, etc...
https://transparentgov.net/json2tree/esri/server/folder2.html?org=https%3A%2F%2Fpublic.gis.lacounty.gov%2Fpublic%2Frest%2Fservices&url=https%3A%2F%2Fpublic.gis.lacounty.gov%2Fpublic%2Frest%2Fservices&arcgis_online_token=&timeout=9000&select_folder=9&select_layer=3&select_folder_text=Basemap_and_Grids+%3Csup%3EMapServer%3C%2Fsup%3E&select_layer_text=3+%26%23x21E2%3B+USNG+1000M+%3Csup%3EFeature+Layer%3Csub%3E+esriGeometryPolygon%3C%2Fsub%3E%3C%2Fsup%3E
Use each grid number as file name of downloaded POI geojson file.
Open ArcPro, add all those grid-number.geojson into single layer, then you have all 400,000 point in a single POI layer .
This is bigger size grid 1000 x 1000 meter
I collect 287 commonplace in this bigger box