What you are doing will determine an appropriate solution.
I can only assume that raster to point is for some reason, since I would not normally want a bunch of closely spaced points holding the data.
I would use RasterToNumPyArray if the spatial location of all the data were coincident. You can make a multidimensional array, The location (x, y) values of the cells can be determined and the multiple layers hold information from different layers occupying the same space.
If the rasters are adjacent I probably would only use points if the Z values were floats/doubles and I had to retain every last decimal place, otherwise I would convert to polygon using an appropriate reclassification.
Your end use isn't specified, so no specific recommendations can be made
... sort of retired...