That's not the point he's trying to make. Smaller cluster sizes leads to larger amounts of file-system metadata keeping track of where those clusters are laid out on disk, as well as the overhead of generating, accessing, and updating those data structures.
NTFS uses run lists (comparable to extends in ext4) so you would only have more metadata if your drive gets fragmented.
(Which I hope for the sake of comparability is not the case here)
If files are typically bigger than 16KB, 16KB clusters can potentially save space vs 4KB clusters, by needing smaller cluster indices. Not sure if that's the case for NTFS though.
