The point that ML requires lots of compute on few bytes makes senses as ML performs mostly repetitive matrix operations, but doesn't a series of SQL queries on

The point that ML requires lots of compute on few bytes makes senses as ML performs mostly repetitive matrix operations, but doesn't a series of SQL queries on database data with a "star schema" also require lots of compute on the same bytes?