In the example,
create table a
(
col1 int,
col2 int,
col3 as col1,
)
col3 is technically a computed column, but uses only an extremely low amount of cpu time, especially since it does only a simple in-memory copy operation, but since it's not read from a disk buffer, is probably actually more efficient than a normal column. The statement 'definitely uses more resources' may be a stretch. I would agree that an expression utilizing a SELECT or UDF would use more resources and could potentially cause performance issues.