Loading data into Redshift with DBT

Why we load the S3 data into the Redshift again? it already queryable via Redshift Spectrum? I guess it's for the performance? Transform the S3 raw...