So I would argue that most businesses use tabular data that can be used to train classical models (XGB, etc), and reach the same performance (if not better) tha

So I would argue that most businesses use tabular data that can be used to train classical models (XGB, etc), and reach the same performance (if not better) than deep learning.