The issue isn't usually competency, can vs. cannot. It is usually a matter of should or should not. I have seen databases thrown at problems where they decidedly should not have, and have seen the product development cycle suffer where it should not have. While some engineers and analysts relish CTE spaghetti -- solving complex logic utilizing enormous wads of SQL, others are more wary and rightly look to tackle complexity with more accountable and powerful tools when possible. I admit that tools like DBT can help address some of the shortcomings of SQL, but it still does not have the generality and conciseness of modern languages.