To charitably read and steel-man the GP's comment, one could argue that while undergraduate education is fine - there are better patterns of research - such as industry (bell labs, spaceX, google ml research), or government (oak ridge, NASA center). That the mixed mission between doing research and teaching at the same time in a T1 research university doesn't work as well as more formally separating those functions.

I'm pretty sure I don't agree with the above ^, it probably discounts the value of grad-school research as an apprenticeship model amongst other drawbacks. But it does have some merit to bear consideration. Surely the current model isn't perfect - where could we improve?