Well the first and biggest problem is that phd students are INSANELY cheap teaching labor. Math PhD programs run at a loss in absolute terms, but almost never compared to the alternative of having to pay market rate for summer instructors and TA/grading labor.
More to your point though: who in their right mind wants tenure after doing a useful PhD in math?
Math phds can be extremely remunerative if you focus on studying useful topics, avoid abstract nonsense, pick up domain knowledge, and stay the heck away from low paying teaching jobs like “professor” and “instructor”. Finance, biotech, and tech all pay good math phds comparable to or better than big tech pays mediocre CS phds. Mid six to low seven.
It’s a surprisingly low bar, since most math phds are incredibly romantic about choice of research area.
> Math phds can be extremely remunerative if you focus on studying useful topics, avoid abstract nonsense, pick up domain knowledge, and stay the heck away from low paying teaching jobs like “professor” and “instructor”.