Investing in opensource systems like Coq, Agda, etc, seems more promising. They both work on improving their experience based on the latest achievements. Mathematica is very expensive and proprietary for the most of the world.
There's a difference in the kind of things that proof assistants like Coq can do and what computer algebra systems like Sage can do. One can still use CAS like Sage to prove things and not just do numerical computations, but it's a bit of a different approach