In academic research mathematics, LaTeX is totally dominant. That's why I included various bits of support for LaTeX editing all over the place in https://cloud.sagemath.com, and why in SageMath (http://sagemath.org) we have a LaTeX function that takes almost any Python object and gives it a LaTeX representation. And also LaTeX formula support is integrated into http://mathoverflow.net/ and even Wikipedia (for the math-related articles).

Current CS undergrad: LaTeX is mandatory or strongly encouraged (slight grade penalty for not using it) in pretty much every undergrad math class post-Calculus, as well as CS classes like Discrete Math, Algorithms, Complexity Theory, etc.

It also turns out great-looking papers for professors who are okay with PDF and count words rather than pages (which is IMO the way to do it.) 100% of my academic output is plain text under version control; some of it is code, some of it is LaTeX.

It may not be used much outside of academia but everyone in anything remotely quantitative at my school (Math, Economics, Statistics, Public Policy and other stats-dependent fields, etc) will learn it at some point.

I'd humbly suggest that one should consider the Venn diagram of writers in general. Out of the entire population who is writing with regard to careful formatting? The large subset, i'd assert, would be academics. And to further prove the unreliable nature of impressions, I see in university the least use in engineering, (where MSWord seems to dominate). Also Europe seems to use LaTeX more than in the states, (again MSWord?); but perhaps that's because of the subtle demands of diacritical marks and the like.