You cannot retain "it has many extensions to reach different goals" with something new. By definition something new that is not TeX/LaTeX will start with no extension.
Having been a core contributor to the GNU TeXmacs project, which aims to be something like "TeX except for the 21st century", I think have some good insight on the succession of LaTeX.
I am not aware of anything on the horizon that is any likely of dethroning LaTeX. The people of influence in the mathematical community are just too used to using LaTeX to seriously consider the effort required to learn any tool that could be equally flexible.
Maybe in 20 years, when all the folks that got their PhD when LaTeX was cool are retired will something else replace it. Hopefully, it will be something like TeXmacs, but without the unnecessary complexity.
