Consider that compiling Boost gets them up to compilers like gcc and MSVC for the current standard of C++. There's still the upcoming C++0x standard that both gcc and MSVC have made big progress in supporting.
Also, something like OpenMP is light on the actual compiler support (compared to, say, handling Boost or some of the upcoming C++0x features). Most of the work goes into the runtime system. The expertise needed to build a good runtime system for OpenMP is different than the expertise to build a good C++ compiler. There's significant overlap, no doubt, but I doubt this development will "free up" anyone.
Considering this in combination with the recent announcement of libc++ (http://libcxx.llvm.org/), and the fact that the chair of the library working group for C++0x is on the team, I'd venture to guess that support for C++0x is probably a pretty high priority. (At least, probably higher than OpenMP.)
Also, something like OpenMP is light on the actual compiler support (compared to, say, handling Boost or some of the upcoming C++0x features). Most of the work goes into the runtime system. The expertise needed to build a good runtime system for OpenMP is different than the expertise to build a good C++ compiler. There's significant overlap, no doubt, but I doubt this development will "free up" anyone.