Well, of course all architectures must be cross compiled on x86-64!
Why the heck are they even considering using native compilation, which is just absurd, given that it gives no advantage, and that of course all non-x86-64 CPUs just suck horribly?
And what does many core have to do with it?
Compilation is mostly embarassingly parallel, so there's no issue with SMP or clusters.
Just run distcc on an x86-64 cluster and cross-build everything...
BTW, in case they are worried about it, if a build system tries to execute a program it builds (= it's broken), just have qemu-arm set up to run it automatically.
If really needed, they can even just run a whole ARM distribution via qemu-arm on x86-64, and simply substitute gcc, as and ld with native x86-64 binaries, since they are usually the only performance critical programs.
Sometimes one wonders how things work at all, when the people in charge are so clueless.