Why NVIDIA does not allow to build samples on PX2 itself?

When I tried to use CMake to compile these samples, I got the error info that showed such kind of operation pattern are not supported. Then I checked the related CMake file and found out that if the architecture aarch64 ( PX2 itself ) was detected then the build process will be terminated. So why it is forbidden intentionally?

Dear @yuanyuanyuan,
A team may have few DRIVE PX2 and have couple of developers who wants to develop AV applications independently. We expect the developers to use cross compilation so that developers can do develop/test on host and deploy on target later.

But there is no reason keeping those who has a lot of PX2s from native building? I do not think this make sense from my view point.