If a company makes a new MCU with some exciting new instruction set, they need to make a compiler available which supports that instruction set and make that compiler available to their customers.
With LLVM as the base, the vendor could make their toolchain proprietary, making it impossible to integrate it back into LLVM, which means the vendor toolchain will exist until the ISA gets wide-spread enough for volunteers to invest the time required to make a separate production-quality LLVM back-end from scratch.
With GCC as the base, the vendor must at least make their GCC fork available to their customers under the GPL. This, in theory, allows the GCC community to "just" integrate the back-end developed by the vendor into GCC rather than starting from scratch.
Now I don't know how effective this is, or how much it happens in practice that the GCC project integrates back-ends from vendor toolchains. But in principle, it's great that vendors have to make their toolchains FOSS because it reduces the need for vendor toolchains.
Code getting open source is not impossible. Companies do it all of the time because it's expensive to rebase.
Previous reality: companies write fully proprietary code to avoid GCC
Current reality: companies choose Clang over GCC because of the license and then contribute many of their changes back.
Because Apple certainly isn't alone.
The one vendor who forks LLVM and doesn't contribute their biggest patches back is Apple, and if you want bleeding edge or compliance you're not using Apple Clang at all.
If you say "isn't it great vendor toolchains have to contribute back to upstream?" I'm going to say "no, it sucks that vendor toolchains have to exist"