At its inception the appeal was the "multi-stage optimisation system" and the virtual instruction set design (from https://llvm.org/pubs/2002-12-LattnerMSThesis.html)
I seem to remember that "machine-independent binaries" was Apple's first use of LLVM: distributing LLVM IR and having it be converted to machine code on the user's computer, back when they were supporting Motorola and Intel chips. And I think consequently that's how LLVM got a lot of its momentum.