To some extent, it does own it. Ever tried to push even a mildly controversial patch to LLVM or Clang upstream? If Apple guys are against it, it's a full stop, no matter how many other parties are interested.

And their reasons to be against a particular patch are often as simple as that it'll break compatibility with their internal, unpublished changes to LLVM.

Learned it the hard way...

Chris Lattner is the "Apple guys" now, so..

What's the relevant definition of "owned" here? As long as Apple can modify LLVM however it wants, and as long as it's sure no corporate actor can restrict access rights to it in the future, it effectively owns it, altough not exclusively.

Having exclusive rights on a strategic business brick is important when this brick is enough to give someone else a stronghold on the whole vertical market. Although having a good compilers is one of the many requirements in Apple's business model, it's neither the defining feature, nor what competitors miss in order to threaten Apple.

So yes, in this context, Apple non-exclusively owns LLVM.

owns is the wrong word. No need to redefine it to fit the sentence, they should have chosen a different word.

Perhaps the word is 'pwn'?

Of course it can.

Open source doesn't mean the process is collaborative. It just means the code is available for you to use, modify and fork. There are numerous open source projects (mostly those sponsored by companies) for which you would never be able to change the direction of.

Look at Node/IO for how things would need to work.

I reckon the authors point is that they control how they use it. They might not own it all (but they sure hold the copyright to much of it), but no other entity outside of their control is going to pull the rug from under their feet.