This is the main reason I'm excited about this. I really hope they continue the very open approach they've used so far, but even if they start going binary blob for some of it like nvidia and (now to a lesser extent) amd have at least they're likely to properly implement KMS and other things because that's what they've been doing already.