QM nonlocality isn't useful for instant communication though. If it is a physical process, you still cant influence it on one end in order to communicate faste

QM nonlocality isn't useful for instant communication though. If it is a physical process, you still cant influence it on one end in order to communicate faster than the speed of light. And nothing is "observed instantly" since that phrase doesn't have meaning in SR. Two space-like separated observations, which cannot communicate with each other in SR, will have correlated measurements. Which measurement comes first or if they happen at the same time, will depend on your reference frame and motion. So once a single measurement has been made somewhere in the universe, all observers (light-like or space-like separated) will measure entangled values that agree with that measurement, but the entangled state was created indeterminately. If the observations are space-like separated you cannot say which one "caused" the collapse of the wave function. And like I said you can't use it for communication. The fact that you can't assign which observation was cause and which one was effect is probably tightly tied to the fact that you can't use it to communicate -- which side is the sender and which side is the receiver? That depends on the reference frame, which produces nonsense, so to avoid a paradox it is banned.