cos αE = exp(-TR/T1)
where αE is Ernst flip angle. This angle maximises the MR signal, when a TR is used which prevents full recovery of Mz. You can't just keep using 90° excitation pulses to maximise signal because, although after your first excitation you'd have loads of signal in the x-y plane, you'd have hardly any in the z-direction ready for the next excitation. For a particular TR, there will be some middle ground where a sufficiently large amount of magnetisation is rotated into the x-y plane for measurement, but a significant amount remains in the z-direction for the next pulse. The Ernst angle is the "ideal" angle.
So why don't we always use the Ernst angle for a chosen tissue of interest? The answer is, we are looking for contrast between tissues, not just a maximum SNR from one tissue.