Since we're on the subject, I'm growing a bit sceptical of the AB effect.
AFAIK, in the actual experiment we have an "infinite" string of electrons whose spins are aligned in one direction, so their magnetic fields cancel each other. We replace these electrons with a classical magnetic potential, and it looks like the passing electron can feel this potential, even though the (classical) field strength vanishes.
At what point can stacking N electrons on top of each other be described by a classical potential? It's not a phase transition. For any finite N, the system is just a bunch of electrons interacting by bouncing photons back and forth.