I'ld say the resulting (far) field of the fancy coil arrangement is in the vertical direction and the magnetic ball is spinning around a horizontal axis. Since the fan control PCB will just supply two uni-polar phases, the inner (more or less conventional) coil and the outer one are wired so the two outputs of the fan controller will produce magnetic fields in opposite direction. Since the inductivity of the air-core coils is probably quite low, it's well possible to reach 500k RPM with a tiny ball magnet, suspended on an air cushion. A single, primitive air coil (or just an inductor of an old speaker cross-over), controlled from a totem pole or bridge amplifier would have done the job as well.
I've reached 180kRPM without any difficulty with a "real" rotor, suspended on two hybrid ball bearings, driven by a special three-phase coil arrangement. I could easily have gone higher but since this rotor was part of a 5kW turbo generator prototype that was designed to spin at 160kRPM, and I didn't want to destroy it due to the centrifugal forces, I behaved myself and limited the speed there.
Neverthesess, spinning magnets at that high speed can be quite dangerous. They are made of brittle, sintered material and the application of a nickel plating (for increased corrosion resistance) makes them appear much more rigid than they actually are. Assuming the magnet in the video is 6mm in dameter, 500kRPM means it's got a circumferential verlocity of almost 160m/s (this is almost half the speed of sound). If the magnet fractures at such speeds, its shrapnels can cause significant damage or injury. It's a lot of fun playing with such setups, but especially in cases like this where the rotational axis may be unknown / undefined, it is difficult to take protective measures (shields) and stay out of the plane of rotation.