stationary. Hence the 'as long as it is moving' bit above. Because the motion allows for the coils to generate enough of a current to drive the compensation. So you need a support system to bring the assembly up to a certain minimum speed above which it will stably levitate.

Right. Everyone should just read up on Earnshaw's theorem to know what all the boundary conditions are.

https://en.wikipedia.org/wiki/Levitron

I wonder how long you could get one of those to spin in a vacuum.

Halbach arrays with compensating coils have been proposed for some interesting applications, such as low loss flywheels for electrical storage. I don't know if that ever got commercialized but I do recall that some prototypes were made by a US company. I can't find a reference to it though.

>I wonder how long you could get one of those to spin in a vacuum.

Not too much longer apparently...

https://www.youtube.com/watch?v=mn7IedCgva0