My thought is that the weight of the flywheel matters. To be able to absorb / release a tremendous amount of power, I imagine this flywheel would need to be very heavy. If so, wouldn't it non-linearly differ in efficiency from the near vacuum & magnetic bearing designs we've built before?
Perhaps the proposal is to build a large array of smaller flywheels that don't have to deal with the problem I imagine?
I'm not familiar with any part of engineering here -- just a curious soul -- sincerely asking!