So HPD's post does suggest good old bearing issues.
I assume that the turbine wheel has its own bearing(s) so maybe these would have failed first if the problem was to do with exhaust side temperatures.
Keep in mind that EGT is north of 1000C while the magnets Curie temps are under 800C and they most likely operate a lot lower than that.
The fact that the relatively cooler MGUH bearing failed could mean that the friction in the bearing is too high , leading to high heat generation and race distortion.
This could be due to the preload being too high and/or insufficient oil flow. Normally preloads are kept high enough to maintain bearing stiffness and load capacity at high speeds, however the downside is an increase in friction torque (and therefore heat). This is typically countered by increasing the oil flow which further increases the power losses.
As a result there is a very fine balance between mechanical failures (hertzian stress), thermal failures (race distortion) and bearing frictional losses.
Maybe Honda were too aggressive in reducing bearing losses and neglected reliability. That or they are failing to determine the mechanical loads on the bearings - which is often a challenge with high speed rotating machinery.