#84 spindle has bearing id of 1.376 and a shoulder of 1.72. That's a difference of 0.344 in diameter. The shoulder is 0.172 bigger than the bearing in radius.
Here's the bearing, per Timken:
The critical measurement is R, which is 0.14 max. So an L68149 on a #84 has an overlap of 0.032 (min), and a contact surface of ~.14 sq in (min). It has a thrust rating 1890 lbs, which would result in 13500 psi (aka 13.5 ksi).
A cheap and common mild steel alloy, cold drawn 1018, has a yield strength of 54 ksi. But the spindles are likely better steel. I've found references to 1040 and 1045 being used for spindles (71 and 77 ksi).
If the spindle is deforming, it's because the bearings were installed too loose and it's getting hammered.
These spindles have been around forever and are very common. If the design were a problem, it would be a widespread issue. It's not.