There are two models of the InTurn™. The new ULTRA has a one piece Forged Chrome Moly Steel (4140) spindle with a D1-6/A6 mount.
The MEGA shown in this thread has a two piece spindle with a D1-4/A4 spindle.
D1 and A are standards with D1 being CamLoc and A being bolt on.
On the MEGA, spindle is two piece with the flange being a heavy press fit, heated to 550F and pressed on with a 20 ton press at that temp. It is exceedingly unlikely that you could dislodge such a press fit with the several times the 2HP available for the MEGA.
In addition to the press fit, the hardened steel cams for the CamLoc system extend thru the flange and into the spindle itself so the absolute maximum the flange could ever move is the .001" - .002" clearance in the Cam holes in the flange. The very first prototype MEGA was abused by my own hand for a few years before being sold when replaced by the new ULTRA. A few crashes bent the chuck's backing plate, but the spindle nor D1 flange was ever effected. A bent backing plate is easily skimmed to re-true it. Eventually, I made a thick cast iron backing plate which never bent even in severe crashes.
Early on, I spoke with the engineers at SKF about preloading the ball bearings. Deep Groove balls (as opposed to AC) are not specifically designed to be preloaed, but they have a lateral load spec and as long as that is not exceeded, there is no problem with preloading. Because the spindle is hollow, it has a very large OD which results in a huge bearing that will never see the loads it is capable of in this application, so there is plenty of excess load capability, even for drilling 1" dia hole into steel.
The preload is light and intended only to remove lateral play inherent in the bearings. The amount of preload is far less than what the bearings experience in operations like drilling into the center of a workpiece.
The preload is retained by the steel locking collar on the back of the spindle. The force exerted by backing out a stuck drill bit, heavy cuts on the back side of a workpiece, or using the wrong parameters for TPI on a large tap will exert significantly more force on the retaining collar that just holding the light preload.
The maximum predictable force on the locking collar is easily contained by the friction of the collar against the spindle, which is provided by a fine thread clamping screw. For the forces presented by the MEGA, a threaded connection is not needed, nor was there ever a treaded connection on this piece.
Perhaps you are confusing the MEGA with the new ULTRA. The ULTRA build videos clearly show the threaded retaining collar, but that is required by the optional tapered roller bearings, which are adjusted by clearance and not force applied. The 'micrometer' type of accuracy provided by the 20TPI locking collar is needed to provide this fine adjustment. The collar is still clamped in place using a large fine thread bolt.
In both machines, the locking collar doubles at the main pulley mount so there is also torque applied to the joint. Both applications work out fine on paper with 3x the horsepower ratings of the respective machines. There are lots of MEGAs out there and to my knowledge, only one has dislocated the locking collar. Most likely I did not set the collar tight enough from the get-go, but the user was able to re-establish the pre load and lock the collar down.
There are only a few of the new ULTRAs and only one had the tapered roller option, so it is a bit early to declare victory on that scheme.