There's not really much opportunity for it to get stressed as such. It's only maybe 4mm at most of the end of the rod that the epoxy is on. If the rod gets twanged, it tends to flex most near where the figure is gripping it. The tiny length at the end doesn't really flex at all. And as far as the tip itself getting knocked, it's not really any worse than the same putty would be if applied to on the surface of the figure. In terms of proportional volume vs. bond surface it's more like, say, a coat lapel than an extended arm.
But this also brings up some of the additional benefits of carbon fiber: it doesn't flex nearly as much or as readily as metal*, and the surface has a natural grain (the rod is actually a compressed bundle of hair-thin fibers), so adhesion is better than it would be with metal. Especially on a sanded area, where you're dealing with a slope of roughly exposed fiber ends.
*Though this can work against you if the figure is holding a long rod in one hand with the arm extended away from the body: the rod becomes a lever if it gets pranged. With brass, the "spring" of the metal can act as a shock absorber, whereas with carbon fiber most of that force just gets transmitted directly to the figure, so a figure with such a pose can be at an increased risk of a bent wrist/arm or broken glue joint.