HP really has nothing to do with the amount of strain placed on components during a shift point, the amount of TQ rise to accelerate the slow component is where the stress is. Since TQ tends to fall off with rpm you actually have less to deal with at higher rpms.
Inertia also plays into the equation, the faster the driven components are going the less strain there is to accelerate the mass faster. This works in conjunction with the the way the TC works as the faster you spin in it the closer the driven and drive components get and the more cohesive the fluid coupling is. There is simply have more cushion there at 3000 rpms than at 2000 rpms when a change in drive ratio occurs.
Add to that the trans apply pressures are almost double at 3000 rpms as opposed to 2000 rpms and the defuel is much sharper at high rpms. The ECU has a scaled calibration of how much to de-fuel based on speed and rpms, it knows nothing about TQ. It just knows that at higher speeds and rpms it needs to de-fuel more than it does at lower rpms.
Input shafts, output shafts,etc, don't break at high rpms and speeds they break in low speed high TQ conditions, pretty much what every day driving emulates with the stock shift algorithms. Pretty standard operation to not shift gears on top of the TQ peak, rather, shift at higher HP and allow the ratio change to facilitate the acceleration with the TQ.
In addition, your analogy of comparing speed shifting to high rpm shifting is faulted. The faster the shift happens the better it is for any transmission, it is the means of speeding the shift that causes the damage. Manuals don't break gears when you speed shift them, it tears out the synchros because speed shifting forces the issue and something will have to wear. You break gears and knock bearings out by short shifting these manual transmissions, just like destroying clutch packs and breaking input shafts in an auto, in high TQ conditions, rather than use some rpm to get them out of the danger zones.
Short recap: less TQ rise, higher inertia, less component speed difference, higher apply pressures, and greater defuel application at high rpm. Why would the transmission NOT function better and longer?
That is just stock, all things being equal. When you move off into tuning for drive-ability and usable TQ, the problems get WORSE a LOT faster.