There are at least a few additional options to those three to consider here:
- The manufacturer doesn't so much want it to break, but rather, they expend no effort avoiding short lifespans, even where trivial. With enough components the chances then of including one or an interaction of two that happen to last not much beyond your guarantee grows rapidly.
- smaller/lighter/simpler or in some other way less durable components are cheaper (due to lower waste), or possibly more efficient (due to less unnecessary mass to move/cool/heat/whatever), so there's a tradeoff that leads to more easily visible and testable attributes being prioritized even if the tradeoff wouldn't be worth it were everything transparent and all customers completely informed.
- throwaway items actually are more efficient, because small savings accrue from not worrying about longevity, and repair has costs and wastes of its own (i.e. the premise is misleading)
- some combination of all of the above.
Personally, I suspect the difference between planned obsolescence and unplanned is in practice nil - you need quite the dysfunctional organisation (and possibly planned dysfunction ;-) ) to get everyone involved to so completely disregard longevity.