True. Also don't forget the opportunity cost of the time lost: 1) building a replacement and waiting for a spare slot in which to launch it, 2) engineering, manufacturing, and other resources which could have been used for (and likely were planned for) something else, but now are used for the replacement, and 3) the project the asset was used for, and its output (revenue, research results, etc.), now put on hold.
Many organizations don't have the spare time and capacity to afford all of that; they or their project may simply be done.
Keep in mind that several copies of a satellite will be built during production. Some are mockups of a few subsystems (e.g. structure), some are used for tests (e.g. thermal, electrical, interference), often there's also a "flight spare" which is an exact copy of the satellite in case something goes wrong with the flight hardware (e.g. broken during testing, launch failure).
E.g. Curiosity has such a spare (I'm not sure if it's the same as the flight hardware or close enough) which will be sent in a future mission (I think 2017, though I don't have any sources at hand) with (some) different experiments on board.
And if I recall, some of the SpaceX launches are pricier (though not that level) for the more sensitive payloads--as in there is a Secret-Spy-Satellite optional charge package.
The extra charge is due to the paperwork requirement for these launches... which were imposed after a series of launch failures by LM and Boeing rockets carrying very expensive US Government satellites.
But do realize that the replacement cost of an instrument is much lower than the original creation cost; most of the cost of these things is designing and tooling, and not part of the unit cost of the actual article.
It heavily depends on the cost of the asset. If a satellite costs close to 1bn, it worth to use $110/launch but more reliable rocket.