I think Bitcoin does have a soft floor because that cost would form the natural price target for a miner trying to trade, but the cost is avoidable by switching to proof of stake as done in PeerCoin and Nxt. The crypto problems solved in Bitcoin don't have the intrinsic utility of a precious metal; you could make a better argument for something like PrimeCoin, which attacks a major computation problem, having this kind of utility.
So as I see it Bitcoin is already obsolete; there are new coins that achieve the transaction capabilities for cheaper - they can be used in more potential applications, thus they're more interesting speculative instruments.
The specific benefit of proof of work mining comes in insuring more people have a chance to enter during initial distribution, but it still massively favors the early entrants and especially the people who have the hardware environment to mine efficiently.
So as I see it Bitcoin is already obsolete; there are new coins that achieve the transaction capabilities for cheaper - they can be used in more potential applications, thus they're more interesting speculative instruments.
The specific benefit of proof of work mining comes in insuring more people have a chance to enter during initial distribution, but it still massively favors the early entrants and especially the people who have the hardware environment to mine efficiently.