Personally, I'd be looking at a transition strategy. History suggests that when Apple makes a move like this, it's only a matter of time before 'legacy' support is deprecated.
i.e. with the switch to OS X, first developers had the option of their apps running in a 'Classic' sandbox, which disappeared with the move to Intel.
The next step was Carbon -this was created in response to developer complaints over being forced to use Cocoa/Objective C, and was a set of APIs that made it easy to port earlier applications to OS X. That finally died in the transition to 64 bit, but the warning signs were there before.
The key point will be when Apple start developing new APIs with features that only work with Swift, and redeveloping their own code to use it - this won't be for a few years yet, but you see the start of the process with Playgrounds.
It is pretty evident that this is not - like MacRuby - an optional alternative to writing in Objective-C. This is, like C#, the official future.
Of course, most CIO I know are still persuing lowest-common denominator cross-platform solutions, and I don't see Switft changing that. Few applications need deep platform integration or performance.