I don't have much to add to what Gordon and Simon have said, except that as a Mac OSX developer myself, the Intel switch was obviously going to require devs seeded with cross compilers at some point and in sufficient time to allow major corporations to be able to ship universal binaries out of the box in their next releases. And that seed was never going to be sufficiently private that the MacIntel plan would stay under wraps for long. For MacIntel to work, there has to be native apps that will offset the undoubted performance impact of the Rosetta layer.
That said, compiling existing Cocoa Mac OSX apps for Intel is likely to be a piece of piss for most of them. Vienna 2.0 compiled with nothing more than a bunch of compiler warnings brought about by the switch to GCC 4.0. I don't know if it will actually run as I've no way of finding out for a while but I'm pretty optimistic. There's a category of code that will need revisions but they're limited to those that assume byte ordering, AltiVec, Preference plugins and assembler stuff to name but a few.
Perhaps there's more interesting stuff coming, possibly at the Paris Expo. Possibly significant price reductions on existing hardware. But I'm also confident that PPC machines have years of life left in them. (But given my financial situation and the developer necessity of such an approach, I'll probably wait and buy a MacIntel box as my next Apple PC.)