Microsoft has taught that control of your platform comes from the language and interfaces you extend to your users.  It's no accident that DirectX is the prime reason that Microsoft is the king of the hill in PC gaming to this day - OpenGL in the mid/late 90's was woefully inadequate for dealing with the massive increase in GPU capabilities that begin with 3dfx and continues to today.  It is conceivable that if OpenGL had remained dominant that it would be at least marginally trivial to support the Linux/OSX markets.  But that argument is only tangential to Apple's decision. 

 

Java on the desktop has been a non-starter for the most part.  Except a number of purpose-built tools (Eclipse, SoapUI, etc) - it's never really reached mainstream users.  Java doesn't keep up with advances in desktop engineering - it has no concept of supporting advanced concepts and "failing gracefully."  It's hard to build an app to fit the native desktop metaphor.  Projects like JDesktop help, but are still small potatoes compared to natively integrated software.  Apple has also never really been big in the server marketplace, which is where Java really shines.  

 

I see why Apple really doesn't want to support their own JVM - it makes no long term sense when native platform support is what they want people to focus on.  Why should Apple have to do  the work of making Java work when Microsoft doesn't have to?  Arguably, from an installed-base point of view, Apple has more potential Java users than IBM or HP do - they've officially reached mainstream.   With the open JDK there's no need to hide behind the solid walls of Oracle, either - it's not as if the community isn't going to step up and fill in that gap. 

 

And if they don't, that just goes to show how right Jobs was - no one cares about Java on the Mac.  Which is sad, because I do.