Having a portfolio of games you can demonstrate is definitely a good thing. It doesn't matter so much whether those games are Java or C++, so long as they demonstrate your ideas and abilities. However I think you should set your expectations that if you make it into in the industry, you probably won't be using Java.
Casual web-based games have moved away from Java to Flash, because a considerably higher percentage of casual web surfers have Flash installed than have a JRE.
Console and PC games do not use Java to any extent. They are primarily C++ right now.
High end smartphones (Symbian, Windows Mobile, iPhone, etc) have native SDKs, and if they support Java at all tend to discourage its use. Blackberry is Java based, but there are relatively few games for the Blackberry. I'd have to speculate there is not much demand for games on that device.
The one place where Java is commonly used for games is in low end mobile phones. This has not been a terribly successful category, only a few companies have made money. There are two problems with this space, so far as I can tell:
* write-once-run-anywhere never worked out. The QA and testing load to support a reasonable selection of devices is demanding, and never-ending since new phones come out all the time.
* To distribute your game you need a business relationship with the wireless carrier, and they demand a 50% (or more) cut.
One last thing: the games industry is famous for programmer burnout. You should search Google for stories about developers at Electronic Arts. Games nowadays are extraordinarily large projects which must meet a particular deadline to have a chance at making their sales targets, and they ratchet up the hours as needed to meet that deadline. Developer burnout is handled via the never-ending supply of young developers enamored with developing games for a living, i.e. replace burned out engineers with fresh ones.
Make sure you know what you're getting into.