Big questions like this are probably best asked on the xml-dev list. People will probably go off at all sorts of tangents.
I have a very bad track record at predicting the future. I don't expect to see radical change in the XML scene over the next few years. XQuery may or may not become big; XForms similarly. I'd like to see standardisation of pipeline processing but that doesn't mean it will happen.
I think the XSLT language will remain pretty stable after 2.0; I expect there will be expansion in function libraries, but I'm not sure where these are likely to come from. There are one or two "big ideas" I would still like to see as language enhancements, for example higher order functions, but I don't think there's enough momentum in the W3C process to make it happen for a while.
As for marketable skills, I've never approached my personal development in that post-Thatcherite way. I learn things either because they look interesting or because I need them on the next project.
Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference