Esp. considering the terrible interoperability of these standards on different web platforms. I would suggest going pragmatic for now, with possibly a separate idealised track to be able to exploit the xhtml stuff when it stabilizes in the web clients.

Keeping some development going on the xhtml stuff is good, but the main thrust should be using what works now, not what might work down the road. I'd suggest doing something like an 80/20 split html/xhtml, with appropriate communication between the two. In short, get the xhtml people working on the more back end oriented stuff for the moment (the stuff that fits well in ideal land) and if/when it comes together (when MSFT realises they are only hurting themselves by wacking off on the incompatibilities as they don't have market power there) then you can move towards moving that stuff into actual production.

NB- working on incomplete information, of course.