if you don't want to learn  EJB3, whatever you replace it with will change  before you finish the book. IT is all under experiment every minute

the problem that you are spotting on is what we all are facing everyday with IT, only looking at JavaEE, it has scoped so widely that it seems very difficult to be really skilled in every aspect of it and even keeping up-to-date required passion, dedication and of course time
but generalizing a comment like that is pretty bold

if a project is based on (for argument sake) ajax or web services and after a while much more reliable alternatives are found you'll find yourself in a position of decision making

if you don't see much projects around you that are going to be based on EJB 3, its because if a project is developed with J2EE for years and is already running in the field it wouldn't seem mature to replace it with new stuff just because its new or its cool, but at the same time  you should keep up with the tech cause you might need to evolve existing projects or cope with the industry as technology  evolves

but I understand  sometime it sound a bit frustrating when you get to a point of applying EJB3 and then you figure EJB4 is on the way :)

in summary, learning it will not be a waste of time (I guess it applies to everything not just EJB 3 :) , even if in future  releases, EE looks much different from what we have now, it would be beneficial to leverage the knowledge as EE grows
but as a compliment to you, it takes a lot of courage to mention such a thing in this forum