Have you already told him that you have concerns? If so, you can't change his mind. If he was able to listen, he would have listened by now.
What you can do is let him make the decision, let it be carried through, and then work to show that it was a bad decision that needs to be changed.
So, once the stack has been selected, push to start work on tasks that will reveal its weak points. Say he's decided to implement a real-time trading system in shell script: build a quick prototype and start some performance testing. Say he's decided to implement an ESB with Microsoft tools in CORBA-dominated environment; tackle some of the integrations first. Think about your objections to the stack, and how you can demonstrate them.
Note that i am absolutely not suggesting you sabotage the project. Don't make things look worse than they are. Pick a weak spot, do your honest best work on it, and let the weakness of the decision shine through.
If the problems you foresee really exist, you'll reveal them, and it will be impossible to hide them - you'll have developers on all sides wailing, and QA screaming, and project managers gnashing their teeth.
If, on the other hand, your fears are groundless, which they might be, you'll discover that too.
It's worth mentioning that even if you think the right decisions have been made, a "hard things first" approach is generally a good idea anyway, for exactly the same reasons.