That's what they *say*, anyhow. In practice though, most managers have all sorts of other concerns, some of which are blown way out of proportion or sometimes completely irrational. The trick is to address these too.
One of the biggest things is the manager's version of premature optimization: worrying about scale way too early. Not just scaling technologically (although that might come up) but also hiring and training lots of developers. All too often, they think that training developers in a new language will be too difficult and that they need a market of developers comparable to Java to be able to hire for their team.
The funny thing is that this is almost the opposite of what happens in practice: using a language like Haskell actually makes it *easier* to hire good developers. Haskell motivates programmers to apply more actively and self-select, which is nice. This thread[1] has some testimony that might be useful to show that hiring Haskellers is actually easy.
As far as training goes, it's also not too difficult. The IMVU blog post mentioned earlier covers this:
> "Today, training an engineer to be productive in our Haskell code is not
much harder than training someone to be productive in our PHP
environment. People who have prior functional programming knowledge
seem to find their stride in just a few days."