Rather than saying "We should reimplement our core products in $FOO" you can often achieve the result of using $FOO while hitting less resistance by implementing one-off projects or support systems in $FOO instead.
e.g. I worked at a Japanese megacorp and, despite us being a Java shop, one of the sales guys sold a customer on "We're certainly the right people to write documentation for your Perl program." This contract only required a bunch of HTML files as a deliverable, rather than getting a team of programmers to speed on something which would be maintained for 30 years, so I successfully convinced my boss that I'd jump on the Perl grenade if he let me use any means necessary to create the HTMLed documentation. I ended up doing it in Ruby. Then I had a "Hey boss remember that time I delivered a $X0,000 project in 1/4th the budget and made you look like an effing hero? I want to use that tech again" case study.
Though for the amount of time I spent on evangelizing Ruby and modern Javascript-heavy UIs at my previous company I could have launched my own. Eventually I did. It turned out to be a much better use of my time.