Re: But It's Shiny!
we like doing interesting stuff
It's been a long and slippery slope.
Back in the old days there wasn't much alternative to rolling your own: Phoenix involved writing an alternative to IBM's woeful TSO and repurposing a bunch of PDP-11s as async terminal concentrators. Teaching was largely done in BCPL, which was of no specific commercial value as a skill, but the compiler writer was conveniently on hand. Other commercially niche languages that formed part of an undergraduate course included LISP and Algol 68.
There is an argument for having your researchers and students contribute to real projects with impact on real users - it can be a great teaching and learning experience - but you'd be unlikely to get research funding to support operational services and a decent developer could get significantly better paid elsewhere, so it no longer works economically. Equally, given the rate at which programming languages come and go, there isn't any reason why you might want to teach those that are currently commercially valuable - it's the principles that count - but try telling that to someone paying 9K a year for their course.
I'm sure it's a similar story with Microsoft Office - students expect to be given it "free" as part of their course - so it's made available when it would be a useful lesson for most students that other office suites are in fact available, some without cost.