Because it could be really, really cool. As an engineer I like "hard" problems, and when I find a project interesting I go after it -- regardless of how "flashy" it is to the general public.
I think in Google's case, this is certainly true if you examine how much work has gone into their infrastructure (and how many research papers have been written on those topics).
If you push the "extra ice" button on the soft drink vending machine, you won't get any ice. If you push the "no ice" button, you'll get ice, but no cup.