So, in 2005 I was involved in a company that thought "man, it would be great to build a search engine for source code". I even started putting together components for it, such as a model for finding text inside of larger documents that had statistical properties similar to code, auto-detecting the language, so you could find code snippets inside of blog posts. However, by the time we got organized enough to actually…
If anyone can come up with a law that basically forces companies who give up on free products to open source the code they no longer use, that would lead to amazing opportunities!
You also need serious $$$ in infrastructure to run it for the web masses.
This is getting to be a problem for FOSS. I mean, once you could get some universities, volunteers, etc, donate some hosting for repository mirrors and stuff, and it was enough.
Nowadays, a service that caters to the web crowds needs serious many to run with some capacity (and redundancy). So, say, GNU cannot just start a FOSS Gmail, or a FOSS Github service running on free servers and controlled by the community. Also see: Diaspora.
So, FOSS is mostly confined to the desktop or as server backend infrastructure, but cannot compete as SaaS.