It's worth mentioning that there are a few languages which are new to the web that are mostly encoded in Unicode -- Amharic comes to mind. There is Amharic content produced both within and outside of Ethiopia in Unicode. As opposed to Hindi, for which there are a multiplicity of legacy encodings.
Encouraging Wikipedias in such languages may early may pay dividends, since it's likely that users that read existing content will be have fonts installed already (input is another issue).
-Patrick Hall