(geek level of this post: 45%)
This is interesting. Non-Roman alphabets (Chinese, Arabic, etc.) have never been usable in domain names.
This presents two big problems:
Web programmers (like me) will have to rewrite a lot of parsing and validation code for anything that uses internet hostnames, including URLs and email addresses. (Not to mention the massive problems of updating all levels of DNS for it.)
Windows users will actually think the domain is ????????.com instead of 食店の店舗情報や.com.