Actually, this is subtler than it looks.
The code above would give the incorrect answer for a lower case character whose code point was above U+FFFF (such as U+1D4C3, MATHEMATICAL SCRIPT SMALL N). String.charAt would return a UTF-16 surrogate pair, which is not a character, but rather half the character, so to speak. So you have to use String.codePointAt, which returns an int above 0xFFFF (not a char). You would do:
Character.isUpperCase(s.codePointAt(0));
Don't feel bad overlooked this; almost all Java coders handle UTF-16 badly, because the terminology misleadingly makes you think that each "char" value represents a character. UTF-16 sucks, because it is almost fixed width but not quite. So non-fixed-width edge cases tend not to get tested. Until one day, some document comes in which contains a character like U+1D4C3, and your entire system blows up.