There are many questions on this site regarding Unicode and wchar_t. I guess I have grasped the concept, but then found something that proves most (if not all) answers wrong if it is true. On this page, Microsoft claims that one wchar_t character can hold any Unicode character (emphasis mine):

A wide character is a 2-byte multilingual character code. Any character in use in modern computing worldwide, including technical symbols and special publishing characters, can be represented according to the Unicode specification as a wide character. Developed and maintained by a large consortium that includes Microsoft, the Unicode standard is now widely accepted.
A wide character is of type wchar_t. A wide-character string is represented as a wchar_t[] array. You point to the array with a wchar_t* pointer.

Since this statement is from Microsoft directly, I am quite worried now:

How can a “two-byte multilingual character code” hold any character of the Unicode character set that already contains around 150,000 code points (characters)? [ Plus, if we take into account the private use code points, surrogates, code points that already have been reserved, and so on, it would be over a 1,000,000 code points? ]

I hope that this question is not a duplicate because its core is that Microsoft itself states something which seems to be plain wrong, and I really would like to know what I have misunderstood specifically on the Microsoft page I have linked.

By the way, then there is this page that contradicts the first one and eventually tells the truth:

Windows represents Unicode characters using UTF-16 encoding, in which each character is encoded as one or two 16-bit values.

So obviously, we sometimes need two wchar_t characters (4 bytes) to represent Unicode code points. Well, that would make sense somehow, but given the contradicting documentation, I am complete unsure now.

If somebody is interested in how that question originated:

In one of my projects, I have a string that must have a character at a certain fixed position replaced by another character. This happens in a loop and must be done as fast as possible. This is a no-brainer with normal char[] strings. But the string in question is of type wchar_t[], and I don’t have control over the replacement characters.

Depending on which from the above Microsoft statements is true, this is either a no-brainer, too (if the first statement is true), but if the second statement is true, it would become quite a mess: I the could not just replace the wchar_t character at the respective index by the replacement character, because the original character might be one wchar_t while the second might need two wchar_t, or vice versa.

That’s why I’d like know which documentation is true.