Earlier quoted context omitted.
B-trees (and related data structures) are pervasive in the real world. Pretty much any application that requires high speed access from disk uses them. Think about filesystems, databases, etc... They're hugely useful. But they're not useful for the same purpose as a hash table, which is also a pervasive data structure, but for a different problem area: constant time access to in-memory data.
Hash tables can be used on disk as well.
That's certainly true. They're also useful for distributed caches, but that doesn't make B-trees any less useful when the situation calls for it.