In Cassandra, both UPDATE and INSERT statements are write operations, which means that these operation don't do any reading before writing. This implies that, if we're executing an UPDATE operation on a primary key and the row for that primary key doesn't exist, UPDATE will insert a new row.
Similarly, if we're performing an INSERT operation and the row mentioned in the INSERT operation identified by the primary key already exists in the database, that row will be overwritten by our INSERT operation.
This is why we call them UPSERT operations as well.