Modified data is being addressed as a new version and therefore completely downloaded to the backup repository at the next backup run. The incremental runs are based on complete data objects and not on parts of changed data content. The same behaviour will take place once data objects are archived on the Exchange side, since they are stored as a new object with a new ID.
If Mailboxed are migrated between on-prem Exchange Organisations and Exchange Online, these mailboxes and all its data objects receive new IDs. This is even happening in a hybrid setup.
If the Mailbox is protected with Veeam Backup for Microsoft 365, the migrated data is being downloaded again at the next backup run.