Thereβs a difference between the two: a single row might exist on-disk as more than one tuple at any time, with only one of them visible to any single transaction.
The transaction doing an update now sees the new version of the row, the new tuple just inserted on-disk. As long as this transaction has yet to commit then the rest of the world still sees the previous version of the row, which is another tuple on-disk.
While in some contexts tuples and rows are equivalent, we must be careful to use them in the right context.