At the time of backup, CBT of the source VM was disabled.

Source VM resides on NFS datastore during backup. For NFS datastores, Druva cannot perform a * query to find the allocated data blocks. Hence, the Hybrid Workloads VMware agent does not detect the sparse block and backs up the sparse block as a zero data block. Therefore, the thin disk is converted into a thick disk during restore.

At the time of backup, there was no sparse block present on the disk - Hybrid Workloads VMware agent does not change the disk type explicitly, but the change is from VMware when there is no sparse block on the disk of the restored VM.