When you delete a physical machine, the platform will put all VMs in the “Not allocated” state, without changing the VMs on the hypervisor. You cannot deploy the captured VMs again unless you recreate them from an instance template.
If you later add the physical machine again or upgrade to cluster as physical machine, the platform will automatically update all VMs from the hypervisor, returning captured VMs to a deployed state. In the API these recovered VMs will have the same ID and URL as before.