Because in larger clusters, the cluster can fix any issues if one hosts goes offline (load balancing) or a database gets corrupted.

In a single cluster configuration, anything going wrong can destroy your cluster's capability. Think of it as RAID, the more drives, the more you can do for high availability.