RJ can progressively add capabilities to an existing cluster so that all the services aren't required during Iteration Zero

Iteration Zero should be used to set up the initial, perhaps limited, environment based on the understanding at the beginning of the project. As capabilities are needed, it should be easy to add them to the environment - e.g. It is known persistence will be required for the project but the choice of persistence provider (SQL vs NoSQL, Cloudant vs Mongo vs Cassandra, etc) is deferred until Iteration 2 when more information is available. Or perhaps in Iteration 6 it becomes clear that a Redis cache should be introduced.

This would apply to software/services or tools

Whatever the reason, it should be easy to introduce new capabilities as the project progresses. Ideally, this functionality will be provided using GitOps (although not *necessarily* implemented with ArgoCD or operators at this time) where the new desired state is described in Git and the pushed change triggers the reconciliation process with the existing environment. This is something that Terraform does naturally as part of its operational model.