Hoping we will talk through this at an upcoming WG meeting. When we talk about there being multiple edges, I think we may have to call those out and make design distinctions accordingly. For example, with my customers; I see HPC at the edge. I see mixed air gap and remote managed edge where the primary deployment model and runtime model is still cloud-native. I see multi-modal edge with vms living side by side with containers. I see device edge which looks like IoT but is a cloud native single node stack.
What I don't see is a requirement for cloud backed edge. Cloud (DC) often comes in as a "requirement" because it provides means to configure and manage at scale but it's not a hard requirement.