This is very true. My circumstances are a little unusual where using Azure costed more than running it internally on VMs, and running k8s or an equivalent didn't really add much value since I would have had to manage that, and my workload is uniform where each VM runs the same services so just running a podman pod was easier. There was no need for dynamic scheduling and scaling would just be launching more VMs and running more podman pods, and the entire deployment is just an Ansible playbook that preps the VM after boot then launches the containers. It didn't make sense to have another kind of YAML file to deploy the containers.