It's a managed Kubernetes service that you can run anywhere you like, you can still leverage the public cloud (even on hybrid environments), but you are not tied to a specific public cloud Kubernetes managed offering.
I see that you are using the Cluster API which is currently v1alpha1. I would be curious to hear your thoughts on the state of this and how it has been to work with so far. It looks like you use Kind to bootstrap everything is that correct?
Quickstart now has 7 target provider examples: https://cluster-api.sigs.k8s.io/user/quick-start.html
I'm not using Cluster API, but the oneinfra's defined set of API's, that are reduced in purpose, to reduce the complexity of spawning new clusters. Let me know if something doesn't click, I can help!
One thing that I knew oneinfra wouldn't do, funnily enough, was to create or manage infrastructure itself.
This is just an "in-house" Kubernetes managed service :)
I still see cpu burning at around 40% or more with k8s clusters (in a full setup with encryption and service mesh and all things you need for production) without even rendering one useful payload - this seems a totally wrong strategy in a world where we want to reduce energy usage - seriously, how do we accept such an enormous step back?