When upstream Karpenter finishes the early implementation of the newly accepted CapacityQuota autoscaling API, we need to downstream it and implement it as a VAP protected resource in AutoNode enabled guest clusters.

The resource needs to select all Nodes in the cluster, and have a configured limit according to a cluster-admins setting. This cluster-administrated setting will likely need to be put on the hostedCluster.spec.autoNode API, requiring a new field that will need to be reconciled onto the underlying CapacityQuota by the karpenter-operator, which should be VAP protected (similar to existing implementations of VAP protected resources in hypershift).