The GKE examples do this a lot: https://github.com/pulumi/examples/blob/master/gcp-ts-gke/cluster.ts#L10-#L14

// We can't create a cluster with no node pool defined, but we want to only use

// separately managed node pools. So we create the smallest possible default

// node pool and immediately delete it.

Why do we only want to use seperately managed node pools? The GKE introduction docs don't do this

Resources that are created automatically are not managed by Pulumi. You want to manage everything in Pulumi, since otherwise you're managing things in multiple places. Not only is this harder work, it also means you can't configure things just the way you want to. Pulumi is declarative not imperative: you can't implement a workflow like "find default node pool, change property X to foo". Instead, you have to "define node pool with property X equal to foo".

This is a great answer, thank you