Skip to main content
Phase 1 provisions the foundational infrastructure for the CoreWeave reference architecture: a VPC and a CKS Kubernetes cluster. This phase is for platform engineers and infrastructure operators who are bootstrapping a new CoreWeave environment and want to follow the reference architecture end to end. After this phase completes, you have a running cluster and can download its kubeconfig to proceed to Phase 2, which adds node pools and DFS storage.
Ensure you meet the prerequisites before starting, including the CKS Admin Identity and Access Management (IAM) role.

​
Clone the repository

The reference architecture is distributed as a Terraform configuration in a public GitHub repository. Clone the CoreWeave reference architecture repository, navigate to the terraform/ folder, and copy the example Terraform variables file: 
git clone https://github.com/coreweave/reference-architecture.git
cd ./reference-architecture/terraform/
cp terraform.tfvars.example terraform.tfvars
The terraform.tfvars.example file contains example values you can use as a starting point. You’ll edit terraform.tfvars to set your values in the next section.

​
Configuration

In this section, you set your API token and configure your Terraform variables.

​
Set your API token

Set the CoreWeave API token as an environment variable: 
export TF_VAR_coreweave_api_token="[YOUR-COREWEAVE-API-TOKEN]"
Setting the API token as an environment variable is the recommended approach because it keeps the token out of files that might be committed to version control.

​
Configure variables

Edit terraform.tfvars and set the following values:
VariableDescription
zoneDeployment zone, for example US-EAST-02A.
vpc_nameA name for your VPC.
vpc_prefixesCIDR blocks for pod, service, and internal LB networks. See VPC CIDR sizing for guidance. The example values in terraform.tfvars.example are valid defaults.
cluster_nameA name for your CKS cluster (maximum 30 characters).
Leave create_nodepool and create_dfs_pvc as false for Phase 1. You enable these variables in Phase 2 when you add node pools and storage.
If you choose to add your CoreWeave API token to the terraform.tfvars file, do not commit terraform.tfvars to the repository. Add it to .gitignore or use environment variables exclusively for security.

​
Create and verify your VPC and CKS cluster

In this section, you apply your Terraform configuration to create your VPC and CKS cluster, verify the outputs, and download the resulting kubeconfig.

​
Apply your Terraform configuration

Initialize your Terraform configuration, preview the planned changes, and apply them to create your VPC and CKS cluster: 
terraform init
terraform plan
terraform apply
Cluster creation takes approximately 45 minutes. Terraform displays progress as it provisions the VPC and cluster.

​
Verify outputs

After terraform apply completes, verify the core outputs to confirm Terraform created your VPC and CKS cluster successfully: 
terraform output vpc_id
terraform output cks_cluster_id
terraform output cks_api_server_endpoint
terraform output cks_status
The cks_status output shows the cluster as running.

​
Download kubeconfig

Download the kubeconfig for your new cluster from the CoreWeave Console. The kubeconfig authenticates kubectl to your cluster, and you also need it for Phase 2, which covers node pools and DFS.
  1. Log in to the CoreWeave Console.
  2. On the Clusters page, find your new cluster and click the vertical ellipsis (More options) next to the cluster name.
  3. Click Download kubeconfig.
  4. Save the kubeconfig file to your local machine.
After downloading, set the KUBECONFIG environment variable to the path to the downloaded kubeconfig and verify connectivity: 
export KUBECONFIG=[PATH-TO-DOWNLOADED-KUBECONFIG]
kubectl cluster-info

​
Next steps

You now have a running CKS cluster in a CoreWeave VPC, with kubectl configured to access it. Continue to Phase 2, Add node pools and storage, to add node pools and DFS storage.
Last modified on June 4, 2026