Skip to main content
A data binding reads information about an existing cloud resource without bringing it under ubx management. It’s the equivalent of Terraform’s data sources. Data bindings live inside a stack body, next to regular resource declarations.

​
Syntax

binding_name = data provider.namespace.Type {
  // filter/lookup properties
}
Data sources use the same dotted type paths as resources, with the data keyword in front.

​
Example

Look up the latest Amazon Linux 2 AMI, then use it: 
stack compute {
  ami = data aws.ec2.Ami {
    owners      = ["amazon"]
    most_recent = true
    filters = [
      { name = "name",                values = ["amzn2-ami-hvm-*-x86_64-gp2"] },
      { name = "virtualization-type", values = ["hvm"] }
    ]
  }

  instance = aws.ec2.Instance {
    ami           = ami.id        // resolved at apply time
    instance_type = input.instance_type
  }
}

​
Look up an existing VPC

stack application {
  existing_vpc = data aws.ec2.Vpc {
    filters = [{ name = "tag:Name", values = ["prod-vpc"] }]
  }

  subnet = aws.ec2.Subnet {
    vpc_id     = existing_vpc.id
    cidr_block = "10.0.99.0/24"
  }
}

​
Data source outputs

Attributes read from a data source are pending values — they are fetched at apply time from the cloud provider, and the typechecker tracks them the same way it tracks resource outputs: 
ami.id
ami.architecture
existing_vpc.id

​
In generated code

For the Pulumi Python engine, a data binding compiles to the provider’s get_* lookup function; for the Terraform and OpenTofu engines it becomes a data block in main.tf.