Snowflake IaC without switching to Terraform

Scenario: you already use CDK, your team is happy, and you use CDKTF for non-AWS resources.

Then HashiCorp (now owned by IBM) sunsets CDKTF, and suddenly it feels like you must run both CDK and plain Terraform/HCL in the same pipelines.

That is the panic moment. But that is not where things end up.

Using CDK Terrain, you can consume the official Snowflake Terraform provider and stay in a CDK-first workflow.

My proof-of-concept repo:

https://github.com/kaihendry/snowflake-cdktn

Hashicorp drops the Cloud Development Kit

HashiCorp deprecated CDKTF and archived the upstream repository the same day, on December 10, 2025

• https://github.com/hashicorp/terraform-cdk/commit/a0c56138e2b4c7b6c128bd7ba08a8e317df852af
• https://github.com/hashicorp/terraform-cdk

CDK Terrain is the community continuation. In practice, it preserves the CDKTF workflow and gives a more CDK-consistent way to keep using the Terraform provider ecosystem.

Why I still want CDK over plain Terraform

• Type checking and IDE support: autocomplete, compile-time checks, and first-class language tooling.
• Real refactoring: rename symbols, extract abstractions, and reorganize IaC like normal software.
• Testability: infrastructure can be unit-tested with standard test frameworks.
• Better abstraction for complex systems: CDK constructs are reusable, composable modules instead of repeating verbose config.
• Level 2 constructs: CDK is much more flexible to express Organisational patterns and guard rails that would be very difficult to do with Terraform.

For complex infrastructure, CDK gives me the software engineering ergonomics I want, while Terraform providers still do the provisioning.

If you need to deploy Snowflake at scale, CDK is still a viable option.