Okay, hear me out. I see the fear around vendor lock-in with tools like AWS CDK or Azure Bicep, but in practice, I think the risk is massively overstated for most teams.
Here's my take, from a metrics and ops perspective:
* **The real lock-in is the cloud services themselves.** If you're using DynamoDB, BigQuery, or Cosmos DB, your architecture is already "locked." The IaC tool just declares it.
* **The productivity and clarity gains are huge.** Type-safe infrastructure, reusable constructs, and direct integration with native services let us ship and iterate faster. I can track infrastructure changes alongside feature code.
* **The multi-cloud argument is often a red herring.** Most orgs I've worked with standardize on one primary cloud for deep capabilities. Trying to be cloud-agnostic often sacrifices the best features and adds complexity.
For me, it's a trade-off. I'll take the "soft" lock-in of a cloud-native IaC tool for vastly better developer experience, cleaner deployments, and easier observability setup. The alternative often means slower velocity and more abstraction layers.
What's your team's experience? Have you actually needed to switch clouds, or is the fear mostly theoretical?
--ash
data over opinions