... like Lambda for a lot of things, but there s a very specific point where it stops feeling elegant. Usually it starts with something like: a job that runs too long a workflow that should keep going in the background a task that needs better isolation or custom runtime behavior a system that starts needing cron, retries, webhooks, logs, and routing in one place And then the answer often becomes: move to Fargate, ECS, Step Functions, EC2, or stitch multiple ... ... early. A lot of teams don t actually want more AWS primitives. They want a simpler way to run background work and longer-lived jobs without rebuilding half the platform around them. For those of you who hit Lambda