The hypothetical 200ms is for cold starts. That means when the docker container that runs behind the lambda first starts up. When a lambda is invocated, that d

The hypothetical 200ms is for cold starts. That means when the docker container that runs behind the lambda first starts up. When a lambda is invocated, that docker container isn’t immediately destroyed. It’s kept around for reuse. So there isn’t a realistic scenario where you would get a million cold starts in a month.