Personally I suspect that the chaotic tooling will waste more dev time than what the sysadmin time will cost you. So that it will be a net loss even before you consider lock-in and fees.
While a good argument in favor of the vendor locked-in serverless services can be made at the start of a project, it looks like a net loss during any other phase of the project.
It comes down to if you want to have generic and reusable sysadmins vs. cloud serverless specialists.
