Ask HN: What to do when code becomes an ad-hoc database?

A hypothetical company uses infrastructure-as-code tools to provision all of their infrastructure. Managing the resources becomes tedious, and an engineer builds automation that allows all of the infrastructure code to be generated by pipelines from modules. Now a user clicks a button in order to provision infrastructure. The engineers only look at a few lines of module code once in a while. The issue is that the automation is very complicated and costly, and the generated code and all of it's associated git repositories etc have essentially become ad-hoc data in makeshift databases; what once was human-readable code is now primarily concerned with tracking the state of the infrastructure.

It seems like we've come full circle. If I can automate my infrastructure in this way, can I skip IAC and just use a database to track my state? Who has experience with this kind of problem? I can't help feeling like we're really re-inventing the wheel and making a huge complicated mess here.