What you are talking about is called Modular Monolith(some info here https://www.infoq.com/news/2020/01/monolith-architectural-drivers/?useSponsorshipSuggestions=true&itm_source=presentations_about_architecture-design&itm_medium=link&itm_campaign=architecture-design however there is tons of documentation out there)

In my opinion, the problem is not only the number of repositories to deal with but solid investment in automating all the repetitive tasks around creating a repo, preparing a standard CI, add monitoring and observability and so on.
If you read how large organizations are handling large codebase with monorepos, the first thing they have done is investing a lot in automation even creating their own tools (like Google, Twitter, Facebook, you name it), and also with that, not every developer is happy about the pitfalls.

I believe it’s always a trade-off, when starts to hurt, finding the time and resources to alleviate the pain and improve the situation should be a primary goal for an organization.
Iteratively improving the automation part should be a constant effort for the teams considering they are using on a daily basis the CI pipelines.

My 2 cents 😁