Uber had 8000 repos at one point with 2000 engineers - https://highscalability.com/lessons-learned-from-scaling-ube...
And it was each team owning multiple internal repos of their own deployments/libraries, and not, primarily, clones of public repos.