I agree with that, although I believe that prediction (I can't seem to dig up where it was) was more about how many new articles need to be created entirely, i.e. how many subjects Wikipedia should cover don't even have a 1-sentence stub yet. I think there are probably still several million of those, but new-article growth is inevitably slowing as low-hanging fruit has been quite well harvested.

I agree that there are a ton of the current 4 million articles that are nowhere near complete, in many cases not even 10% complete. That's an area that could use considerably more work than new-article creation imo.