If you feel discouraged...

Consider the following. If you filter on the Lichess Open DB to 2300+ and 20 ply, and remove any non-standard games or bullet games, approx. 200 GB a month is reduced to more like 750 MB a month. Along those lines, the Chesscom Published Data API makes it possible to recreate the same thing, relatively close to being complete, and it reduces down to about 500 MB a month.

That's one set of numbers. Consider also that the Lichess Open DB contains about five billion games, from 2013-2024, whereas Chesscom's archives are adding a billion games a day.

This would seem to imply that almost everyone at Chesscom is lower-rated. The average is supposed to be about 850, and it keeps dropping, as more and more members join up.