Our idea of code review groups is that people actively work together. This implies that everyone contributes by leading a review or adding feedback.
If a group scales vertically, then more and more members join. Huge groups cannot operate as flexible as small ones and it becomes more and more difficult to integrate everyone into active roles.
Eventually not everyone can actively contribute anymore. You can try to compensate this by splitting members into active contributors to a session and a more passive audience. One possibility is to share a screen and video session with active contributors and stream the whole channel through a streaming platform. The audience can still follow the session and add live feedback through chat. Consider adding moderators to the chat to push questions to the active group and for audience management.
But even with all these techniques the vertical scaling does not work well. Too many people are forced to follow a process defined by the active group, who force their way of performing a review and adding "blind spots" by forcing the audience to follow their thought patterns.
Splitting groups if too many members join is a way to horizontally scale. People can keep their mode of operation without additional technical or organizational overhead.
Exchanging ideas and new techniques between groups has to be managed in addition to sessions. One such possibility is writing a blog, exchanging meeting minutes or if people occasionally join other groups to share their experiences.