That is only for 2 participants. For 3 or more (video participants), it uses the videobridge SFU.
Source: https://github.com/jitsi/jitsi-meet/blob/master/doc/manual-i...
Source: https://github.com/jitsi/jitsi-meet/blob/master/doc/manual-i...
That said, it is worth pointing out that this thread is specifically about videobridge (i.e. scaling beyond a full mesh).
Video shenanigans can be a dark art, but not that magical ;-)