I think the fediverse will continue to grow, and it would mostly be because of projects like Loops and Pixelfed and other new services. But I think that Bluesky has a lead on us and it will continue to grow, mostly because bigger players with money can afford to actually build on the AT Protocol, which has higher hardware requirements.
Eventually, thanks to bridges and multi-protocol platforms, there will be a multi-protocol network, with big players who have money using AT Protocol, and people who want to start a server on a lower budget using ActivityPub.
I agree that there needs to be some more “killer apps” for the fediverse for it to really take off. Luckily there are some people working on that. Loops and Pixelfed seem to be some recent highlights and are growing fast. And there are other apps that are being worked on that could change things.
One of the problems is that ActivityPub is somewhat limited to what Mastodon has implemented. A lot of other platforms want to implement some cool features, but none of those are supported by Mastodon. This limits the rest of the fediverse since they have to cater to the lowest common denominator.
It is one of the main reasons why Hubzilla still uses Zot6 as its primary protocol, and uses ActivityPub to communicate with everyone else. Hubzilla has features that the rest of the fediverse does not support, such as nomadic identity, privacy, and access control. And related to that, Bluesky also has features that ActivityPub does not support, or if ActivityPub does support it, is not implemented by the larger platforms.
I think this is the primary reason why the fediverse is falling behind. People look at what Mastodon has implemented and think that ActivityPub is weak compared to Zot6 or AT Protocol.