You also cannot separate the success of AT Protocol in it's developer engagement from the existence of the Personal Data Server (or, Personal Data Storage), building applications is much easier when you don't need to deal with user registration, authentication and authorization, data storage, etc. When all you need to concern yourself with as an application developer is "I write data to the personal data server for the authenticated account, using this lexicon", development is much much easier.

Right now in the ActivityPub ecosystem, we see all the same problems as those that the Solid ecosystem sees (both rely on JSON-LD and data effectively being schemaless), but on top of that, the current architecture of the Fediverse is these huge monolithic servers with their own proprietary APIs. If you spend enough time in the ATmosphere looking at AT Protocol, you'll note that they just don't have that issue. Sure, there's lexicon clashes where someone extends someone else's lexicon, and there's all the problems of schema design, but you don't need to think about so much lower level infrastructure — it just exists.

Only a relatively small handful of people end up needing to think about that lower level infrastructure: the people implementing custom PDSes, Relays, and AppViews, but that last one is changing with the advent of https://slices.network which makes it trivial to build an AppView for your application.