I’m sorry to say that, but you get you’re definition wrong. “decentralized” means “which has no center anymore”. ActivityPub is decentralized. The usual criticism of the Fediverse by peer to peer networks such as Secure Scuttlebutt or Dat is not that ActivityPub is not decentralized, but that it will eventually “recentralize”, like client/server models tend to do, when one instance capture all the traffic (like Gmail with SMTP, we already see signs of that with mastodon.social, but we’re still very far from it to be a center). I think that maybe you’ve been exposed to that argument and misunderstood it?
What you really want to say is that ActivityPub is not p2p. You can criticize the fact that there is a server/client model behind it, which means that users don’t really own their data and can lost it if the server goes down - that’s a valid criticism.
To which I would answer that it’s a tradeoff. :) ActivityPub is built on top of HTTP, the well known protocol on which the web is built. This makes it dirt simple to build an ActivityPub app. The difference of adoption rate between SSB, Dat or IPFS and ActivityPub has nothing to do with luck. It’s HTTP and JSON, it’s just simpler (and easier) to build on top of ActivityPub. Not only that, but it’s a w3c standard. Which means, for people like me who have been burnt by building apps on top of the Beaker Browser only to see it abandoned, that we can trust there won’t be any rug pull. That matters.
And of course, you can also… run your own server (look into self-hosting if you’re interested in that, there’s a vibrant community here on Lemmy about that). If you run your server, then you own your data and the other servers become your peers. The idea that only others (presumably big companies) can have servers is a very centralized way of thinking.