I think there's actually a middle ground between Big Fedi and Small Fedi, as mentioned by @evan — perhaps we call it "medium fedi" or "just right fedi"
Companies will probably also want tools that help their engagement be understood: are we in positive standing or negative? Are we contributing positively to the fediverse or having a negative effect & getting blocked / banned.
Companies may also be the reason E2EE DMs finally happen, because they'll want good secure channels to communicate with customers on, without having to leave the fediverse.
We do need well behaved corporate accounts, because that's really useful on social media (e.g., telekom_hilft on Twitter was able to resolve issues 1000x better than any other channel with Telekom)
But we also don't want to optimise for corporations, and if your company does horrible shit you'll likely find yourself on the wrong side of social discourse: this is normal.
That is to say, there are some good ideas on both sides of the "big fedi" and "small fedi".
Most of all we need "sustainable fedi".
A fediverse where we can both keep people safe, but also have a sizeable population because people don't want to be alone on the fedi.
We need moderation that scales and has repeatable practices, but we also need humans involved, and understanding how moderation can harm marginalised communities & reinforces cis-hetero-normative white ideals.
We need to acknowledge that an instance with more and more users does lead to worse moderation, but also need to ensure that humans are involved in moderation decisions and it's not just a machine.
A server with 1 billion users is almost certainly going to be unmoderated at a high quality because there's simple too much work involved. But that doesn't necessarily mean a 100 or 1000 user server is well moderated either.