From the top of my head, I would say internalize MEV and having more customizable space to innovate are the main ones to choose to launch an appchain over an ETH rollup.

The majority of rollups today don’t capture any value out of their users’ activity because this would seem like a toxic behaviour, so they implement mechanism like FCFS + private mempools only available to their centralized sequencers. They can ‘minimize‘ MEV since no other party can extract value besides them. If they were to decentralize the sequencers, this would be another story, although before decentralizing i think they’d implement something like encrypted mempools.

Think of the classic example of Uniswap. The MEV they’re generating is flowing back to Ethereum validators, and UNI token holders do not receive anything. Why do they choose to stay on Ethereum, then? Network effects. This would be the same case if they choose to launch a rollup; the only thing left for them is to launch an appchain if they want to internalize the value they generate.

Appchains can have a more explicit say about *how* mev will be extracted, *what* kind of MEV is allow to be extracted, *which* parties will be receiving it (users in form of rebates, validator’ rewards, or burning it) in a coordinated and distributed (yes, distributed rather than decentralized) way instead of just relying on the rollup foundation team. This is possible because of the freedom that launching your own appchain implies; customize any execution environment can be useful for some kind of applications (e.g. dYdX, Maker is also planning to build an appchain).

But it’s also true that cosmos’ chains have validators’ cap, so it’s not like everyone can participate; just professional validator teams will have the power.

Token utility can be another option to launch your own appchain. What is the utility of a rollup’ token? Just governance, if that’s an utility.. What they can actually vote for?

Appchains can give their tokens more broader use cases such as using them for gas payment, and the governance in this cases can actually be useful because of the customization that i talked above.

I think that might be a technical answer. The other could be brand-alignment. Some chains want to have full sovereignty over their brands, some chains that uses CometBFT (Tendermint) and the Cosmos SDK don’t even want to be aligned with Cosmos brand (dydx, berachain)

The appchain thesis is dead, long live the appchain thesis.