How does a zero-knowledge rollup settle back to the l1 main chain?

[MatthewER]

I’m trying to understand how a **zero-knowledge rollup** actually settles back to the main chain in practice. I’ve read the high-level explanations, but when I think about the specific steps of finalizing a batch of transactions and updating the L1 state, the mechanics get fuzzy for me.

[Layla54]

In a zero-knowledge rollup I watched a batch go from offchain to L1. They collected transactions and ran a zk proof over the batch and then posted a single L1 transaction with the proof and a new state root. The L1 verifier contract checked the proof and updated its stored root. The block containing that L1 transaction is the settle moment and the data for the batch went on chain so anyone could reconstruct the new on chain state.

[Chloe_S]

I am not sure about timing. Once that L1 transaction lands and the proof verifies is the new root treated as final instantly. In practice I saw a couple of blocks pass with no reorg risk but Ethereum can flip things so we wait for confirmations. Is that right or am I missing something about finality?

[Victoria_D]

A colleague told me the on chain part is just a root pointer. It does not carry all the tx data the batch is shown elsewhere. The on chain contract stores the root and a data availability commitment and the proof verification uses a verifier key.

[NatalieSP]

Sometimes I drift toward thinking the hard part is data availability and sequencing not the proof I once saw a batch published but the data not wired to the L1 side so it caused confusion then they fixed it by resubmitting