Written by Pablo Gómez
Founder & CEO at Metlabs · Reviewed on October 11, 2026
A rollup is not an independent chain. An operator called a sequencer orders user transactions, groups them into compressed batches and posts that data to layer 1. It historically did so via calldata; since the Dencun hard fork (13 March 2024, EIP-4844) it uses blobs, which are far cheaper. After each batch it publishes the new state to a bridge contract on Ethereum, which acts as the settlement layer.
The two main types differ in how they prove that the state transition is correct. Optimistic rollups assume it is valid and allow it to be challenged with fraud proofs. ZK rollups prove it mathematically with validity proofs. This architecture is the core of Ethereum's rollup-centric roadmap, set out by Vitalik Buterin in October 2020, where layer 1 settles and rollups execute.
For anyone tokenizing an asset, a rollup solves a concrete problem. It enables per-transaction costs orders of magnitude lower and far more throughput with guarantees depending on the rollup architecture, because anyone can rebuild the state from the data posted on layer 1. It is the environment where smart contracts, stablecoins and regulated tokens run with the liquidity and cost a production product needs. Real examples include Arbitrum One (Offchain Labs), OP Mainnet and Base (Coinbase, on OP Stack), zkSync Era (Matter Labs), Starknet (StarkWare) and Scroll.
Sources
Also known as
- Rollups





