An optimistic rollup is a Layer 2 scaling solution that bundles many transactions off-chain, posts the resulting state changes to Ethereum L1, and optimistically assumes the posted state is correct — unless someone successfully challenges it within a dispute window. The “optimistic” part refers to the fact that the system does not require a cryptographic proof of correctness up front; it trusts the submitter and allows anyone to challenge if they think the submitter is wrong. If a valid challenge is raised, the disputed computation gets re-executed on L1 and, if the challenger was right, the fraudulent state gets reverted and the submitter’s bond gets slashed.
This design is much simpler to implement than a zero-knowledge rollup, which needs to generate cryptographic proofs for every state transition. The tradeoff is the dispute window, which is typically 7 days on live rollups. During those 7 days, withdrawals from the rollup to L1 are in a kind of pending state — you can withdraw “immediately” via third-party bridges that front the liquidity and collect the withdrawal later, but a true protocol-level withdrawal has to wait for the window to close. This friction is the most visible user-facing cost of the optimistic design.
The Main Players
Arbitrum, launched by Offchain Labs in August 2021, is the largest optimistic rollup by TVL and by transaction volume. Its technology (originally Arbitrum Classic, then Arbitrum Nitro since 2022) compiles EVM bytecode to WASM for efficient fraud proofs, and it is the most EVM-compatible rollup on Ethereum — contracts that work on mainnet will almost always work unchanged on Arbitrum. The Arbitrum DAO governs the chain and the ARB token distributes governance rights.
Optimism, launched by OP Labs around the same time, is the other major optimistic rollup. Its Bedrock upgrade in 2023 rewrote the stack to be simpler, cheaper, and extensible, and the resulting OP Stack is what Base (Coinbase’s L2), Worldchain, Unichain, and several others are built on. The collection of OP-Stack chains is sometimes called the Superchain, and the vision is an ecosystem of interoperable rollups sharing a common tech stack and eventually shared security.
Base, which uses the OP Stack, launched in August 2023 and has become the single largest OP-Stack chain by both activity and user count, largely because Coinbase’s brand and user base gave it a head start that other L2s did not have. Base has become a default home for consumer-facing crypto apps in a way that Ethereum mainnet never quite managed, and its growth has been one of the more visible success stories of the rollup-centric scaling roadmap.
The Fraud Proof Situation
A thing worth knowing: for the first several years of their existence, most optimistic rollups did not actually have fraud proofs turned on. The security guarantee of “anyone can challenge a bad state” was conditional on the challenge mechanism existing and being functional, and in practice the team behind each rollup retained the ability to update state unilaterally during the period before fraud proofs were shipped. This was acknowledged publicly and tracked by L2BEAT, which gave early rollups lower trust ratings as a result.
Fraud proofs started shipping in 2024. Arbitrum activated its BoLD fraud-proof system on mainnet, and Optimism has been working toward similar deployment on its stack. As of 2026, the major optimistic rollups are progressively transitioning to a state where the fraud-proof mechanism is actually live and not just theoretical, which is the milestone that makes the “inherits Ethereum security” claim more than a hand-wave. L2BEAT is still the best place to check the current status of any specific rollup — it is the de facto auditor of where each L2 actually stands on decentralisation and trust-minimisation.
Why Optimistic Won First
ZK rollups have the cleaner long-term design — instant withdrawals, no challenge window, mathematically proven state transitions — but optimistic rollups shipped first because they are much easier to build. Generating a ZK proof for an EVM execution trace is computationally hard and the proving systems are still maturing. Running an optimistic rollup, by contrast, is mostly just running an EVM off-chain and periodically submitting state roots, which is well-understood engineering. The result has been that optimistic rollups have had a 2-3 year lead in production usage, and by the time ZK rollups got usable EVM compatibility, optimistic chains had already acquired the user base, the liquidity, and the developer momentum.
The expectation among most observers is that ZK will eventually displace optimistic as the dominant architecture, because instant withdrawals and stronger trust assumptions are clearly better once the prover overhead is manageable. But “eventually” may be several more years, and in the meantime optimistic rollups are where most of the actual Layer 2 activity lives.