An optimistic rollup is a Layer 2 blockchain that batches thousands of transactions offchain, posts them to Ethereum, and assumes they are valid unless someone proves otherwise during a challenge window. It is the cryptoeconomic opposite of a ZK rollup: instead of proving every batch is correct upfront, optimistic rollups rely on fraud proofs and watchers to catch cheating after the fact. That tradeoff — simpler tech, harder withdrawal UX — is what Arbitrum and Optimism bet on, and it is why they still settle the majority of Layer 2 value in 2026.
​

This guide covers how optimistic rollups work, the role of fraud proofs and challenge windows, how Arbitrum and Optimism diverge in practice, where Base and the OP Superchain fit, and how optimistic rollups stack up against ZK rollups on cost, finality, and security. If you want the Layer 2 primer first, start with what is a Layer 2 blockchain.
​

Arbitrum and Optimism both launched in 2021 and have spent four years hardening fraud proofs, sequencer performance, and exit games. By 2026 they collectively settle tens of billions in TVL, run the most mature EVM-equivalent environments outside Ethereum mainnet, and host most of the mainstream DeFi TVL on L2. Live dashboards at L2Beat and Dune's L2 metrics track the split. The headline: optimistic rollups lead in TVL, ZK rollups are gaining fast, and most new chains launching in 2026 pick ZK — but the installed base of optimistic infrastructure is enormous.
​

Every optimistic rollup has three pieces: a sequencer that orders and executes transactions offchain, a batcher that posts compressed transaction data plus state roots to Ethereum, and a fraud-proof system that lets any watcher challenge a dishonest state root. The sequencer is the performance layer (fast execution, low latency); the Ethereum contracts are the security layer (final settlement, fraud adjudication, deposit/withdrawal accounting). The Ethereum Foundation's optimistic rollup docs walk through each piece with diagrams.
​

The critical word is "optimistic." The sequencer posts a state root without any cryptographic proof that it is correct. Ethereum accepts it provisionally. For the next several days — the challenge window, typically seven days — any watcher can dispute the state root by submitting a fraud proof. If the proof verifies, the fraudulent sequencer loses its bond and the state rolls back. If the window closes without a successful challenge, the state is final.
​

Fraud proofs come in two main flavors: single-round (the fraud proof includes the full execution trace, Ethereum replays it) and interactive multi-round (Ethereum adjudicates a dispute narrowed down to a single disputed opcode). Arbitrum's interactive fraud-proof protocol uses multi-round dissection, which makes proof cost roughly independent of the length of the disputed computation. Optimism shipped its own fault-proof system in 2024, detailed in the OP Stack fault proofs documentation, after running for years with a privileged roll-back mechanism as a stopgap.
​

The existence of working fraud proofs is not theoretical — it is the security model. L2Beat's rollup risk framework scores every rollup on whether fraud proofs are live, who can run them, and what the exit game looks like if the sequencer censors a withdrawal. In 2026, both Arbitrum and Optimism have permissionless fraud proofs in production, a milestone that took years to reach.
​

The challenge window is the tradeoff optimistic rollups make. Because fraud is proven after the fact, Ethereum has to wait long enough that any honest watcher has time to detect and submit a challenge. Seven days is the consensus value — long enough to cover weekends, eclipse attacks, and censorship attempts. During that window, a withdrawal from the rollup back to Ethereum is not yet final, and users either wait it out or pay a liquidity provider to front the funds against the pending withdrawal.
​

In practice, third-party bridges and intent-based routers handle almost all real-world withdrawal UX. Users rarely wait the full seven days because intent-based protocols let a solver take the pending withdrawal as collateral and front destination-chain liquidity immediately. The seven-day wait becomes a settlement delay for the solver, not a UX problem for the user.
​

Like ZK rollups, optimistic rollups post compressed transaction data to Ethereum so anyone can reconstruct state and, if needed, submit a fraud proof. Before Ethereum's Cancun upgrade, that data went in calldata at full gas price. EIP-4844 introduced blobs, a cheaper data lane specifically sized for rollup payloads, and cut optimistic rollup fees by roughly an order of magnitude overnight. Every major optimistic rollup migrated to blobs within weeks of the upgrade landing.
​

Arbitrum is the largest optimistic rollup by TVL and transaction volume. It runs Arbitrum Nitro, a WASM-based execution environment that compiles Geth down to WASM for fraud-proof adjudication. This lets Arbitrum inherit Ethereum's codebase almost directly — extremely high EVM equivalence — and makes its fraud-proof protocol independent of the specific VM. The Arbitrum developer portal is the canonical reference. Arbitrum also ships Arbitrum Orbit, a framework for launching Arbitrum-derived chains that settle back to Arbitrum One, Ethereum, or custom data-availability layers.
​

Optimism ships the OP Stack, an open-source rollup framework that other teams fork to launch their own chains. Optimism itself, Base, Mode, Zora, and a long tail of app-chains all run the OP Stack, sharing a common codebase and — increasingly — shared sequencing through the Superchain. The OP Stack documentation covers the architecture, and the fault-proof documentation covers how challenge games work now that permissionless fraud proofs are live.
​

Base is Coinbase's OP Stack chain and, by 2026, one of the highest-throughput L2s in production. Its stablecoin footprint is particularly deep — USDC is native on Base and most retail onramp flows settle there before moving elsewhere. The broader Superchain concept is Optimism's long-term pitch: a set of interoperable OP Stack chains that share security, a governance model, and eventually native cross-chain messaging. Chains that join the Superchain inherit a common fraud-proof system and contribute protocol revenue to the collective.
​

Technical finality on an optimistic rollup is seven days. Practical finality depends on who you trust. Intent-based bridges close the gap by letting solvers front destination-chain liquidity against pending rollup withdrawals — users see fast withdrawals, solvers carry the timing risk. The L2Beat summary tracks each rollup's "state validation" score, which is the closest thing to an apples-to-apples finality comparison.
​

The cryptoeconomic contrast is sharp. Optimistic rollups assume validity and use fraud proofs as an afterward-audit; ZK rollups prove validity upfront with a cryptographic proof. The tradeoffs that follow:

The full side-by-side is in our ZK rollup 2026 guide. In 2026 the industry consensus is that ZK is the long-term destination, but optimistic rollups' installed base, ecosystem depth, and EVM-equivalence advantages mean they remain the default for new deployments that prioritize developer velocity.
​

Three security pillars underpin optimistic rollups: (1) the fraud-proof protocol must be correct and permissionlessly runnable, (2) at least one honest watcher must monitor every batch within the challenge window, (3) data availability must be preserved so watchers can reconstruct state. All three are testable and all three are live in 2026 on Arbitrum and Optimism. L2Beat's risk framework scores each rollup on these dimensions.
​

The remaining honest risks: sequencer centralization (most L2s still run a single sequencer, though decentralization is on the roadmap), MEV capture by the sequencer, and governance upgradeability (most rollup contracts can be upgraded by a security council, a necessary evil during the early years). ethresear.ch is where the frontier of these discussions happens in public.
​

Optimistic rollups carry most of the stablecoin TVL on L2 as of 2026. USDC is native on Arbitrum, Optimism, and Base; USDT is deeply liquid on Arbitrum; and the OP Stack's shared infrastructure means USDC flows between Base and Optimism with minimal friction. For teams moving stablecoins between optimistic rollups, ZK rollups, and L1 — which is most production teams — Eco's stablecoin execution network routes across 15 chains including Arbitrum, Optimism, and Base, handling solver selection, liquidity, and settlement under a single API.
​

After EIP-4844, simple transfers on Arbitrum and Optimism run in the single-digit cents; DeFi interactions are typically 10-50 cents. L2Fees.info tracks representative transactions live. The sequencer and data-availability costs scale differently from ZK rollups: optimistic rollups pay for data but not for proving, while ZK rollups pay for both (though proving costs continue to drop). Under sustained load, the two families end up in a similar fee range, with specific workloads favoring one or the other.
​

Pick an optimistic rollup if you need: maximum EVM equivalence with minimal contract porting, access to the deepest L2 liquidity pools and developer ecosystems, mature tooling (Hardhat, Foundry, block explorers, indexers all work out of the box), or the largest stablecoin liquidity on L2. Pick a ZK rollup if fast finality, cryptographic security guarantees, or long-term post-quantum concerns matter more. Most production teams in 2026 end up deploying on both and routing liquidity between them.
​

Optimistic rollups assume transactions are valid and use fraud proofs plus a seven-day challenge window to catch cheating. ZK rollups prove validity upfront with a cryptographic proof Ethereum verifies per batch. Optimistic rollups have better EVM equivalence today; ZK rollups have faster finality and stronger long-term security assumptions.
​

The challenge window must be long enough for an honest watcher to detect a fraudulent state root and submit a fraud proof. Seven days covers weekends, network outages, and censorship attempts. In practice, intent-based bridges and liquidity providers make user-facing withdrawals near-instant by fronting funds against the pending rollup withdrawal.
​

Yes. Both chains shipped permissionless fraud proofs after years of running in a trusted-operator mode. L2Beat's risk framework tracks the status of fraud-proof systems across every optimistic rollup and shows both Arbitrum and Optimism at the highest security tier.
​

The OP Stack is Optimism's open-source rollup framework. Base, Mode, Zora, and many others fork the OP Stack to launch their own chains. The Superchain is the initiative to give all OP Stack chains shared security, shared sequencing, and native cross-chain messaging. Base is the largest OP Stack chain by TVL and stablecoin volume.
​

EIP-4844 introduced blobs, a cheaper data-availability lane for rollup payloads, and cut optimistic rollup fees by roughly 10x overnight. Simple transfers on Arbitrum and Optimism now run in the single-digit cents. L2Fees.info tracks live fee comparisons.
​

Most rollup contracts are upgradeable by a security council during the early years — a necessary concession while fraud-proof systems mature. Upgrade keys, timelocks, and council composition are tracked by L2Beat's risk framework. The long-term direction is to harden and eventually remove upgrade keys once the fraud-proof system has been live long enough to be trusted unchanged.
​