Arbitrum Nitro (2022): A Second-Generation Optimistic Rollup

Authors Lee Bousfield, Rachel Bousfield, Chris Buckland, Ben Burgess, Joshua Colvin, Edward W. Felten, Steven Goldfeder, Daniel Goldman, Braden Huddleston, Harry Kalodner, Frederico Arnaud Lacs, Harry Ng, Aman Sanghi, Tristan Wilson, Valeria Yermakova, and Tsahi Zidenberg (Offchain Labs)
Year 2022
Work Arbitrum Nitro: A Second-Generation Optimistic Rollup
Field Ethereum Layer 2 optimistic rollup
Official source Official Nitro whitepaper in Arbitrum documentation

Editorial note: This page summarizes the protocol design and distinguishes it from developments after the paper. It is not a node-operator manual, security audit, or financial recommendation.

Summary

Arbitrum Nitro: A Second-Generation Optimistic Rollup describes the Layer 2 design that replaced Arbitrum Classic’s stack. Nitro is not a blockchain or token name. It is a technology stack used by chains such as Arbitrum One; Arbitrum One and Arbitrum Nova are separate chains with different data-availability choices.

The paper’s central design is to separate transaction ordering from deterministic execution, use Geth components for Ethereum compatibility, and compile the same state-transition logic for ordinary execution and dispute proving. Nitro aims to make routine execution more efficient while preserving a way to challenge and resolve disputed results on Ethereum.

How Nitro works

A Sequencer orders transactions and commits to that ordering. The State Transition Function then applies the transactions deterministically, one by one. Ordering and transaction-data availability still matter: other nodes need access to the inputs required to check the chain’s result.

Nitro’s execution core uses Go-Ethereum (Geth) code as a library. ArbOS surrounds it with Layer 2 functions such as processing Sequencer batches, accounting for Layer 1 gas, and supporting bridge messages. The phrase “Geth sandwich” describes this Geth-and-ArbOS arrangement; Nitro is not simply an ordinary standalone Geth node.

For normal operation, the State Transition Function is compiled to native code for a node’s machine. The same source code is compiled to WebAssembly for proving. A transformation produces Nitro’s proving format, called WAVM. If validators disagree about an execution result, an interactive proving protocol can narrow the dispute to a step that Layer 1 can check. A “fraud proof” here is a mechanism for demonstrating an incorrect result, not a guarantee that every transaction receives human review.

Security assumptions and AnyTrust

The paper’s security model assumes that Ethereum remains safe and live and that at least one Nitro protocol participant behaves honestly. “Optimistic” describes efficient execution when participants follow the incentives and no disputed result needs proving. Those guarantees depend on the protocol assumptions and a chain’s configuration; the Nitro name alone is not a security certification.

The paper also describes AnyTrust, a variant that sends data to an external data-availability committee. This can reduce data costs, but it adds a trust assumption that the committee will make data available under its rules. Arbitrum One, which posts transaction data to Ethereum, and Nova, which uses the AnyTrust approach, are not the same design.

Costs, performance, and claim limits

Arbitrum’s August 2022 migration announcement claimed throughput 7–10 times higher than before Nitro and said calldata compression would reduce the data posted to Layer 1. That was the project’s launch claim, not an independent benchmark guaranteeing every application will be faster or user fees will fall by a fixed ratio. Actual costs also depend on Ethereum gas prices, transaction size, network demand, and chain configuration.

Developments after the paper

The paper describes Nitro’s interactive dispute protocol as it existed at publication. Later Arbitrum documentation covers BoLD, an interactive dispute protocol that can enable permissionless validation and challenges on chains that have activated and configured it. BoLD does not mean every Nitro-based chain automatically has identical validator settings or guarantees.

The technical paper also does not establish an unrestricted open-source license for Nitro code. The current Nitro repository lists Business Source License 1.1 with an Additional Use Grant for specified uses. That software license is separate from the paper’s own reuse rights, which are not stated as a separate license in the source we reviewed.

Social Media Sentiment

A small April 2023 Reddit thread asked how Arbitrum One, Nova, and Nitro differ. Its most-upvoted answer clarified that Nitro is a technology stack, not a chain, then contrasted One’s Layer 1 data publication with Nova’s committee-based approach. The thread drew only around a dozen votes, so it is a useful example of beginner questions, not a representative picture of current sentiment or market opinion.

Last updated: 2026-10

Related Terms

See Also

Sources