On September 14, 2021, two weeks after Arbitrum One opened to the public, its sequencer stopped. A burst of transactions had triggered a bug. For roughly 45 minutes, the service that put transactions in order could not do its job. Offchain Labs, the company behind it, reported that funds were never at risk. For someone waiting to transact, however, the distinction was painfully practical: safe money could still be stuck waiting.
- Arbitrum moves execution off Ethereum to reduce congestion and fees.
- Orbit and Stylus give builders more control over chains and code.
- ZeroDev tackles wallets; enterprise services help institutions ship.
The incident supplies a useful introduction to Offchain, as the company now calls itself. Blockchain adoption is a procession of bottlenecks. Remove one and another becomes visible. Offchain began with Ethereum’s limited capacity. Its expanding portfolio suggests that capacity was only the first item on a considerably longer list.
The queue was the clue
The company grew out of Princeton research by Ed Felten, Steven Goldfeder and Harry Kalodner. They founded Offchain Labs in 2018. Felten, now chief scientist, had been a Princeton professor and Deputy US Chief Technology Officer. Goldfeder became CEO; Kalodner became CTO. Their subject was an awkward engineering question: how much work must a blockchain actually perform itself?
Ethereum offers a shared environment for programs called smart contracts. But shared execution becomes expensive when everybody wants it simultaneously. Arbitrum’s approach moves computation elsewhere, then posts transaction data and commitments back to Ethereum. Its optimistic rollup design allows incorrect claims to be challenged. Users share the cost of reaching the base chain rather than asking Ethereum to execute every operation directly.
The sequencer outage exposed a separate requirement. A system can preserve the rules protecting money while its preferred transaction-ordering service stops responding. Offchain fixed the bug and published a postmortem. For builders, that distinction belongs in the product specification: keeping assets safe and keeping an application available require different checks.
Bring the code you already have
Arbitrum One opened publicly on August 31, 2021. Its attraction was partly what developers could keep. Ethereum compatibility let teams bring existing contracts and familiar tools. Uniswap, SushiSwap and Aave were among the application partners identified by investor Lightspeed at launch. Switching infrastructure is easier to contemplate when it does not require replacing the workshop.
Ed Felten
Steven Goldfeder
Harry KalodnerOffchain subsequently widened the workshop. Nitro became the underlying technology stack. Orbit gave teams tools to create custom chains, with choices such as gas tokens and governance. Stylus, activated on mainnet in September 2024, added WebAssembly contracts, allowing languages including Rust and C++ to work alongside Ethereum’s established contract environment.
These products put Offchain among Ethereum scaling and chain-development alternatives such as Optimism’s OP Stack, Polygon and zkSync. Its particular combination is compatibility, configurable infrastructure and additional programming languages. A team choosing between them still has to examine liquidity, tooling, security assumptions and operating costs. A feature list cannot make that decision.
The wallet gets a vote
In 2022, Offchain acquired Prysmatic Labs, the team behind Prysm, an Ethereum consensus client. That extended its work down into the software helping Ethereum agree on its own history. In August 2025, it acquired ZeroDev, reaching in the other direction: toward the account a person actually uses.
“account abstraction is no longer the sole bottleneck for Web3 adoption”ZeroDev, announcing its acquisition, August 2025
ZeroDev’s announcement explained the reasoning. Smart accounts alone could not fix wallet interfaces, bridging and other awkward handoffs. Joining a larger infrastructure team offered a way to address those pieces together. At acquisition, ZeroDev reported powering more than five million smart accounts across more than 50 chains. Those are accounts, not five million distinct customers.
For an application developer, the practical possibilities include sponsoring transaction fees, batching actions and granting narrowly defined permissions. A game can authorize a limited session instead of interrupting every move with another approval. A financial app can design account recovery and access around its customers. The implementation still needs careful permission design; a smoother interface does not excuse an overpowered account.
Someone still has to pay
Offchain also offers embedded engineering and infrastructure services to enterprises. Its account of Robinhood Chain describes technical guidance, Arbitrum expertise and production support. The customer buys help getting a financial system into operation, including decisions that extend beyond deploying contracts.
The funding history needs careful reading. Lightspeed identifies the August 2021 Series B it led as $100 million. Contemporary announcements also carried a $120 million funding headline. The company’s current site reports approximately $124 million raised. None of these figures is a transaction-fee quote or a measure of annual sales.
The Expansion Program’s stated 10% share of qualifying chains’ net revenue. This is an ecosystem arrangement, not a corporate sales figure.
Likewise, Arbitrum’s network economics and Offchain’s accounts deserve separate treatment. The company, the DAO and the Foundation have different roles. A prospective chain operator must budget for settlement, infrastructure and engineering alongside any licensing obligations. Launching a custom chain transfers operational decisions to the builder, including decisions a shared network previously absorbed.
The chain that got smaller
Nova makes the trade-offs unusually concrete. Launched for high-volume uses such as gaming and social applications, it later faced a less persuasive economic case. Arbitrum governance approved moving it toward maintenance, with a June-to-September 2026 migration window. Reduced usage and strategic value no longer justified the same operational overhead.
There is another present-tense qualification: official documentation currently reports a temporary security pause on new Stylus activations on One and Nova. Developers must check that notice before planning a launch. Shipping history does not guarantee that every feature is available today.
The transferable lesson is to follow the obstruction. Preserve familiar developer tools. Watch the user’s entire journey. Retire infrastructure whose costs outgrow its purpose. Offchain’s portfolio now stretches from consensus to accounts because a cheaper blockchain still needs people who can comfortably use it. The next bottleneck may be waiting at the very next click.