Ethereum’s layer-2 scaling ecosystem has matured from a theoretical solution to Ethereum’s gas fee problem into a functioning multi-network ecosystem that processes more transactions than Ethereum’s base layer on most days. The four networks that dominate L2 activity — Arbitrum, Base, Optimism, and zkSync Era — collectively process several million transactions per day, have attracted tens of billions in total value locked, and are home to the majority of DeFi and consumer DApp activity that Ethereum users are conducting at scale. The growth narrative is accurate and well-documented.
What is less well-documented, and significantly more differentiated across networks, is the economic sustainability of L2 operations. Running an Ethereum L2 involves paying fees to Ethereum’s base layer for posting transaction data (data availability costs), operating sequencer infrastructure, and funding the development and security programs that maintain the network’s operation. The revenue that offsets these costs comes primarily from the spread between the gas fees users pay on L2 and the actual cost of settling those transactions to Ethereum’s base layer — the “sequencer margin” that is the core economic unit of L2 operations.
The sequencer margin, and whether it is sufficient to sustain L2 operations profitably, varies dramatically across networks and has been significantly affected by Ethereum’s EIP-4844 (proto-danksharding) implementation in March 2024, which reduced the cost of posting L2 transaction data to Ethereum by approximately 90%. The data availability cost reduction was good for L2 users — it enabled lower transaction fees — but it compressed the unit economics of L2 sequencer operations significantly. Networks that had built cost structures around the pre-4844 data availability pricing needed to grow volume substantially to maintain revenue at lower per-transaction margins.
Arbitrum: The Revenue Leader With a Governance Question
Arbitrum generates the largest absolute revenue of any Ethereum L2, driven by the highest transaction volume and the longest-established DeFi ecosystem of any optimistic rollup. Arbitrum One and Arbitrum Nova together process several hundred million transactions monthly, with DeFi protocol TVL that includes significant positions from established protocols including GMX, Uniswap, Aave, and Camelot.
Arbitrum’s protocol revenue — the sequencer margin after data availability costs — has been consistently tracked by Token Terminal and DefiLlama and shows a network that is operationally profitable: fee revenue exceeds the direct costs of sequencer operation and data posting on most measurement periods. The ARB token, however, trades at a significant discount to the implied value that would be suggested by Arbitrum’s revenue if it were fully accruing to token holders. The disconnect reflects the fact that Arbitrum’s governance has not yet implemented a fee-sharing mechanism that would route sequencer margin to ARB stakers or token holders — a governance decision that has been proposed and debated but not executed.
The governance question matters because Arbitrum DAO controls a substantial treasury — approximately $3–4 billion in ARB tokens at various price levels — and has been spending on grants and ecosystem development at a pace that has generated scrutiny from some token holders. The combination of protocol-level profitability and governance-level spending creates a financial picture where the network is sustainable at the protocol layer but may be consuming treasury at the governance layer faster than the protocol revenue supports. Understanding Arbitrum’s economics requires reading both the sequencer margin data and the DAO treasury data — they are telling different stories about the same network.
Base: Coinbase’s L2 and What Its Revenue Model Reveals
Base, launched by Coinbase in August 2023, has grown to become the highest-transaction-volume L2 by daily activity in many measurement periods, driven by consumer DApp adoption, memecoin trading, and the social-application ecosystem that has developed on the network. Base’s economic model is distinct from Arbitrum’s in one critical structural way: Base does not have a native token, and all sequencer revenue accrues directly to Coinbase rather than to a protocol treasury or token holders.
This makes Base the most transparent example of what L2 sequencer economics look like when there is no token distribution to obscure the cash flow. Coinbase has disclosed that Base generates meaningful revenue for the company — sequencer margin that has been described in investor presentations as a growing contribution to Coinbase’s net revenue. The specific numbers are embedded in Coinbase’s reported financials rather than in a standalone protocol disclosure, but analysts tracking Base’s transaction volume and estimated sequencer margin have calculated quarterly revenue contributions that are material to Coinbase’s technology-segment reporting.
Base’s no-token model has implications for the broader L2 ecosystem. It demonstrates that an L2 can sustain meaningful transaction volume and generate real revenue without a token launch — removing one of the assumed incentive mechanisms for L2 user acquisition. It also demonstrates that a corporate parent with distribution (Coinbase’s 100+ million registered users) can successfully seed L2 adoption without the grant programs and liquidity mining that Arbitrum and Optimism used to attract initial users.
Optimism: The OP Stack and the Network Effect Question
Optimism’s strategic position in 2026 is defined more by the OP Stack — its open-source L2 development framework — than by Optimism Mainnet’s own transaction volume. The OP Stack is the technical foundation for Base, and for several other networks including Zora, Mode, and the emerging “Superchain” ecosystem that OP Labs is building. The thesis is that Optimism’s value is network-level rather than chain-level: as more chains deploy on the OP Stack, Optimism’s governance position and the potential for cross-chain fee-sharing within the Superchain increases.
The economic tension in this model is that Optimism Mainnet’s own transaction volume has been partially cannibalised by Base — users who might otherwise have been on Optimism Mainnet are on Base instead, where Coinbase’s distribution has driven adoption. Optimism Mainnet’s sequencer revenue is lower than Arbitrum’s and has grown more slowly. The OP token’s value case therefore depends more heavily on the Superchain fee-sharing thesis than on Optimism Mainnet’s direct financial performance.
The Superchain fee-sharing mechanism — where a percentage of sequencer revenue from all OP Stack chains flows to the Optimism Collective’s treasury — has been proposed and partially implemented but is not yet at the scale that would make it the dominant value driver for OP tokens. The bet investors in OP are making is that the Superchain ecosystem grows to a scale where the collective fee-sharing produces Optimism Collective treasury inflows that justify the OP token’s market cap. This is a longer-horizon, more uncertain bet than Arbitrum’s “already profitable sequencer, unresolved governance distribution” story.
zkSync Era: The ZK Rollup Economics and What They Reveal
zkSync Era, developed by Matter Labs, represents the largest zero-knowledge rollup by TVL and transaction volume. ZK rollups have a different cost structure than optimistic rollups: they require generating cryptographic proofs for each batch of transactions, which adds compute cost that optimistic rollups do not incur. The trade-off is that ZK rollups do not need a fraud proof period (the 7-day challenge window that optimistic rollups require before assets can be withdrawn), making finality faster and potentially enabling more use cases that require real-time settlement certainty.
zkSync Era’s economics in 2026 are characterised by proof generation costs that are significant but declining as hardware efficiency improves and proof systems are optimised. The network has been working toward proof generation cost structures that allow sequencer margins comparable to optimistic rollups, but the proof cost remains a meaningful component of zkSync Era’s cost base that Arbitrum and Base do not have. The ZK technology premium — the benefit of faster finality and cryptographic security guarantees — has not yet translated into materially higher fees that would offset the higher cost structure. zkSync Era competes on fees with networks that have lower cost bases, which has compressed its sequencer margin relative to its proof generation costs.
The ZK ecosystem thesis is that proof generation costs will continue declining — following a trajectory similar to how storage costs have declined — to the point where the ZK technology premium becomes costless and the finality advantage becomes a genuine differentiator. The timeline on that cost trajectory is the primary uncertainty in evaluating ZK rollup economics in 2026.
Data Availability and the Celestia Question
One structural development that cuts across all L2 economics in 2026 is the emergence of alternative data availability layers — primarily Celestia and EigenDA — that offer lower data availability costs than Ethereum’s own blob storage introduced in EIP-4844. Several L2 networks have begun using or are evaluating alternative data availability layers, which would further reduce their operating costs but would also change their relationship with Ethereum’s security model.
The economics are significant: data availability costs on alternative layers can be 90%+ lower than Ethereum blob costs, which are themselves 90% lower than pre-EIP-4844 calldata costs. An L2 that uses Celestia for data availability rather than Ethereum blobs can potentially offer much lower transaction fees or operate at higher sequencer margins. The trade-off is that transactions settled on an alternative data availability layer do not inherit Ethereum’s full security model — they depend instead on the security of the data availability layer, which is a different and generally lower security guarantee than Ethereum’s validator set provides.
The L2 economic story in 2026 is therefore a dynamic one: the cost structure of L2 operations is continuing to decrease as data availability alternatives mature, which benefits users through lower fees but compresses sequencer margins in ways that affect each network’s treasury sustainability and token economics differently. Reading the on-chain financial data for L2 networks — sequencer revenue, data availability costs, treasury balances — is the only way to track these economics accurately as they evolve.
FAQ
What is the sequencer margin for an Ethereum L2? The sequencer margin is the spread between the gas fees users pay on the L2 and the actual cost of settling those transactions to Ethereum’s base layer (data availability costs). It is the core revenue unit of L2 operations. After EIP-4844 reduced data availability costs by approximately 90%, sequencer margins per transaction decreased significantly, requiring networks to grow volume to maintain absolute revenue.
Which L2 is most financially sustainable? Arbitrum generates the largest absolute revenue and is operationally profitable at the sequencer level. Base generates significant revenue for Coinbase but doesn’t have a protocol token through which that revenue accrues to external stakeholders. Optimism’s financial case depends increasingly on the Superchain fee-sharing thesis. zkSync Era’s ZK proof costs make its margin structure more complex than optimistic rollups at current proof generation costs.
Why does Base not have a token? Coinbase chose to launch Base without a native token, with sequencer revenue accruing directly to Coinbase. This makes Base the clearest example of L2 sequencer economics without the distortion of token distribution programs. It demonstrates that L2 networks can grow without a token launch when the operator has sufficient distribution advantages.
What is the OP Stack and why does it matter for Optimism’s economics? The OP Stack is Optimism’s open-source L2 development framework used by Base and other networks in the emerging “Superchain” ecosystem. Optimism’s thesis is that as more chains deploy on the OP Stack, a Superchain fee-sharing mechanism will route collective sequencer revenue to the Optimism Collective treasury. This is a longer-horizon bet than direct Optimism Mainnet sequencer revenue.
What are alternative data availability layers and how do they affect L2 economics? Networks like Celestia and EigenDA offer data availability at costs 90%+ lower than Ethereum blob storage. L2s that use these alternatives can offer lower transaction fees or maintain higher margins, but at the cost of not inheriting Ethereum’s full security model. The adoption of alternative data availability continues to evolve L2 cost structures in ways that affect each network’s economics differently.
Sources
- Token Terminal — Arbitrum protocol revenue, fees, and financial metrics
- DeFi Llama — Ethereum L2 chains: TVL, fees, and revenue comparison
- Coinbase — Base L2 revenue contribution in investor materials 2026
- Optimism — Superchain fee sharing: mechanics and current implementation status
- zkSync — ZK proof cost economics and roadmap update 2026
