The L2 Fee War: How Ethereum Rollups Compete on Cost, Speed, and Decentralization

Ethereum's Layer 2 landscape has entered a new phase. After EIP-4844 slashed data posting costs by over 90% in March 2024, and the Pectra upgrade doubled blob capacity in May 2025, the major rollups are locked in a fee war where transaction costs have collapsed to fractions of a cent. The question is no longer whether L2s can be cheap: it is how they sustain themselves when fees approach zero.

This article maps the competitive dynamics across Arbitrum, Optimism, Base, zkSync Era, Scroll, and Linea. We compare their fee structures, revenue models, grant strategies, and sequencer economics to understand where the L2 market is heading and what it means for users building on these networks.

The Post-Blob Fee Landscape

Before the Dencun upgrade (March 13, 2024), rollups posted transaction data as calldata on Ethereum L1, competing with every other transaction for block space. A typical L2 transaction cost $0.20 to $1.00, with the bulk going to L1 data availability fees.

EIP-4844 introduced a separate blob fee market specifically for rollup data. The effect was immediate: L2 fees dropped 90 to 95% within days. Where a token swap on Arbitrum might have cost $0.50, it fell below $0.05.

Pectra and Fusaka: Expanding and Repricing Blobs

The Pectra upgrade (May 7, 2025) pushed costs even lower. EIP-7691 doubled the blob target from 3 to 6 per block and raised the maximum from 6 to 9. Daily blob capacity expanded from roughly 5.5 GB to 8.15 GB, while the fee adjustment curve became more aggressive on the downside: blob fees now decrease approximately 14.5% per block when demand is low.

The result: daily rollup blob spending fell by roughly 51% despite a 20.8% increase in blob purchases. L2s were posting more data to Ethereum while paying less for it. By late 2025, blob fees had spiraled down to 1 wei per blob: effectively free.

The Fusaka upgrade (December 2025) responded by introducing EIP-7918, a blob base fee floor that prevents blob costs from dropping below 1/15,258th of the L1 execution base fee. This recalibration addressed a growing concern that L2s were free-riding on Ethereum's security without contributing meaningful fee revenue back to validators.

The Ethereum tradeoff: L2 payments to Ethereum L1 dropped from $113 million in 2024 to roughly $10 million in 2025. Ethereum chose to subsidize its L2 ecosystem at the expense of L1 fee revenue, betting that rollup adoption will drive long-term demand for ETH as a settlement asset. Fusaka's fee floor represents a partial course correction.

Current Fee Comparison Across Major L2s

As of mid-2026, the spread between the cheapest and most expensive major L2 is only a few cents for most operations. The following table shows median fees across the leading rollups.

The numbers reveal a near-commodity market. Simple transfers cost under $0.06 on every major L2. Swap fees run 3 to 5x higher due to increased computation, but even the most expensive L2 swap is a fraction of Ethereum L1 costs. For stablecoin transfers, Base leads at roughly $0.002 per transaction: functionally free for most use cases.

Competitive Strategies: How Each L2 Differentiates

With fees converging toward zero, the rollup wars have shifted from technical cost optimization to ecosystem strategy. Each major L2 has adopted a distinct approach.

Base: The Coinbase Volume Play

Base dominates L2 activity. With approximately 12.9 million daily transactions as of mid-2026, it processes over 60% of all L2 transactions and more daily volume than Ethereum mainnet itself. The strategy is straightforward: Coinbase funnels its 9.3 million monthly active trading users directly onto Base, creating a distribution advantage no other L2 can match.

Base's sequencer generated roughly $75 million in revenue during 2025, representing 58% of the entire L2 sector's fee income. After L1 data posting costs, Base retained approximately $55 million in net profit: the only L2 consistently operating in the black. By September 2026, Base has overtaken Arbitrum in total value locked, reaching over $13 billion on L2BEAT.

In February 2026, Coinbase announced that Base would migrate off the OP Stack to its own proprietary infrastructure, signaling a shift from ecosystem participant to independent platform. The OP token dropped 28% within 48 hours of the announcement. Base had been contributing 2.5% of sequencer revenue (or 15% of net profits, whichever is greater) to the Optimism Collective, and its departure removes the Superchain's largest revenue source.

Arbitrum: Grant-Funded Ecosystem Growth

Arbitrum maintains one of the largest L2 TVL positions at over $11 billion and has invested heavily in ecosystem incentives to defend its competitive standing. The Arbitrum DAO has deployed roughly 154 million ARB tokens across three major programs.

• STIP (Short-Term Incentive Program): 71.4 million ARB distributed to 56 projects
• LTIPP (Long-Term Incentive Pilot Program): 45 million ARB, approved with 97.5% DAO support in January 2024
• STIP Bridge: 37.6 million ARB for protocols that missed the initial round

The DAO has since expanded its commitments further: a Gaming Catalyst Program (225 million ARB over three years) and a DeFi Renaissance Incentive Program (80 million ARB). The total incentive spend makes Arbitrum the most aggressive L2 in using token-denominated grants to attract and retain protocols.

Arbitrum was also the first major L2 to reach Stage 1 on L2BEAT's classification, meaning it has permissionless fraud proofs and a functioning challenge mechanism. Base and OP Mainnet have since also reached Stage 1. This decentralization progress is a meaningful differentiator for users who prioritize security guarantees over raw cost savings.

Optimism: The Superchain Franchise

Rather than competing on a single chain, Optimism has pursued a franchise model. The OP Stack powers over 30 chains in production, including World Chain, Unichain, Ink, Soneium, Mode, and Zora. Collectively, OP Stack chains have accounted for roughly 62% of all L2 transactions.

The Superchain model generates revenue through sequencer profit-sharing: each chain contributes a percentage of its sequencer income to the Optimism Collective. However, the strategy took a severe hit when Base announced its migration off the OP Stack in February 2026. Base had been responsible for over 96% of Superchain gas fees. OP Labs subsequently laid off 20% of staff in March 2026, with CEO Jing Wang signaling a shift toward "doing fewer things well." The long-term viability of the franchise model depends on whether enough chains join to replace Base's lost revenue.

zkSync Era: The ZK Endgame Bet

zkSync Era is betting that zero-knowledge proofs will eventually provide superior security and cost properties. ZK rollups replace the 7-day challenge period of optimistic rollups with cryptographic validity proofs, enabling faster finality and trustless withdrawal verification.

The tradeoff is cost. ZK proof generation requires significant computation, which rollup operators amortize across many transactions. At current volumes, this amortization keeps zkSync Era's fees slightly higher than optimistic competitors. As transaction throughput grows, per-transaction proof costs should decrease: the ZK bet is fundamentally a bet on scale.

zkSync's ZK token airdrop in June 2024 distributed 3.675 billion tokens (17.5% of supply) to 695,232 wallets, but the reception was largely negative. Critics flagged Sybil farming and inconsistent eligibility criteria. The chain remains at Stage 0 on L2BEAT's security classification, meaning it is still operator-dependent for critical safety functions.

Scroll and Linea: Fast Followers

Scroll and Linea represent the second tier of ZK rollups. Both maintain competitive fee structures (Linea at $0.04 and Scroll at $0.06 for ERC-20 transfers) and have attracted TVL in the hundreds of millions on L2BEAT's value-secured metric. Their strategies focus on EVM compatibility and developer tooling rather than novel economic models, aiming to capture market share through execution rather than tokenomics or distribution advantages.

The Race to Zero: Where Does Revenue Come From?

Total L2 sequencer revenue in 2025 was $129 million, down 53% from $277 million in 2024. The decline reflects how aggressively blob fees compressed both costs and the margins L2s could charge above cost.

The concentration is stark: Base alone captures nearly 60% of all L2 sequencer revenue. If you exclude Base, the entire remaining L2 sector generated just $54 million in 2025: a modest sum for dozens of competing networks.

Revenue Sources Beyond Execution Fees

As execution fees approach commodity pricing, L2s are exploring alternative revenue streams.

• Priority fees: on Base, priority fees account for approximately 86% of sequencer revenue ($156,138 per day average), dwarfing base fees. Users pay extra for transaction ordering, effectively creating a premium lane even on a low-fee chain
• MEV extraction: centralized sequencers can capture MEV from transaction ordering. While Ethereum L1 has a mature MEV supply chain via MEV-Boost and PBS, L2 MEV remains largely opaque: most sequencers self-order transactions without a public auction mechanism
• Ecosystem revenue sharing: applications on Base generated $369.9 million in revenue in 2025, creating indirect value for Coinbase through platform lock-in even when the chain itself operates at thin margins
• Token value capture: Arbitrum and Optimism use their native tokens (ARB, OP) for governance and incentive distribution. The tokens accrue value through ecosystem growth rather than direct fee capture, though neither has implemented fee-switch mechanisms that would direct protocol revenue to token holders

The sequencer centralization problem: Every major Ethereum L2 currently runs a single, centralized sequencer. This gives the sequencer operator full control over transaction ordering, MEV extraction, and censorship. While most L2s have roadmaps for sequencer decentralization, none have delivered a fully decentralized sequencer in production. Users trust the operator not to exploit ordering advantages.

Stablecoin Transfers: The Payments Lens

For payment applications, the L2 fee war has a direct practical impact. Sending USDC or USDT on Base costs roughly $0.002: less than a tenth of a cent. Even on the more expensive zkSync Era, a stablecoin transfer runs about $0.05. Compare this to Ethereum L1 at $2 to $15, or traditional wire transfers at $25 to $50.

These economics have made L2s viable for stablecoin payment rails. Base in particular has processed over $32 trillion in cumulative stablecoin volume through mid-2026, driven by Coinbase's integration pipeline. For cross-border remittances and B2B invoice settlement, the fee compression makes L2 stablecoins competitive with traditional payment rails on pure cost.

However, the cheapest L2 stablecoin transfer still carries complexity that traditional rails do not: users need to bridge assets to the right L2, manage gas tokens for fees, and navigate liquidity fragmentation across chains. Cross-chain USDC transfers via services like Across add 0.04% relayer fees, and bridge aggregators introduce additional smart contract risk. The total cost of a stablecoin payment is not just the L2 execution fee: it includes bridging, gas token acquisition, and the cognitive overhead of chain selection.

The Consolidation Thesis

L2BEAT tracks 73 rollups with combined TVL exceeding $48 billion. But the top three chains (Base, Arbitrum, and OP Mainnet) control approximately 90% of transaction volume. The long tail of L2s: chains with sub-$100 million TVL and minimal transaction activity: increasingly resembles what some analysts have called "zombie chains." Kinto has shut down, Loopring closed its wallet service, and Blast's TVL collapsed by 97% from its peak.

The fee war accelerates this consolidation. When every L2 charges under $0.05 for a transfer, fees stop being a differentiator. What matters instead is distribution (Coinbase's user base for Base), ecosystem depth (Arbitrum's DeFi liquidity), security maturity (Stage 1 status on L2BEAT), or developer tooling (OP Stack's modularity). Chains without at least one strong competitive moat will struggle to attract users in a commodity fee environment.

The App-Chain Counterargument

Not everyone agrees that consolidation is inevitable. The app-chain thesis argues that large applications will launch their own dedicated L2s to control sequencing, capture MEV, and customize gas economics. Unichain (Uniswap's L2), World Chain (Worldcoin), and Soneium (Sony) represent this trend. MegaETH, which launched in February 2026 targeting 100,000 TPS with sub-10ms block times, represents the performance-maximalist variant: purpose-built infrastructure that sacrifices generality for speed. If successful, the L2 landscape may fragment further into purpose-built chains rather than consolidating around general-purpose platforms.

Implications for Ethereum's Roadmap

Ethereum's rollup-centric roadmap deliberately pushes execution to L2s while L1 focuses on settlement and data availability. The fee war validates this architecture: users are transacting on L2s at a fraction of L1 cost, and over 65% of new smart contracts in 2025 were deployed on L2 rather than L1. But it creates a revenue problem for Ethereum validators.

With L2 payments to L1 collapsing from $113 million to $10 million in a single year, daily ETH burn has dropped to 50 to 70 ETH per day, well below the roughly 1,700 ETH daily issuance. ETH has become mildly inflationary as a result. Future upgrades like danksharding will expand blob capacity further, though Fusaka's fee floor signals that Ethereum is trying to balance cheap L2 data with sustainable validator economics.

What This Means for Payments

The Ethereum L2 fee war has produced genuinely cheap transaction infrastructure. Sub-cent stablecoin transfers are now the norm, not the exception. For developers building payment applications, the execution cost is effectively solved.

But cheap execution does not automatically mean simple payments. The Ethereum L2 ecosystem introduces complexity that payment use cases are particularly sensitive to: which chain to deploy on, how to handle cross-chain liquidity, which bridge to trust, how to abstract gas fees from end users via paymasters or account abstraction. Each layer of abstraction adds latency, cost, and potential points of failure.

This is where the L2 approach to scaling creates a structural tension for payments. Payment systems value simplicity, finality, and predictability. The Ethereum L2 stack offers low fees but requires navigating a fragmented landscape of competing chains, bridging protocols, and sequencer trust assumptions.

The Bitcoin Alternative

Spark takes a different approach to the low-cost payment problem. Rather than stacking rollup layers on top of a base chain, Spark enables instant, self-custodial transfers directly on Bitcoin using statechains and FROST threshold signatures. There is no separate fee market, no blob economics, and no sequencer revenue extraction. Transfers settle between parties without on-chain transactions, and users can exit to Bitcoin L1 at any time without permission.

For payment applications, this eliminates the chain-selection and bridging complexity that Ethereum L2s introduce. A stablecoin like USDB on Spark can be sent for near-zero cost without choosing between six competing rollups or trusting a bridge contract. The tradeoff is a different trust model (1-of-n operators rather than Ethereum's rollup verification) and a younger ecosystem. But for the specific use case of low-cost, instant payments, the architectural simplicity is a genuine advantage.

Developers looking to build on Spark can explore the Spark SDK and documentation, or try General Bread, a Spark-powered wallet for dollar-denominated Bitcoin payments. For a deeper comparison of Ethereum rollup lessons applied to Bitcoin scaling, see our research on what Bitcoin can learn from Ethereum's L2 evolution.

Conclusion

The Ethereum L2 fee war has driven transaction costs to near zero, but it has also exposed the limits of fee competition as a moat. With every major rollup charging under $0.06 for a token transfer, differentiation has shifted to distribution, ecosystem depth, security maturity, and revenue model sustainability. Base leads on volume and profitability thanks to Coinbase's user funnel. Arbitrum leads on DeFi depth and was first to achieve decentralization milestones. Optimism is rebuilding its chain franchise after Base's departure. zkSync is betting on ZK proofs as a long-term technical advantage.

For users, the fee war is unambiguously positive: cheap transactions across multiple competing chains. For L2 operators, it is a profitability challenge that will likely drive consolidation toward the strongest two or three platforms. And for payment builders, it creates a choice: navigate the complexity of Ethereum's rollup landscape for its deep liquidity, or use simpler architectures that optimize directly for payment-specific requirements.

This article is for educational purposes only. It does not constitute financial or investment advice. Bitcoin and Layer 2 protocols involve technical and financial risk. Always do your own research and understand the tradeoffs before using any protocol.