Ethereum's Layer 2 ecosystem has consolidated into a de facto duopoly. As of May 2026, 73 active rollups collectively secure more than $48 billion in total value locked, but two networks — Coinbase's Base and Offchain Labs' Arbitrum One — control approximately 77% of all L2 DeFi liquidity. Add OP...
"Keep true to your values and the reasons why you're here. If we're just replacing one centralized system with another, certain individuals might benefit, but the world won't." — Steven Goldfeder, CEO, Offchain Labs (Arbitrum)
Ethereum's Layer 2 ecosystem has consolidated into a de facto duopoly. As of May 2026, 73 active rollups collectively secure more than $48 billion in total value locked, but two networks — Coinbase's Base and Offchain Labs' Arbitrum One — control approximately 77% of all L2 DeFi liquidity. Add OP Mainnet and the top three command 83% of TVL and process roughly 90% of all L2 transactions.
The remaining 50-plus rollups are fighting over approximately $8 billion in residual TVL — a figure that has been shrinking since mid-2025. At least 10 Web3 projects shut down in Q1 2026 alone. ZKsync Lite, Ethereum's first ZK-rollup, is scheduled for deprecation this year. Blast's TVL collapsed 97% from its $2.2 billion peak. The data points to a structural consolidation where distribution — not technology — determines survival.
This report examines how the L2 market reached this concentration, the economic forces sustaining the duopoly, and what it means for the 50+ rollups that have effectively lost the war for users.
L2BEAT's staging dashboard lists 73 active rollups as of May 2026. Their combined TVL exceeds $48 billion. The distribution of that capital is sharply asymmetric:
| Network | TVL (Est.) | L2 DeFi Share | Daily Txns | |---|---|---|---| | Base | $12.8B | ~46.6% | ~15M | | Arbitrum One | $15.9B | ~30.9% | ~1.5M | | OP Mainnet | $1.9B | ~5.5% | ~800K | | zkSync Era | $4.1B | ~3.5% | — | | Starknet | $1.5B | ~1.7% | — | | Linea | $3.4B | ~1.2% | — | | Scroll | $2.1B | ~0.9% | — | | All others | ~$6.3B | ~9.7% | — |
Sources: L2BEAT, SpotedCrypto, BlockEden; TVL and DeFi share metrics may diverge due to differing methodologies (total value secured vs. DeFi-specific TVL).
Three networks — Base, Arbitrum, and Optimism — process nearly 90% of all L2 transactions. Base alone handles over 60% of L2 transaction volume, according to data compiled by EarnPark. This is a market structure that resembles cloud computing (AWS, Azure, GCP) far more than the decentralized pluralism Ethereum's rollup-centric roadmap envisioned.
In January 2026, Arbitrum One, OP Mainnet, and Base each achieved Stage 1 rollup status on L2BEAT's classification system, enabling permissionless fraud proofs. This security milestone further entrenched their position: institutional capital follows security guarantees, and Stage 1 status provides a measurable trust signal that smaller rollups lack.
Base is the fastest-growing L2 by both transaction volume and TVL. It processes approximately 15 million daily transactions and maintains over 1 million daily active addresses. Its structural advantage is Coinbase's 110 million verified retail users — a distribution pipeline no competing rollup can replicate.
Base's average daily revenue over the past 180 days sits at approximately $185,291, according to CoinLaw, with priority fees constituting 86.1% of that figure. Coinbase retains sequencer revenue directly, making Base one of the few rollups with a clear, profitable business model. The network generated approximately $55 million in profit during 2025, according to EarnPark's analysis — the only L2 to achieve sustained profitability.
Jesse Pollak, Base's creator and a VP at Coinbase, stated in April 2026 that "agents are defined in software and operating software, they want money as software," signaling Base's strategic expansion into AI-agent payment infrastructure. Base recorded nearly 100 million stablecoin transactions in April 2026 alone.
Arbitrum One leads in raw TVL at $15.9 billion, with a concentration of DeFi-native liquidity. It maintains 250,000–300,000 daily active users, skewing toward institutional DeFi participants rather than retail. Offchain Labs' BoLD (Bounded Liquidity Delay) fraud-proof system removed centralized validator requirements, and the network is described by CEO Steven Goldfeder as "very, very close" to Stage 2 decentralization.
Arbitrum generates approximately $55,025 in daily revenue — roughly 3.4x less than Base — reflecting lower transaction volumes but higher average transaction values. The Arbitrum DAO manages a $215 million gaming catalyst program and monthly protocol revenue has ranged between $3–8 million, creating a foundation for potential value accrual if governance implements fee-sharing mechanisms for ARB token holders.
Median transaction fees across the leaders remain low: Base and OP Mainnet at $0.02–$0.03, Arbitrum at $0.04, zkSync Era at $0.05, and Scroll at $0.06.
The data on smaller rollups is consistent and grim. According to a BlockEden analysis published in February 2026, most new rollups launched in 2024–2025 followed identical trajectories: incentive-driven activity surges preceding token generation events, followed by rapid post-TGE declines as liquidity migrated to established ecosystems. TVL across smaller rollups declined an average of 61% since June 2025.
Specific casualties:
Within the first 90 days of 2026, more than 10 Web3 projects publicly announced cessation of operations, according to Odaily's tracking. The pattern is consistent: without differentiation, without users, and without sustainable economics, projects cannot survive the end of incentive cycles.
Optimistic rollups command approximately 80% of total L2 TVL. The ZK rollup cohort — zkSync Era, Linea, Scroll, Starknet, World Chain — collectively holds approximately 20%, or $9.6 billion in absolute terms.
The throughput gap remains the principal constraint. ZK proof generation is computationally intensive: zkSync Era delivers approximately 28 TPS; Starknet approximately 19 TPS. Both figures trail optimistic rollup benchmarks substantially.
Rather than competing for retail DeFi share at a structural disadvantage, ZK rollups are pivoting. zkSync Era, through its Prividium subsidiary, is targeting Deutsche Bank and UBS for privacy-preserving tokenized asset settlement, according to SpotedCrypto's May 2026 analysis. The thesis: ZK privacy properties command a compliance-driven premium in regulated institutional contexts where cryptographic finality matters more than throughput.
Whether this pivot generates meaningful revenue is unproven. ZK rollups settling more stablecoin volume than optimistic rollups combined — a claim made by multiple sources — suggests the institutional thesis has some traction, but auditable revenue data is scarce.
Rollup-as-a-Service providers — Conduit, Caldera, Gelato, Alchemy — made deploying a rollup trivially easy. A production-grade L3 rollup can be launched via Conduit for $3,000/month in approximately 15 minutes. Caldera quotes average gas fees of $0.001 per transaction on its platform.
This ease of deployment contributed directly to oversupply. The current landscape of a dozen-plus RaaS providers is unsustainable, according to industry analysis, with expectations for acquisitions and a shakeout that leaves three to four dominant platforms. Conduit maintains deep OP Stack and Superchain alignment; Caldera differentiates with its Metalayer cross-chain messaging; Gelato uniquely supports ZKsync's Elastic Chain framework.
The economic irony is clear: RaaS platforms lowered the cost of launching a rollup to near zero, but they could not solve the distribution problem. A rollup that costs $3,000/month to run but attracts $500,000 in TVL and generates $200/day in fees is a zombie chain — technically alive, economically dead.
The profitability picture in L2s is stark:
This maps directly to the economic value framework: in a fee-fragmented ecosystem, value accrues to the layer with distribution leverage. Coinbase's ability to funnel 110 million accounts into Base is the equivalent of AWS bundling compute with its retail marketplace. The technology is sufficient; the distribution is decisive.
The ARB token illustrates the governance-token problem. It carries a 2% annual inflation cap and confers voting rights over the DAO treasury and protocol upgrades, but it does not currently receive fee revenue. Unless governance implements a staking or fee-sharing mechanism, ARB remains what analysts describe as a "political asset" rather than infrastructure equity.
The Ethereum L2 market in May 2026 resembles an industry that has undergone its first real shakeout. The rollup-centric roadmap produced 73 live networks but concentrated economic activity in two. 21Shares projects the ecosystem will consolidate further around "a leaner, more resilient set of networks" by year-end.
The surviving rollups share common traits: exchange-backed distribution (Base), deep DeFi liquidity and DAO governance (Arbitrum), or Superchain network effects (OP Mainnet). ZK rollups are attempting a differentiated survival strategy through institutional finance, but that bet remains unproven at scale.
For the 50-plus rollups in the long tail, the math is unforgiving. Without differentiated use cases, without sustainable fee revenue, and without distribution partners, the trajectory is a slow fade — reduced development activity, migrated liquidity, and eventual effective abandonment while technically remaining operational.
The L2 wars are not ending. They ended. The question now is whether the duopoly hardens or whether a structural shift — shared sequencing, native interoperability, or a new distribution channel — reopens the market.