Layer 2 Wars
Executive summary
Ethereum's L2 ecosystem has consolidated into a tier of dominant general-purpose rollups led by Base, Arbitrum and Optimism, with a mid-tier of zk-rollups (zkSync, Linea, Scroll, Starknet) and Polygon's pivoted CDK/AggLayer architecture splitting the remainder. As of July 2026 Base leads on user count, while the value-secured race tightened abruptly: Arbitrum One showed roughly $17.4bn against Base's $11.4bn in early July, but on July 13 L2Beat removed about $7bn of team-controlled RAIN tokens — a prediction-market asset it called 'heavily manipulated' — from Arbitrum's count, leaving the two chains roughly level at the top; Optimism's OP Stack ecosystem (including Base, World Chain, Unichain and others) collectively dominates economic activity; zkSync, Linea, Scroll and Starknet retain technical credibility but lag in distribution; Polygon's CDK and AggLayer represent a re-architected attempt to consolidate fragmented liquidity. Cumulative L2 value secured was roughly $41bn in early July 2026, closer to $34bn after the July 13 RAIN adjustment (L2Beat) — compressed in USD terms by ETH's roughly 45% year-to-date decline — with daily transactions in the tens of millions. The 2026-2027 questions are whether interop frameworks (AggLayer, Superchain interop, Across) defragment liquidity meaningfully, whether sequencer decentralisation matures, and whether L2s compete or compose toward a unified rollup mesh.
Background and current state
The Layer-2 thesis — that Ethereum scales by moving execution off the base chain to rollups while preserving security via fraud proofs (optimistic) or validity proofs (ZK) — has been the defining scaling architecture since 2021. By April 2026 the architecture has matured into a recognisable competitive landscape. Optimistic rollups dominate by raw activity and mindshare: Base (Coinbase's L2, launched August 2023) leads on monthly active users, secures roughly $11.4bn as of July 2026 (L2Beat), and underpins a significant share of Coinbase user activity through deep wallet integration. Arbitrum One has retained leadership on certain DeFi-heavy metrics, showing roughly $17.4bn secured in early July 2026 — though on July 13 L2Beat cut that figure by about $7bn by removing team-controlled RAIN tokens it described as 'heavily manipulated' (roughly $2.6bn of RAIN remains counted, under continuing review), leaving Arbitrum and Base roughly level at the top of the value-secured table — a dominant share of L2 perpetuals trading (GMX, Hyperliquid bridge flows; Vertex wound down in 2025) and the largest pre-Base ecosystem mass. Optimism Mainnet itself is far smaller (about $1.5bn secured as of July 2026) but hosts the OP Stack governance hub and benefits indirectly from the Superchain ecosystem (Base, World Chain, Unichain, Mode, Zora and others). Polygon, having pivoted from PoS-only to CDK and AggLayer architecture, retains roughly $4bn secured (July 2026) but has restructured its identity around interoperability rather than monolithic L2 dominance. The ZK-rollup tier — zkSync Era, Linea (Consensys), Scroll, Starknet — collectively holds only about $1bn secured as of July 2026 but has retained technical credibility for institutional and validity-proof-sensitive applications. Cumulative L2 value secured stood at roughly $41bn in early July 2026, closer to $34bn after the July 13 RAIN adjustment (L2Beat), with USD-denominated figures compressed by the 2026 ETH price decline and daily transactions across the ecosystem in the tens of millions.
Key actors and market structure
Five competitive blocs structure the L2 economy. First, the Coinbase-Base ecosystem. Base benefits from the deepest distribution moat in the category: Coinbase's 100m+ retail user base, integrated Wallet, fiat-onramp and creator/onchain primitives (Onchain Summer programmes, creator-first product surfaces). Base sequencer revenue flows to Coinbase's bottom line at moderate scale — roughly $19m gross in Q4 2025, about $60m annualised, booked within Coinbase's 'other transaction revenue' line. The Base team has emphasised consumer applications, social and creator-economy primitives, and a markedly different cultural posture from Arbitrum or Optimism's developer-centric framing. Second, the Arbitrum / Offchain Labs ecosystem. Arbitrum's strength is on DeFi composability and developer maturity; the Arbitrum DAO governs a meaningful treasury and the Stylus stack supports non-EVM execution. The Arbitrum Orbit framework, open for mainnet deployments since late 2023, underpins a small ecosystem of Orbit chains targeting specific verticals. Third, the Optimism-Superchain ecosystem. The OP Stack is the most successful L2 software stack by adoption, with Base, World Chain (the Worldcoin/Sam Altman initiative), Unichain (Uniswap's L2), and a long tail of OP-Stack chains constituting a federated 'Superchain' designed for shared sequencing and security. The OP collective governs the stack and aligns economic interests. Fourth, the Polygon ecosystem. Polygon's pivot from PoS sidechain to CDK and AggLayer represents a hard restructuring: PoS continues but is downgraded; CDK is the rollup framework; AggLayer is the unified-state interop layer designed to defragment chains built on different architectures. Polygon retains substantial enterprise relationships (Stripe, Disney, Nubank-type integrations) and meaningful Latin American distribution. Fifth, the ZK-tier. zkSync, Linea, Scroll, Starknet and a few smaller players retain the technical-leadership argument for validity-proof-secured rollups. Their ecosystem traction is real but materially smaller than the OP-Stack collective; the trajectory depends on whether ZK rollups prove faster, cheaper or more secure in operationally meaningful ways.
Mechanism and economics
L2 economics rest on the spread between user-paid transaction fees on the L2 and the cost of posting data to Ethereum. Post-EIP-4844 (March 2024), data is posted via blobs at a fraction of historical cost; with the further EIP-7691 changes in 2025, blob throughput has doubled, reducing L2 costs further. The result: L2 fees per user transaction are typically $0.001-0.05, with the L2 sequencer capturing 60-90% of the user fee as net revenue. Sequencer revenue is the principal economic engine. Base's sequencer fees are booked within Coinbase's 'other transaction revenue' and ran roughly $19m gross in Q4 2025 (about $60m annualised); Arbitrum and Optimism distribute their sequencer revenue through DAO treasuries and grant programmes. The economics for ZK rollups are slightly different: the proof-generation cost is non-trivial but has fallen significantly with proving-system optimisation (Plonky3, AIR-based provers, hardware acceleration) and is now sub-cent per transaction at scale, making ZK rollup unit economics competitive with optimistic rollups. The competitive dimension is how each rollup deploys its sequencer revenue: Base reinvests through Coinbase distribution and growth programmes, Arbitrum and Optimism via DAO treasury allocation (developer grants, ecosystem incentives, retroactive funding), and Polygon via the AggLayer development and Polygon 2.0 token-holder distributions. The interop economics of AggLayer and Superchain Interop are still proving out: shared sequencing and unified bridges promise atomic cross-rollup composability but require coordination among independent governance entities, with revenue-share questions unresolved.
Recent milestones (2024-2026)
February 2024 - Polygon's AggLayer v1 mainnet goes live (February 23), connecting Polygon zkEVM and OKX's X Layer to begin the unified-state interop test. March 2024 - EIP-4844 activates blob transactions, dropping L2 data-posting costs by 90%+. May 2024 - Base TVL crosses $5bn; user count crosses 5m. June 2024 - ZKsync's v24 upgrade begins the Elastic Chain (later Elastic Network) era, the ZK tier's response to OP Stack federation. September 2024 - Arbitrum Stylus mainnet activates; non-EVM execution comes online. October 2024 - Worldcoin ecosystem migrates to its own OP-Stack chain (World Chain), accelerating OP Stack distribution; Uniswap announces Unichain as an OP Stack chain, signalling DeFi-native L2 strategy. December 2024 - Polygon ships the AggLayer v0.2 pessimistic-proofs testnet. February 2025 - Unichain mainnet launches. May 2025 - the Pectra upgrade activates (May 7), including EIP-7691's doubling of blob throughput; L2 costs per transaction fall again. Q4 2025 - Ethereum L1 gas usage stabilises low; over 95% of Ethereum-ecosystem user transactions occur on L2s. H1 2026 - cumulative L2 TVL crosses $55bn in Q1 before the ETH price decline compresses USD figures; Optimism's Superchain native interop reaches devnet (April 2026) with mainnet targeted for later in 2026; Across Protocol emerges as a leading cross-rollup bridging primitive.
Key risks and open questions
The L2 ecosystem faces three core risks. First, sequencer centralisation. Almost every major L2 still operates a centralised sequencer (Base, Arbitrum, Optimism, Linea, zkSync, Scroll). Decentralised sequencing is on the roadmap for several but no major L2 has yet shipped a production decentralised sequencer with operational liveness comparable to the centralised version. The risk is twofold: regulatory exposure (a centralised sequencer is a clear point of intermediation that regulators can target) and operational (sequencer downtime forces users to wait or to exit through the L1 escape hatch). Second, fragmentation. Even with EIP-4844 and improved interop, L2s fragment user attention, liquidity and developer effort. AggLayer, Superchain Interop and Across attempt to defragment, but each operates within its own scope (AggLayer for CDK chains, Superchain for OP Stack chains, Across as a generalised bridge) and cross-scope interop remains immature. The risk is that the L2 ecosystem stays fragmented, with users facing materially different UX across rollups and developers forced to deploy multiple times. Third, value-capture dilution. As L2s grow, ETH's role as the L2 settlement asset is reinforced (positive for ETH), but the ETH-denominated fee revenue per transaction is small and falling, which may not generate sufficient ETH burn to offset issuance. The 'ETH is moneyness' narrative depends on whether L2 expansion concentrates economic value in ETH or disperses it across L2 native tokens (ARB, OP, MATIC/POL, ZK and competitors).
Regulatory landscape
L2 regulation is largely upstream — through the ETH and broader-rollup regulatory frame — but several specific questions are emerging. Sequencer operation as a regulated activity: a centralised sequencer is functionally an intermediary processing user transactions and capturing fees; in jurisdictions like the EU and UK, this could attract MTF (multilateral trading facility) or similar registration requirements. To date, no major regulator has formally classified sequencer operation as licensed activity, but the question is live. The Coinbase-Base relationship is a particular focus: Coinbase as a SEC-registered exchange operating an L2 sequencer creates a clear regulated nexus, and SEC posture under the post-2025 administration has been broadly accommodating but the legal framework remains underspecified. Tokenised governance (ARB, OP, POL, ZK tokens) faces standard securities-law questions about token issuance, vesting, distribution and governance utility; the Howey-test analysis varies but the post-2025 US enforcement posture has reduced acute risk. Internationally, MiCA's classification of crypto-assets covers L2 native tokens under its Title II 'crypto-asset' regime, requiring whitepaper disclosure but otherwise light regulation. The bigger regulatory pressure may come from cross-border data-localisation rules: as L2s scale and become payments-relevant, the question of which jurisdiction's rules govern L2-mediated transactions becomes less abstract.
Outlook through 2027
Three vectors will determine the 2026-2027 outcome. First, distribution moat compounding. Base's Coinbase-distributed advantage looks structurally durable; absent a Coinbase-specific setback, Base's user-count lead is likely to expand. The competitive question is whether Arbitrum and Optimism's DeFi and developer-ecosystem advantages translate into sustained TVL leadership despite Base's user-count edge. Second, interop maturation. AggLayer's ability to connect non-Polygon chains, Superchain Interop's ability to extend beyond OP Stack, and the broader cross-rollup-bridge ecosystem (Across, LayerZero, Wormhole) will determine whether the rollup mesh feels unified or balkanised to end-users. A successful interop layer compresses the competitive landscape — users become rollup-agnostic and choose based on UX, cost and applications rather than ecosystem lock-in. Third, ZK-rollup operational scale. The technical case for ZK rollups (faster finality, validity-proof security, no fraud-proof challenge windows) is strong, but the user and TVL gap to optimistic rollups remains significant. If ZK rollups close that gap by 2027 — through better dev tooling, cheaper proofs and meaningful institutional adoption — the L2 landscape rebalances toward technical-merit competition. If they don't, the optimistic rollups (especially OP Stack-based) consolidate dominance and the ZK tier remains a niche. The plausible 2027 picture: 3-4 dominant general-purpose rollups (Base, Arbitrum, Optimism Mainnet, plus one ZK winner — likely Linea or zkSync) commanding 80%+ of activity, AggLayer and Superchain Interop largely unifying liquidity for end-users, and ETH's L1 functioning primarily as a settlement and DA layer with most consumer activity occurring on rollups.
Watch points
- Base sequencer revenue and user count trajectory; Coinbase corporate disclosures
- Arbitrum Stylus and Orbit ecosystem traction beyond core EVM use
- Optimism Superchain Interop expansion and any non-OP-Stack chain participation
- AggLayer cross-architecture connections beyond Polygon CDK chains
- Decentralised sequencer ship-dates from any major L2
- ZK rollup TVL and user-count growth versus optimistic rollups; Linea/zkSync trajectory
TL;DR
Ethereum's L2 ecosystem has consolidated around Arbitrum (DeFi-leading, ~$10.4bn secured after L2Beat's July 13 removal of ~$7bn in team-controlled RAIN tokens), Base (user-count leader, ~$11.4bn), the OP Stack collective (Base/World Chain/Unichain federation), Polygon's AggLayer pivot (~$4bn) and the ZK tier (zkSync, Linea, Scroll, Starknet) at roughly $1bn combined; cumulative L2 value secured is in the mid-$30bns as of mid-July 2026 (L2Beat), with EIP-4844/EIP-7691 having driven per-transaction costs to fractions of a cent, leaving sequencer decentralisation and interop the central 2026-2027 questions.
Get DeFi Intel research in your inbox
Weekly long-form coverage of papers, incidents, jurisdictions, chains, tokens and the people building them. Free tier covers headlines; Pro adds the analyst-grade breakdowns.
Sources & verification
- Per-chain and aggregate value secured (Arbitrum One ~$17.4bn pre-adjustment, Base ~$11.4bn, OP Mainnet ~$1.5bn, Polygon PoS ~$4.1bn, zkSync+Linea+Scroll+Starknet ~$1.0bn, all-L2 total ~$41bn; early July 2026): L2Beat
- L2Beat removal of ~$7bn in team-controlled RAIN tokens from Arbitrum's TVS (July 13, 2026; ~$2.6bn RAIN still counted, under review): The Crypto Times
- ETH price decline (~$3,356 on Jan 15, 2026 to ~$1,782 on July 13, 2026): CoinGecko
- Vertex wind-down (August 2025): The Block
- Base sequencer revenue (~$19m gross in Q4 2025, booked in 'other transaction revenue'): Talos — Coinbase Q4 2025 earnings outlook
- Unichain announcement (October 2024) and mainnet (February 2025): Uniswap Labs; CoinDesk
- Pectra upgrade with EIP-7691 blob-throughput doubling (May 7, 2025): Ethereum Foundation
- AggLayer v1 mainnet (February 2024): Binance Square/Polygon; ZKsync Elastic Chain (v24 upgrade, June 2024): The Block
- Superchain interop status (devnet April 2026, mainnet targeted later in 2026): Optimism
Value-secured figures last verified July 15, 2026. User counts and forward scenarios are editorial analysis from earlier coverage; a prior Base monthly-active-address figure and Base sequencer-revenue run-rate were corrected against Coinbase Q4 2025 disclosures.