Of the 73 active Ethereum Layer 2 rollups tracked by L2BEAT, two networks — Arbitrum One and Base — now hold approximately 77% of the $48 billion in combined L2 total value locked. The remaining 71 rollups split the rest, and a growing number are ceasing operations entirely. More than 60 crypto p...
"For every new rollup launching, several more are quietly shutting down." — Will Papper, Co-founder, Syndicate Labs
Of the 73 active Ethereum Layer 2 rollups tracked by L2BEAT, two networks — Arbitrum One and Base — now hold approximately 77% of the $48 billion in combined L2 total value locked. The remaining 71 rollups split the rest, and a growing number are ceasing operations entirely.
More than 60 crypto projects have shut down in the first half of 2026, according to data compiled by Crypto Briefing. Among them: Syndicate Labs ($27M raised, a16z-backed), Zero Network (Zerion's gasless L2), Lattice's Redstone chain, Everclear (formerly Connext, $500M monthly volume), and shared sequencer project Astria ($18M raised). The pattern is consistent: technically functional projects that failed to generate sustainable fee revenue in a market where two incumbents absorb most user activity.
This consolidation reflects a structural repricing of the Layer 2 thesis. The Dencun upgrade's 90%+ fee reduction made rollups cheap to use but also cheap to compete with, triggering fee wars that pushed most operators into losses. Base was the only L2 that turned a profit in 2025, earning approximately $55 million from sequencer fees. The rest subsidized operations through treasury spend, VC runway, or token incentives — none of which proved durable.
L2BEAT data as of May 2026 shows the following TVL distribution across Ethereum's Layer 2 ecosystem:
| Network | TVL (approx.) | Market Share | |---------|--------------|--------------| | Arbitrum One | $16.9B | ~35% | | Base | $12.8B | ~27% | | OP Mainnet | $1.9B | ~4% | | Starknet | $617M | ~1.3% | | Linea | $421M | ~0.9% | | zkSync Era | $404M | ~0.8% | | Other 67 rollups | ~$14.9B | ~31% |
Base, Arbitrum, and Optimism together process approximately 90% of all L2 transactions, according to research from 21Shares published June 24, 2026. Base alone handles over 60% of L2 transaction volume, driven by Coinbase's 100M+ user distribution channel.
The combined L2 ecosystem processes roughly 5,600 transactions per second on a typical weekday, with developers projecting capacity above 24,000 TPS as blob throughput increases through 2026.
May 2026 marked the most concentrated period of project closures in the L2 sector. On a single day — May 21 — three infrastructure projects announced shutdowns simultaneously:
Syndicate Labs — The a16z-backed startup, which raised over $27 million to build on-chain developer tools for DAOs, closed after five years. Co-founder Will Papper stated that demand had shifted toward custom execution environments built from scratch, making Syndicate's general-purpose rollup framework commercially unviable. The company chose an orderly wind-down to fulfill customer obligations.
Everclear (formerly Connext) — The cross-chain settlement protocol shut down its foundation, research lab, protocol, and frontend despite processing $500 million in monthly volume. The team stated: "Despite reaching $500M in monthly volume, the cross-chain solvers segment never developed the commercial depth we needed." The CLEAR token fell 48% to $0.0002332 in a single session following the announcement.
Zero Network — Zerion's gasless Ethereum L2, built on ZK Stack technology, announced wind-down after 18 months of operation. The network had experienced a three-week block production halt in late December 2025. Users were given until July 31, 2026 to withdraw funds. Zerion redirected resources to its wallet and API business.
Earlier in May, Lattice announced a phased shutdown of its blockchain gaming infrastructure, with the Redstone L2 network ceasing operations on May 16. Founded in 2021, Lattice developed the MUD framework and autonomous world tooling but failed to achieve a sustainable business model. Remaining tools were open-sourced.
In December 2025, Astria Network — once the leading shared sequencer project — halted at block 15,360,577 after raising $18 million. The project's sole major integration, with the Flame network, was rolled back before shutdown.
Other notable closures in H1 2026 include DeFi app Legend ($15M raised), Fantasy.top, and a16z-backed Entropy ($25M raised) and Yupp ($33M raised). Collectively, three a16z-backed projects alone account for $87 million in funding that did not produce surviving products.
The fee compression that began with Ethereum's Dencun upgrade in March 2024 fundamentally altered L2 economics. EIP-4844 introduced blob transactions — a dedicated, cheap data lane for rollups — reducing L2 data posting costs by over 90%. The Pectra upgrade further expanded blob throughput from 3 target / 6 max blobs per block to 6 target / 9 max (EIP-7691).
The result: L2 transaction fees fell from dollars to cents. Ethereum's daily gas fee revenue dropped from over $30 million to approximately $500,000. This made L2s accessible for consumer applications but destroyed the fee revenue model for operators that lacked scale.
The economics are stark. Coinbase's Base chain earned over $94 million in sequencer revenue but returned only $4.9 million to Ethereum in blob fees — a margin that is viable only because Coinbase operates the sole sequencer and routes its 100M+ user base through the network. For an L2 without a captive user base, replicating this margin structure is not feasible.
The Fusaka upgrade (December 2025) introduced EIP-7918, a blob base fee floor that ensures minimum ETH burn even during low-demand periods. This addressed Ethereum's revenue leak but did nothing to improve economics for subscale rollups.
The two dominant L2s followed different paths to market leadership.
Base executed the most rapid TVL growth in the L2 category: from approximately $2.1 billion in October 2024 to $12.8 billion by May 2026, a roughly 6x increase in 19 months. The network's advantage is structural — it is the primary on-chain destination for Coinbase's regulated user base, giving it a distribution moat that no standalone rollup can replicate. Base processes millions of transactions daily, and its DeFi and consumer app ecosystems (including social applications) are the most active in the L2 space.
Arbitrum One maintains the deepest on-chain liquidity and the most mature DeFi protocol stack in the L2 category, including GMX, Aave, Curve, and Camelot. Its $16.9 billion TVL reflects institutional and DeFi-native capital concentration. Arbitrum's developer tooling ecosystem remains the most extensive among L2s.
Together, these two networks hold approximately $29.7 billion in TVL — more than the remaining 71 rollups combined.
The networks between the top tier and the shutdown tier are losing ground steadily.
Linea (Consensys) saw bridge deposits fall from $976 million in November 2025 to $367 million in May 2026 — a 62% decline over six months.
Starknet holds approximately $241 million in TVL — less than one-twentieth of Base's. Its monthly chain fees have fallen over 95% since November 2023. StarkWare, the company behind Starknet, has restructured operations in response to the revenue decline.
zkSync Era holds approximately $404 million in TVL. Like Starknet, it has struggled to attract sustained DeFi liquidity despite technical differentiation in zero-knowledge proof architecture.
The gap between these networks and the top two is structural, not cyclical. Without a captive user base (Base's Coinbase integration) or deep DeFi composability (Arbitrum's protocol ecosystem), mid-tier L2s face a cold-start problem that token incentives have failed to solve.
The concentration of activity on two networks surfaces a persistent architectural concern: sequencer centralization. Every major L2, including Arbitrum, Base, and Optimism, operates a single centralized sequencer run by its core team.
This creates three compounding risks: transaction censorship, MEV extraction by the sequencer operator, and single-point-of-failure liveness risk. Base's sequencer is operated exclusively by Coinbase. Arbitrum's is run by Offchain Labs. Optimism's is run by the Optimism Foundation.
Decentralization timelines remain indefinite. Arbitrum's BoLD permissionless fraud proofs are rolling out in 2026, but full sequencer decentralization has no confirmed mainnet date. Base has committed to progressive sequencer decentralization under the Superchain roadmap but remains single-operator. Espresso Systems, the remaining shared sequencer project after Astria's shutdown, targets permissionless proof-of-stake sequencing in late 2026.
The economic reality: centralized sequencers are more profitable. In a market where only one L2 turned a profit in 2025, the incentive to decentralize — and share or forgo sequencer revenue — is minimal.
The Ethereum Layer 2 market is undergoing a consolidation that follows the pattern observed in most infrastructure markets: rapid proliferation during the technology's expansion phase, followed by concentration around a small number of scaled operators once unit economics tighten.
The 73 active rollups tracked by L2BEAT will likely compress further. 21Shares projects the landscape will consolidate around "ETH-aligned, high-performance, and exchange-backed networks." The data supports this thesis. Networks without structural distribution advantages — a captive user base, deep protocol integrations, or institutional capital relationships — are running on finite treasury runway.
The economic value in the L2 sector has migrated to sequencer operation at scale. This is not a technology problem; it is a market structure outcome. The rollups that survive will be those that solved distribution before they solved decentralization.