Of 73 active Ethereum rollups tracked by L2BEAT, two networks — Base and Arbitrum One — now control approximately 77% of all Layer 2 DeFi liquidity. Total value locked across the rollup ecosystem exceeds $48 billion as of May 2026. The remaining 71 networks split roughly $11 billion among them, w...
"We don't need more copypasta EVM chains." — Vitalik Buterin, Ethereum Co-Founder, February 2026
Of 73 active Ethereum rollups tracked by L2BEAT, two networks — Base and Arbitrum One — now control approximately 77% of all Layer 2 DeFi liquidity. Total value locked across the rollup ecosystem exceeds $48 billion as of May 2026. The remaining 71 networks split roughly $11 billion among them, with the bottom 50 holding negligible economic activity.
The consolidation accelerated following three structural events in early 2026: Coinbase's Base departed the Optimism Superchain in February, triggering a 28% collapse in OP token price within 48 hours; OP Labs cut 20% of its workforce in March; and a 21Shares research report concluded that most Ethereum L2s are unlikely to survive the year. Blast's TVL fell 97% from its $2.2 billion peak. Loopring closed its wallet. Kinto shut down entirely. The L2 landscape is bifurcating into a small number of economically viable networks and a long tail of what industry analysts now call "zombie chains."
This report examines the data behind the consolidation, maps the competitive positioning of surviving networks, and evaluates the economic sustainability of the remaining rollup ecosystem.
The numbers are unambiguous. According to L2BEAT and DefiLlama data as of mid-May 2026:
| Network | TVL (Approx.) | L2 DeFi Share | Daily Transactions | |---------|--------------|---------------|-------------------| | Base | $10.7–12.8B | ~46.6% | ~15M | | Arbitrum One | $15.9–16.9B | ~30.9% | — | | OP Mainnet | $1.7–1.9B | ~7% | — | | Starknet | ~$617M | ~2.5% | — | | Linea | ~$421M | ~1.7% | — | | zkSync Era | ~$404M | ~1.6% | — | | All others (67 networks) | ~$1–2B combined | <10% | — |
Base, Arbitrum, and Optimism together process approximately 90% of all L2 transactions. The ecosystem's L2 scaling factor reached 49.98x relative to Ethereum mainnet on May 16, 2026, with 666.04 unique operations per second — but that throughput is concentrated in three networks.
Base processes approximately 15 million daily transactions and generated $82.6 million in annual revenue in 2025, with approximately $55 million in net profit after L1 data posting costs and Optimism Collective revenue-sharing obligations. According to multiple industry analyses, Base was the only Layer 2 to achieve profitability in 2025.
Arbitrum's DAO income ran at approximately $12 million annualized as of March 2026, supplemented by Timeboost gas-fee ordering revenue exceeding $2 million since its April 2025 launch. Its ArbOS 51 "Dia" upgrade in January 2026 increased capacity while reducing fee volatility.
The gap between the top three and the rest is not narrowing. Usage across non-dominant L2s dropped 61% since June 2025, according to the 21Shares analysis.
The single most consequential event in L2 market structure in 2026 was Base's departure from the Optimism Superchain, announced February 18.
In a blog post titled "The Next Chapter for Base," Coinbase's L2 team stated its intention to consolidate its technology stack into a Base-managed codebase rather than depending on OP Stack's shared upgrade schedule. The stated rationale: Base wanted to double its hard fork cadence from approximately three major upgrades per year to six, and move to Trusted Execution Environment (TEE) and ZK finality proofs on a timeline the shared OP Stack roadmap could not accommodate.
The financial impact on Optimism was severe. Base had accounted for an estimated 96.5% of the Optimism Collective's gas fee revenue. The OP token fell 28% within 48 hours of the announcement. Optimism's entire revenue-sharing model — the economic foundation of the Superchain thesis — was effectively hollowed out by the departure of its largest contributor.
Three weeks later, on March 12, OP Labs CEO Jing Wang announced a 20% workforce reduction — 20 employees. "This is not about finances. OP Labs is well capitalized with years of runway," Wang wrote. "This is about doing fewer things well, making decisions faster, and reducing coordination overhead."
The Superchain still operates, but its viability as a shared economic model depends on attracting new chains capable of replacing Base's revenue contribution — a task that no current Superchain participant appears positioned to accomplish.
The term "zombie chain" has entered standard industry vocabulary in 2026. It describes networks that remain technically operational but carry minimal user activity, negligible TVL, and no clear path to economic sustainability.
The pattern is consistent across dozens of networks launched during the 2023–2025 rollup boom:
Specific casualties documented in 2026:
Networks launched between 2023 and 2025 that relied on token incentive programs saw TVL collapse 70–90% within weeks of program expiration, according to the BlockEden.xyz analysis. As the 21Shares report stated: "Points-fueled TVL is not real demand — it is rented attention that evaporates the moment incentive programs end."
Zero-knowledge rollups were widely expected to overtake optimistic rollups on technical merits — faster finality, stronger cryptographic guarantees, and theoretically lower long-term costs. As of May 2026, that thesis has not materialized in market share terms.
The combined TVL of Starknet (~$617M), Linea (~$421M), and zkSync Era (~$404M) totals approximately $1.4 billion — roughly 3% of the L2 ecosystem's $48 billion. All three trail OP Mainnet ($1.7–1.9B), which itself sits far behind Base and Arbitrum.
Prover decentralization — a core promise of the ZK model — remains in early stages. zkSync has begun experimenting with external prover networks, but practical decentralization of proof generation had not been achieved as of the most recent reports.
The competitive disadvantage is primarily one of distribution, not technology. Base benefits from Coinbase's 100+ million verified user base. Arbitrum captured early DeFi liquidity and maintains deep protocol integrations. ZK rollups, despite theoretical advantages, have not secured equivalent distribution channels or application ecosystems.
L2BEAT's "Stage 1" classification — the institutional-grade protocol security benchmark — now includes six networks: Arbitrum One, Base, OP Mainnet, Starknet, Scroll, and Ink. Starknet's inclusion validates its technical architecture, but the designation has not translated into proportional TVL gains.
Ethereum's Dencun upgrade in March 2024 reduced L2 data posting costs by approximately 90%, triggering fee wars that compressed margins across all rollups. For networks without substantial organic transaction volume, the math became unsustainable.
The fee reduction was intended to make L2s cheaper for users. It succeeded. But it also eliminated the pricing floor that had allowed mid-tier rollups to generate modest revenue from limited activity. With blob fees near zero, rollups need high transaction volume to generate meaningful sequencer revenue.
Base achieves this through Coinbase integration. Arbitrum achieves it through DeFi protocol density. Optimism achieves it through the residual Superchain ecosystem, though its economics post-Base are significantly weaker. For the remaining 70 networks, the economic model requires either:
The Dencun-era fee compression effectively converted the L2 market from a technology competition into a distribution competition. Networks with built-in user acquisition channels — Coinbase for Base, institutional DeFi for Arbitrum — pulled away. Networks dependent on incentive programs lost their only competitive lever.
Not all new L2 activity is declining. MegaETH launched its mainnet on February 9, 2026, targeting 100,000 TPS with sub-10ms block times. Its token generation event on April 30 was tied to verified on-chain performance milestones. TVL reached $580 million within days of the MEGA token launch.
MegaETH's positioning targets a different market segment — real-time execution for high-frequency DeFi and latency-sensitive applications — rather than competing for general-purpose L2 traffic. During a January 2026 stress test, the network processed 10.7 billion transactions at a sustained 35,000 TPS, though independent third-party verification of those figures had not been published as of April 2026.
Linea, developed by Consensys, has adopted an "ETH-aligned" economic model that redirects fees back to Ethereum through burns or validator rewards. This approach may improve sustainability by leveraging Ethereum's user base rather than competing for independent liquidity.
Buterin himself, in his February 2026 comments, endorsed two models: "tightly coupled app-specific systems that rely on Ethereum for key functions" and "institutional or application-driven chains that post cryptographic proofs or state commitments back to Ethereum." Both descriptions point toward a future where surviving L2s serve specific functions rather than operating as generic EVM clones.
The Ethereum L2 ecosystem in May 2026 resembles a classic platform market approaching winner-take-most equilibrium. Of 73 active rollups, three generate the vast majority of economic activity. The remaining 70 operate in various states of decline, subsidy dependence, or outright shutdown.
The consolidation was accelerated by three factors: Dencun's fee compression, which eliminated margins for low-volume networks; the exhaustion of airdrop-driven growth models; and the departure of Base from the Superchain, which demonstrated that shared infrastructure models cannot hold when the dominant participant has the scale to operate independently.
For the broader Ethereum ecosystem, the consolidation carries mixed implications. Fewer viable L2s means less fragmented liquidity and reduced bridging risk for users. But it also means greater dependence on a smaller number of sequencer operators, with Base's sequencer controlled by Coinbase and Arbitrum's by Offchain Labs. The decentralization properties that motivated the rollup-centric roadmap have not yet materialized at the economic layer.
As the 21Shares report concluded, a "leaner, more resilient" set of networks is likely to define Ethereum scaling by year-end 2026. The data suggests that set may be smaller than most anticipated.