Three Ethereum Layer 2 networks — Base, Arbitrum One, and OP Mainnet — now control approximately 83% of all L2 DeFi total value locked, up from roughly 65% a year ago. L2BEAT tracks 73 active rollups securing over $48 billion in combined TVL, but the distribution is lopsided: Arbitrum One holds a...
"We're in a consolidation phase for general-purpose layer twos, not layer twos broadly." — Ben Fisch, CEO, Espresso Systems
Three Ethereum Layer 2 networks — Base, Arbitrum One, and OP Mainnet — now control approximately 83% of all L2 DeFi total value locked, up from roughly 65% a year ago. L2BEAT tracks 73 active rollups securing over $48 billion in combined TVL, but the distribution is lopsided: Arbitrum One holds approximately $17–18 billion (40–44% share), Base holds $12–13.5 billion (30–47% share depending on the metric), and OP Mainnet trails at $1.3–1.9 billion. The remaining 70 rollups split less than $17 billion among them, and the tail is thinning fast.
RootData's mid-year tally counts 99 crypto projects that have formally shut down in 2026. Among them: Loopring, a pioneering zk-rollup whose TVL collapsed from $760 million in November 2021 to $8 million before it ceased operations on June 28, 2026. Lattice's Redstone chain shut down on May 16. Syndicate Labs, backed by $20 million from Andreessen Horowitz, wound down its rollup-infrastructure business in May. 21Shares projects that most mid-tier rollups will not survive the year, leaving what it calls a "leaner, more resilient" set of 5–10 meaningful networks by December.
L2BEAT data as of mid-2026 shows the following TVL distribution across major Ethereum rollups:
| Network | TVL (approx.) | Share of L2 DeFi TVL | |---------|--------------|---------------------| | Arbitrum One | $16.9–18.0B | 40–44% | | Base | $12.8–13.5B | 30–47% | | OP Mainnet | $1.3–1.9B | ~4% | | zkSync Era | $4.5B | ~9% | | All others (~69 rollups) | <$11B combined | <17% |
Optimistic rollups (Arbitrum, Base, Optimism) hold roughly 80% of DeFi TVL and 77% of secured value across the rollup ecosystem in 2026, according to multiple tracking sources including L2BEAT and DeFi Llama. The concentration ratio has tightened materially since 2024.
Base leads in daily transaction volume at approximately 12.89 million transactions per day with nearly 400,000 daily active addresses. On multiple occasions in early 2026, Base exceeded 6 million daily transactions. The top three networks together process close to 90% of all L2 transactions.
The mid-tier tells a different story. Linea's bridge deposits fell from $976 million in November 2025 to $367 million by May 2026, per data cited by The Block. Linea Bridge TVL currently sits at $258 million, down 13% in the past 30 days. Starknet's TVL stands at roughly $241 million — less than one-twentieth of Base's.
Median per-transaction fees in Q1–Q2 2026:
| Network | Median Fee | |---------|-----------| | Base | $0.02 | | OP Mainnet | $0.03 | | Arbitrum One | $0.04 | | zkSync Era | $0.05 | | Scroll | $0.06 |
Base emerged as the cost leader among major networks. According to 21Shares' mid-year report, Base was the only L2 that turned a profit in 2025, earning approximately $55 million. Coinbase does not break out Base revenue as a standalone line item, but its quarterly filings describe Base as a "growing contributor" to the company's technology segment.
The profitability gap matters because it determines which networks can sustain operations without perpetual subsidy. Smaller rollups that relied on liquidity-mining incentives and airdrop-farming cycles to attract users have seen activity collapse once those programs expired. Capital drained back toward venues with the deepest liquidity and largest user bases, according to The Block's 2026 Layer 2 Outlook.
StarkWare's revenue from Starknet plunged from a peak of nearly $6 million per month in late 2023 to approximately $48,000 in April 2026 — a decline exceeding 99%, per CoinDesk reporting. The company announced layoffs and a reorganization into two business units. StarkWare CEO Eli Ben-Sasson told employees the company needs to "take our technological superiority and convert it into meaningful revenue, meaningful usage."
RootData's 2026 dead-projects list records 99 crypto projects that have formally ceased operations through mid-year, spanning exchanges, lending protocols, NFT marketplaces, L2 networks, wallets, and developer tools.
Notable L2-adjacent shutdowns:
Loopring (zk-rollup DEX): Ceased all trading and relayer operations on June 28, 2026. TVL had collapsed from $760 million to $8 million. LRC token fell from $3.75 to approximately $0.01. The team cited weak user adoption, limited business development, and competition from newer zkEVM networks.
Lattice / Redstone (OP Stack chain for onchain gaming): Announced shutdown in April 2026. Redstone chain ceased service on May 16, 2026. Founded in 2021 to build "autonomous worlds," the team could not achieve a sustainable business model after five years.
Syndicate Labs (rollup infrastructure): Wound down in May 2026 after five years and a $20 million Series A from Andreessen Horowitz. The company cited a market that had "consolidated around a handful of dominant Layer-2 networks." It considered pivoting to rollup-as-a-service consulting but concluded the market was moving toward custom execution environments.
Three blockchain infrastructure projects shut down on the same day in May, according to CoinReporter, accelerating a consolidation trend that had been building for months. CryptoTimes tallied over 100 crypto project closures in H1 2026 alone, ranging from BitMEX to Leap Wallet.
The consolidation follows a pattern common in network-effect-driven markets. Three structural advantages separate the winners from the rest:
Distribution. Base benefits from Coinbase's 110+ million verified users and direct fiat onramps. Arbitrum captured the DeFi-native developer community early. OP Mainnet serves as the reference implementation for the OP Stack, which powers multiple chains including Coinbase's Base itself. Networks without a built-in distribution channel — an exchange, a wallet, or a protocol ecosystem — have struggled to sustain user acquisition once token incentives expire.
Liquidity depth. Arbitrum and Base together hold 77% of all L2 DeFi liquidity. This concentration is self-reinforcing: deeper liquidity attracts more traders, which generates more fees, which attracts more liquidity providers. Mid-tier rollups that lost their incentive programs saw capital migrate to the two deepest pools.
Sequencer economics. Running a sequencer is a fixed-cost business. Networks need sustained transaction volume to cover infrastructure costs and generate surplus. At Base's fee level of $0.02 per transaction and 12.89 million daily transactions, the math works. At a smaller rollup processing 50,000 daily transactions, it does not.
Alice Hou, a former research analyst at Messari, stated: "I think only a few L2s with clear financial demand will be able to sustain themselves over time."
Arbitrum DAO published its first-half 2026 income report in early September, providing the most detailed public accounting of any L2 network's economics. The DAO accrued $6.19 million from four income lines in H1 2026, with protocol gross margins above 97%.
Revenue sources:
The Robinhood Chain, launched on Arbitrum's Orbit stack, contributed $531,641 in 30-day revenue share to ArbitrumDAO's treasury by early September 2026. Daily revenue spikes on Robinhood Chain exceeded $1.9 million. Licensing fees accounted for 35% of the DAO's income in July, the first month Robinhood Chain was live on mainnet.
Average monthly stablecoin transfer volume on Arbitrum exceeded $70 billion in H1 2026. Stablecoin holders on the network increased 40% to 10.5 million. These are the metrics of a functioning settlement layer, not a speculative playground.
Ethereum's Glamsterdam upgrade, expected on mainnet between September and December 2026, targets a gas limit of 200 million, up from the current 60 million. The upgrade includes ten EIPs, notably ePBS (enshrined proposer-builder separation) and block-level access lists. It promises a 78.6% reduction in gas costs for both simple transfers and complex contract interactions.
For L2s, Glamsterdam expands the data-availability budget. The total number of data blobs will increase to 72 or more per block, widening the propagation window that rollups use to post transaction data back to Ethereum.
The paradox: this benefits large rollups disproportionately. Networks already posting high volumes of blob data — Base and Arbitrum — will see their per-transaction data costs fall further, improving already-positive margins. Smaller rollups that post infrequently see less marginal benefit. The upgrade hardens Ethereum L1 as a settlement layer, making the L2-centric roadmap more durable, but it does not solve the demand problem for chains that lack users.
zkSync Era represents a partial counterpoint to the optimistic-rollup dominance. With approximately $4.5 billion in TVL and over 100,000 daily active users as of mid-2026, it has recovered from a difficult 2025 to claim a meaningful share of the market.
However, zkSync continues to operate at thinner sequencer margins than optimistic rollups. ZK proof generation costs remain the primary constraint on its economics at current transaction volumes, according to Yellow Research. The technology's theoretical advantages in finality speed and withdrawal times have not yet translated into sufficient user-activity differentials to close the gap with Base and Arbitrum.
MegaETH, a newer entrant, is targeting near-real-time execution to compete with fast monolithic chains. Whether it can achieve distribution at scale remains unproven. The history of 2026 suggests that technical differentiation alone — without distribution and liquidity — is not sufficient to sustain an L2 network.
Three networks control 83% of L2 DeFi TVL. Arbitrum One ($17–18B), Base ($12.8–13.5B), and OP Mainnet ($1.3–1.9B) dominate. The remaining 70 rollups split less than $17 billion.
99 crypto projects have shut down in 2026. Notable L2-related closures include Loopring, Lattice/Redstone, and Syndicate Labs, all citing market consolidation as a primary factor.
Only Base was profitable in 2025 among L2s, earning approximately $55 million. StarkWare's monthly revenue fell 99% from peak, forcing layoffs.
Arbitrum DAO earned $6.19 million in H1 2026 with 97%+ gross margins, demonstrating viable unit economics. Licensing revenue from Robinhood Chain is accelerating.
Glamsterdam will reduce L1 gas costs by ~78.6%, benefiting high-volume rollups disproportionately and potentially widening the gap between leaders and laggards.
Distribution, liquidity depth, and sequencer economics — not technology — determine L2 survival. Networks without a built-in user base are running out of runway.
The Ethereum L2 market in 2026 resembles the cloud-computing market circa 2016: a period of rapid consolidation after an initial proliferation of providers. Three networks have captured the structural advantages — distribution, liquidity, and scale economics — that make their positions self-reinforcing. The remaining 70 rollups face a binary outcome: find a defensible niche (gaming, enterprise settlement, application-specific execution) or join the growing list of shutdowns.
21Shares projects 5–10 meaningful L2 networks by year-end. The data suggests that figure may be generous. Arbitrum's DAO revenue report shows what a sustainable L2 business model looks like: diversified income streams, high gross margins, and expanding institutional use (Robinhood Chain). Base demonstrates what distribution-led growth looks like: Coinbase's user base converts directly into transaction volume.
For the rest, the math is unforgiving. Running a sequencer costs money. Attracting users without subsidies requires an existing audience. Maintaining liquidity requires fee revenue. Networks that lack all three face the same trajectory as Loopring, Lattice, and Syndicate Labs.
The question is no longer which L2 technology is superior. It is which L2 business model is sustainable.