ETH$3,842.17+1.84% ARB/L2TVL $14.7B+0.92% OP MAINNETTVL $7.4B-0.31% BASETVL $9.2B+3.41% zkSYNCTVL $1.6B+5.07% STARKNETTVL $0.84B-1.12% SCROLLTVL $0.71B+2.66% LINEATVL $1.04B+0.48% MANTLETVL $0.52B-2.18% BLASTTVL $1.29B+4.83% L2 TPS847.2+12.4% AVG FEE$0.0031-8.92% GAS L123 GWEI+1.10%
Volume IV / Issue 14 Saturday 2 May 2026 Edition: Global

layer-2.report

Data journalism & analytics for the Ethereum scaling layer

Cover Story/Rollup Economics

The $42.8 Billion Question: Has Layer 2 Already Won?

After eighteen months of relentless growth, optimistic and zero-knowledge rollups now custody more value than 70% of the Top-50 chains combined. We unpack the data that defines the new settlement order.

By Helena Marchetti · Senior Markets Correspondent ·
Cancun upgrade Q1 2026 ATH
Fig. 1 — Total Value Locked across all major Layer-2 networks, measured at quarter close. Shaded area indicates aggregate denominated in USD. Source: layer-2.report internal index.

For most of the previous decade, the conventional argument against Ethereum scaling solutions ran along a familiar groove: rollups would absorb traffic, but never custody. Capital, the thinking went, was sticky to the base layer because the base layer was where guarantees lived. That argument is no longer empirically defensible.

Aggregate L2 Total Value Locked closed Q1 2026 at $42.8 billion, an 18-month compound growth rate of 142%. During the same window, value held in non-Ethereum L1s declined by 7.4%. The migration is no longer a thesis; it is a balance-sheet event.

What remains contested is the geometry of the migration. Optimistic rollups, the technical incumbents, retain the dominant share of TVL through Arbitrum and Optimism. Yet user activity — measured in unique addresses and daily transactions — has tilted decisively toward a new cohort: Base, zkSync Era, and Linea, each of which crossed two million weekly active addresses in the past quarter.

“The question is no longer whether Layer 2 wins. The question is whose model of Layer 2 we live inside for the next decade.”

— Andreas Volker, Chief Strategist, Helix Capital

The economic implications are non-trivial. Sequencer revenue across the L2 cohort exceeded $382 million in Q1, a number that would have placed the category inside the top-15 software businesses by gross margin a decade ago. That revenue is increasingly distributed back to token holders and stakers through fee-sharing schemes that, on paper, resemble nothing so much as the dividend programs of late-twentieth-century industrial corporations.

Our analysis, drawing on on-chain data from 23 active L2 networks and proprietary settlement metrics, suggests three structural shifts now firmly in place: (1) the dominance of EIP-4844 blob space as the binding economic constraint, (2) the bifurcation between optimistic-rollup conservatism and ZK-rollup capital efficiency, and (3) the emergence of so-called ‘based rollups’ as the leading candidate for the next architectural paradigm.

The remainder of this report unpacks those three shifts in turn, and concludes with our forward indicators for Q3 2026.

Continue reading the full report →
$0.0B Total L2 TVL
0 Combined TPS
0 Active Networks
$0.000 Median Tx Cost
0% 18-mo CAGR
$0M 90-day Sequencer Rev.
Section 02 / Deep Analysis
Architecture

Optimistic Rollups Enter the Maturity Phase

Arbitrum and Optimism, having survived their fraud-proof activations, now resemble settled software businesses more than emergent protocols.

Combined TVL across the optimistic-rollup cohort exceeds $28.1 billion, representing 65.7% of the L2 total. Yet the year-over-year growth rate of the cohort is decelerating: 41% in the trailing twelve months versus 187% for the ZK cohort over the same period. The composition of activity tells a similar story. Optimistic rollups host the deepest pools of stablecoin liquidity, the most institutional bridges, and the longest-tenured DeFi blue-chips. They are, in a phrase, the L2 equivalent of mature dividend stocks.

What the architecture loses in growth velocity it gains in optionality. The Stage 2 decentralization roadmap published in late 2025 commits both leading optimistic chains to permissionless validation by Q4 2026. If executed, that step removes the primary criticism of the model — centralized sequencing — and converts the cohort into something approaching a credible settlement layer in its own right.

Pieter de Vries 8 min
Cryptography

The ZK Acceleration: Proving Costs Collapse 80%

Zero-knowledge rollups are no longer the future tense of Ethereum scaling. As of Q1 2026, they are the present-tense growth engine.

zkSync Era, StarkNet, Linea, and Scroll have collectively settled over 500 million transactions — a figure that places the ZK cohort within striking distance of the optimistic-rollup transaction count. The economics underpinning that growth are not narrative-driven. Median proof generation time has fallen from 218 seconds in early 2024 to 42 seconds today, a function of GPU parallelism, recursive proof composition, and dedicated proving hardware.

The downstream effect on user experience is concrete: ZK rollups now offer trust-minimized exits in under an hour, against the 7-day challenge window required by their optimistic counterparts. That single property has already begun to redirect institutional flows toward the ZK cohort, particularly from market makers whose capital efficiency calculus is exit-window-bound.

Yara Solberg 11 min
Frontier

Based Rollups: Sequencing Returns to L1

A small but increasingly serious cohort of architects argue that the next paradigm reattaches the sequencer to Ethereum’s validator set itself.

Based rollups — in which transaction sequencing is performed by L1 validators rather than by a dedicated rollup operator — recover a property the current cohort sacrificed: censorship resistance equivalent to the base layer. The cost is throughput variance and a more complex MEV market. As of this issue, six teams have testnet implementations in flight; two have committed to mainnet during 2026.

The thesis is not without critics. Sequencing-as-a-service providers argue that the optimization surface of a dedicated sequencer cannot be matched by an L1 validator running on commodity hardware. That argument is empirically testable, and the next twelve months will test it.

Marcus Adeyemi 7 min
Section 03 / The L2 League Table

The following table tracks the ten largest Layer 2 networks by Total Value Locked. Sparkline indicators show 30-day TVL trajectory; weekly active addresses (WAA) are taken as a 7-day rolling mean. All figures denominated in USD; data current as of the most recent block at issue close.

# Network Architecture TVL (USD) 7d Δ 30-day Trend WAA Median Fee
01 Arbitrum One Optimistic $14.74B +0.92% 3.84M $0.018
02 Base Optimistic $9.21B +3.41% 5.12M $0.011
03 OP Mainnet Optimistic $7.43B -0.31% 2.07M $0.014
04 zkSync Era ZK $1.62B +5.07% 2.91M $0.0024
05 Blast Optimistic $1.29B +4.83% 1.18M $0.009
06 Linea ZK $1.04B +0.48% 2.06M $0.0031
07 StarkNet ZK $0.84B -1.12% 0.78M $0.0019
08 Scroll ZK $0.71B +2.66% 0.91M $0.0028
09 Mantle Optimistic $0.52B -2.18% 0.46M $0.012
10 Manta Pacific Optimistic $0.38B +1.34% 0.34M $0.0042

† Data composited from public block explorers and node-level instrumentation operated by the layer-2.report data desk. Methodology: layer-2.report/methodology. WAA ≡ weekly active addresses (7-day rolling). Architectures categorized by canonical specification.

Section 04 / Daily Brief

The Day in Layer 2

A morning roundup of pricing, governance, and on-chain anomalies, compiled by the layer-2.report data desk between block height 22,148,021 and 22,153,447.

Trading on the optimistic-rollup cohort opened the session in modestly positive territory, with Arbitrum’s native token printing a 0.92% gain on volume of $74M during European hours. The advance was driven principally by a 14.7% increase in DEX-denominated open interest, suggesting derivative flow rather than spot accumulation. Base, the day’s relative outperformer, gained 3.41% on volume more than 2.4× its trailing-30-day average, with the move concentrated in the post-US-open window.

The ZK cohort presented a more dispersed picture. zkSync Era extended its multi-week recovery, advancing 5.07% as proving-time metrics published overnight showed a 12% sequential improvement. StarkNet declined 1.12%, retracing a portion of the prior week’s rally on what desk traders described as routine profit-taking ahead of the protocol’s scheduled v0.13 upgrade.

In governance, the Optimism Collective opened voting on Proposal RFC-2026-04, which would reallocate 4.2% of retroactive funding toward ZK-circuit development. Early signals suggest passage; quorum was reached within 19 hours of proposal posting, with the ‘for’ tally currently at 73.4% of cast votes.

On-chain anomaly detection flagged a 4,200 ETH bridge from Arbitrum to Base early in the European session, identified by clustering heuristics as belonging to a single counterparty. The transfer is consistent with a previously observed pattern of routine market-maker rebalancing rather than directional flow.

Blob fees on the L1 closed the session at 23 gwei, within the 21-26 gwei range that has characterized the post-EIP-4844 equilibrium for the past nine weeks. Blob fill rates registered 87.3% during peak hours — the highest sustained reading since the protocol upgrade activated — reinforcing the narrative that L2 traffic now operates within a binding data-availability constraint.

Looking ahead to tomorrow’s session, the desk is monitoring three signals: (i) the response of L2 sequencer revenue to the recent uptick in L1 gas, (ii) the conclusion of voting on the Optimism RFC referenced above, and (iii) whether the Base outperformance persists into a fourth consecutive session, a threshold that historically marks the start of structural rather than tactical flow.

Section 05 / Opinion
Editorial

The Case for Patience with Based Rollups

In a market that prizes velocity above almost every other property, the slowest-moving architectural proposal in the rollup space may be the one most worth taking seriously.

AV

Andreas Volker

Editor-at-Large

The first principle of any economic infrastructure is that incentives ought, eventually, to align with the interests of the parties they secure. Measured against that principle, the existing rollup cohort fails a test: their sequencers are operated, almost without exception, by the protocol team rather than the asset holders. The arrangement is permitted by economic realities rather than endorsed by them.

Based rollups invert the arrangement. In binding sequencing to L1 validators, they recover a property — censorship resistance equivalent to the base layer — that the existing cohort accepts the cost of forfeiting. The cost of that recovery is real: throughput becomes variable, MEV markets become more complex, and protocol revenue is shared with the validator set rather than retained by the rollup operator. None of those costs is fatal.

The argument against based rollups, when made in good faith, is principally an argument from optimization: that a dedicated sequencer can serve users better than an L1 validator running on commodity hardware. That is, in fact, true today. It will likely remain true for some defined window of years. But the more interesting question is not which model wins on a given Tuesday in 2026, but which model is structurally legitimate at scale and across decades. By that measure, the case for patience with based rollups is harder to dismiss than market-cap rankings would suggest.

This publication does not, as a rule, take positions on which architectures will win. We do, however, take the position that the question of legitimacy is worth more attention than it currently receives. The next twelve months will tell us whether based rollups can ship. The decade after will tell us whether they should.