layer2.report Issue No. 047 · Q2 2026
Quarterly Briefing Published 02 May 2026 · 14 min read

Layer-2 Throughput Crosses 14M Daily Transactions as Settlement Costs Compress to a Two-Year Low

In Q2 2026 the aggregate Layer-2 ecosystem processed a record 14.2 million transactions per day, a 38% year-over-year increase, while average settlement fees fell to $0.018 — the lowest reading since calldata blob pricing was introduced. This briefing aggregates on-chain telemetry, audit findings, and bridge-risk indicators across 24 networks.

0.0M Daily Transactions ▲ 38.4% YoY
$0.000 Avg. Settlement Cost ▼ 64.1% YoY
$0.0B Total Value Secured ▲ 21.9% YoY
0.0% Avg. Network Uptime ▲ 0.4 pp
Section 01 · Performance

Throughput, Latency, and the Compression of Calldata Costs

Across the 24 monitored networks, throughput growth accelerated through the quarter while finality times trended downward. The dataset below isolates rollup-level performance from generalised execution chains.

The headline figure — 14.2 million daily transactions across the L2 set — masks a meaningful divergence between optimistic and zero-knowledge architectures. Optimistic rollups grew 31.2% quarter-over-quarter, propelled by improved batch posting cadence and the broader rollout of fault-proof systems. Zero-knowledge rollups, by contrast, expanded 47.8% as proving costs fell and circuits matured to support a wider opcode surface.

Median transaction finality on the leading ZK networks now sits below seven minutes for full-state proofs, compared with the seven-day challenge window that still defines optimistic settlement. This gap is increasingly material for institutional flows, where bridge withdrawal latency directly impacts capital efficiency.

Network Architecture TPS (90d) Avg. Fee QoQ Δ
Arbitrum One Optimistic 42.18 $0.014 +18.4%
OP Mainnet Optimistic 28.71 $0.011 +12.6%
Base Optimistic 61.04 $0.009 +44.2%
zkSync Era ZK 19.22 $0.024 +27.1%
Starknet ZK 14.66 $0.031 +33.8%
Linea ZK 11.92 $0.022 +52.0%
Scroll ZK 8.41 $0.026 +61.7%
Polygon zkEVM ZK 6.33 $0.019 -4.2%

[1] Throughput readings are the trailing 90-day mean of confirmed transactions per second, sampled at one-minute intervals from public RPC endpoints. [2] Average fee is reported in USD at the median ETH price for the quarter. [3] QoQ change uses Q1 2026 as the reference period.

Section 02 · Security

Audit Coverage, Prover Liveness, and the State of Fault Proofs

Security posture is improving in aggregate, though the distribution remains uneven. Eleven of 24 networks have live fault-proof or validity-proof systems with no privileged escape hatch.

The maturation of fault-proof systems on optimistic rollups represents the most significant security development of the quarter. Three additional networks transitioned to permissionless challenge mechanisms, removing single-entity withdrawal control. Validity-proof networks continued to focus on proof-system audits, with two completed Tier-1 reviews of recursion circuits.

Bridge security remains the dominant risk vector. Of the eight incidents catalogued this quarter, six involved canonical or third-party bridge infrastructure rather than rollup state machines themselves. The aggregate value at risk in bridges with fewer than three independent audits stood at $4.1B at quarter-end.

Tier 1

Permissionless Fault Proofs

11 / 24 networks

Networks with live fault-proof systems and no privileged sequencer override.

Tier 2

Multi-Auditor Coverage

19 / 24 networks

Two or more independent audit firms with public reports in the trailing 12 months.

Tier 3

Bug Bounty Active

22 / 24 networks

Public bounty program with critical-tier payout exceeding $1M.

Watchlist

Single-Auditor Risk

5 / 24 networks

Networks relying on a single audit firm or with stale (12mo+) audit coverage.

[1] Audit ratings reflect a synthesis of L2BEAT stages, public report inventories, and on-chain governance configuration. [2] Tier definitions are documented in the methodology appendix.

Section 03 · Ecosystem Risk

Concentration, Bridge Exposure, and Governance Surface Area

The top three networks now custody 71.4% of all Layer-2 value. Concentration risk is partially offset by improved governance transparency, though privileged upgrade keys remain the dominant systemic concern.

Total value secured concentration remains elevated. Arbitrum One, OP Mainnet, and Base together secure $30.5B of the $42.7B aggregate. While individual network audits are robust, correlated downside in any of the three would impact a majority of the L2 user base. Governance attack surface — measured by upgrade timelocks, multisig thresholds, and emergency-pause keys — improved on six of 24 networks this quarter.

Bridge architecture continues to dominate the catastrophic-loss tail. Canonical bridges with deterministic state-root verification carry materially lower risk than third-party LP-based bridges, but the latter still account for the majority of cross-rollup value flow. The risk register below tracks the most material exposures.

Bridge Concentration
86 / 100
High
Sequencer Centralisation
74 / 100
High
Governance Surface
58 / 100
Moderate
DA Layer Dependency
41 / 100
Moderate
Audit Recency
22 / 100
Contained

[1] Risk scores normalised 0–100 with higher values indicating higher exposure. [2] Methodology mirrors the ECB's systemic-risk dashboard structure, adapted for distributed-ledger telemetry.

Section 04 · Archive

Prior Briefings and Methodology References

Quarterly briefings remain available in their original form. Each issue links to the dataset snapshot used at publication, and methodology revisions are versioned in the appendix.

  • No. 046 Q1 2026 Calldata Blob Pricing After One Year: A Settlement-Cost Audit
  • No. 045 Q4 2025 Fault-Proof Adoption Reaches Critical Mass on Optimistic Networks
  • No. 044 Q3 2025 The Economics of ZK Recursion: Proving Costs and Their Trajectory
  • No. 043 Q2 2025 Sequencer Decentralisation: Eight Approaches, Three Winners
  • No. 042 Q1 2025 Cross-Domain Liquidity: Mapping the Bridge Topology

[1] All historical issues are immutable and reflect the data available at publication. [2] Methodology v3.4, current as of this issue, supersedes v3.3 published in Q4 2025.