Transaction volume on major Layer 2 networks rose from a measured scaling experiment into the dominant settlement layer for retail-sized activity. The migration is not uniform: throughput concentrates around sequencer economics, liquidity gravity, and application-specific bursts. This report tracks where the expansion is durable, where it is subsidized, and where the next bottleneck has already appeared.
The settlement shift is visible in the shape of ordinary transactions.
The defining change in the last twelve months is not simply that Layer 2 networks processed more transactions. It is that the category of transaction changed. Transfers, swaps, game actions, and social writes that would have been priced out of Layer 1 now appear as continuous baseline demand. Arbitrum's daily activity has held a rising channel even after incentive epochs rolled off.
The reportable signal is durability. Networks with stable fee markets and deep application liquidity show lower variance after campaigns end. Networks dependent on points programs still register dramatic peaks, but their troughs return close to the pre-campaign line.
Lower fees changed user behavior before they changed revenue models.
Fee compression widened the feasible design space for applications. The strongest evidence is the disappearance of the old transaction cadence: users no longer batch actions into one careful operation when a sequence of low-cost interactions is available. Base and Optimism show the clearest behavioral delta , with average actions per active address rising faster than active address count.
| Network | Avg fee | Actions / address | QoQ change |
|---|---|---|---|
| Base | $0.018 | 7.4 | +31% |
| Arbitrum One | $0.026 | 5.9 | +18% |
| Optimism | $0.021 | 6.2 | +24% |
| zkSync Era | $0.033 | 3.8 | +9% |
Revenue, however, has not followed volume in a straight line. The cheapest networks produce the most visible consumer activity and the thinnest immediate margins. The economic question has moved from blockspace scarcity to retention, routing, and the value of the sequencer relationship.
Liquidity still decides which execution environments feel real.
Technical capacity does not automatically become economic depth. The networks that feel immediate to users are those where stablecoins, blue-chip collateral, and market makers arrived before application demand. Liquidity remains the hidden substrate underneath the transaction chart.
Where liquidity is shallow, the user pays indirectly through slippage, bridge delay, or fragmented balances. This is why some technically impressive environments remain analytically peripheral: their charts show capacity, but their tables show insufficient financial gravity.
The next constraint is not throughput. It is trust surface.
The throughput story is maturing into a governance and risk story. Users experience Layer 2 as cheaper execution, but institutions underwrite a more complicated stack: sequencer control, upgrade keys, proof maturity, data availability dependencies, and bridge assumptions.
The mismatch between adoption and trust minimization is the central unresolved finding. Activity has moved faster than decentralization. The best networks now compete less on peak TPS than on how rapidly they can make their operating assumptions legible, constrained, and eventually boring.
Appendix / methodology
Transaction counts are aggregated from public explorers, sequencer batch submissions, and network data portals. Duplicate replays, failed spam bursts, and incomplete outage windows are removed where corroborating sources permit. Fee data reflects user-paid execution fees converted at daily median ETH prices.
Liquidity measures combine canonical bridge balances, stablecoin supply, and protocol-reported collateral. The report uses rolling medians unless otherwise stated. Charts are presented as normalized indices to emphasize relative direction over absolute rankings. Colophon: Space Grotesk, Inter, JetBrains Mono; ruled in cool grays with one signal blue.