Median transaction fees on Arbitrum One, Base, Optimism and zkSync Era have all dropped below a single U.S. cent this week, the lowest sustained level since Dencun originally introduced blob-based data availability in 2024. The relief follows months of tight blob throughput, during which spikes in NFT mints and memecoin trading repeatedly pushed Layer-2 fees back into double-digit cents, and arrives directly on the back of the Pectra upgrade's blob-target expansion that activated on Ethereum mainnet last week.
Pectra's parameter changes effectively double sustained blob throughput, raising the per-block target from three blobs to six and lifting the burst ceiling from six to nine. The change finally gives rollups headroom that comfortably exceeds current organic demand. A simple ERC-20 transfer on Base now costs roughly the same as posting a tweet — except it settles to Ethereum and is auditable on a public ledger. A Uniswap V3 swap on Arbitrum, the highest-fee on-chain action that meaningfully matters to retail users, runs at under three cents at the median.
The numbers in context are striking. Median Base fees, which spent most of the spring oscillating between $0.04 and $0.20, have stabilized at roughly $0.005. Optimism is at $0.003. Arbitrum, whose throughput is the highest of the three, sits at $0.002 for a basic transfer. zkSync, where the fee mechanism is structurally different, has produced similar compression. Aggregate daily transaction count across the four major rollups is at all-time highs, more than twenty million per day, with the cost reduction passing through to users almost directly in the form of lower per-action expense rather than being captured by sequencer margin.
Whether the new equilibrium holds depends on demand growth. Several Layer-2 teams say their internal projections suggest blob capacity is now ahead of organic transaction growth for at least the next two quarters. The framing matters because the Dencun-era equilibrium broke down precisely when episodic demand spikes — memecoin launches, large NFT mints, on-chain experiments by major application developers — exceeded the lower blob-throughput ceiling. With the headroom widened, the same kinds of spikes should now be absorbable without driving median fees back into double-digit cents.
For the broader execution-layer ecosystem, the post-Pectra fee level is the cleanest current evidence that Ethereum's tiered execution stack is delivering its design promise. Where in 2023 the conversation was whether rollups could meaningfully scale Ethereum at all, the conversation in mid-2025 is whether the chain has scaled too aggressively into a regime of fee oversupply, with the implication that Layer-2 sequencer revenue compresses unfavourably for the operators who built their unit economics around higher fees. Coinbase's Base team has publicly acknowledged that sequencer revenue is now roughly half of its first-quarter 2024 levels.
The next data points to watch are the blob-fee market behaviour during the next demand spike — a major NFT mint, a memecoin launch event, or any sustained episode of network congestion — and the on-chain user-cohort metrics that track whether the cheaper fee level expands the addressable user base. The Ethereum core developers are already discussing further blob-target expansions in subsequent forks, with PeerDAS and danksharding the multi-year north stars; the question for the second half of the year is whether the immediate post-Pectra equilibrium produces the kind of transaction-volume growth that justifies further parameter tuning, or whether organic demand falls short of the new capacity ceiling and additional expansion becomes a less urgent priority.