Pectra expands Ethereum blob capacity and reduces rollup costs

Pectra increased the target blobs per block to 6 and the maximum to 9. This change increased daily data capacity to 8.17GB. Before the upgrade, the network maintained a target of 3 blobs and a maximum of 6, which provided 5.5GB of capacity per day. Rollups purchase 20.8% more blobs daily than they did pre-upgrade. Since Pectra went live, the daily number of blobs purchased by rollups increased from 21,200 to 25,600. Despite this increase, the average number of blobs per block stays 33% below the new target of 6. This means the network currently uses only 3.3GB of data capacity daily, which represents 40% of the maximum available capacity and 61% of the target rate. In the 60 days leading into the upgrade, rollups purchased 21,200 blobs per day, providing 2.7GB of capacity daily compared to 3.3GB today. Each blob carries up to 128KB of data. Nodes must hold onto rollup blob data for at least 18 days before they can remove it from their machines. This increase in purchased blobs pushed the amount of data consensus layer nodes must maintain to a new high estimate of 44.6GB.

EIP-7762 and the discovery problem

EIP-7762 addresses price discovery for blob space. It sets the MIN_BASE_FEE_PER_BLOB_GAS to 2**25. This change prevents the long delay seen when trying to climb from 1 wei to a significant price. When demand for blobs exceeds supply, the base fee must climb many factors of 2 to reach the prevailing price, but reaching those levels from 1 wei takes a long time because the growth is exponential. EIP-7762 also resets excess blob gas to 0 to avoid a base fee spike during the fork. One approach to setting this minimum involves looking at the cost of simple transfers when the base fee is 1 GWEI, which equals roughly 5 cents USD at current prices. You already know that low fees benefit users, but I find the minimum fee floor necessary to prevent market stagnation. If the base fee stays at 1 wei, the market lacks the responsiveness needed when demand spikes.

Rollup fee reductions for sequencers

Rollup sequencers saw significant cost improvements. Fees for Base, Arbitrum, and Optimism dropped 40% to 60% within weeks of the activation. Total daily costs for purchasing and executing blobs declined by 51%. In the first quarter of 2026, the median fee for Arbitrum was $0.002183 while Optimism saw fees of $0.000027. Rollups purchase 20.8% more blobs daily than they did pre-upgrade. The amount of ETH burned from rollup data posting through blobs dropped by 71% after Pectra activation. Since Pectra went live, rollups have paid no more than $0.0000092 daily. This represents a near 100% decline in blob object fees paid to Ethereum by rollups, as the daily average fell from $16,250 to less than one-thousandth of a penny. While the amount of ETH burned from blob activity fell, the amount of ETH burned from type-3 execution layer transactions remains high. The reduction in blob costs improved the profit margins of rollups in both relative and absolute terms.

Scaling and the target demand paradox

The target demand paradox creates issues for scaling. If the target falls outside the range of effective demand, the market cannot reach equilibrium. For the EIP-7691 configuration, effective demand falls between 6.57 and 6.72 blobs. Since the target remains 6, the market might fail to discover a fair price even if usage meets the target. EIP-7918 allows thresholds to adapt with execution costs or blob quantity, but EIP-7762 does neither. The reliance on EIP-7762 to fix the discovery problem remains an imperfect solution for a scaling network.

The scaling quadrant defines how these thresholds behave. EIP-7918Q adapts with blob quantity, while EIP-7918E adapts with the execution base fee. Full EIP-7918 adapts with both. As data availability scales, blob consumers can scale up consumption to amortize fixed execution costs like proof verification or batch posting. As the quantity of blobs per block increases, the upper end of the long-run ETH-denominated price per blob will fall. Will the network force consumption past 6.57 blobs to reach equilibrium?