What Blob Storage Actually Changed About Ethereum’s Fee Market
When EIP-4844 shipped with the Dencun upgrade in March 2024, most coverage focused on the immediate price drop for rollup transactions - and that drop was real. Base, Optimism, and Arbitrum users saw fees fall by orders of magnitude almost overnight. But the more lasting change wasn’t the price reduction; it was the introduction of a separate fee market for a new type of data entirely.
Before Dencun, rollups posted their transaction data as calldata to Ethereum’s execution layer. Calldata is permanent - it lives in the blockchain forever - and it competed directly with every other transaction for block space. That meant rollup data costs tracked whatever L1 congestion looked like at any given moment. During high-traffic periods, rollups became nearly as expensive to use as L1 itself.
Blobs changed the underlying structure. EIP-4844 introduced a new transaction type that carries data attachments - blobs - priced through their own independent fee mechanism. The protocol targets a specific number of blobs per block (currently six, with a hard cap of nine), and the blob base fee adjusts up or down using an EIP-1559-style algorithm depending on whether recent blocks hit that target. Crucially, blob data is not stored by execution nodes permanently. It’s available for roughly 18 days - long enough for fraud provers and validity verifiers to do their work - then pruned.
Two Markets, Not One
This matters because Ethereum now runs two functionally separate resource markets inside the same block. Execution gas covers computation and permanent calldata. Blob gas covers temporary data availability. Neither market directly feeds into the other’s pricing.
The practical effect: rollup fees stopped being a direct function of L1 speculation. A surge in NFT minting or DeFi activity that drives up execution gas doesn’t necessarily touch blob fees at all, because blobs have their own supply curve.
That said, the system isn’t frictionless. When blob space fills up consistently - which happened during several high-activity periods in 2024 - blob fees spike independently. Rollups are still subject to congestion, just their own kind of congestion rather than sharing it with every other Ethereum user.
The Danksharding Direction
EIP-4844 is explicitly a stepping stone. Full danksharding would expand blob capacity significantly and introduce data availability sampling, allowing light clients to verify blob availability without downloading entire blobs. The blob fee market architecture is already designed to accommodate that expansion - the target and max blob counts are protocol parameters that can be adjusted through future upgrades without rearchitecting the fee mechanism itself.
What Dencun actually shipped was a production-grade fee market for temporary data, running live on mainnet. That’s the part worth paying attention to, separate from any particular fee number on any particular day.