EIP-4844, also known as proto-danksharding, is an Ethereum upgrade that gives Layer-2 rollups a dedicated, temporary data channel called a blob, separate from the regular execution gas market. By moving rollup transaction data off the expensive calldata path, the upgrade cut typical Layer-2 fees by more than 90% almost overnight.
Why posting data to Ethereum used to be so costly
Before this upgrade, a rollup like Arbitrum or Optimism had to publish every batch of compressed transactions as permanent calldata on Ethereum's main execution layer. That calldata competed directly with DeFi swaps, NFT mints, and every other on-chain activity for the same block space, so when the network got busy, rollups paid the same inflated gas prices as everyone else. Users footed that bill in the form of higher L2 transaction fees, even though the underlying L2 execution itself was cheap.
How the blob mechanism works
A blob is a packet of up to 128 KB of data that Ethereum consensus nodes store for roughly 18 days, long enough for fraud proofs or validity proofs to be checked, then discard. Because blobs are pruned rather than kept forever, they do not add to the permanent state that every full node must store indefinitely. Validators still attest that the data was available when needed, which is all a rollup needs to guarantee the security of the transactions it settles.
A separate market with its own price
Blob gas is priced through its own supply and demand curve, independent of standard execution gas. When regular gas spikes to 50 gwei during an NFT mint frenzy, blob gas can sit near its minimum of 1 gwei, because the two markets only interact at the margins. This decoupling is what allows rollups to keep fees low for users even while the base layer is congested elsewhere.
What this means for rollup economics
Sequencers now keep a larger share of the fees they collect, since their single biggest cost, L1 settlement, dropped sharply. That has made it viable for rollups to subsidize small transactions and experiment with near-zero fee products. The tradeoff is that blob capacity is still limited, so during periods when many rollups post data at once, the blob market can saturate and briefly push fees higher until the next batch of blobs expires and capacity frees up. Traders watching L2 fee trends should keep an eye on blob utilization data, since a sustained rise in blob demand is often the earliest signal of a coming fee increase.