What is a Blob-Carrying Transaction?
A blob-carrying transaction is the new Ethereum transaction type introduced by EIP-4844 (also called “Proto-Danksharding”). It allows a transaction to attach a large data “blob” — a Binary Large Object — that rollups use to post their transaction data. Before EIP-4844, Layer 2 rollups had to publish their data as regular calldata, competing for the same expensive block space as every other transaction. Blob-carrying transactions give rollups a separate, far cheaper data market, which is why L2 fees dropped by 10–100x after the upgrade went live on Ethereum mainnet in March 2024.
The bottom line: blob-carrying transactions are the mechanism that made rollups genuinely cheap. They are the “proto” step toward full danksharding, delivering most of the cost benefit with a fraction of the engineering complexity. See also EIP-4844 and Blob.
How Blob-Carrying Transactions Work / Technical Details
The Problem: Calldata Was Too Expensive
An optimistic or ZK rollup must publish its transaction data on Ethereum so that anyone can reconstruct the L2 state and challenge fraudulent or invalid transitions. Pre-EIP-4844, this data went into transaction calldata, which is priced per byte and competes in the same gas market as smart-contract execution. As L2 usage grew, calldata costs dominated rollup operating expenses and user fees.
The Solution: A Separate Blob Market
EIP-4844 introduces a new transaction type — type-3 (blob) transactions — that can attach one or more blobs:
- Each blob is a fixed ~128 KB data payload
- A block can include a bounded number of blobs (the target and max blob counts per block)
- Blobs are stored separately from execution state and pruned after ~18 days — they are not needed forever, only long enough to challenge fraudulent withdrawals
- Blob space has its own gas market, priced in a new unit called a blob-base-fee, separate from the regular base fee for execution
Why Separate Pricing Matters
By giving blobs their own fee market, EIP-4844 follows the same EIP-1559 design as execution gas: a target blob usage with an adjustment mechanism that raises blob fees when demand is high and lowers it when demand is low. The crucial point is that a surge in rollup data posting no longer drives up the cost of an ordinary token transfer — the two markets are decoupled. This is what produced the dramatic L2 fee reductions.
What Blobs Are Not
- Blobs are not executable. The EVM cannot read arbitrary blob data during execution; a rollup’s smart contracts can only verify a commitment (a KZG commitment) to the blob’s contents, not execute from it.
- Blobs are not permanent. They expire and are pruned, because rollups only need them during the challenge window for optimistic withdrawals.
- Blobs are not general storage. They exist specifically for rollup data availability, not for arbitrary DApp data.
Notable Examples and Impact
The March 2024 “Dencun” Upgrade
EIP-4844 went live with the Dencun hard fork on March 13, 2024. The impact on L2 fees was immediate and dramatic: major rollups like Arbitrum, Optimism, Base, and zkSync saw transaction fees fall by an order of magnitude or more. Transactions that previously cost dollars dropped to cents.
L2 Adoption Surge
Cheaper data posting reshaped the L2 landscape. Base, in particular, leveraged cheap blobs to drive massive user growth and transaction volume, becoming one of the busiest chains by activity. The cost reduction also enabled new use cases — micropayments, high-frequency DeFi, and gaming — that were infeasible under the old calldata model.
Blob Fee Spikes
Because blob capacity is limited (a fixed number of blobs per block), periods of intense rollup activity can still spike the blob-base-fee. During popular L2 launches or airdrop farming, blob fees have risen sharply before settling — a reminder that blobs are cheaper than calldata, but not free.
The Path to Full Danksharding
EIP-4844 is the “proto” version. Full danksharding will scale blob capacity further using data-availability sampling and peer-to-peer blob distribution, enabling vastly more L2 throughput. EIP-4844 reuses the same transaction format and data structures so that the upgrade path is smooth: rollups adopted blob-carrying transactions once and keep working as capacity grows underneath them.
How to Understand and Use Blob Transactions
For L2 Users
- You don’t interact with blobs directly. Your L2 wallet transaction is bundled by the rollup, which posts aggregated data to Ethereum as a blob. You simply enjoy the low fees.
- Fees can still spike during congestion. When the blob market is hot (e.g., a major airdrop), L2 fees rise — just far less than they would have under calldata.
For Rollup Developers
- Submit data as blobs, not calldata. Using the type-3 transaction format and KZG commitments is what captures the cost savings.
- Budget for blob-base-fee volatility. Capacity is bounded; design fee mechanisms that pass blob costs through to users transparently.
- Handle blob expiry. Since blobs are pruned after ~18 days, ensure your challenge and data-reconstruction logic operates within that window or relies on your own archival storage.
For the Ecosystem
The move to blob-carrying transactions validated the rollup-centric roadmap: rather than scaling the L1 directly, Ethereum provides cheap data availability for many L2s to scale on top. Future upgrades (peer-DAS, full danksharding) extend this design rather than replacing it.
Frequently Asked Questions
Q: Do blob transactions make L1 (Ethereum mainnet) transactions cheaper? A: Not directly. They make L2 transactions cheaper by giving rollups a separate data market. Ordinary L1 transactions still use execution gas. The benefit is that rollup data posting no longer competes with and inflates L1 gas prices.
Q: What happens to old blob data? A: It is pruned after roughly 18 days. Blobs only need to live long enough to allow challenging fraudulent rollup withdrawals; after that, rollups are responsible for their own archival storage.
Q: Are blobs the same as danksharding? A: No. EIP-4844 (proto-danksharding) introduced blobs with limited capacity and a simple fee market. Full danksharding will scale that capacity enormously using data-availability sampling while keeping the same blob transaction format.