EIP-4844, commonly called “Proto-Danksharding” or “the blobs upgrade,” is one of Ethereum’s most impactful upgrades since The Merge. Activated in the Dencun hard fork on March 13, 2024, it introduced a new transaction type that allows Ethereum L2s to post compressed data to Ethereum at a fraction of the previous cost. Before EIP-4844, L2s posted their transaction data as Ethereum calldata — expensive, permanent, and priced by the byte at the full L1 gas rate. After EIP-4844, L2s use blobs — large binary data chunks that are cheaply priced via a separate blob fee market and automatically pruned from Ethereum nodes after ~18 days (they don’t need to stay forever since they’re only needed for the rollup verification window). The result: L2 transaction fees collapsed 10-100x within hours of the Dencun activation. Standard Arbitrum/OP Mainnet transactions that cost $0.20-0.50 dropped to $0.01-0.05.
Background: The Scaling Roadmap
The development phases are outlined below.
Why L2s Needed Cheaper Data
- Executing transactions off-chain (fast and cheap)
- Posting compressed transaction data to Ethereum L1 (for security/verifiability)
- Posting state root proofs (for fraud/validity proofs)
The bottleneck was step 2: posting compressed transaction data as calldata to Ethereum L1 was expensive. Calldata costs 16 gas per non-zero byte on Ethereum — and L2s were spending significant portions of their costs just on L1 data posting.
Pre-EIP-4844 reality:
- In busy periods, L2 fees tracked L1 fees because L1 calldata costs dominated
- The “cheap L2” promise was only fully realized during low L1 fee periods
- L2 fees regularly hit $0.50-$5 during Ethereum congestion
The Danksharding Vision
- Full danksharding: 64+ shards of data, distributed across validators, dramatically expanding Ethereum’s data bandwidth
- Proto-Danksharding (EIP-4844): The first step — implements blob transactions and the blob fee market, before the full distributed data-architecture is built
“Proto” because it implements the transaction format and fee market of the full design, but without the distributed sampling (EIP-4844 blobs are still downloaded by all full nodes, just pruned after 18 days).
How Blobs Work
The following sections cover this in detail.
Blob Transactions (Type 3 Transactions)
- Contains one or more “blobs” of up to 128 KB each
- Each blob is binary data in KZG polynomial commitment form
- Target: 3 blobs per block; max: 6 blobs per block (EIP-7691 later raised this)
- Blob fee market: separate EIP-1559-style fee market (base blob fee + tip)
KZG Commitments
- Commits to the blob data in a short (48-byte) form
- Allows anyone to verify a specific value within the blob without downloading the full blob
- Enables future data availability sampling where verifiers sample random chunks
This is the cryptographic primitive that will enable full Danksharding — but EIP-4844 introduced the commitment scheme before the full sampling network.
Blob Pruning
- 18 days >> 7-day fraud proof window for optimistic rollups (enough time for challenges)
- 18 days >> ZK rollup verification windows
- Not stored permanently reduces full node storage requirements
- Historical blob contents stored in voluntary “blob archiver” infrastructure, not required of all nodes
Why 18 days: Long enough for rollup verification to complete, short enough to not burden nodes with permanent storage of all rollup data forever.
Impact on L2 Economics
The following sections cover this in detail.
Fee Reductions
| L2 | Pre-Dencun avg fee | Post-Dencun avg fee | Reduction |
|---|---|---|---|
| Arbitrum One | $0.20-0.50 | $0.01-0.05 | ~10-20x |
| Optimism | $0.15-0.40 | $0.01-0.04 | ~10-15x |
| Base | $0.10-0.30 | $0.005-0.02 | ~15-30x |
| zkSync Era | $0.10-0.40 | $0.01-0.03 | ~10-30x |
| Starknet | $0.05-0.20 | $0.001-0.01 | ~50-100x |
Starknet saw the largest reduction because ZK validity proofs benefited most from cheaper data posting.
New Economic Dynamics
- L2s became so cheap that activity migrated from L1 to L2
- Ethereum L1 fee revenue (and ETH burn rate) decreased
- ETH briefly became inflationary post-Dencun
- Longstanding question: “Is ETH value accrual dependent on L1 fees, or does the modular model accrue value differently?”
EIP-7691: Increasing Blob Count (2025)
Following EIP-4844’s success, EIP-7691 was included in the Pectra upgrade to increase blob capacity:
- Raised target blobs per block: 3 → 6
- Raised max blobs per block: 6 → 9
- Further reduced L2 costs and increased Ethereum’s data availability throughput
The Danksharding Roadmap Ahead
Full Danksharding (future):
- PeerDAS: Peer Data Availability Sampling — allow validators to verify blob availability by downloading only a random sample (not full blobs)
- Blob capacity expansion: Target 32-64+ blobs per block
- Separate blob market layer: Full separation of execution and data availability
Full Danksharding remains years away, but EIP-4844 established the cryptographic and economic foundation.
Social Media Sentiment
EIP-4844 is considered unambiguously positive by the L2 ecosystem and widely by Ethereum developers. The 10-100x fee reduction was immediate and measurable — “I paid $0.003 for this swap” screenshots proliferated on same day as activation. It is commonly cited as the most impactful Ethereum upgrade since The Merge in terms of immediate user experience improvement. The main debate it sparked was about ETH value accrual: if L2s use cheap Ethereum as a data layer without paying high L1 fees, does ETH still accrue value? The “ultrasound money” narrative (ETH becoming deflationary from burn) weakened post-Dencun. Most Ethereum supporters accept this as the correct tradeoff — security should be cheap to use, and ETH value derives from being the canonical collateral and gas token of the ecosystem, not from extracting high rents from L2s.
Last updated: 2026-04
Related Terms
Sources
Buterin, V. (2021). An Incomplete Guide to Rollups. Vitalik.ca Blog.
Feist, D. (2022). Danksharding. Ethereum Research Forum.
Adlerjung, N., Al-Bassam, M., & Buterin, V. (2018). Fraud and Data Availability Proofs. arXiv.
Ethereum Foundation. (2023). EIP-4844: Shard Blob Transactions. eips.ethereum.org.
Thibault, L. T., Sarry, T., & Hafid, A. S. (2022). Blockchain Scaling Using Rollups: A Comprehensive Survey. IEEE Access.