EIP-4844 targets data, not execution. It introduces blob-carrying transactions to provide cheap, ephemeral data storage for L2s, separating data posting costs from the EVM's computational gas market.
the-ethereum-roadmap-merge-surge-verge
Blog
Why EIP-4844 Doesn’t Touch Execution Gas
A technical breakdown of why Proto-Danksharding is a data availability upgrade for rollups, not a mainnet execution layer gas reduction. It's about L2 economics, not your mainnet swap.
introduction
THE SEPARATION OF CONCERNS
The Misconception
EIP-4844 is a data availability upgrade, not an execution layer change, leaving L2 gas fees fundamentally unchanged.
Execution gas is a separate market. L2 transaction fees are the sum of data posting costs (now blobs) and execution/state growth costs (still L1 calldata). EIP-4844 slashes the first component only.
The bottleneck shifts to execution. With cheap data secured, the next constraint is L1 block gas limit for proving and finalizing L2 state updates. This is why Arbitrum and Optimism still face non-zero base fees.
Evidence: Post-4844, data costs fell ~90%, but L2 fees didn't drop proportionally. The remaining fee is the execution overhead for L1 sequencers and provers like those used by zkSync Era.
key-trends
DECOUPLING DATA FROM EXECUTION
The Surge's Real Target: Rollup Economics
EIP-4844's proto-danksharding is a surgical strike on data availability costs, not a general-purpose gas reduction for on-chain apps.
01
The Problem: L2s Pay for Permanence, Not Computation
Rollups like Arbitrum and Optimism spend ~80-90% of their transaction cost posting data to Ethereum's expensive calldata. Execution on the L2 is already cheap; the bottleneck is paying Mainnet to remember the data forever.
- Cost Structure: L2 User Fee = (Execution Cost) + (Data Publishing Cost).
- Real Bottleneck: Data publishing dominates, often costing $0.10-$0.50+ per transaction at scale.
~90%
Cost is Data
$0.10+
Base Tx Cost
02
The Solution: Blobs as a Separate Auction Market
EIP-4844 introduces blob-carrying transactions with a separate fee market from execution gas. This creates a dedicated, volatile marketplace for data bandwidth, decoupling it from NFT mints and DeFi liquidations.
- Market Isolation: Blob gas prices fluctuate independently, preventing congestion spillover.
- Ephemeral Storage: Data is stored for ~18 days, not forever, aligning cost with actual utility for fraud/validity proofs.
~18 days
Data Storage
10-100x
Cheaper Data
03
The Result: Hyper-Scalable, Sustainable L2 Business Models
With data costs reduced by 10-100x, the economic model for rollups flips. Profit margins improve, enabling sustainable sequencer revenue and paving the way for mass adoption via true sub-cent transactions.
- New Competition: Focus shifts to execution performance and UX between zkSync, Starknet, and Base.
- VC Takeaway: Infrastructure value accrues to L2s and bridging/sequencing stacks like Espresso Systems and Astria.
>100 TPS
Per L2 Target
<$0.01
Target Tx Cost
deep-dive
THE DATA LAYER
Architecture, Not Accident: Separating Data from Execution
EIP-4844's design isolates data availability from execution to create a scalable, modular foundation for L2s.
EIP-4844 is a data layer upgrade. It introduces blob-carrying transactions that provide cheap, temporary data availability for L2s like Arbitrum and Optimism. It does not modify the EVM's execution logic or gas costs for computation.
Execution and data are separate concerns. The EVM processes logic; blobs store data. This separation prevents execution-layer bloat and keeps base layer gas predictable for apps like Uniswap or Aave.
Blob gas is a parallel market. It uses a separate fee market from execution gas, decoupling L2 data posting costs from mainnet congestion. This is why EIP-4844 doesn't touch execution gas.
Evidence: Post-EIP-4844, L2 transaction fees are 90% data cost. Arbitrum and Base now submit proofs referencing blob data, not calldata, slashing L1 settlement costs without altering Ethereum's core execution engine.
EXECUTION VS. DATA AVAILABILITY
Gas Market Segmentation: Pre vs. Post EIP-4844
A comparison of how gas fees are allocated and priced for different network resources before and after the implementation of EIP-4844 (Proto-Danksharding).
| Gas Market Segment | Pre EIP-4844 (Legacy) | Post EIP-4844 (Proto-Danksharding) | Impact |
|---|---|---|---|
Primary Resource | Execution & Data in same block | Separate Blobs for data | Decouples pricing |
Fee Market Mechanism | Unified 1559 auction for all gas | Dual markets: Execution (1559) & Blob (1559) | Prevents data from crowding out execution |
Pricing Metric | Gas (21k gas per non-zero byte) | Blob Gas (~0.125 gas per byte) | ~99% cheaper calldata for L2s |
Target Resource Usage | 12.5 MGas / 30M gas block | 3 Blobs (0.375 MB) / block | Increases data bandwidth 10x |
Persistence Duration | Permanent on-chain (forever) | ~18 days (then pruned by nodes) | Reduces node storage burden |
Direct Beneficiary | Ethereum validators | Ethereum validators & L2s (via lower costs) | L2 transaction fees drop 10-100x |
Interacts with EVM | Yes, via CALLDATA opcode | No, accessed via precompile (0x0A) | Execution gas is untouched; pure scaling layer |
Congestion Driver | All transactions compete | L2 batch submissions isolated | Execution gas volatility reduced for users |
counter-argument
THE EXECUTION BOTTLENECK
The Steelman: "But Cheaper L2s Should Ease Mainnet Congestion..."
EIP-4844 reduces L2 data costs but does not address the fundamental gas cost of executing transactions on Ethereum mainnet.
EIP-4844 targets data, not execution. The upgrade introduces blob-carrying transactions for cheaper L2 data availability, but the execution gas for processing a transaction's logic remains unchanged on Layer 1.
Cheaper L2s increase mainnet demand. Lower L2 posting costs via blobs will increase L2 transaction volume, which in turn increases the frequency of L1 settlement and proof verification calls, putting upward pressure on base layer execution gas.
The bottleneck shifts to state growth. With more affordable data, L2s like Arbitrum and Optimism will process more user transactions, accelerating the growth of the L1 state that all nodes must store and compute against.
Evidence: Post-4844, mainnet gas for an L1 bridge finalization or a zkSync Era validity proof verification is identical. The cost to run a Uniswap swap or mint an NFT on Ethereum itself is unaffected.
takeaways
EXECUTION VS. DATA
TL;DR for Builders and Architects
EIP-4844 (Proto-Danksharding) is a data-layer upgrade, not an execution-layer change. This distinction is critical for system design.
01
The Problem: Execution Gas is a Consensus & State Problem
Lowering execution gas requires changing EVM opcode pricing or consensus rules for state growth, which is politically fraught and risks breaking smart contracts. EIP-4844 sidesteps this entirely by creating a new, temporary data channel.
- Targets Rollups: Separates data availability (DA) cost from L1 execution.
- Preserves Stability: No changes to the core EVM gas schedule or state transition function.
- Enables Parallel Scaling: L2s like Arbitrum, Optimism, zkSync get cheaper data, while L1 execution remains unchanged.
0%
EVM Change
100%
DA Focus
02
The Solution: Blob-Carrying Transactions
Introduces a new transaction type that carries large data 'blobs' (~128 KB each) priced separately via a blob gas market. This creates a dedicated, volatile fee market for data, decoupled from the execution gas used for computation and storage.
- Separate Fee Market: Blob gas prices fluctuate based on rollup demand, independent of Uniswap swaps or NFT mints.
- Ephemeral Storage: Blobs are pruned after ~18 days, avoiding permanent L1 state bloat.
- Cost Predictability: Builders can forecast data costs without competing with L1 DeFi activity.
~128KB
Per Blob
10-100x
Cheaper DA
03
The Architect's Take: A Clean Abstraction
EIP-4844 is a textbook example of layer separation. It treats data as a first-class resource with its own economics, enabling L2s to act as primary execution environments. This mirrors the philosophy behind Celestia's modular stack.
- Future-Proofs for Danksharding: Blobs are the foundational primitive for full sharding.
- Optimizes for Rollups: Protocols like StarkNet and Base see immediate fee reductions.
- Simplifies Roadmap: L1 can now focus on security and decentralization, while scaling happens at L2.
Modular
Design
L2-First
Scaling Path
ENQUIRY
Get In Touch
today.
Our experts will offer a free quote and a 30min call to discuss your project.
NDA Protected
Confidentiality24h Response
Time to ValueDirectly to Engineering Team
No Sales Fluff10+
Protocols Shipped
$20M+
TVL Overall