Blobs are the breakthrough. They introduce a new transaction type that carries large data packets, which are cheap to post and automatically deleted after ~18 days. This separates expensive, permanent consensus storage from cheap, temporary data availability.
the-ethereum-roadmap-merge-surge-verge
Blog
Proto Danksharding Design Goals in Plain English
EIP-4844 (Proto-Danksharding) is a pivotal, incremental upgrade for Ethereum's data layer. It introduces 'blob-carrying transactions' to provide cheap, temporary data availability for rollups, directly lowering L2 transaction costs without executing a full sharding implementation.
introduction
THE BANDWIDTH BOTTLENECK
The Scaling Lie Everyone Believes
Proto-Danksharding (EIP-4844) directly attacks the fundamental data availability bottleneck that limits all Layer 2 rollups.
Rollups need data, not consensus. Optimism and Arbitrum currently post transaction data as expensive calldata on Ethereum. Blobs reduce this cost by ~100x, directly lowering L2 transaction fees for users.
It's a capacity upgrade, not a TPS boost. The lie is that sharding increases Ethereum's execution speed. It doesn't. It massively increases the data bandwidth for rollups like zkSync and StarkNet to prove transactions, enabling their scale.
Evidence: A single blob holds ~125 KB. With a target of 3 blobs per block, this creates ~1.3 MB/sec of dedicated data bandwidth for rollups, a 100x increase over the current calldata system.
market-context
THE CALDATA BOTTLENECK
Why Rollups Are Choking on Their Own Success
Proto-danksharding (EIP-4844) directly addresses the unsustainable data cost crisis created by L2 scaling.
Data is the new gas fee. Rollups like Arbitrum and Optimism compress transactions but must post all data to Ethereum for security. This 'calldata' consumes expensive L1 block space, creating a hard ceiling on L2 scalability and cost reduction.
Blobs are a dedicated data lane. Proto-danksharding introduces blob-carrying transactions, a separate data channel with its own fee market. This decouples L2 data availability costs from the volatile execution gas market dominated by Uniswap and NFT mints.
Blob data is ephemeral. Unlike calldata, blobs are stored for ~18 days, not forever. This temporary storage, verified by nodes, provides the security guarantee rollups need while drastically reducing Ethereum's long-term state bloat.
Evidence: Post-EIP-4844, L2 transaction fees are dominated by execution, not data. On Arbitrum, posting a batch's data cost dropped from ~$100 to under $0.01, proving the model works.
key-trends
EIP-4844'S CORE MECHANICS
The Three Pillars of Proto-Danksharding's Design
EIP-4844 isn't just a scaling upgrade; it's a foundational redesign of Ethereum's data layer to enable a rollup-centric future.
01
The Problem: Rollups Are Data-Starved
Rollups like Arbitrum and Optimism post all transaction data to Mainnet, paying ~$0.25 per transaction in calldata fees. This hard-caps scalability and keeps L2 fees artificially high.
- Cost Bottleneck: Data is the single largest cost component for rollups.
- Throughput Limit: Ethereum's ~80 KB/s data bandwidth is saturated.
~$0.25
Avg L2 Tx Cost
80 KB/s
Data Cap
02
The Solution: Introduce Blob-Carrying Transactions
EIP-4844 creates a new transaction type that carries large ~128 KB data blobs. This data is separated from execution and stored temporarily (~18 days), decoupling data availability from state growth.
- Cheaper Data: Blobs are priced via a separate fee market, targeting ~$0.001 per transaction.
- Future-Proof: Blob count scales to 16 per block, enabling ~1.3 MB/s of dedicated data bandwidth for rollups.
~128 KB
Per Blob
-99%
Cost Target
03
The Bridge: KZG Commitments & Data Sampling
To keep light clients and future Danksharding viable, blobs are committed to using KZG polynomial commitments. This allows nodes to verify data availability with tiny proofs without downloading the full blob.
- Trustless Verification: Enables secure bridging for Celestia, EigenDA, and other modular DA layers.
- Path to Danksharding: This cryptographic primitive is the prerequisite for full 2D data sampling and scaling to 64 blobs/block.
48 Bytes
Proof Size
64
Future Blobs/Block
PROTO DANKS HARDING
Calldata vs. Blobs: The Cost & Capacity Breakdown
A data-driven comparison of Ethereum's primary data storage layers, highlighting the paradigm shift introduced by EIP-4844 for rollups like Arbitrum, Optimism, and zkSync.
| Feature / Metric | Execution Layer Calldata | Consensus Layer Blobs (EIP-4844) |
|---|---|---|
Primary Purpose | Smart contract execution input | Cheap, temporary data for L2 proofs |
Data Persistence | Permanent on-chain state | Pruned after ~18 days (4096 epochs) |
Target Throughput (per block) | ~90 KB (gas-limited) | ~768 KB (3 blobs * 128 KB) |
Current Avg. Cost per KB (USD) | $1.50 - $5.00 | $0.01 - $0.03 |
Cost Reduction for Rollups | Baseline (1x) | ~100x cheaper |
EVM Execution Access | Fully accessible (CALLDATA opcode) | Inaccessible (only commitment verified) |
Key Enabler For | Direct contract calls, some L1 apps | Ultra-low-fee L2s, Volition modes, Celestia DA |
deep-dive
THE DATA
Deconstructing the Blob: A Data-First Architecture
Proto-danksharding re-architects Ethereum's data layer to make L2 rollups cheaper and more scalable.
Separates Execution from Data. Proto-danksharding introduces blob-carrying transactions, where data is posted to the beacon chain but not processed by the EVM. This creates a dedicated data availability channel that is priced independently from gas, decoupling L2 costs from mainnet congestion.
Targets Rollup Economics. The primary design goal is L2 fee reduction. By providing a cheaper, dedicated data layer, rollups like Arbitrum and Optimism slash their fixed costs for posting transaction data, directly lowering fees for end-users.
Enables Future Scaling. Blobs are the foundational primitive for full danksharding. The architecture pre-configures a data availability sampling framework, allowing the network to securely scale data capacity to 16 MB per slot without requiring nodes to download everything.
Evidence: Current rollups spend 80-90% of their operational cost on Ethereum calldata. Blobs reduce this cost by an order of magnitude, a prerequisite for scaling to 100,000+ TPS across the L2 ecosystem.
risk-analysis
IMPLEMENTATION PITFALLS
What Could Go Wrong? The Pragmatic Risks
Proto-danksharding's elegant design introduces new attack surfaces and economic dependencies.
01
The Blob Fee Market: A New MEV Vector
Blob data is priced via a separate EIP-4844 fee market, decoupling it from execution gas. This creates a new, volatile commodity for block builders to optimize.
- Builder Collusion: Builders could withhold blobs to inflate prices before large L2 batch submissions.
- Cross-Domain MEV: Searchers may front-run L2 sequencers by manipulating blob space availability.
- Fee Volatility: Sudden demand spikes from major NFT mints or airdrops could temporarily price out smaller rollups.
~10-100x
Price Swings
New Vector
MEV Surface
02
Data Availability Sampling: The 75% Honest Assumption
Clients use Data Availability Sampling (DAS) to verify blob availability without downloading it all. This security model has a critical threshold.
- Sybil Attacks: An adversary controlling >25% of sampling nodes can trick the network into accepting unavailable data.
- Network-Level Attacks: Eclipse attacks or network partitioning could isolate honest samplers.
- Implementation Bugs: Flaws in the sampling logic or peer discovery could break the security guarantee, risking $10B+ in bridged assets.
>25%
Failure Threshold
Critical
Security Class
03
The 18-Day Time Bomb & L2 Synchronization
Blobs are pruned from consensus nodes after ~18 days. This forces a hard deadline for L2s and data availability committees (DACs) to permanently store the data.
- L2 Catastrophe: If an L2 node goes offline for 19 days, it cannot reconstruct its own state, breaking the chain.
- DAC Centralization Risk: Reliance on a small set of entities (like EigenDA, Celestia) for long-term storage creates a single point of failure.
- Proposer-Builder Separation (PBS) Conflict: Builders have no incentive to store old blobs, pushing the cost and risk onto others.
18 Days
Prune Deadline
Synchronization Risk
L2 Hazard
04
Client Diversity in a Post-Danksharding World
The complexity of blob validation, DAS, and new transaction types increases the risk of client bugs and consensus failures.
- Spec Complexity: More moving parts (blob txs, KZG commitments, sidecars) increase the chance of divergent implementations.
- Validation Overhead: Resource requirements for sampling may centralize node operation to well-funded entities.
- Historical Precedent: Past incidents like the Besu/Nethermind bug show how client diversity mitigates risk; new complexity threatens this.
High
Integration Risk
>5 Clients
Critical Mass
future-outlook
THE ROADMAP
Beyond the Blob: The Path to Full Danksharding
Proto-Danksharding (EIP-4844) is a tactical, incremental step designed to unlock cheaper L2 data today while building the scaffolding for a fully sharded future.
Blobs are a data commitment, not storage. EIP-4844 introduces a new transaction type carrying large, temporary data packets called blobs. The Ethereum consensus layer only stores these blobs for ~18 days, but their cryptographic commitment persists forever. This separates the cost of data availability from the cost of permanent storage, which is the core economic innovation.
The design enables a clean upgrade path. The blob data structure and KZG commitment scheme are direct subsets of the full Danksharding specification. Building with KZG polynomial commitments now forces client and tooling development (like Prysm, Lighthouse, Erigon) to implement the exact cryptography required for the final system. This avoids a costly, disruptive hard fork later.
It creates an instant market for cheap data. Rollups like Arbitrum, Optimism, and zkSync immediately replace their expensive calldata posts with cheap blob posts. This creates a real-world, fee-market-driven testing ground for blob propagation, networking, and pruning logic long before the full 64-shard system is live. The blob fee market itself is a critical subsystem to validate.
Evidence: Post-EIP-4844, L2 transaction fees dropped by over 90% for many operations, as data posting shifted from ~100 gas per byte (calldata) to a new, separate fee market. This proves the data availability bottleneck was the primary cost driver, not execution.
takeaways
PROTO DANKS SHARDING
TL;DR for Time-Poor Architects
EIP-4844 isn't about scaling execution; it's a surgical upgrade to decouple data availability from execution costs, setting the stage for rollup dominance.
01
The Blob-Carrying Transaction
Introduces a new transaction type that carries large, temporary data packets called blobs. This separates the pricing and lifecycle of data from standard calldata, creating a dedicated marketplace for rollup data.\n- ~0.125 MB per blob, ~2 MB target per block\n- Data is pruned after ~18 days, as L2s only need short-term availability for fraud proofs\n- Creates a separate fee market, insulating L1 gas from L2 data posting spikes
~100x
Cheaper Data
~2 MB/block
Data Capacity
02
KZG Commitments: The Cryptographic Anchor
Uses KZG polynomial commitments to create a compact proof (the commitment) that the full blob data is available and correct. This is the core trust mechanism enabling light clients and validators to verify data without downloading it all.\n- 48-byte commitment per blob vs. 128 KB of full data\n- Enables efficient data availability sampling (DAS) for future full Danksharding\n- Removes need for complex fraud proofs for data availability at this stage
48 bytes
Commitment Size
Trustless
Verification
03
Decoupling to Democratize Rollups
The primary goal: make L2 transaction costs dominated by execution, not data availability. By creating a cheap, abundant data layer, it removes the biggest barrier to rollup adoption and innovation.\n- Targets <$0.01 L2 transaction fees for optimistic rollups like Arbitrum and Optimism\n- Enables viable ZK-rollup scaling for general-purpose EVM chains (e.g., zkSync, Scroll)\n- Fundamentally shifts Ethereum's scaling roadmap to a modular, rollup-centric future
>100 TPS
Effective L2 Capacity
Cent-Driven
L2 TX Fees
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