EIP-4844 blob economics and 2026 market impact · full view from Dencun to Pectra and Fusaka

TL;DR

EIP-4844 (proto-danksharding · 2024-03 Dencun launch) introduced blobs — rollup-dedicated temporary storage · priced independently from calldata · after 18 days only the commitment remains onchain
Blob mechanic: each block target=3 / max=6 blobs (Dencun original) · each blob 128 KiB (131072 bytes) · total max ~768 KiB/block (Dencun)
Pectra EIP-7691 (2025-Q2) moved target → 6 / max → 9 · L2 calldata cost -50% (see Pectra EIP-7691: blob doubling and L2 economic transmission)
Fusaka PeerDAS (EIP-7594 · planned for 2026-Q4 ) adds another 10x capacity → theoretical ~60 GB/day blob throughput
Blob fee market: EIP-1559 style · independent base fee · fee rises exponentially above target and decays below target · decoupled from the gas fee market
Post-Dencun L2 cost reduction: rollup user gas on Arbitrum / Optimism / Base and others fell ~100x · sequencer revenue was net positive because volume increased
Blob market participants: rollup sequencers (blob submitters) · Ethereum builders / proposers (blob packagers) · DA users (data consumers)
Competition with alt-DA (see DA-layer landscape comparison): blobs narrowed alt-DA’s price advantage from 50x to 3-5x · after Pectra to 1.5-2x · after Fusaka it may disappear
Route: pair with systems index · Pectra EIP-7691 / DA comparison

Wiki route

This entry sits under systems index. Read it against Pectra EIP-7691: blob doubling and L2 economic transmission for the 2025-Q2 blob target doubling that directly extends EIP-4844, and Pectra upgrade overview for the broader Ethereum upgrade context. For the L1/L2 strategic framing that makes blob economics structurally important, see Vitalik L1/L2 strategy anchor. For the DA layer competition that blob market reshapes, see DA-layer landscape comparison. For the rollup ecosystem consuming blobs, see rollup market share 2026 and ZK-EVM rollup maturity matrix 2026. For the institutional staking lane that holds the validator economy behind blob processing, see Pectra EIP-7251: institutional staking and BFT validator economics overview. For restaking-secured alternative DA, see EigenLayer overview and restaking AVS landscape matrix.

Blob data structure

Core definitions (EIP-4844 spec):

1 blob = 4096 field elements × 32 bytes = 131072 bytes = 128 KiB
Each transaction can carry 0 or more blobs (the practical limit is the total number of blobs per block)
Blobs are inaccessible inside the EVM · data availability can be proven only through commitments (KZG point evaluation precompile · introduced by EIP-4844 )
Blob data is pruned after ~18 days (4096 epochs · target_blobs_per_block × MIN_BLOB_GASPRICE_UPDATE_FRACTION) from the P2P network · only the KZG commitment remains permanently onchain

blob commitment:

KZG commitment scheme (trusted-setup based · Ethereum uses the Powers of Tau ceremony 2023 )
Each blob has 1 48-byte commitment · a versioned hash (SHA256(commitment) · 32 bytes) is added onchain
Rollup contracts can prove that “the value at point z in the blob is y” through the KZG point evaluation precompile (for ZK rollup state transition proofs)

Blob fee market (EIP-1559 style · independent)

Core mechanism (EIP-4844 spec):

Each block has an excess_blob_gas state variable · it accumulates “blob gas above target”
blob_base_fee = MIN_BLOB_GASPRICE * exp(excess_blob_gas / BLOB_GASPRICE_UPDATE_FRACTION)
target_blobs_per_block = 3 (Dencun) / 6 (Pectra) / 6 (early Fusaka · adjustable later)
max_blobs_per_block = 6 (Dencun) / 9 (Pectra) / planned 18+ (Fusaka PeerDAS)
Each blob consumes GAS_PER_BLOB = 131072 gas
Per 1 blob, pay 131072 × blob_base_fee wei · independent from transaction gas

Fee dynamics:

If blob count in a block = target · excess_blob_gas is unchanged · fee is flat
If block > target · excess_blob_gas increases · fee rises exponentially
If block < target · excess_blob_gas decreases (min 0) · fee falls exponentially
Design target: match blob fee elastically to demand · decouple from calldata fee · avoid affecting user gas

MIN_BLOB_GASPRICE = 1 wei:

When demand is low · blobs are almost free (each blob ~131072 wei, a negligible amount)
When demand is high · blob fee can reach 10-100 gwei · $0.10-$1+ per blob

Blob submission flow (sequencer view)

Typical flow for an L2 sequencer submitting a batch to L1 :

Sequencer collects, orders, and executes L2 txs (locally)
Package batch data (L2 tx + state diff) · compress it (snappy / zstd / bzip2)
Create an EIP-4844 transaction (type 3) · include 0-N blobs (each blob 128 KiB) + KZG commitment + versioned hash
Payment: transaction gas (normal) + blob gas (blob_base_fee × 131072 × number of blobs)
L1 block includes the transaction · validators receive blob data via P2P (blobs do not enter the block body · they use a separate P2P channel)
Rollup contract verifies the KZG commitment (ZK rollups verify that the proof references the commitment · OP rollups can challenge within the challenge window)
After 18 days, blob data is pruned · only the commitment remains onchain

Sequencer cost structure (post-Dencun):

Transaction gas: ~$1-5 (EIP-4844 transaction overhead)
Blob gas: ~$0.10-$1 / blob (depending on demand) · typically 3-6 blobs submitted per batch
Total cost per batch: ~$5-30 · allocated across ~10000+ L2 txs · per L2 tx about $0.0005-$0.003
Pre-Dencun comparison (using calldata): per L2 tx about $0.05-$0.20 (depending on L1 gas)

Comparison with calldata

Dimension	Calldata (pre-Dencun / still optional)	Blob (EIP-4844)
Price	gas_per_byte × gas_base_fee (high)	blob_base_fee × 131072 × number of blobs / blob bytes (low, ~100x)
Retention period	Permanent (state)	~18 days (P2P prune) + permanent commitment
EVM accessible	Yes (via CALLDATALOAD and other opcodes)	No (only through KZG point evaluation precompile)
Use case	State access · onchain data consumption	rollup data availability · short-term reconstructability
Fee market	Shared with user gas · congestion contagion	Independent base fee · no congestion contagion

Initial Dencun launch period (2024-03 - 2024-Q2)

Pre-Dencun state (before 2024-03 ):

OP rollups such as Arbitrum / Optimism / Base used calldata · daily aggregate calldata cost was ~$2-5M
ZK rollups such as zkSync / Linea / Scroll also used calldata + ZK proof submission
L2 sequencer revenue spent 60-80% on L1 cost · margins were thin
L2 user gas was $0.05-$0.50 / simple transfer

Immediate effect after Dencun launch (2024-03 - 2024-Q2):

Each L2 sequencer switched to EIP-4844 blobs · L1 cost fell by ~80-95%
L2 user gas fell to ~$0.001-$0.05 / simple transfer · user experience improved sharply
L2 sequencer margins suddenly surged (L1 cost collapsed · user gas held · net margin high)
However, sequencer competition intensified · over several weeks, user gas was gradually reduced · some savings were passed to users and some were retained as sequencer revenue

Initial blob fee market volatility:

Early after 2024-03 launch · blob usage was insufficient · blob_base_fee stayed at MIN_BLOB_GASPRICE = 1 wei
After 2024-Q2 · Base / Arbitrum / Optimism used blobs aggressively · average blob usage approached target (3)
In 2024-Q3 , blob max (6) was often pressured · blob fee spiked to 10-100 gwei

Blob fee spike period (2024-Q3 - 2025-Q1)

Fee spike trigger scenarios:

Meme coin waves on Base · DEX swap volume jumped · Base sequencer batch submission frequency increased
Multiple L2 s under high load simultaneously · competition for blob slots
Single-block max=6 blobs often exhausted · excess_blob_gas accumulated · fee rose exponentially

Empirical data (2024-Q4 peak period):

Blob base fee spiked to 200-500 gwei · $5-15 per blob
L2 sequencer single-batch cost rose from $5-30 to $50-200
Some sequencers temporarily chose to retreat to calldata when blobs were too expensive · or reduced batch frequency (increasing user finality time)
User gas could rise 5-10x at peak · still far below pre-Dencun

Community discussion:

Discussion of EIP-7691 (blob target doubling) accelerated on EthMagicians / ethresear.ch
Dankrad Feist, Vitalik, and others publicly advocated several times for accelerating the PeerDAS (EIP-7594) roadmap
ACD confirmed inclusion of EIP-7691 in Pectra

After Pectra (2025-Q2 - 2026-Q2)

Pectra EIP-7691 launch effect:

Blob target 3 → 6 · max 6 → 9
L2 sequencers no longer trigger blob fee spikes under normal load (because supply doubled)
User gas fell ~30-50% (see Pectra EIP-7691 detailed analysis)
Blob fee stays at MIN (1 wei) most of the time · short spikes occur only under burst load such as ETF launches or meme coin spikes

**Current state in 2026-Q2 **:

Average daily blob count ~30000-50000 blobs (distributed across ~7200 blocks/day) · average ~4-7 blobs/block (slightly below target 6 )
Average blob fee ~1-5 gwei · occasional spikes to 50-100 gwei
Total daily blob cost ~$50000-$200000 (across all rollups) · close to alt-DA total revenue
Actual average cost per blob ~$0.10-0.30 · 1 GB DA cost ~$100-300 (per day)

Sequencer margin recovery:

L2 sequencer net margin stabilized after Pectra · no need to worry again about sudden blob spikes
Leading L2 sequencers such as Base · Arbitrum · Optimism each have annualized revenue of ~$50-150M (2026)
Mid-sized L2 sequencers such as Linea · Scroll · zkSync each have revenue of ~$10-30M

Market dynamics with alt-DA

See DA-layer landscape comparison. Impact of the blob market on alt-DA:

Pre-Dencun:

Celestia / EigenDA / Avail / NEAR DA price advantage was ~50x versus calldata
Alt-DA captured cost-sensitive rollup customers · forming the modular DA narrative

Post-Dencun (2024-Q2 - 2025-Q1):

Blob price advantage narrowed alt-DA’s price advantage to ~5-10x
Most ETH-aligned rollups still chose blobs (ecosystem alignment + security)
Alt-DA shifted to non-ETH-aligned customers (Cosmos · Move · sovereign rollups) or high-throughput child chains (Mantle · Manta)

Post-Pectra (2025-Q2 - 2026-Q2):

Blob capacity doubled · alt-DA price advantage narrowed further to ~3-5x
Celestia revenue grew ~50% YoY, but customer types diverged (mainstream L2 did not grow · sovereign / Move rollups grew)
EigenDA borrowed EigenLayer security + high throughput · maintained ~$120M ARR (see restaking AVS matrix)

Post-Fusaka (2026-Q4+):

PeerDAS increases blob capacity by another 10x · alt-DA price advantage may remain only ~1.5-2x
Alt-DA must shift to pure differentiated value (sovereign · privacy · specific throughput), not a pure cost war

Blob market participants

Sequencers (blob submitters):

Leaders: Base · Arbitrum · Optimism · zkSync · Linea · Scroll · Polygon zkEVM
Mid-sized: Taiko · Blast · Mode · Mantle (some use EigenDA) · etc.
Long-tail: multiple small OP Stack / ZK Stack rollups
In total, ~30-50 rollups form the demand side of the blob market

Builders / Proposers (blob packagers):

MEV builders such as Flashbots · bloXroute · Manifold · Eden package blobs in addition to ordering transactions
Each block’s proposer (randomly selected validator) accepts a builder bundle and includes blob transactions
Blob packaging income is part of builder revenue

Validators (blob forwarders):

More than approximately 1 million Ethereum validators distribute blob data through the P2P network
With DAS (after PeerDAS), light clients only need sample verification and no longer need to download every blob

DA users (data consumers):

Primarily rollup customers (challenge provers · ZK verifiers · state reconstruction)
Then indexers / explorers (blob data during the 18 -day ephemeral period) · long term requires archive nodes (Blobscan / Ethscan, etc.)

Blob vs calldata vs alt-DA · 9 -dimension comparison

Dimension	Calldata	Blob (EIP-4844)	EigenDA	Celestia	NEAR DA
Cost per GB	$500-2000+ (calldata-equivalent ETH gas per GB)	~$100-300/GB-day	$60-150/GB-day	$20-50/GB-day	$10-30/GB-day
Retention period	Permanent (L1 state)	18 days P2P + permanent commitment	EigenDA dispatcher setting (usually 14 days+ commitment)	Celestia full history retained	NEAR history retained
Security level	Ethereum L1 (highest)	Ethereum L1 + DAS (post-PeerDAS)	EigenLayer ~$6B opt-in	Celestia ~$1-2B staking	NEAR ~$1-2B staking
Throughput	Constrained by L1 gas limit (~50-100 KB/block actual)	0.75 MB/block (Dencun) → 1.1 MB (Pectra) → ~11 MB (Fusaka PeerDAS)	15 MB/s sustained	2 MB/block / 6s = 28 GB/day	Theoretical 100 MB/s
Finality	L1 finality (~12-15 min)	L1 finality (~12-15 min)	~10-30s + slashing	~6s (Tendermint)	~2-3s (NEAR)
Adoption share	Pre-Dencun mainstream · post-Dencun very small	95%+ ETH-aligned rollup default	~20 rollups · modular-led	~30+ chains · non-ETH-aligned led	~10 chains · niche
Fee market design	Shared with user gas (congestion contagion)	Independent EIP-1559 base fee	EigenDA dispatcher has custom pricing	Celestia custom pricing (denominated in TIA)	NEAR storage fee model
EVM accessible	Yes (CALLDATALOAD)	No (only via KZG point eval precompile)	No	No	No
Regulatory surface	Ethereum L1 sufficiently decentralized	Ethereum L1 sufficiently decentralized	EIGEN token + restaking gray area	TIA token gray area	NEAR token reviewed across multiple jurisdictions

Main conclusions:

Cost ranking: Blob > EigenDA > Celestia > NEAR DA · but the price gap narrows with Pectra/Fusaka
Security-level ranking: Blob = Ethereum L1 (highest) > EigenDA (borrowed ETH stake) > Celestia ≈ NEAR DA (independent PoS)
Throughput ranking: EigenDA / NEAR > Pectra Blob > Celestia / Avail > Fusaka Blob (jumps to EigenDA-equivalent level)
Finality ranking: NEAR > Celestia > EigenDA > Blob (Blob is slowest, but when combined with ZK proofs, waiting for finality is unnecessary)

Before vs after Pectra EIP-7691

State	Pre-Pectra (Dencun)	Post-Pectra (2025-Q2+)
Blob target	3 / block	6 / block
Blob max	6 / block	9 / block
Max throughput	~0.75 MB / 12s ≈ 5.4 GB/day	~1.1 MB / 12s ≈ 8 GB/day
Average blob fee	5-50 gwei (volatile)	1-5 gwei (most of the time)
L2 sequencer cost	$20-100 / batch (volatile)	$5-30 / batch (stable)
L2 user gas	$0.001-0.05 (spikes in high vol)	$0.0005-0.02 (further 30-50% decline after Pectra)
Alt-DA price advantage	~5-10x	~3-5x

Fusaka PeerDAS forecast (planned for 2026-Q4 )

EIP-7594 (PeerDAS · Peer Data Availability Sampling):

Validators do not download all blobs · they verify availability through DAS sampling
Per-validator network bandwidth requirement falls to ~1/8 (sampling only, no full download)
Blob capacity can expand 10x (from max=9 to max=72+) · without increasing validator network load
Accompanying blob fee market adjustments such as EIP-7642 / 7691

Expected impact:

Blob max 9 → 72 (initial) · ultimate goal 256+
Throughput ~1.1 MB/block → ~11 MB/block (initial)
Alt-DA price advantage may remain only ~1.5-2x · some alt-DA customers return to L1 blobs
Full danksharding (theoretical max 32 MB/block · ~230 GB/day) is a long-term vision · unlikely to complete in 2027-2028

Relationship between L2 sequencer revenue and user gas

Sequencer revenue formula (simplified):

Revenue = (user gas × tx count - L1 blob cost - operational cost - MEV redistribution)
User gas unit price can fall when volume rises (competition) · but total revenue depends on elasticity

Empirical data (2026-Q2 estimate · leading L2s):

Base: Daily user gas total ~$200-400k · Daily blob cost ~$30-80k · Sequencer net ~$100-300k/day ≈ $30-100M ARR
Arbitrum: Daily user gas total ~$150-300k · Daily blob cost ~$20-50k · Sequencer net ~$80-200k/day ≈ $30-70M ARR
Optimism: Daily user gas total ~$50-100k · Daily blob cost ~$10-20k · Sequencer net ~$30-70k/day ≈ $10-25M ARR
zkSync · Linea · Scroll and others: each $5-20M ARR

Aggregate sequencer revenue across all ETH-aligned rollups is ~$150-300M ARR (2026). This is the stable state after Pectra · higher than aggregate sequencer revenue pre-Dencun (2023) · but per-tx revenue has fallen sharply.

Beneficiaries among institutional and consumer products such as Stripe / Coinbase Smart Wallet

Stripe USDC on Base: after Pectra, a single USDC transfer is ~$0.0005 · economics improved materially for Stripe · Stripe launched Base-only merchant settlement
Coinbase Smart Wallet: Account Abstraction (see ERC-4337) + low Base gas · Coinbase achieved sub-cent UX · user experience feels close to Web2
PayPal PYUSD on Base: similarly benefits · 2025-2026 PYUSD circulating supply on Base grew ~3x
RWA tokenization: Franklin Templeton FOBXX pilot on Polygon zkEVM reduced settlement cost through Pectra blobs · however institutions still prefer L1 (see DA-layer comparison)

Asymmetric benefits for ZK rollups vs OP rollups

OP rollups (Arbitrum · Optimism · Base · Blast, etc.):

Blobs directly reduce calldata cost · all batch data goes into blobs
State can be reconstructed within the challenge window · the 18 -day blob retention period is sufficient
Benefits are most direct · user gas decline is most visible

ZK rollups (zkSync · Linea · Scroll · Polygon zkEVM · Taiko, etc.):

Blobs reduce state diff submission cost · but ZK proof submission cost (verifier gas) is unchanged
ZK proofs account for ~30-50% of sequencer cost · blobs only ~30-50%
Benefits are relatively smaller but still material (see ZK-EVM rollup matrix 2026)

Impact on validator economy and staking yield

See BFT validator economics overview and Pectra EIP-7251 institutional staking:

Blob fee is an additional source of validator revenue (blob_base_fee is not burned · proposer collects it)
2026 average daily blob fee revenue ~$50-100k · allocated across ~7200 blocks · about $7-15 per block · a tiny share of validator daily revenue
ETH burning (EIP-1559 base fee burn) is unaffected by blob fee · blob fee is not burned
Long-term blob economics do not directly affect ETH staking yield · but they have indirect effects (blobs make L2 thrive · dApp activity on L2 brings L1 economic activity back)

Change in Builder / Proposer roles

Builders now must package blob transactions plus normal transactions · complexity increases
Blob P2P propagation path differs from transactions · builders must coordinate
The proposer role is largely unchanged, although blob reception delays can affect attestation timing
In long-term PBS (proposer-builder separation), builders specialize in blob handling

Sources

EIP-4844 spec · https://eips.ethereum.org/EIPS/eip-4844
EIP-7691 spec · https://eips.ethereum.org/EIPS/eip-7691
EIP-7594 PeerDAS spec · https://eips.ethereum.org/EIPS/eip-7594
Ethereum.org Danksharding roadmap · https://ethereum.org/en/roadmap/danksharding/
Blobscan · https://blobscan.com/
L2Beat scaling summary · https://l2beat.com/scaling/summary
DefiLlama rollups · https://defillama.com/chains/Rollup
ethresear.ch sharding category · https://ethresear.ch/c/sharding/
Dankrad Feist blog · “Data availability sampling and danksharding”(public)
Vitalik blog · “Layer 2s as cultural extensions of Ethereum”(public)
Ethereum Foundation KZG ceremony documentation