Ethereum blob usage hit an all-time high in the first week of September 2026, averaging 6.7 blobs per block on a daily basis and 5.9 on a three-day moving average. The milestone marks the highest demand for Layer 2 data posting since the blob mechanism launched with the Dencun upgrade in March 20...
"This week ethereum blob usage hit an all-time high: 3 day moving avg of 5.9 blobs/block & 6.7 daily avg!!" — Trent van Epps, Protocol Guild Organizer
Ethereum blob usage hit an all-time high in the first week of September 2026, averaging 6.7 blobs per block on a daily basis and 5.9 on a three-day moving average. The milestone marks the highest demand for Layer 2 data posting since the blob mechanism launched with the Dencun upgrade in March 2024. Yet the network operates at roughly 40–50% of its current 14-blob target capacity, leaving substantial headroom before congestion pricing takes effect.
The record arrives amid a deepening structural tension in Ethereum's economics. Layer 2 networks now process 5-to-10 times more daily transactions than mainnet. Average L2 fees have collapsed 99.16% since Q1 2024, falling from $0.180 to $0.0015. Users benefit. Ethereum's fee revenue does not. Daily ETH burns have fallen to 50–70 ETH against roughly 1,700 ETH in daily staker issuance, pushing the network's annual inflation rate to 0.2–0.8%. The "ultrasound money" thesis — the idea that EIP-1559 burns would make ETH supply deflationary — has been inverted by the very scaling success it was designed to enable.
Since September 3, 2026, Ethereum's blob demand has exceeded all prior peaks, according to on-chain data highlighted by Protocol Guild organizer Trent van Epps. The daily average reached 6.7 blobs per block, surpassing the late-2025 highs that first triggered capacity concerns.
Blobs are the dedicated data slots introduced by EIP-4844 (Dencun upgrade, March 2024) that allow Layer 2 rollups to post transaction batches to Ethereum at a fraction of prior calldata costs. Each blob carries approximately 128 KB of data. When demand exceeds the target, a separate EIP-1559-style fee market kicks in, raising blob prices exponentially to ration capacity.
At 6.7 blobs per block against a 14-blob target and 21-blob maximum, the network is operating at approximately 48% of target capacity. This means the exponential pricing mechanism has not activated — blob fees remain at or near floor levels.
According to an academic paper published on arXiv in June 2026, each additional target blob correlates with 10.75 additional transactions per second across L2 networks and approximately 17.34% fee reductions every 30 days.
Blob capacity has expanded through four distinct phases:
| Upgrade | Date | Target Blobs | Maximum Blobs | Target Data/Block | |---------|------|-------------|--------------|-------------------| | Dencun | March 2024 | 3 | 6 | 384 KB | | Pectra | May 2025 | 6 | 9 | 768 KB | | Fusaka BPO-1 | December 2025 | 10 | 15 | 1,280 KB | | Fusaka BPO-2 | January 2026 | 14 | 21 | 1,792 KB |
Developers are now considering a further increase to a 21-blob target and 32-blob maximum, according to the Ethereum Foundation. PeerDAS (Peer Data Availability Sampling), deployed with Fusaka, enables this expansion by allowing nodes to verify blob data through sampling rather than downloading complete copies, reducing per-node bandwidth requirements.
Each capacity expansion has followed the same pattern: demand fills available space within 4–9 months, fee pressure builds, the target is raised, and utilization drops to 20–30% before climbing again. The current record suggests the cycle is repeating against the 14-blob target.
The scaling has achieved its stated goal. According to the arXiv paper analyzing Arbitrum, Base, and Optimism — the three highest-volume Ethereum L2s — average transaction fees declined 99.16% from $0.180219 in Q1 2024 to $0.001512 in Q1 2026.
Individual network fees in Q1 2026:
For comparison, Solana's average transaction cost was $0.000496 in the same period. Optimism undercut it by 94.6%.
These fees are low enough to enable micropayment and agentic use cases that were economically unviable at 2024 prices. Coinbase reported in its Q2 2026 earnings that over 90% of AI-driven stablecoin transaction volume now runs on Base, with stablecoin transaction volume on the chain up 7x year-over-year and annualized transfer volume reaching $32 trillion.
The fee collapse creates a direct economic problem for Ethereum's Layer 1.
Before EIP-4844, L2 rollups paid execution-layer gas prices to post data as calldata to Ethereum mainnet. This generated meaningful fee revenue that was partially burned via EIP-1559. In peak months of 2023, Ethereum collected over $300 million in monthly fees, according to Yellow Research.
Post-Dencun, L2s post data as blobs at prices set by a separate, low-demand fee market. The result: daily ETH burns from blob fees frequently measure in single digits, according to Yellow Research, compared to hundreds of ETH per day in the pre-blob era.
The case of Coinbase's Base chain illustrates the asymmetry. Base generated over $94 million in sequencer profit while returning just $4.9 million to Ethereum in blob fees, according to multiple analyses. That represents a 5.2% pass-through rate. The sequencer batches user transactions, pays Ethereum's blob fee for data settlement, and retains the spread. Coinbase operates the sole sequencer and captures 100% of the margin.
With L2 networks collectively processing over $40 billion in TVL and handling 5-to-10 times more daily transactions than mainnet, the economic center of gravity has shifted. As Yellow Research noted: "The result is a network that processes more data than ever before while collecting less ETH revenue from that data."
The Ethereum community recognized the value leakage problem. Fusaka, activated on December 3, 2025, included EIP-7918, which establishes a minimum blob base fee tied to the execution-layer base fee. Specifically, the blob base fee cannot fall below the execution base fee divided by approximately 16 (the precise denominator is 15.258).
Before EIP-7918, the blob fee floor was 1 wei — essentially zero. During low-demand periods, validators processed KZG commitment verifications for blob data at fees that did not cover the computational cost.
The impact was immediate. According to BingX flash data, the blob base fee surged 15 million-fold following Fusaka's activation.
According to analysis by Max Wadington at Fidelity Digital Assets, had EIP-7918 been active since the Dencun upgrade, Ethereum would have generated an additional $78.6 million (24,641 ETH) in cumulative blob-fee revenue through October 2025. That translates to roughly $6.02 in additional cost per blob. For Base specifically, Fidelity estimated $30.6 million in additional annual fees.
Fidelity characterized EIP-7918 as transforming "blobs from a subsidized utility into a potentially scalable revenue stream." The mechanism provides a guaranteed minimum stream of ETH being burned even during quiet periods, partially addressing the post-Dencun collapse in burn rates.
Whether the floor is high enough to restore deflationary dynamics remains unclear. The daily burn threshold for deflation is approximately 1,700 ETH (matching issuance). Current burn rates, even with the floor, remain well below that level during normal demand.
Ethereum's blob market does not operate in a vacuum. Alternative data availability (DA) layers now compete for rollup data posting.
According to BlockEden.xyz analysis, Celestia offers DA at costs roughly 96% cheaper than Ethereum blobs. A concrete comparison: at Ethereum blob prices, Eclipse's 83 GB of posted data would have cost over $300,000; on Celestia, the cost was approximately $6,000.
The DA market as of mid-2026:
Rollups choosing Celestia or EigenDA sacrifice Ethereum's full security model but gain 90%+ cost reductions. The choice creates a tiering effect: high-value settlement (institutional, large transfers) gravitates toward Ethereum blob security; lower-value, high-throughput applications may opt for cheaper alternatives.
This competition constrains how aggressively Ethereum can price blob fees. If the EIP-7918 floor or future capacity constraints push blob costs too high, rollups have viable alternatives — fracturing the demand that the record-setting 6.7-blob-per-block metric represents.
Ethereum validators operate in an environment of compressed returns. Native staking APR has fallen to 2.78% across approximately 897,000 active validators, with 38.9 million ETH staked (31.98% of supply), according to KuCoin Research. MEV adds 0.5–1% for a total effective yield of roughly 3.3–3.8%.
Liquid staking protocols (Lido, Rocket Pool) offer 3.0–3.5% net yield after protocol fees. With ETH trading at approximately $2,490, the annualized dollar return on a 32-ETH validator ($79,680 at stake) is approximately $2,215–$2,610 before hardware and bandwidth costs.
The Pectra upgrade (May 2025) introduced EIP-7251, raising the maximum effective validator balance from 32 ETH to 2,048 ETH, allowing large stakers to consolidate operations. The validator exit queue collapsed 99.9% from a peak of 2,670,000 ETH in September 2025 to 32 ETH in January 2026, removing exit-driven selling pressure.
For validators, blob fee revenue is additive but marginal. The economic model remains dominated by issuance rewards — a subsidy mechanism, not fee revenue. This aligns with the foundational observation that blockchain network activity is sustained primarily by inflationary issuance rather than self-sustaining fee economics.
Ethereum's blob usage record is genuine evidence of Layer 2 adoption growth. The data pipeline is working: rollups are posting more data, users are transacting at sub-cent costs, and the network has capacity to absorb further demand increases.
The economic implications are less favorable for ETH as an asset. The value generated by this activity accrues disproportionately to L2 sequencer operators — entities like Coinbase — rather than to Ethereum's base layer or its validators. EIP-7918 imposes a fee floor, but the floor is modest relative to the issuance rate, and competitive DA layers prevent aggressive repricing.
The fundamental question is whether volume can compensate for price. Ethereum needs blob demand to fill and exceed its current 14-blob target consistently enough to trigger the exponential fee mechanism. At 48% utilization, that threshold remains distant. Until it is reached, Ethereum's Layer 1 economics continue to operate on a subsidy model — the blob usage record notwithstanding.