Ethereum's Fusaka hard fork, activated December 3, 2025, introduced EIP-7918 — a blob base fee floor that raised the minimum cost of posting Layer 2 data to Ethereum by roughly 15 million times relative to its prior 1-wei minimum. The mechanism ties blob pricing to L1 execution gas, eliminating a...
"PeerDAS in Fusaka is significant because it literally is sharding. Sharding has been a dream for Ethereum since 2015, and data availability sampling since 2017, and now we have it." — Vitalik Buterin, Ethereum Co-Founder
Ethereum's Fusaka hard fork, activated December 3, 2025, introduced EIP-7918 — a blob base fee floor that raised the minimum cost of posting Layer 2 data to Ethereum by roughly 15 million times relative to its prior 1-wei minimum. The mechanism ties blob pricing to L1 execution gas, eliminating a period in which rollups consumed data availability at near-zero cost while nodes bore the computational burden of KZG verification.
Seven months post-activation, the upgrade's economic effects are measurable. ETH supply inflation sits at approximately 0.23% annually, down from the trajectory set by the post-Dencun fee collapse but still above the deflationary threshold the network briefly achieved in 2023. L2 transaction fees have dropped below $0.01 on chains like Base and Arbitrum, even as blob capacity more than doubled through two subsequent Blob Parameter Only (BPO) forks. The net result is a protocol that is simultaneously cheaper to use and harder to free-ride on — an unusual combination in fee market design.
Fidelity Digital Assets described Fusaka as "a decisive shift toward a more strategically aligned and economically coherent roadmap," noting that nearly every EIP in the upgrade has the potential to drive more revenue to ETH holders. Whether that revenue materializes depends on whether L2 transaction volume scales fast enough to overcome the expanded supply of blob space.
Ethereum's March 2024 Dencun upgrade introduced EIP-4844, creating a separate fee market for "blobs" — large binary objects used by rollups to post transaction data to L1. The design succeeded at its primary goal: cutting L2 costs. But it produced a side effect that undermined Ethereum's economic model.
Blob fees operated on an independent EIP-1559-style adjustment mechanism. When blob demand fell below the per-block target (then six blobs), the base fee ratcheted down toward its absolute minimum: 1 wei. In practice, the fee stayed at or near 1 wei for extended periods. Rollups posting data to Ethereum paid fractions of a cent for megabytes of storage and verification that cost nodes real computational resources — specifically, KZG polynomial commitment verification.
The consequences compounded. First, nodes bore costs they were not compensated for. Second, because blob fees are burned (not paid to validators), the collapse in blob pricing meant the burn channel created by EIP-4844 produced almost no deflationary pressure. Daily ETH burn fell from thousands of ETH during 2023 network peaks to as low as 50-70 ETH after Dencun, according to ultrasound.money data. With issuance running at approximately 1,700 ETH per day, the network flipped from deflationary to inflationary.
EIP-7918, authored by Ethereum researcher Anders Elowsson, addresses the free-riding problem through a proportional reserve price. The core logic:
blob_base_fee >= (BLOB_BASE_COST × execution_base_fee) / GAS_PER_BLOB
The mechanism requires the blob base fee to equal at least 1/15.258 of the L1 execution base fee. When the computed reserve price exceeds the nominal blob base fee, the adjustment algorithm treats the block as over-target and stops pushing the fee downward.
According to Elowsson's technical notes published on the Ethereum Foundation's HackMD, the design serves three purposes: (1) it prevents blob fees from collapsing to economically meaningless levels during low-demand periods; (2) it ensures that the cost of KZG verification is at least partially covered by blob fees; and (3) it creates a guaranteed minimum stream of ETH being burned, even when blob space is underutilized.
The practical effect was immediate. On activation, the blob base fee rose from 1 wei to approximately 15 million wei — a 15-million-fold increase. In absolute terms, this remains a small number (roughly 0.000015 gwei), but the mechanism ensures the fee scales proportionally with L1 gas prices. When L1 execution fees rise during congestion, the blob floor rises with them.
The mechanism does not affect rollups during periods of genuine demand. When blob utilization approaches or exceeds the target, the equilibrium fee naturally sits above the floor. EIP-7918 prevents only the downside collapse — what Elowsson described as "blob-fee instability when execution gas dominates."
Fusaka's second major component is EIP-7594, which implements Peer Data Availability Sampling (PeerDAS). The protocol allows nodes to verify blob data availability by downloading and checking only a subset of the data, rather than the full payload. Each full node stores roughly one-eighth of total blob data under the new scheme.
PeerDAS enables higher blob counts per block without proportionally increasing node bandwidth and storage requirements. The capacity expansion was rolled out in three stages:
| Fork | Date | Target Blobs | Max Blobs | |------|------|-------------|-----------| | Fusaka | Dec 3, 2025 | 6 | 9 | | BPO-1 | Dec 17, 2025 | 10 | 15 | | BPO-2 | Jan 7, 2026 | 14 | 21 |
According to reporting by Unchained, the BPO-2 fork completed the initial parameter optimization phase. Core developers are collecting performance data before proceeding with BPO-3 and BPO-4, which aim to reach 128 blobs per block — roughly a 21x increase from the pre-Fusaka maximum of 6.
The combined effect of PeerDAS and the BPO sequence is a data availability layer that can handle substantially more rollup throughput. Ethereum's L2 ecosystem currently processes approximately 5,600 transactions per second in aggregate, according to L2Beat data. Developers project that full utilization of expanded blob capacity could support over 24,000 TPS across rollups.
The data on L2 costs and activity since Fusaka tells a story of cheaper transactions alongside concentrating market structure.
Fee reductions: Layer 2 transaction fees on Base and Arbitrum dropped below $0.01 after Fusaka cut data availability costs by 90% to 99%, according to Optimism's post-Fusaka analysis. A typical L2 transaction that cost approximately $0.50 in late 2025 fell to between $0.20 and $0.30 immediately after Fusaka, and continued declining as BPO forks expanded capacity.
TVL concentration: As of mid-2026, 73 active rollups collectively hold over $48 billion in total value locked. But the market is dominated by three chains. Arbitrum One holds approximately $16.9 billion (40-44% market share), Base follows at roughly $12.8 billion, and Optimism holds $8 billion. Together, these three handle close to 90% of all L2 transactions, per data from Everstake and L2Beat.
Throughput gains: Optimism's blog reported that OP Stack chains process 3.5 times more transaction data than pre-Fusaka levels. Base projects it can reach 10,000-20,000 TPS with further optimization.
The economic tension is clear. Cheaper blob space benefits L2 users directly. But cheaper blob space also means lower blob fees burned, reducing ETH's deflationary pressure. EIP-7918's floor prevents the fee from hitting zero, but it does not guarantee that the burn rate exceeds issuance.
As of mid-July 2026, Ethereum's supply sits at approximately 121.88 million ETH, according to ultrasound.money. The network runs a modest annual inflation rate of approximately 0.23% — better than Bitcoin's post-halving ~1.6% issuance rate, but not deflationary.
The ultrasound money thesis requires mainnet gas activity to push average execution fees above approximately 16 gwei on a sustained basis. Ethereum currently generates roughly $227,000 per day in total fees and approximately $55,700 per day in chain revenue, according to available on-chain data — levels that correspond to a calm market far below the threshold needed for net deflation.
Approximately 28-30% of all ETH is locked in staking, reducing the tradeable float. This makes the headline supply figure somewhat misleading in terms of market impact. But the directional fact remains: Ethereum is currently minting more ETH than it burns.
The Fusaka upgrade represents a partial fix. EIP-7918 ensures blobs always contribute some burn, even during low-demand periods. Fidelity Digital Assets projected that blob fees could contribute 30-50% of total ETH burn by 2026, depending on L2 transaction growth. But achieving net deflation requires either a sustained increase in L1 execution demand, a reduction in validator issuance (a politically contentious topic within the community), or an order-of-magnitude increase in blob utilization.
ETH's price reflects the uncertainty. The token traded at approximately $1,918 as of late July 2026, well below its 2024 highs, according to Fortune and CoinDesk data.
The blob capacity roadmap extends well beyond the current 14/21 target-max configuration. Core developers have indicated that BPO-3 and BPO-4 aim to reach 128 blobs per block, which would provide approximately 21 times the data throughput available before Fusaka.
The next major hard fork after Fusaka is Glamsterdam, planned for 2026. According to ethereum.org's roadmap documentation, Glamsterdam will focus on execution layer improvements alongside continued blob scaling. Specific EIPs have not been finalized.
For the L2 ecosystem, the path forward involves a feedback loop. More blob capacity lowers per-transaction data costs, which should drive more L2 adoption, which generates more blob demand, which pushes blob fees above the EIP-7918 floor and generates meaningful ETH burn. Whether this virtuous cycle materializes depends on factors outside the protocol's control: application demand, regulatory environment, and competitive pressure from alternative Layer 1 chains.
Metrika, an institutional blockchain analytics firm, noted that financial institutions should pay attention to Fusaka not for its headline scaling numbers but for what it signals about Ethereum's governance: "a narrower set of goals that more directly reinforce scalability, usability and, increasingly, value accrual to ether itself."
Fusaka is the first Ethereum upgrade to explicitly prioritize value accrual to ETH alongside scaling. The EIP-7918 blob fee floor and PeerDAS capacity expansion represent a coordinated attempt to resolve a contradiction that emerged after Dencun: cheap L2 transactions are good for users but bad for ETH's monetary properties.
The mechanism design is sound. Linking blob fees to execution gas prevents free-riding without pricing out rollups. The graduated BPO approach to capacity expansion provides real-world performance data before each increase. And PeerDAS reduces per-node costs, making higher blob counts sustainable.
But the economic question remains open. Ethereum's deflationary thesis depends on demand growth outpacing supply expansion — the same challenge faced by any infrastructure provider that reduces unit costs. Seven months of data shows cheaper transactions and growing L2 adoption, but not enough burn to offset issuance. The next 12 months, as BPO-3/4 expand capacity toward 128 blobs and Glamsterdam refines the execution layer, will determine whether Fusaka's fee redesign was a sufficient correction or merely a partial one.