The Ethereum EIP-7918 burn floor, activated inside the Fusaka upgrade on 3 December 2025, is the protocol’s most direct answer yet to a problem that has been draining the ultrasound money thesis since March 2024: after Dencun, rollups got so cheap to run that they stopped burning anything worth measuring.
Daily ETH burn collapsed from thousands of coins during congested mainnet periods to as low as 50 to 70 ETH after the Dencun upgrade introduced blob transactions. With issuance running around 1,700 ETH per day, the equation flipped. According to ChainCatcher citing ultrasound.money data, Ethereum’s annual inflation turned positive after Dencun, and net supply growth settled somewhere between 0.2% and 0.8% depending on the measurement window. ETH supply crossed back above its Merge-era level. The deflation that made the ultrasound story legible was gone.
How EIP-7918 Works and What It Actually Fixes
Before EIP-7918, the minimum blob fee was 1 wei, set by EIP-4844 at launch. In practice, with blob capacity massively oversupplied relative to demand, fees collapsed to that floor and rollups consumed base-layer data for near-free. The burn from blobs was rounding-error territory.
EIP-7918 ties the minimum blob fee to the execution base fee. Authored by Anders Elowsson, Ben Adams, and Francesco D’Amato (created 25 March 2025), the proposal sets a preliminary BLOB_BASE_COST of 2^14 (16,384). Per Elowsson’s research note, the reserve blob base fee works out to 1/8 of the execution base fee (2^14 divided by 2^17), placing the blob reserve price slightly below the cost of a simple ETH transfer. Rollups can no longer get blob space essentially free during quiet periods on the execution layer.
The authors’ framing on Ethereum Magicians is that this is not a temporary patch. The mechanism is designed to adapt with scaling without further manual adjustment, and to remove extended periods of above-target blob consumption that the old mechanism could not price quickly enough when starting from near-zero.
The Fidelity Modelling: $78.6 Million Left on the Table
Fidelity Digital Assets ran the retrospective. Had EIP-7918 been live since blobs launched, it would have generated an estimated additional $78,646,739 (24,641 ETH) in cumulative blob-fee revenue across 93% of days since Dencun, using an ETH price of $3,984 as of 28 October 2025. Of that, $9,793,737 (3,037 ETH) accrued in the narrower window since the Prague-Electra upgrade on 7 May 2025.
The per-blob cost math puts the L2 burden in perspective. Fidelity calculated an additional cost of roughly $6.02 per blob ($78,646,738 divided across 13,057,849 blobs since Dencun). For Base, the most active L2 platform, submitting at approximately 580 blobs per hour, that translates to roughly $30.6 million in additional annual fees. Base paid approximately $5.2 million in total blob fees over the past year while generating roughly $94 million in revenue from user transaction fees. The base layer was providing the security that made that $94 million possible and capturing a fraction of it.
Blob Supply Keeps Growing, Which Complicates the Picture
EIP-7918 was not deployed in isolation. The Fusaka mainnet activation at 21:49:11 UTC (block slot 13,164,544) was followed by two Blob Parameter Only (BPO) forks: BPO1 on 9 December 2025 and BPO2 on 7 January 2026, according to CCN. Those forks progressively raised blob throughput targets to a 14/21 (target/maximum) configuration. The result, per Spark’s blob fee market research, is that utilisation dropped to 20–30% of capacity as supply again outpaced demand.
That is the central tension EIP-7918 cannot fully resolve. The floor prevents fees from collapsing to 1 wei, but it cannot force utilisation up. If the protocol keeps expanding blob capacity faster than rollup demand fills it, the floor becomes the ceiling, and the additional burn remains modest.
Ethereum EIP-7918 Burn Floor: Floor, Not Restoration
The ultrasound money thesis needed burning to outpace issuance on a sustained basis. EIP-7918 does not recreate the congested-mainnet era where thousands of ETH burned daily. It ensures the protocol captures something from rollup activity rather than nothing, and that something grows with execution-layer demand.
The structural conflict remains. Ethereum’s roadmap chose cheap, high-throughput settlement over base-layer fee density. A congested, expensive mainnet burned more ETH. A scaled, efficient one burns less, and the roadmap explicitly optimises for the latter. EIP-7918 narrows the gap; it does not close the contradiction.
Ethereum’s roughly 0.2% net inflation in calmer periods already runs below Bitcoin’s fixed-schedule 0.8%. With roughly 28–30% of ETH supply locked in staking, the tradeable float is tighter than the headline supply number implies. Neither of those facts requires ultrasound money to be true; they are just the actual state of the asset in 2026.
The two numbers that will settle the argument over the next twelve months are net ETH supply growth and base-layer fee revenue. If EIP-7918 combined with rising rollup throughput pushes fee revenue back up from its roughly $10 million lows while net supply growth flattens, the value-capture case firms up. If blob capacity keeps expanding faster than demand and fee revenue stays depressed, the burn floor was a maintenance patch on a structurally changed network, not a revival of a monetary thesis.
