EIP-8363: the formula already burns 60.46% with the Sep-2026 ETH

Applied to the active balance of the Ethereum beacon chain —the consensus layer where validators live— measured on September 11, 2026, at 43,081,487 ETH distributed among 909,773 validators, the EIP-8363 formula burns 60.46% of the consensus reward. Since August, the proposal has been framed as a long-term threat: burning validator issuance until the net yield reaches zero "when half of all ETH is staked" (staking = locking ETH to validate and earn rewards). The constant governing this burn is not a percentage of the supply but a fixed balance of 60,250,000 ETH, and the counter feeding it is already at 71.50% of that figure. Since the formula's exponent is 3/2, that 71.50% progress translates into a 60.46% burn. None of this is active —EIP-8363 (Ethereum Improvement Proposal) remains a draft— but the number deciding how much is burned is not moved by a vote: it is moved by every new deposit the network activates, at a rate the protocol itself limits to 256 ETH per epoch. This article breaks down the mechanics piece by piece: what magnitude the formula reads, what pushes it, at what speed, what would remain for the validator, and where the proposal actually stands in the Ethereum process.

What exactly does EIP-8363 propose and on what number is the burn calculated?

A mobile data plan with progressive speed throttling serves as a starting point: while consumption remains below a certain threshold, the connection runs at full speed; beyond that, the operator cuts an increasing percentage until it reaches a minimum. The mapping to EIP-8363 is direct. The cumulative consumption is the network's total active balance; the cut is the burned fraction; the plan's threshold is a constant called SATURATION_BALANCE.

The technical name for this cut is tapered issuance burn. For every validator duty (attestation, block proposal, sync committee), the protocol calculates the reward corresponding to current rules and immediately deducts a fraction b that is destroyed. The fraction is b = (D / SATURATION_BALANCE)^(3/2), capped at 1, where D is the total active balance.

Two limits of the analogy matter more than the similarity. First: there is no allowance that resets every month. The counter is cumulative and collective, increasing whenever anyone deposits. Second, and this is what the specification explicitly states: the protocol has no notion of the total ETH supply. The only thing it observes is get_total_active_balance(state), the sum of the effective balances of validators in an active state. The practical consequence is that the "50% of supply" mentioned in headlines appears nowhere in the code: what appears is an absolute number of ETH.

The authors of the draft are pintail, Jérôme de Tychey, dapplion, pa7x1, Ladislaus von Daniels, and Justin Drake. The file creation date is July 14, 2026; on August 4, the pull request incorporating it into the EIP repository was opened, and this second date is the one that has circulated as if it were the first.

Why does the EIP-8363 formula yield 60.46% with the active balance from September 11?

The yield curve for this proposal is published within the EIP itself: there are issuance and yield charts among the document's attachments, the closed form ỹ(f) = (1 − b(f)) · y(f) is present, and four numerical points are cited in the text. What follows is an evaluation of that curve at the active balance measured today on the network, which is the number the formula actually reads.

Measurement Method. Querying the /eth/v1/beacon/states/head/validators?status=active_ongoing endpoint of a public beacon node and summing the effective_balance fields of the returned validators. Result at slot 15,190,426 on September 11, 2026: 43,081,487 ETH active across 909,773 validators. As an independent contrast on the same day, the validatorqueue.com dashboard showed 909,711 active validators and 43.0 million ETH.

The chosen metric is not interchangeable with those appearing on dashboards. The sum of beacon balances published by ultrasound.money for that same moment was 43,239,500 ETH: 158,013 ETH more than the sum of effective balances. The difference stems from the step-rounding of effective balances and balances that do not yet count as active. For the EIP formula, only the first of the two figures is valid, as it is what get_total_active_balance returns.

Regarding issuance, the EIP declares its own assumptions: annual consensus issuance is 64 · √(D) Gwei per epoch with 225 epochs per day. From this, a closed model —proprietary calculation— for the annual gross yield is derived. Before using it for points the document does not publish, it is worth contrasting it with the four it does publish.

It matches within 0.5% across all four. With the model validated, this is the curve evaluated at the measured point and in reference scenarios. The ratios use a supply of 122,032,253 ETH as the denominator as of September 11, 2026, derived from the identity: execution balances + beacon balances − beacon deposits; the supply only affects the ratio column, never the calculation of b.

The takeaway from the first row is that 50% of the supply does not mark the start of the burn. It marks its end. The stretch between today's state and that end is worth 39.78 burn points, and the first 60.46 have already been covered.

Why is the "50% of supply" in EIP-8363 headlines actually 49.372%?

The EIP explains that SATURATION_BALANCE is set at the fork "at approximately half of the current ETH supply." A balance of 60,250,000 ETH implies a reference supply of 120,500,000 ETH. The supply measured on September 11, 2026, is 122,032,253 ETH: 1,532,253 ETH above the figure used to calibrate the constant. Therefore, the effective saturation ratio today is 49.372%, not 50%.

The difference seems cosmetic but is not, due to the direction in which it moves. A constant fixed in absolute ETH and a varying supply produce a permanent drift of the effective threshold. And the proposal itself pushes this drift: burning issuance reduces the supply, and with a smaller supply, the same balance of 60,250,000 ETH represents a higher percentage. A mechanism justified as a "cap at 50% of ETH staked" would, once active, have an effective cap that moves away from 50% upwards as it fulfills its function.

The formulation in absolute balance is reasoned in the specification itself: the protocol cannot query the supply, which is a real limitation of the consensus layer. What is lost with it is precisely the property —the cap at half of all ETH— that the headline takes for granted.

How much burn do the 1,875,196 ETH in Ethereum's entry queue add?

The entry queue is the piece that moves all others. The same query to the beacon node, at the /eth/v1/beacon/states/head/pending_deposits endpoint and in the same slot, returns 1,875,195.89 ETH distributed across 22,963 pending deposits awaiting activation. The validatorqueue.com dashboard gave 1,881,451 ETH that same day, with an estimated wait of 32 days and 16 hours, and an exit queue of 2,635 ETH.

This ETH is not attributed to any specific type of participant in this article: it is a figure measured in the chain state, with no source allowing it to be split between liquid staking providers, corporate treasuries, exchange-traded products, or individual validators. Without such a document, any distribution is hypothetical.

What can be calculated precisely is its effect on the formula. The protocol activates deposits at a maximum rate of 256 ETH per epoch, which is 57,600 ETH per day. From there, everything that follows is a proprietary calculation using the EIP curve:

1. Each day at the maximum activation rate adds 0.121 percentage points of burn. The balance moves from 43,081,487 to 43,024,643 ETH and b rises from 60.46% to 60.34%.
2. Thirty days of a full queue are worth 3.67 points. The burn would reach 64.14% without any client team, developer, or ETH holder making a single decision.
3. The full queue is worth 3.99 points and takes 32.6 days to clear. With the 1,875,196 ETH already activated, the balance rises to 44,956,683 ETH, b reaches 64.45%, and the permanent net yield drops from 1.002% to 0.882%.

And an arithmetic ceiling that should be read as an upper limit rather than a forecast: at the maximum rate the protocol allows for activation, it would take 298 days from September 11, 2026, to bring the active balance from 43,081,487 to the 60,250,000 ETH saturation point. This would require the queue to remain full for that entire time, which is not guaranteed. It marks the physical maximum speed of the counter, which is the only thing the protocol controls.

The historical series of this counter helps calibrate the real pace. On June 15, 2026, we analyzed the ETH paradox with 39.5 million ETH staked, around 32.5% of the supply, while exchange-traded products were losing capital. In that article, EIP-8363 was not on the table —the draft did not yet exist— and the staking record was read as a sign of conviction. Applying today's formula to that figure, b would have been around 53% —slightly less strictly speaking, as that data was ETH staked and not active effective balance—. From June 15 to September 11, in 88 days, the implicit burn has risen by about seven points. Conversely, in April 2026 we wrote that the entry queue was around 4 million ETH with waits over 60 days: the September 11 measurement places it at less than half, 1,875,196 ETH and 33 days.

What would an Ethereum validator earn under the permanent EIP-8363 curve, and does MEV compensate for it?

With the measured balance, the annualized gross consensus yield comes to 2.534% and the net under the permanent curve to 1.002%. In units that can be checked in an account: a 32 ETH validator would go from 0.811 to 0.321 ETH per year in consensus rewards; a 1,000 ETH position, from 25.34 to 10.02 ETH annually.

At the network scale, annual consensus issuance with today's balance is 1,091,687 ETH (proprietary calculation using the formula declared in the EIP). The permanent curve would destroy 660,083 ETH per year and leave 431,604 ETH of net issuance, 0.354% of the supply compared to the current 0.895%.

The common objection is that MEV (Maximal Extractable Value from transaction ordering within a block) would compensate for the loss. The EIP itself answers this with two figures: MEV is worth at most 0.20% annually against the staked base, and issuance accounts for at least 93% of a validator's annual yield. Adding that 0.20% to both sides, total income would drop from 2.737% to 1.209%: a 55.8% drop compared to the 60.2% taken by the isolated consensus portion. The cushion is worth four percentage points on the drop. Those who feel it most are validators with high fixed costs per ETH managed, one of the three repeated objections in the Ethereum Magicians thread from August 4 and 5, along with the relative weight of MEV and the effect on institutional demand. We wrote about how this income base is distributed between operators and depositors when analyzing the $27.6 billion managed by Lido.

Does the 18-month transition included in EIP-8363 serve as a cushion?

The proposal does not apply the permanent curve all at once. It starts with TRANSITION_BASE_REWARD_FACTOR at 128 —double the current issuance factor— and lets it decay to 64 over 123,300 epochs, which at 225 epochs per day is 548 days: the 18 months mentioned in headlines.

The cushion works better the less ETH is staked, and therein lies its problem with the measured state. Evaluating the initial epoch of the transition with the September 11, 2026 balance, the net yield comes to 2.004% compared to today's 2.534% gross. In other words: on the first day of the transition period, the validator would already earn 20.9% less than now, and from there it would decrease over 18 months to 1.002%.

The document says it in its own words when describing the design: the transition curve crosses the current curve at a staking ratio of around 31%. Below that point, the transition pays more than the current system; above it, less. The ratio measured on September 11, 2026, is 35.30% and the entry queue is pushing upward, so the transition period would start already on the side where it subtracts.

Where does EIP-8363 stand in the process after ACDC #184, and is it going to Glamsterdam or Hegotá?

The proposal was presented at ACDC #184 (All Core Devs – Consensus, the periodic meeting of teams maintaining Ethereum's consensus layer clients) on August 6, 2026, with 30 minutes on the agenda —more than any other item in the block dedicated to Hegotá. The formal request for inclusion was published by Jérôme de Tychey on August 5 at 14:41 UTC, one day before the window closed, and this haste was one of the procedural objections raised.

The automated summary of the call, published as an artifact in the ethereum/pm repository, records the sequence: Greg, from Lido, requested the voluntary withdrawal of the proposal from the Hegotá list; de Tychey declined; a third participant recorded in the summary, Ansgar, noted that community consensus is a prerequisite for any stage advancement. The decision noted at that same minute is literal and deserves to be read in full: "EIP-8363 PFI is valid per process (permissionless); no CFI or SFI decision made; community consensus explicitly stated as prerequisite for future acceptance."

That sentence sets the exact stage. PFI (Proposed for Inclusion) is a permissionless stage: anyone can request that an EIP enter the list, and the process admits it. It is not equivalent to a vote of support from developers nor a stage that is "earned." CFI (Considered for Inclusion) and SFI (Scheduled for Inclusion) are, and no decision was made regarding those.

The documentary status on September 11, 2026, queried in EIP-8081 —the Hegotá fork meta— gives an honest measure of where the proposal stands:

• 50 EIPs on the PFI list for Hegotá, EIP-8363 among them.
• 2 EIPs in SFI: EIP-7805 (FOCIL) and EIP-8141 (Frame Transaction).
• 0 in CFI and 0 in Declined: no proposal from that list has advanced a stage or been discarded.
• Empty activation table across the three networks listed in the document —Sepolia, Hoodi, and mainnet— with no epoch or timestamp. Hegotá has no date.

Two more clarifications, both verifiable against the repository. First: the fork is Hegotá, not Glamsterdam. They are two different upgrades, and the proposal appears in the meta for the second one; Glamsterdam is the previous fork, whose delay we analyzed regarding ePBS (proposer-builder separation) in June. Second: the correct number is 8363. The file in master is eip-8363.md and no eip-8361.md exists. The origin of the error is localized: the pull request adding the proposal to the Hegotá list was opened with a provisional number and titled "Update EIP-8081: PFI EIP-8361"; a subsequent commit corrected the number within the diff, but the title was never fixed, and much of the coverage took the title from the PR.

That same PR is the answer to the withdrawal rumor circulating in some aggregators. It was merged on August 21, 2026, fifteen days after the call in which the proposal's withdrawal was requested, and the entry remains on the list as of September 11, 2026. The sequence, in order: withdrawal was requested, the author rejected it, and the EIP remained in the stage that does not require agreement.

What would need to be re-measured to know if the 60.46% burn of EIP-8363 is still valid?

The premise of this article depends on two numbers, and both are public, so anyone can redo the math and see when it stops being valid. Three specific things to watch:

1. The active balance. If get_total_active_balance drops below approximately 35.4 million ETH, the burn falls below 45% and the statement that it is "already above 60%" ceases to be true. With the exit queue at 800 ETH compared to 1.97 million in the entry queue, the pressure on September 11 was moving in the other direction.
2. The constant. SATURATION_BALANCE is worth 60,250,000 ETH in a document in Draft status. Reparameterizing it is a single line of diff and would change all the numbers in the curve table.
3. The stage. A promotion to CFI would turn the proposal into a matter with a schedule; a move to Declined would make it archival material. With 50 EIPs in PFI and 2 in SFI for a fork without an activation date, the probability of Hegotá incorporating this specific piece is low as long as the stage does not change.

In the meantime, anyone with staked ETH can do a calculation today that wasn't on the table two months ago: what part of their yield comes from issuance and what part from MEV, and what would remain for them with a 60.46% discount applied only to the first part. The arithmetic of the proposal —the fixed balance versus the ratio, the 3/2 exponent, the queue pushing the counter— remains the same even if the draft changes number, fork, or destination; what moves every epoch is the counter's position within it. Regarding the relationship between burning and the network's income structure, you can continue with our analysis of MEV Era III, and those without prior context can find the basics in our guide on what is staking.

Sources and links: EIP-8363 — Tapered Issuance Burn (ethereum/EIPs) · EIP-8081 — Hardfork Meta: Hegotá · PR #12081 — EIP registration · PR #12087 — PFI in Hegotá (merged Aug 21, 2026) · ACDC #184 Agenda · ACDC #184 Automated Summary · Discussion thread on Ethereum Magicians · Public beacon node (proprietary measurement) · validatorqueue.com · ultrasound.money