Notice: all figures are derived from a proprietary extraction dated Oct-25-2026, and the fraction they measure changes weekly. Informational analysis: does not constitute financial advice. CleanSky does not receive commissions or referral payments from the products mentioned.
In the seven days ending Oct-25-2026, Ethereum issued 20,816.80 ETH and burned 213.18: the burn covered 1.02% of the gross issuance. The two terms of this fraction are not formed in the same way. Issuance is derived from a protocol formula based on the effective staking balance —43,827,546 ETH— and was 2,974.26 ETH per day, 98.66% of its ceiling. The burn consists of 30.45 ETH per day in base fee, the portion of the gas fee that the protocol destroys since EIP-1559 (August 2021); for every 100 ETH paid in fees that week, between 24 and 28 were burned depending on the data series: 24 according to the ultrasound.money dashboard and the DefiLlama aggregator, and 28 according to the implicit burn reported by data provider Coin Metrics. Between the Oct-21 and Oct-25 extractions, the seven-day window issuance moved by 0.04% and the burn by 6.00%. There is not a single 7-day or 30-day window with negative net supply between Oct-25-2025 and Oct-25-2026 in either of the two series; the last deflationary week was Nov-10 to Nov-17-2024.
How much did Ethereum issue and burn in the seven days leading up to Oct-25-2026?
The denominator is the step where it is easy to make a mistake. The supply variation published by a dashboard is net: it already has the burn discounted, so gross issuance is that variation plus the burn. In the seven-day window closed on Oct-25-2026, the supply measured by ultrasound.money rose from 122,139,419.81 to 122,160,023.43 ETH, a net increase of 20,603.62, and the burn was 213.18 ETH: gross issuance is 20,816.80. Dividing by the net figure gives 1.03% instead of 1.02%, and the difference grows as the burn increases.
| Window to Oct-25-2026 (rolling close at 08:30 UTC) | Burn (ETH) | Gross Issuance (ETH) | Burn / Issuance | Factor for zero net supply |
|---|---|---|---|---|
| 7 days | 213.18 | 20,816.80 | 1.0241% | 97.65 × |
| 30 days | 2,072.94 | 89,210.72 | 2.3236% | 43.04 × |
| Since the Merge (Sep-15-2022) | 2,017,944.47 | 3,656,928.51 | 55.18% | 1.81 × |
What determines Ethereum's issuance and how much can it fluctuate?
Consensus layer issuance does not depend on how many people use the network. It comes from the beacon chain reward formula, fixed since the Altair fork: for every epoch (32 slots of 12 seconds, approximately 6.4 minutes; 225 epochs per day), the protocol pays validators 64 times the square root of the effective staking balance measured in gwei, the billionth part of an ETH. With the 43,827,546 ETH balance as of Oct-25-2026 —35.88% of the supply— this results in 13.398 ETH per epoch, 3,014.64 ETH per day and 1,101,098.81 per year: a ceiling that assumes all validators are performing their duties without penalties.
The two independent series measure 2,974.26 ETH per day (ultrasound.money, from issuance per slot) and 2,978.92 (Coin Metrics, calendar days until Oct-24): 98.66% and 98.82% of the ceiling, respectively. The missing 1.2-1.3% represents undelivered work, likely late attestations and offline validators. The only lever that lowers issuance is having less ETH in staking, and the square root dampens this effect: to halve issuance, three-quarters of the balance would need to be withdrawn.
What determines the EIP-1559 burn and what portion of the fee is actually burned?
The other term is decided by how much gas is consumed and at what price. Each block burns its base fee multiplied by the gas used; the tip to the block proposer is not burned. A reading of the 7,175 blocks on Oct-20-2026 via RPC (the interface through which an Ethereum node answers queries) shows 217,874,198,953 gas units, an average base fee of 0.2290 gwei per block —simple average; weighted by gas used it is 0.2262— and 49.29 ETH burned that day.
How much of the fee ends up burned depends on the source. In the seven-day window, the implicit burn from Coin Metrics relative to the fees in its own series is 28.49%; ultrasound.money's burn relative to those same fees is 24.06%, and DefiLlama's ratio between burned revenue and fees is 24.14%. Three or four out of every five dollars paid to use Ethereum go to the block proposer. By month, and according to DefiLlama, burned revenue was 45.23% of fees in October 2025 and 31.3% in the 24 closed days of October 2026. This burn is not that of EIP-8363, the Draft proposal that calculates validator rewards as they are now and then discounts and destroys a portion: that would change how the fraction is accounted for. The current mechanism since 2021 burns fees, which is the numerator.
How much did ETH burn and issuance change between Oct-21 and Oct-25-2026?
The two extractions are 3 days and 19 hours apart, spanning 27,173 blocks.
| Data Point | Oct-21-2026 | Oct-25-2026 | Variation |
|---|---|---|---|
| 7-day Burn (ETH) | 201.11 | 213.18 | +6.00% |
| 7-day Gross Issuance (ETH) | 20,808.34 | 20,816.80 | +0.04% |
| 7-day Burn / Issuance | 0.9665% | 1.0241% | +5.96% |
| 30-day Burn (ETH) | 2,405.15 | 2,072.94 | −13.81% |
| 30-day Gross Issuance (ETH) | 89,045.86 | 89,210.72 | +0.19% |
| 30-day Burn / Issuance | 2.7010% | 2.3236% | −13.97% |
| Effective Staking Balance (ETH) | 43,758,037 | 43,827,546 | +69,509 |
| Formula Issuance Ceiling (ETH/day) | 3,012.25 | 3,014.64 | +0.08% |
| ETH Price (ultrasound.money) | $2,468.00 | $2,334.68 | −5.40% |
During that period, the 7-day fraction rose by 6% and the 30-day fraction fell by 14%, with the denominator remaining steady in both. The 30-day figure dropped because the window is rolling: it dropped four days from late September and picked up four from late October, which saw less burning. It lost 0.38 percentage points while the denominator rose by 0.19%.
How many times would fees need to increase for ETH supply to stop growing?
The factor is gross issuance divided by the burn: 97.65 times in the seven-day window and 43.04 in the 30-day window, two periods with different average burn rates. To keep the supply flat, the burn would need to be between 43 and 98 times higher and, since only 24% to 28% of what is paid is burned, fees would need to rise by that same order of magnitude.
They are moving in the opposite direction. According to DefiLlama, in the 24 closed days of October 2026 —from the 1st to the 24th; the 25th is still ongoing and excluded— Ethereum collected $10,429,690 in fees, $434,570 per day, compared to $1,328,444 per day during the 31 days of October 2025: 32.7% per day (comparing monthly totals gives 25.3%, but that crosses months of different lengths). In the burned portion, the drop is even greater: $136,018 per day compared to $600,825, or 22.6%. Meanwhile, the supply is growing at 0.88% annually at the rate of that seven-day window.
When was the last week with negative net supply on Ethereum?
Between Oct-25-2025 and Oct-25-2026, neither of the two series shows a 7-day or 30-day window with negative net supply, and the seven-day minimum for that year is not close to zero either.
| 7-day Window | Series | Net Supply (ETH) |
|---|---|---|
| Nov-10 → Nov-17-2024 | ultrasound.money | −1,221.20 |
| Nov-9 → Nov-16-2024 | Coin Metrics | −610.68 |
| Nov-8 → Nov-15-2024 | Coin Metrics | −227.91 |
| Apr-22 → Apr-29-2026 (year low) | ultrasound.money | +15,610.43 |
| Oct-18 → Oct-25-2026 | ultrasound.money | +20,603.62 |
The two sources label the window with a one-day difference —ultrasound.money marks the point at the start of the day and Coin Metrics at the close— and measure −1,221.20 and −610.68 ETH for what appears to be the same week: the difference is the level disagreement between series discussed in the limitations section. They agree on the month: November 2024. The long-term ultrasound.money data counts about 400 negative seven-day windows in its entire history —a figure sensitive to the noise mentioned in the next section— and none in the last year.
Why does the long-term ultrasound.money series report weekly variations incorrectly?
The since_merge segment of the supply-over-time endpoint labels its points at 00:00, but its daily increments from Sep-26 to Oct-25-2026 range from 198.27 to 5,834.49 ETH, when the actual daily issuance for that period —from the d30 segment, with exact midnight points— ranges from 2,716.52 to 2,964.96. No day in October 2026 issued 198 ETH or 5,834, and the level of the long segment deviates from the d30 level at the same midnight by up to 2,938.82 ETH: it appears the points are not placed at the hour they are labeled.
The seven-day variation from Oct-17 to Oct-24-2026 comes out to 17,841.45 ETH using the long segment and 20,599.01 using the short one: 13.4% less. In the week of Oct-13 to Oct-20, the difference was 8.3%. On Oct-21, the range of those daily increments was 92.01 to 5,834.53 ETH and remains just as wide on Oct-25-2026: the segment has not been corrected. Anyone calculating a specific week from that segment carries the error; however, it still serves for the historical count of negative windows.
What portion of issuance is covered by the toll that 22 L2s pay to Ethereum?
The toll that Layer 2 networks pay to Ethereum to publish their data is mostly burned and is included within those 213.18 ETH without being broken down. The measurement from Oct-13 calculated that 22 L2s paid $195,891.89 to Layer 1 between July 15 and Oct-12-2026 according to growthepie, the Layer 2 data aggregator: $2,176.58 per day over those 90 days. For the week of Oct-25, this is about 3% of the daily burn ($73,512.51) and 0.031% of the daily gross issuance ($6,942,938.63), which would need to be multiplied by 3,190 to cover one day of issuance. This comparison mixes a 90-day window with a 7-day window and two valuation bases —the burn is valued at the price of each block and issuance at the Oct-25 price of $2,334.68 published by ultrasound.money; with the burn at the daily price, the toll would be near 3.06%— and counts everything L2s pay, including tips, as burn. With these concessions, the toll moves 0.031 points on the 1.02% fraction: the second decimal place.
What does this Ethereum burn and issuance comparison not measure?
Five limitations of the measurement.
- Blob burn is not read via RPC. That reading is not included in any figure of the fraction, which uses ultrasound.money's burn: it serves as a contrast for Oct-20-2026. The BPO —blob-parameter-only— forks following Fusaka changed the update fraction for the blob base fee (EIP-8134 sets it at 8,346,193 and EIP-8135 at 11,684,671), so it is estimated by difference: DefiLlama reports $120,879 in burned revenue that day —50.24 ETH at the opening price of $2,405.92— compared to 49.29 from the RPC. This is an approximate bound of 0.96 ETH, which also absorbs the aggregator's price rounding.
- Windows do not align. ultrasound.money windows are rolling with a close at 08:30 UTC, while Coin Metrics uses calendar days: its 24th day closes at 00:00 UTC on Oct-25, 8.5 hours earlier than ultrasound.money. This is why the two measures of gross issuance at 7 and 30 days differ by 0.17% and 0.18% instead of zero.
- The two series do not measure the same burn or the same supply level. At seven days, Coin Metrics' implicit burn is 252.40 ETH compared to ultrasound.money's 213.18, 18% higher, which would result in a fraction of 1.21% instead of 1.02%. Supply variations do match —0.005% difference at 7 days and 0.027% at 28— but the level does not: ultrasound.money's short series is on average 4,327.14 ETH below Coin Metrics, with a range of 506.63 ETH over 28 days, five times the maximum 100 ETH expected between two meters of the same supply; the range jumped from 289 to 506.63 ETH due to a single day, Oct-23. The most plausible hypothesis is different treatment of deposits in flight and withdrawals. Coin Metrics' daily data also contains an artifact of +506.95 ETH on Oct-23 and −456.78 on Oct-24, which offsets within the window. Hence, variations are cited, but never cross-referenced levels.
- Effective balance is from an aggregator. The public beacon API returns 403 on
validator_balances, so the 43,827,546 ETH figure comes from ultrasound.money; contrasting it with the formula is circular, as the formula starts from that same balance. The independent contrast is Coin Metrics: its 2,978.92 daily ETH is 98.82% of the ceiling that balance implies. - The fraction expires: the 1.02% from Oct-25 was 0.97% on Oct-21. What does not expire is the method —gross issuance as the denominator, two series for variation— and the date of the last deflationary week.
The breakdown for Oct-25-2026 stands as follows. One side of the fraction is set by the protocol: a ceiling of 3,014.64 ETH per day with 98.66% delivered. The other is set by block-by-block demand: 30.45 ETH per day burned, with three or four out of every five fee dollars going to the proposer and daily fees at 32.7% of those from October 2025 according to DefiLlama.
Related articles: How much does an L2 pay to Ethereum? From 0.14% to 189.89%. EIP-8363: the formula already burns 60.46% with Sep-2026 ETH. Ethereum Exit Queue: 72.56% of ETH is consolidation. JPMorgan: why ETH will lag behind BTC in 2026. 2026 Scorecard: the 10 largest blockchains across 5 fundamentals.