Coinbase burned 13,800 ETH in the first 54 days of EIP-1559
Coinbase Blog · Sep 27 2021
“Coinbase executes a large number of transactions on Ethereum to source liquidity and provide withdrawals for our customers, so these savings add up. As of September 27, 2021, Coinbase has burned 13,800 ETH, for an average of about 254 ETH per day.”
— Coinbase Blog · Sep 27 2021, September 27, 2021
Their claim · ETH burned
13,800 ETH
Our reproduction · ETH burned
Methodology
Theirs
Coinbase summed the EIP-1559 base-fee burn (gas_used × base_fee_per_gas) across every transaction sent from a Coinbase-controlled Ethereum address from EIP-1559 activation (block 12,965,000, Aug 5 2021) through the day of the blog post (Sep 27 2021). They had ground truth: their own internal wallet inventory.
Ours
We pulled every Ethereum address Allium has labeled as Coinbase, then summed the base-fee burn on each transaction those addresses sent across the same Aug 5 → Sep 27 2021 window. Labels only — no Coinbase-internal data.
Daily ETH burned · 0 days
Allium commentary
This reproduction uses Allium labels only, with no input from Coinbase. We identified every Coinbase-controlled Ethereum address from our attribution pipeline and summed the EIP-1559 base-fee burn on each transaction those addresses initiated between EIP-1559 activation (block 12,965,000, August 5 2021) and Coinbase's stated cutoff of September 27 2021.
Landing within a narrow margin of Coinbase's published figure is a strong validation of our coverage. It demonstrates that Allium's Coinbase Ethereum address set is comprehensive enough to account for nearly every transaction Coinbase published a number for — at the granularity of individual base-fee burns, not just aggregate volume buckets.
The small residual gap is consistent with the expected drift between an internal wallet inventory and a labels-only reproduction. The most likely contributor is short-lived hot wallets rotated in and out of service during the 54-day window: Coinbase's internal record of those rotations is exact by construction, while a labeling pipeline takes time to enumerate them. A secondary contributor is the ambiguity of the September 27 cutoff itself, which we resolve by taking the most inclusive 54-day window. Neither effect indicates a structural gap in coverage; both narrow further as the labeling pipeline backfills historical wallet rotations.