The phrase "ETH is burned" gets repeated constantly — usually tied to a narrative about ETH becoming scarce, deflationary, or "ultrasound money." But the mechanism behind it is specific, and the supply implications are considerably more conditional than the framing suggests.
Ethereum started burning ETH in August 2021, when EIP-1559 activated as part of the London hard fork. Before that, miners received all transaction fees. After EIP-1559, a portion of every fee is permanently removed from circulation. Not paid to validators. Destroyed.
The burning exists because of a problem with how transaction fees worked — not because someone decided ETH should become scarce. The supply effect is a consequence of the mechanism, not its purpose.
The Problem EIP-1559 Was Solving
Before August 2021, Ethereum used a first-price auction for transaction fees. Users submitted a gas price — effectively a bid — and miners ordered transactions by bid, highest first. This created two persistent problems.
Predicting the right bid was genuinely hard. Too low, and your transaction sat in the mempool for hours. Too high, and you overpaid significantly. During congestion spikes — a popular NFT mint, a DeFi liquidation cascade, anything that suddenly drove demand — fees could jump 10x in minutes with no reliable way to anticipate it.
The second problem was subtler: miners had an incentive to manipulate fees. They could include their own transactions to fill blocks and drive fees up. They could strategically exclude low-fee transactions to create artificial scarcity. The auction format, combined with miner control over block contents, made this structurally possible.
EIP-1559 addressed both.
How the Burn Mechanism Actually Works
The upgrade replaced the single gas price with a two-part fee structure.
The base fee is set algorithmically by the protocol itself, not by users bidding. It adjusts block by block: when a block is more than 50% full, the base fee rises; when it's less than 50% full, it falls. This makes fee estimation tractable — users know roughly what they'll pay because the base fee changes predictably, not chaotically.
The priority tip is optional, paid directly to validators. It's how you signal urgency when you need faster inclusion. Most wallets default to a small tip; during periods of intense demand, users add more.
Here's the critical part: the base fee is burned. It goes to a null address. Validators receive only the tip.
Why burn the base fee rather than pay it to validators? Because a validator receiving the base fee has an incentive to game the fee market — fill blocks with their own transactions to collect fees, or exclude others to drive fees higher. Burning eliminates that incentive cleanly. Validators get only what users voluntarily offer as a tip. The base fee, which the protocol controls algorithmically, goes nowhere.
This design is the reason for the burn. The supply effect follows from it.
What the Burn Actually Means for ETH Supply
Every Ethereum transaction destroys some ETH — specifically, the base fee portion multiplied by the gas used. High activity means higher base fees and more transactions, which means more ETH destroyed.
Whether this makes ETH deflationary depends on the relationship between burns and new issuance. Validators currently earn roughly 1,700 ETH per day in staking rewards (the exact figure shifts slightly with validator count). If daily burns exceed that figure, net supply shrinks. If they fall below, it grows.
During high-activity periods — major token launches, DeFi liquidation events, spikes in L1 activity — burns have exceeded issuance and ETH supply has contracted. During quieter stretches, supply expands.
It isn't a fixed property. "ETH is deflationary" is only accurate sometimes, and only contingent on usage levels the protocol can't guarantee.
Where the Constraints Live
The burn rate is entirely usage-dependent. The protocol guarantees that base fees are burned; it can't guarantee how much base fees will be.
One development that genuinely complicates the supply narrative: Layer 2 adoption. As more activity migrates to rollups like Arbitrum, Base, and Optimism, L1 transaction volume declines. Less L1 activity means lower base fees and fewer burns.
The EIP-4844 "blob" upgrade in March 2024 made this more concrete. Blobs are a separate data lane that L2s use to post batches to Ethereum. Blob fees use a different fee market — one that was initially set very low — and the economics differ from regular transactions. When L2s submit data using blobs instead of calldata, the resulting L1 burn is minimal compared to equivalent activity conducted directly on L1.
This creates a real tension: successful L2 scaling, which Ethereum's roadmap explicitly targets, may structurally reduce L1 burn rates. The "ultrasound money" thesis assumed continued high L1 activity. That assumption deserves scrutiny.
What Would Confirm the Burn Thesis
The mechanism continues working: base fees are burned on every transaction by protocol rule, and that's not changing. Confirmation that the supply model is materially deflationary: sustained weeks where on-chain data shows net negative ETH issuance — not occasional dips, but consistent patterns during normal market conditions rather than only during extreme demand spikes.
What Would Break It
A meaningful shift of L1 activity to L2s, with blob fees remaining low, could reduce burn rates to the point where ETH is structurally inflationary at typical usage levels. A protocol-level change to EIP-1559 (theoretically possible through governance, practically very difficult) could alter the burn mechanic. If validator rewards were increased significantly — for example, through changes to the issuance schedule — issuance could outpace burns even during high-activity periods.
None of these are likely near-term. But framing ETH as permanently deflationary ignores that the math is conditional.
Timing
Now: EIP-1559 is active and every L1 transaction burns ETH. Whether net issuance is positive or negative shifts week to week based on activity. The data is publicly visible in real time.
Next: Blob fee economics are still early. As L2 usage grows and blob space fills more consistently, the fee market for blobs will develop, potentially increasing L1 burns from that channel.
Later: Full danksharding, if implemented, would dramatically expand blob capacity. How this affects long-term burn rates is genuinely uncertain — more capacity could mean lower blob fees and less burn, or more total activity could offset lower per-unit costs.
What This Doesn't Cover
This is an explanation of the burn mechanism and why it exists. It isn't a supply forecast or a price thesis. Whether ETH supply is net inflationary or deflationary at any given moment depends on conditions outside this scope — and anyone presenting deflation as a guaranteed feature of ETH is making a stronger claim than the mechanism supports.
The burn is real. It works as designed. Whether it's enough to drive meaningful supply reduction depends on how Ethereum is used.
