Ethereum has destroyed more than 4.6 million ETH since August 2021. And yet there is more ETH today than ever before. Both are true, and the reason is the most useful thing to know about token burns: a burn tells you what left the supply, not what came in.

Two figures on a dark background. Left: 4,634,719 ETH, destroyed since August 2021, with a flame. Right: 122.07M ETH, the most there has ever been, over a line chart of Ethereum's supply rising from 2021 to today.
As of 25 September 2026: 4,634,719 ETH burned since August 2021, and a supply of about 122.07 million ETH, the highest on record.Burns: BurnTracker. Supply: Coin Metrics Community (SplyCur, daily), CC BY-NC 4.0, charted and rounded by cryptoburntracker.com.

To see how both can be true, we need to be precise about three things: what a burn is, what it is not, and what a burn total can and can’t tell you.

What a token burn is

A token burn permanently removes tokens from use. The burned tokens stay in the blockchain’s history, but nobody can ever spend them again.

The supply drawn as a mint pie chart. One slice is missing: it has gone into a dead address, marked by a flame.
A burn takes a slice of the supply out of use, for good.

To see why that’s possible, start with how tokens work. Your tokens aren’t files on your phone. They’re a balance recorded on a blockchain, held by an address. To move them, you need that address’s private key, the secret that signs every transaction. Whoever has the key controls the tokens.

A dark card labelled with an address, 0x7a3…c91e, holding three tokens, next to a gold key.
Tokens are a balance at an address. The private key is what lets someone move them.

A burn takes that control away for good. It works in one of two ways:

  • The tokens go to an address that no one is known to have a key for. This is called a dead address or burn address. The tokens are still recorded there, but nobody can sign a transaction to move them out.
  • Or the network deletes them outright. The protocol’s own rules destroy the tokens, and they stop existing.

Either way, nobody can ever spend them again: not the project, not an exchange, not you.

Left: a dead address with a flame and a crossed-out key, labelled 'an address no one has a key for'. Right: a dark block, labelled 'or the network deletes them'. Below: Nobody can ever spend them again.
The two ways a burn removes control: a dead address no one has a key for, or deletion by the network.

What a burn is not

Plenty of token moves look like burns in a headline but aren’t. Many projects keep a treasury: a wallet that holds the project’s own tokens, to spend later. Others lock tokens in a contract that only releases them on a set date.

In both cases, someone still controls those tokens. They’re only parked, and they can come back.

Under the heading 'Parked, not burned': a pie chart missing one slice, a treasury vault with its key still held, and a padlock labelled 'released on a set date'.
A treasury and a lock only park tokens. Someone still holds the key, or the lock opens on a set date.

The same goes for buying. A project that buys its own tokens back hasn’t burned anything yet. The purchase becomes a burn only when the tokens it bought are permanently removed.

If someone can still move the tokens, they aren’t burned.

Three ways tokens get burned

Burns come in three main forms. Each works differently, and each has a well-known real example.

1 · Burn addressTokens are sent to an address no one can move them out of. Example: Shiba Inu.
2 · Protocol burnThe network destroys tokens under its own rules. Example: Ethereum.
3 · Buyback and burnTokens are bought on the market, then removed for good. Example: Hyperliquid.

1. Burn address: Shiba Inu

The simplest burn is a transfer: you send the tokens somewhere they can never leave. Shiba Inu’s history holds the most famous example.

The SHIB token contract was deployed on 31 July 2020 with 1,000 trillion SHIB. Within days, about half of all the tokens, some 495 trillion SHIB, were sent to Ethereum’s co-founder, Vitalik Buterin.

A pie of 1,000 trillion SHIB split in half. The gold half sits on a card labelled Vitalik Buterin, 0xab58…aec9b. Below, his quote: I don't want to be a locus of power of that kind.
August 2020: about half of all SHIB went to Vitalik Buterin’s wallet.Source: Etherscan records of the SHIB contract and his address.

He hadn’t asked for them. In a note he left on the blockchain, he wrote that he didn’t want to be “a locus of power of that kind.”

In May 2021, he acted. On 12 May, he gave about 50 trillion SHIB to the India Covid-Crypto Relief Fund. Four days later, on 16 May at 21:48 UTC, he sent 410,241,996,771,871 SHIB, 90% of what he had left, to a dead wallet, in one single transaction. That one transfer removed about 41% of all the SHIB that was ever created.

Three cards in a row: Vitalik Buterin's wallet with a small sliver of SHIB left, the COVID relief fund (India, 12 May 2021), and a dead address with a flame (16 May 2021). Top right: 410.24T SHIB burned in one transaction, about 41% of all SHIB ever created.
May 2021: about 50 trillion SHIB to a COVID relief fund in India, then 410.24 trillion SHIB burned in one transaction.

Now look at SHIB’s burns over time. As of 25 September 2026, BurnTracker counts 410.76 trillion SHIB burned in total. About 99.9% of it is that single transfer. Everything burned since then barely shows on the same scale.

A bar chart of SHIB burned per month from 2020 to 2026. One tall red bar in May 2021, labelled 16 May 2021, 410.24T. Every other month is too small to see.
SHIB burned per month, September 2020 to September 2026. One month holds almost the entire total.Source: BurnTracker SHIB tracker, daily burns.

So a big burn total doesn’t always mean a lot is still being burned. It can be one event, years ago.

2. Protocol burn: Ethereum

In a protocol burn, nobody sends anything anywhere. The network itself destroys tokens under its own rules.

Ethereum has done this since the London upgrade on 5 August 2021, which introduced EIP-1559. Every transaction pays a fee in two parts. The base fee is burned: it isn’t paid to anyone, it simply stops existing. The rest, the priority fee or tip, goes to the validator who adds the transaction to a block.

Under the heading 'Protocol burn': a validator card labelled 'tip: to the validator', and a dead address with a flame labelled 'base fee: burned'.
Every Ethereum fee splits in two: the base fee is burned, the tip goes to the validator.

Each block burns only a little. But every block burns some, and it adds up. In just over five years, Ethereum has burned 4,634,719 ETH, as of 25 September 2026. BurnTracker counts it block by block: for every finalized block, the base fee per unit of gas times the gas used.

An area chart of cumulative ETH burned from 2021 to 2026, rising steeply at first and flattening from 2024, ending at 4.63M ETH.
Cumulative ETH burned since the London upgrade in August 2021: 4.63 million ETH as of 25 September 2026.Source: BurnTracker ETH tracker, daily burns.

3. Buyback and burn: Hyperliquid

The third way involves the market. Tokens are bought, then removed for good. Remember: the purchase alone isn’t the burn. The removal is.

Hyperliquid works like this. Its Assistance Fund converts trading fees into HYPE automatically, and Hyperliquid’s documentation says that HYPE is burned, removed permanently from both the circulating and the total supply.

A flow from left to right: fees, then the Assistance Fund, then a HYPE token burning on a dead address.
Hyperliquid: trading fees go to the Assistance Fund, which turns them into HYPE that is burned.

Because it runs on fees, the burn follows trading activity: busy markets burn more, quiet ones less.

  • Over the last 30 days, it burned about 23 thousand HYPE a day. At today’s price, that’s around $2.2 million a day.
  • Its busiest day was 10 October 2025: 517,563 HYPE. That was worth about $22 million at the time. At today’s price, the same amount would be worth about $48 million.
  • In total, BurnTracker counts about 47.8 million HYPE burned as of 25 September 2026. At today’s price, that’s about $4.5 billion.
A chart of HYPE burned per day from February 2025 to today, with one tall spike on 10 October 2025 (517,563 HYPE, about $22.48M at the time, about $48.44M at today's price) and the last 30 days marked at about 23.1K HYPE a day. Top right: total burned 47.85M HYPE as of 25 September 2026, about $4.48B at today's price.
HYPE burned per day, with the busiest day and the last 30 days marked.Source: BurnTracker HYPE tracker. “Today’s price”: $93.60, CoinGecko in BurnTracker’s snapshot of 25 September 2026. Price on 10 October 2025: the average of that day’s hourly CoinGecko prices, $43.44.

One note on those dollar figures. A value “at today’s price” multiplies the burned tokens by the latest price. It isn’t the money spent on the burn, and it isn’t what the tokens were worth when they were burned.

Does a burn shrink the supply?

Now for the question from the start. If a token burns, does its supply go down? Not necessarily.

Think of the supply as water in a bathtub. New tokens are the tap. Burns are the drain. If the drain is bigger than the tap, the level falls. If the tap is bigger, the level still rises, even while tokens are being burned.

A bathtub of mint water labelled supply, with a tap labelled 'new tokens' pouring in and a drain labelled 'burns' with a flame. Below: More created than burned: it rises, even with burns.
Supply as a bathtub: new tokens flow in from the tap, burns flow out through the drain.
Example · round numbers

A network creates 1,000 new tokens in a day and burns 600. That’s a busy day for burns, and the supply still grows by 400 tokens.

Supply changenew tokens created − tokens burnedPositive: the supply grows, even if tokens are burned. Negative: it shrinks.

A tank, a river or a mint

It also matters what feeds the tap. Tokens fall into three groups:

  • A maximum supply, like a tank. There is a fixed number of tokens. Once the tank is empty, no new ones can be created. Examples: HYPE, SHIB, XRP.
  • No maximum, like a river. New tokens can keep flowing for as long as the network’s rules allow. Examples: ETH, TRX, ICP.
  • Mintable. Whoever controls the token’s contract can create new ones. Examples: the stablecoins USDC and USDT, whose issuers mint new tokens.
Three icons with example coins. A tank marked max: maximum supply (HYPE, SHIB, XRP). Waves of a river: no maximum (ETH, TRX, ICP). A contract card reading mint(): mintable (USDC, USDT).
Where new tokens come from: a tank, a river or a mint.

Ethereum: a record supply after 4.6 million burned

Ethereum has no fixed maximum supply, so its tap is a river. Here is its supply since 2021, with the ETH burned each day underneath.

A line chart of ETH supply since 2021, with markers for the start of the burn (August 2021) and the Merge (September 2022), a low of 120.07M in April 2024 and a highest-ever 122.07M today. Below it, red bars of ETH burned per day, falling from about 2,800 a day to about 500 and then 41.
ETH supply since 2021, with ETH burned per day below.Supply: Coin Metrics Community (SplyCur, daily), CC BY-NC 4.0, charted and rounded by cryptoburntracker.com. Burns: BurnTracker ETH tracker.

Read it from left to right:

  1. August 2021: the burn starts. But new ETH was still being created faster than it was burned: about 13,000 ETH a day was issued before the Merge, against about 6,461 burned. So the supply kept rising, from about 117.19 million ETH at London to 120.52 million at the Merge.
  2. September 2022: the Merge. It cut the amount of new ETH by about 88%. For a while, burns were bigger than issuance, and the supply actually shrank, down to about 120.07 million on 4 April 2024.
  3. Then the burn slowed. It averaged about 2,800 ETH a day from September 2022 to March 2024, around 500 a day from March 2024 to September 2026, and just 41 a day over the last 30 days.
  4. Today: the tap wins again. With the drain that small, Ethereum’s supply is about 122.07 million ETH, the highest it has ever been.

That is how both facts from the start are true: more than 4.6 million ETH burned, and a record supply.

Hyperliquid: a tank that is still being released

HYPE is different. It has a maximum supply of one billion tokens, so its tap is fed by a tank that can’t be refilled.

Two bathtubs side by side. Ethereum's tap is fed by a river: no fixed maximum. Hyperliquid's tap is fed by a tank marked max 1 billion: a tank.
Ethereum’s tap is fed by a river. Hyperliquid’s is fed by a tank of at most 1 billion HYPE.

A tank doesn’t mean the tap is closed, though. HYPE burns about 23 thousand tokens a day, and new HYPE is still being released from the tank:

  • Staking rewards come from a reserve of future tokens. In Hyperliquid’s own example, with 400 million HYPE staked, rewards are about 2.37% a year. That’s roughly 26 thousand HYPE a day, about as much as the burn.
  • Core contributor tokens, 23.8% of the supply, were locked for a year after the launch on 29 November 2024. They’re being released over the next few years; the Hyper Foundation says most vesting schedules finish between 2027 and 2028.
A bathtub labelled 'in circulation'. The tap pours new HYPE from the 1 billion tank, fed by staking rewards (about 26K a day in the docs' example) and core contributor tokens (vesting, mostly until 2027–2028). The drain burns about 23K a day.
HYPE: about 23 thousand burned a day, while staking rewards and vesting contributor tokens flow into circulation.Hyperliquid publishes no official daily or monthly unlock amount, so we don’t give one.

A burn alone can’t tell you if the supply is shrinking. Always look at both sides: how much is burned, and how much is created.

How big is a burn?

Big numbers can be misleading. One billion tokens sounds like a lot. From a supply of 500 trillion, it’s just 0.0002%.

A full mint pie labelled 'supply: 500 trillion'. The slice for 1 billion burned is too thin to see. Beside it: 0.0002%.
One billion tokens out of a supply of 500 trillion: 0.0002%, too thin to see.

That’s why you compare a burn with the supply. The result is the share of the tokens that’s gone for good, and you can compare it across any token, however big its numbers are.

Share burnedtokens burned ÷ supply × 100BurnTracker divides by each asset’s all-time peak supply, explained below.

But which supply? Today’s supply has already shrunk from the burns, so it makes a burn look bigger. Take Shiba Inu: its 410.76 trillion SHIB burned is about 70% of today’s supply of 589.50 trillion, but only 41% of all the SHIB that ever existed, 1,000 trillion.

The formula burned ÷ supply × 100. Below it, the 410T SHIB burned compared two ways: 70% of today's supply (589 trillion SHIB) and 41% of the most that ever existed (1,000 trillion SHIB), with a check mark on the second.
The same SHIB burn: about 70% of today’s supply, 41% of the most that ever existed.Today’s supply: CoinGecko total supply in BurnTracker’s snapshot, 25 September 2026.

That’s why BurnTracker uses each asset’s all-time peak supply: the most tokens that ever existed at once. For a token with a fixed supply, that’s the amount created at launch. For a changing supply, it’s the highest supply on record. It doesn’t shrink when tokens burn, so every asset is measured the same way.

Here is how the tracked assets compare, as of 25 September 2026. At the top is Shiba Inu, with about 41% burned, then TRON with about 18%, LEO with 8% and Luna Classic with nearly 7%. Hyperliquid is at about 4.8%, Ethereum at 3.8%, Internet Computer at half a percent, and XRP at just 0.01%.

Horizontal bars of the share of all-time peak supply burned: Shiba Inu 41.08%, TRON 17.56%, UNUS SED LEO 8.01%, Luna Classic 6.65%, Hyperliquid 4.78%, Ethereum 3.80%, Internet Computer 0.53%, XRP 0.0144%.
Share of all-time peak supply burned, every BurnTracker asset, 25 September 2026.

Same idea, very different impact. But a percentage isn’t the whole story either. In dollars, at today’s prices, the ranking changes: Ethereum’s burned ETH is worth about $12.5 billion, more than five times Shiba Inu’s $2.4 billion. And almost all of Shiba Inu’s 41% comes from that one transaction by Vitalik Buterin, not from steady burning.

Horizontal gold bars of burned value at today's prices: Ethereum $12.53B, TRON $6.04B, Hyperliquid $4.48B, Shiba Inu $2.41B (highlighted), UNUS SED LEO $705.25M, Luna Classic $25.38M, XRP $22.45M, Internet Computer $9.37M.
Burned tokens valued at today’s prices.Prices: CoinGecko, in BurnTracker’s snapshot of 25 September 2026.
BurnTracker, as of 25 September 2026. Dollar values multiply the burned tokens by the latest price; they aren’t money spent on burning.
AssetShare of all-time peak supply burnedBurned value at today’s price
Shiba Inu (SHIB)41.08%$2.41B
TRON (TRX)17.56%$6.04B
UNUS SED LEO (LEO)8.01%$705.25M
Luna Classic (LUNC)6.65%$25.38M
Hyperliquid (HYPE)4.78%$4.48B
Ethereum (ETH)3.80%$12.53B
Internet Computer (ICP)0.53%$9.37M
XRP0.0144%$22.45M

What a burn can’t tell you

There is one thing a burn total can’t tell you. It tells you how many tokens are gone. It can’t tell you what the remaining tokens will be worth.

Heading: How many are gone ≠ what the rest are worth. Left: tokens burned, a count. Right: what the rest are worth, the market decides, with tags for buyers, sellers and tokens available to trade.
A burn total is a count. What the remaining tokens are worth is up to the market.

Price also depends on buyers, sellers, and how many tokens are available to trade. A burn is one input, not a forecast. For more on this, read What a token burn actually changes.

Check it yourself

You don’t have to take anyone’s word for a burn, including ours. Every burn leaves a record on the blockchain that anyone can look up. Here is the Shiba Inu burn from May 2021, as BurnTracker records it.

A BurnTracker burn record card: 16 May 2021, 21:48 UTC; amount 410,241,996,771,871 SHIB; from Vitalik Buterin 0xab58…aec9b; to the dead wallet 0xdead0000…06942069; transaction 0x125714bb…2984f905. Beside it, three ticked checks: the transaction, the amount and token, the destination.
BurnTracker’s record of the 16 May 2021 SHIB burn, and the three things to check.
  1. The transaction. Every transaction has a unique hash, so you can look it up on a block explorer. This one is 0x125714bb…2984f905 on Etherscan.
  2. The amount and the token. Check that the tokens moved are the real token, from its official contract (for SHIB, 0x95ad61b0…c4ce), not a copy with the same name. And mind the decimal places: raw data often counts in a token’s smallest units.
  3. The destination. It should be an address no one is known to have a key for, so nobody can move the tokens out again. Here it’s the dead wallet 0xdead0000…06942069.

Every total on BurnTracker lists the records behind it, and the methodology explains exactly what is counted. The SHIB records are on the SHIB tracker.

Bottom line

A token burn removes tokens for good. But what it means depends on how much is burned, how much is created, and how big the supply is. Check all three before you read anything into a burn total.

Track every burn, with its sources, at cryptoburntracker.com.

Informational only, not financial advice.

Sources

All figures were read on 25 September 2026. Tracker totals change every few minutes; each figure is given “as of” that date.

Corrections and updatesFirst published 25 September 2026. No corrections recorded.