Circulating supply

The chain does not know what is circulating

A blockchain can tell you exactly how many tokens exist and exactly which address holds each one. It cannot tell you whether an address is a treasury, a burn address nobody registered, a founder's cold wallet under a two-year contractual lockup, a bridge escrow, or an exchange omnibus. Circulating supply requires classifying every one of those, and that classification is a human decision applied from outside the protocol.

CoinMarketCap says this in its own glossary rather than burying it: circulating supply is the best approximation of the number of coins circulating in the market and in the general public's hands, and the network at large has no reliable knowledge of how much of the total supply is in active circulation, making the metric an imperfect approximation.¹

Once you accept that, the follow-on questions change. Not what is the circulating supply, but who computed it, from what rules, and against what evidence the project provided.

Bitcoin's number is an estimate too

The cleanest illustration comes from the asset with the most transparent supply schedule in existence. CoinMarketCap's worked example notes that Bitcoin's nominal circulating supply is the amount mined since inception, over 18 million coins in that example, while an estimated 4 million BTC have been permanently lost, placing the true circulating supply closer to 14 million.¹

The mined total keeps climbing with every block. The lost portion never appears in any published supply figure, because there is no way to prove a private key is gone. So the number everyone quotes, including the one that feeds Bitcoin's market cap, carries an unstated overcount of roughly a fifth by the reporting provider's own estimate.

If that is the state of the most auditable supply in the category, treat every altcoin circulating supply as an approximation with wider error bars, not as a figure to build a comparison table on.

Where the calls actually get made

Tokens locked in a smart contract are counted in total supply and excluded from circulating supply. CoinMarketCap gives ICO vesting stages as its example of tokens locked until a particular purpose is achieved.² That case is straightforward because the lock is onchain and anyone can read it.

The difficult cases are the ones with no contract to read. A foundation treasury sitting in a multisig with a published policy but no code enforcing it. A team allocation subject to a lockup written into an employment agreement. Tokens bridged to a chain the provider's indexer does not cover. Market maker inventory on loan. Each is a call, and a project that has not documented its own position will find the call made without it.

The design consequence is that circulating supply is something you can specify rather than something that happens to you. Publish the address list, the lock contracts, the policy for any wallet not under code control, and the schedule that moves tokens between categories. Providers generally follow a project that gives them something verifiable to follow.

Agreement is not the same as accuracy

Reading Ethena from both major providers on 3 August 2026, the supply figures matched exactly: 9,560,937,500 circulating against a total and max supply of 15,000,000,000, a float of 63.74%.³ CoinMarketCap showed the same 9.56 billion against 15 billion on the same day.⁴

That is what well documented supply looks like, and the contrast is sharp: the same pair of reads produced a roughly 6% disagreement on TVL for that protocol. When a project publishes clean, verifiable supply accounting, the providers converge. When it does not, they diverge, and the project inherits whatever they each decided.

It also does not mean the number is correct. Two providers agreeing tells you they applied compatible rules to the same evidence. It does not tell you the rules match economic reality, which is the separate question of how much of that 9.56 billion could realistically be sold this quarter.

Why this figure is commercially load-bearing

Circulating supply sets the market cap denominator, and market cap governs index inclusion, exchange listing tiers, screener rankings and the comparables tables in every investor deck in the category. A supply classification decision made by a data team you have never spoken to can move your token between visibility brackets.

It also sets the reference point for every unlock announcement. If your published float is understated, each unlock looks proportionally larger than it is and reads as a bigger supply shock. If it is overstated, market cap flatters the project now and every future unlock lands with less warning than it should have had.

Neither of those is a market problem. Both are documentation problems, which means both are fixable before launch at close to zero cost.

What we tell founders to publish

One supply page, maintained, with five things on it. Every address holding project-controlled tokens, labelled. The lock contract for each, with the address so anyone can verify it. A written policy for any holding not under code control, because that is the category providers guess at. The unlock schedule with dates and amounts. And a plain statement of which holdings you consider outside circulation and why.

Then hold that definition still. Changing your own circulating supply methodology mid-cycle produces a step change in reported market cap that looks like something happened in the market when nothing did, and it costs credibility with the people you least want to spend it on.