Bitcoin's Supply Schedule: Halving, Block Reward and the 21 Million Cap

Bitcoin is created by miners who find a valid block, and only there. That single sentence carries almost the whole supply model: there is no issuer who can add coins outside the schedule, and any block claiming more than the rules allow is rejected by the network as worthless.

The rules themselves are documented on the Bitcoin Wiki’s controlled supply page and its block page, both consulted on 1 November 2026.

The halving

The number of bitcoin generated per block decreases geometrically, halving every 210,000 blocks — approximately four years. The reward began at 50 and steps down from there.

Because each era pays half of the previous one, the total issued is the sum of a geometric series: 210,000 blocks at 50, then 210,000 at 25, then 210,000 at 12.5, and so on. A series like that converges on a finite total rather than growing without bound — and the wiki’s phrasing is precise about the result: the number of bitcoin in existence will not exceed slightly less than 21 million.

“Slightly less than” is doing real work. It is the difference between a cap and an asymptote.

The 21 million number

The wiki is unusually candid here. It lists two speculative justifications — that the figure falls out of a four-year halving schedule, and that the eventual total number of satoshis is close to the maximum capacity of a 64-bit floating point number — and then states that Satoshi never really justified or explained many of these constants.

That is worth sitting with. The most quoted number in the asset’s history is a design choice with reconstructions after the fact, not a documented decision.

What the wiki does say about intent is that the decreasing-supply algorithm was chosen because it approximates the rate at which commodities like gold are mined. That is a design description, not a monetary argument.

Two clocks that get confused

Bitcoin has two separate adjustment rules, and the halving coverage in most articles merges them.

The reward rule changes every 210,000 blocks: how much new bitcoin a block pays.

The difficulty rule changes every 2,016 blocks: how hard it is to find a block, retargeted so that blocks keep arriving at roughly one per ten minutes, or six per hour. That mechanism is covered separately under difficulty.

Keeping them apart matters. The halving does not make blocks slower. It makes each block worth less in new coins, while the difficulty rule keeps working to hold the spacing steady regardless of how much mining power joins or leaves.

Where supply falls short of the maximum

The theoretical total assumes every coin created is spendable forever. It is not, and the wiki records several distinct ways the real number falls short.

Underpay by the miner. A miner can deliberately pay himself less than the full reward plus fees. The wiki records block 124724, where a solo miner created one satoshi less than he was entitled to — an amount that can never be created afterwards, because the schedule does not go back and top up. It also records block 501726, where a miner destroyed an entire 12.5 BTC block reward through a failure to implement correctly.

Provably unspendable outputs. An output can be made unspendable with OP_RETURN, and the wiki notes that other forms of destruction are not officially recognised as destroying bitcoin — a nuance worth knowing before repeating claims about “burned coins”.

Lost coins. Keys that no longer exist. This is not destruction in a cryptographic sense, but it removes supply from circulation just the same.

Technical peculiarities. Some outputs cannot be spent for structural reasons, including the genesis block’s reward.

The distinction the wiki draws between provable and merely presumed destruction is the useful one: “coins are gone” is a claim about keys, and no one can verify a key that does not exist.

Why the end of issuance is not a cliff

As the reward shrinks, fees become a larger share of what a miner earns. The wiki’s discussion of controlled supply is explicit that bitcoins are created each time a block is found, and the block page adds the incentive structure: peers collect transactions and add them to the block they are working on because of the attached fees.

The consequence is gradual. Long before the reward reaches zero, it is a rounding error next to fee revenue in any period of congestion — and during quiet periods, the reverse is true, which is the part usually left out of long-term security arguments.

What the schedule does and does not establish

It establishes three things with unusual clarity:

• No discretionary issuance. No committee, foundation or miner can vote new supply into existence outside the schedule, because other nodes reject blocks that break the rule.
• A published, auditable rate. The reward for any block height is computable by anyone, and the cumulative supply is verifiable from the chain.
• A known end state. Issuance decays toward zero rather than continuing at a fixed rate.

It does not establish that the price rises, that scarcity translates into demand, or that a fixed schedule is the right monetary policy. The wiki itself carries the argument about deflation, noting the Keynesian case against it and the Austrian response, which is a reminder that the supply rule is a choice with a literature on both sides rather than a settled question.

For anyone holding bitcoin, the practical reading is narrow and useful: the issuance you can verify is the schedule and the chain. Everything else — what the market does with a shrinking float, how miners behave as the subsidy fades — is an inference, and inferences should be labelled as such.