The supply schedule
Twenty-one million, issued on a timetable written in 2009 and unchanged since. It is the most-cited fact about Bitcoin and the least examined.
New bitcoin is created only as a reward to whoever produces a block, and the size of that reward is fixed by a schedule with no discretion in it anywhere.
The arithmetic
The subsidy began at 50 coins per block and halves every 210,000 blocks — roughly every four years at ten-minute blocks. Total issuance is therefore 210,000 × (50 + 25 + 12.5 + …), a geometric series that converges to just under 21 million. The cap is not a separate rule someone chose; it is the sum of the halving schedule.
The halvings have occurred in 2012, 2016, 2020 and April 2024. Since the 2024 halving the subsidy has been 3.125 coins per block. It continues halving until the reward rounds below the smallest representable unit, at which point issuance stops entirely — projected to occur early in the twenty-second century.
What is genuinely unusual about it
Gold is scarce, and its supply still responds to price: higher prices fund more exploration and more mining. Here, higher prices fund more hashing power, difficulty adjusts, and issuance is unchanged. There is no authority that can decide otherwise, no emergency provision, and no process for amending it short of persuading the whole network to run different software. Whether a fixed supply is desirable is a separate argument; the interesting part is that the commitment is credible in a way a central bank's cannot be.
The objection, stated fairly
Monetary economists have a well-developed criticism: a currency whose supply cannot expand transmits every shock into its price, and a money that reliably appreciates gives holders a reason to defer spending. Both effects are real in theory and both are contested in magnitude.
The counter-argument is that the fixed schedule is not trying to be a functioning national currency — it is trying to be an asset nobody can dilute. Those are different objectives and are frequently argued past one another.
Halvings and price
Any account of why a halving should move the price has to explain why a completely anticipated, precisely scheduled change in issuance is not already reflected in the price. That is not a knock-down argument — markets do respond to known events — but it is the question any halving thesis must answer, and it is usually skipped. Past cycles are four data points, which is not a sample.
The security budget problem
Miners are paid from the subsidy plus transaction fees. As the subsidy falls toward zero, fees must fund the entire cost of securing the network.
Whether transaction fees on a chain with limited block space can reach the level required is a genuinely open technical question, debated seriously by people who agree on almost everything else. It is the most substantive unresolved issue in Bitcoin's design, and it resolves slowly — over decades, on a schedule everyone can read. See mining economics.
Lost coins
Coins at addresses whose keys no longer exist are permanently immobile. They still count toward the 21 million, so the effective circulating supply is lower than the issued supply by an amount that can be estimated from on-chain dormancy but never known exactly. See on-chain analysis.