How many bitcoins are there? The 21 million cap, explained

The 21 million cap is a rule every node enforces, not a promise

New bitcoin comes into existence in exactly one way. Each block, the miner who builds it may include a special first transaction that pays a reward, called the block subsidy, to an address of their choosing. The size of that reward is fixed by a short function in Bitcoin Core, the reference software most of the network runs. The function, GetBlockSubsidy, divides the block height by 210,000 to work out how many halvings have happened, starts from 50 BTC, and cuts that in half once for each halving. A comment in the code says it plainly: the subsidy is cut in half every 210,000 blocks, which will occur approximately every four years.

That is the whole mechanism. There is no committee, no annual mint report, and no way for a miner to award themselves more. Every node, meaning every computer running the software and checking blocks for itself, recomputes the permitted subsidy and rejects any block whose reward transaction claims more than the subsidy plus the fees paid by the transactions inside it. Our dossier on halvings and supply reproduces the function line by line and lists the exact boundary blocks.

Because the reward halves at a block height rather than on a calendar date, the familiar “every four years” is an outcome of the ten-minute block target, not a rule. The halving arrives when block 210,000, 420,000, 630,000 and so on is mined, whatever the date happens to be. So far those dates have been 28 November 2012, 9 July 2016, 11 May 2020 and 20 April 2024.

Why the true maximum is 20,999,999.9769, not 21 million

Add up the schedule and you get a familiar geometric series: 10.5 million coins in the first 210,000 blocks, 5.25 million in the next, then 2.625 million, and so on, approaching 21 million without ever quite reaching it. That is where the round number comes from. The code, though, does not deal in fractions of a coin. It counts in satoshis, the smallest unit, of which there are 100 million in a bitcoin, and it halves by shifting an integer one bit to the right. Once the subsidy becomes an odd number of satoshis, the shift throws away the leftover half.

Summed under the current function, the most that can ever be created is 2,099,999,997,690,000 satoshis, or 20,999,999.9769 BTC. The Bitcoin Wiki gives the same figure and notes that the shortfall is a consequence of the integer storage type, not a design goal. The difference is about two hundredths of a coin. It matters only when someone asks for precision, but it is a useful reminder that “21 million” describes a calculation rather than a number typed into the code.

One more guard sits in the function. Once the halving count reaches 64, the subsidy is forced to zero. BIP 42, a proposal by Pieter Wuille dated 1 April 2014, explained that the original code relied on undefined behavior in C++ for very large shifts, and that on the hardware people actually use the reward would have wrapped around and started again after 64 halvings. The fix makes the schedule genuinely finite, centuries before it could have mattered.

How many have been issued so far

The schedule tells you the most that could have been created by any block height. Through block 839,999, just before the fourth halving block on 20 April 2024, the schedule allowed 19,687,500 BTC. Block 840,000 itself adds a maximum subsidy of 3.125 BTC. The pre-halving total is 93.75 percent of the rounded 21 million cap. By the end of the era that begins at block 1,050,000 the figure reaches 20,671,875 BTC, more than 98 percent. Almost all of the supply, in other words, was created in Bitcoin’s first fifteen years, and the remaining sliver trickles out over more than a century.

Actual issuance runs a hair below the schedule, because a miner may claim less than the permitted reward, and some have. Coin Metrics, the data firm whose figures this site uses for price and supply, put the outstanding supply at about 19.69 million BTC on the day of the fourth halving and about 19.97 million BTC on 31 December 2025. That halving cut the reward to 3.125 BTC per block, which is about 450 new coins a day at the target pace of 144 blocks, down from 7,200 a day in 2009.

You can watch the total grow yourself: the reward transaction in every block is public, and our guide to reading a block explorer shows where to find it. What no explorer can tell you is how many of those coins anyone can still spend, which is the last question below.

The final nonzero subsidy is one whole satoshi per block

Because each era pays half the previous one, the subsidy keeps halving until it is a single satoshi per block, and then, at the 33rd halving, it becomes zero. That boundary is block 6,930,000. At ten minutes per block, the Bitcoin Wiki projects the last new satoshi somewhere in the year 2140; its estimates land on different days of that year depending on assumptions about mining pace, which is the honest way of saying nobody can name the date. What is fixed is the height.

After that, miners are paid only by transaction fees, which the white paper anticipated from the start. Whether fees alone will buy enough security is a real question that the code does not answer, and our dossier treats it as an open economic one rather than a missing line of code. For readers today it is mostly a curiosity. The era of meaningful new issuance is the one we are living through, and it is already mostly over.

Issued, lost and liquid are three different numbers

The maximum is a rule. The issued total is a fact you can read from the chain. From there the numbers get softer. Some coins exist on the ledger but can never move. The clearest case is the 50 BTC reward in the genesis block: the original software hard-coded that block but never entered its reward into the database of spendable coins, so no node will accept a transaction that spends it. Some scripts, such as outputs beginning with OP_RETURN, are provably unspendable. Addresses deliberately constructed without a known private key rely instead on the computational difficulty of finding a key; that is a different basis from the genesis database rule.

Then there are coins whose keys have simply been lost, on discarded hard drives, behind forgotten passwords, in wallets whose owners died without leaving instructions. The ledger cannot distinguish a lost coin from a patiently held one, so every figure for lost bitcoin is an estimate built on assumptions about behavior. Subtract the estimate from the issued total and you get something like the spendable supply.

“Liquid” supply narrows further still, to coins that actually trade rather than sit in long-term storage, and it depends even more on definitions. So when someone quotes a single number for how many bitcoins there are, it is worth asking which of the four they mean. The cap is just under 21 million. The number issued at a given height is below that maximum. The number anyone can spend is smaller, and the number that changes hands is smaller again.