Learn

Bitcoin's supply schedule

Circulating supply approaching 21 million

Protocol supply on January 1 of each year, computed from the block subsidy schedule.

Circulating supply (BTC)21M capEffective supply, est. (minus 2.78M–3.79M lost)Halvings (est. = estimated)

Effective supply is an estimate: circulating supply minus Chainalysis's 2017 range of likely lost coins, held constant from 2018 (later losses are not included). Today: about 19.97M at the start of 2026; block 971,632 on 2026-10-15.

The schedule, from the source code

GetBlockSubsidy starts at 50 BTC and shifts the subsidy right by one bit for every 210,000 blocks (nSubsidyHalvingInterval). After 64 halvings it returns zero. Summed in satoshis, that totals just under 21 million BTC.
Past halvings, verified on mempool.space: block 210,000 on Nov 28, 2012 (50 → 25 BTC), 420,000 on Jul 9, 2016 (25 → 12.5), 630,000 on May 11, 2020 (12.5 → 6.25) and 840,000 on Apr 20, 2024 (6.25 → 3.125).
Future halving dates on these charts are estimates at 10 minutes per block from the October 15, 2026 tip, labeled est. Real blocks arrive faster or slower, so the dates will drift.
Dates between known blocks are interpolated, so yearly figures are close approximations rather than exact on-chain counts.

Annual new-supply rate falls with each halving

New coins issued in a year ÷ supply at the start of that year.

BTC new-supply rateUSD M2 growth, yoyUS CPI, Dec/DecHalvings (est. = estimated)

Log scale. Years with zero or negative M2/CPI change cannot be drawn on a log axis and are left blank. 2026 M2 uses Aug/Aug.

Disinflation, not negative issuance

In 2025 the protocol added about 165k BTC, roughly 0.83% of the supply at the start of the year. Each halving roughly halves that rate. It approaches zero but never goes below it: the protocol never destroys coins. For comparison, the chart shows US M2 money supply growth (FRED M2SL) and CPI inflation (FRED CPIAUCSL). M2 and CPI measure different things, and neither is a forecast for bitcoin.

Lost coins and effective supply

Coins whose keys are gone still count in the protocol supply but can never move. In 2017 Chainalysis estimated that 2.78 million to 3.79 million BTC were lost, about 17% to 23% of coins mined at the time; both figures assume Satoshi's coins are gone (Fortune, Nov 25, 2017). We found no other firm's figure we could verify at the original source, so this is the only range used. The “effective supply” band above subtracts it, and is an estimate.

New issuance vs. coins lost per year

Where the orange line drops below a loss scenario, estimated losses exceed new coins and effective supply shrinks in that scenario.

New BTC issued per yearHigh loss: 426.2k/yrMiddle loss: 312.6k/yrLow (illustrative) loss: 31.3k/yrHalvings (est. = estimated)

High

Issuance falls below 426.2k/yr in 2021

Middle

Issuance falls below 312.6k/yr in 2024

Low (illustrative)

Issuance falls below 31.3k/yr in 2036

The loss-rate math

Years from the genesis block (Jan 3, 2009) to the estimate (Nov 25, 2017): 8.89.
High: 3,790,000 ÷ 8.89 ≈ 426k BTC per year. Issuance fell below this in 2021.
Middle: 2,780,000 ÷ 8.89 ≈ 313k BTC per year. Issuance fell below this in 2024.
Low (our assumption): 10% of the middle rate, ≈ 31k BTC per year. Issuance falls below this around 2036 (estimated).

Caveat. Most losses likely happened early, when coins were nearly worthless and wallets, backups and custody tools were immature. A simple average over 2009–2017 therefore probably overstates today's loss rate, which is why we add a much lower scenario. The 10% figure is our illustrative assumption, not a published estimate. No one knows the current rate, and some “lost” coins may move again. This is a model built on estimates, not a prediction, and not investment advice.

So the crossover, where estimated yearly losses overtake new issuance and effective supply starts shrinking, lands anywhere from 2021 to about 2036 depending on the scenario. Good self-custody, with tested backups and an inheritance plan, keeps your own coins out of that statistic.

Sources