Bitcoin's 21 Million Supply Cap: What It Means and How It Works

Bitcoin's protocol caps total issuance at 21 million BTC, enforced by a halving subsidy schedule in Bitcoin Core's GetBlockSubsidy function (source: GitHub). As of April 2026 roughly 19.95 million BTC have been issued (~95% of the cap), leaving ~1.05 million for ~115 more years of decaying issuance. Fixed supply is a verifiable monetary property; it is not a guarantee of price.

Bitcoin's 21M supply cap at a glance, 2026 specifics:

1. The 21M cap emerges from the geometric series 50 × 210,000 / (1 − 0.5) = 21,000,000; the actual asymptotic limit is 20,999,999.9769 BTC because integer-satoshi truncation in the right-shift subsidy halving rounds down each era (source: Bitcoin Wiki)

2. Current era (5th) pays 3.125 BTC per block since the April 2024 halving at block 840,000; ~450 BTC issued per day, ~164,000 per year; next halving to 1.5625 BTC expected around April 2028 at block 1,050,000

3. As of April 2026 ~19.95M-20.01M BTC mined (~95% of cap); ~1.05M remaining; the last fractional satoshi is projected around block 6,930,000 in roughly the year 2140 (Bitcoin Wiki: May 7, 2140 under constant-hashrate assumption from block 367,500, but the projection is probabilistic, not a fixed date)

4. Estimated lost coins range from ~3-4 million BTC (early miner errors, key loss, dormant wallets); accessible supply is meaningfully below 21M though the protocol cap is unaffected

5. Bitcoin's stock-to-flow ratio sits near 121 in 2026 versus ~62-70 for gold (for the full comparison, see Bitcoin vs gold); the cap is a protocol fact not a price model; Bitcoin has experienced 70-85%+ drawdowns from prior peaks while remaining scarce throughout

This guide explains the 21M cap honestly: where the number comes from (geometric series, not the whitepaper), how Bitcoin Core's GetBlockSubsidy function enforces it, the full halving schedule with current 3.125 BTC era, the asymptotic 20,999,999.9769 BTC limit, current 2026 issuance and remaining supply, lost-coin estimates and accessible-versus-protocol supply, why the cap could in principle change but never will in practice, the gold and fiat comparisons that hold up, the stock-to-flow concept and its limits as a price model, and what happens after ~2140 when miners earn fees only. For protocol context, see what Bitcoin is and how Bitcoin mining works at the system level.

Where the 21 million number actually comes from

The 21 million cap is not stated as a single number in the Bitcoin whitepaper. It emerges from three protocol parameters: the original 50 BTC block subsidy, the 210,000-block halving interval, and the geometric series this creates as each era cuts the per-block reward in half.

The arithmetic: each halving era produces 210,000 blocks at a fixed subsidy. The first era pays 50 × 210,000 = 10,500,000 BTC. The second era at 25 BTC pays 5,250,000 BTC. Each subsequent era contributes half the prior era's subsidy. The infinite-series sum is:

50 × 210,000 × (1 + 1/2 + 1/4 + 1/8 + ...) = 50 × 210,000 × 2 = 21,000,000 BTC

That is where 21 million comes from. The number is mathematically derived from three engineering choices, not a top-down economic mandate.

Why those three choices? Satoshi never explained directly 

Early correspondence at the Nakamoto Institute (source: Nakamoto Institute) suggests the round numbers were aesthetic as much as economic. The theory that 21M was chosen to roughly match global M1 money supply in 2009 is plausible but unsourced from Satoshi; the 50 BTC starting subsidy and 210,000-block halving cadence appear in the genesis code without economic justification in surrounding messages. 21M is exact arithmetic; the choice of inputs reads as a "predictable, finite, round" design preference rather than a calibrated monetary target.

Whitepaper Section 6 does not name 21M. It makes the broader claim that "the steady addition of a constant amount of new coins is analogous to gold miners expending resources to add gold to circulation. In our case, it is CPU time and electricity that is expended" and notes issuance can be "completely inflation free" once it ceases (source: Bitcoin.org). The specific 21M figure lives in the code, not the paper.

For what problem Bitcoin was created to address, the fixed-supply choice was an intentional response to discretionary fiat issuance.

The full halving schedule and the current 3.125 BTC era

Bitcoin's block subsidy halves every 210,000 blocks. The schedule below tracks completed and projected eras through the point where integer arithmetic forces the subsidy to zero.

Source: Bitcoin Wiki Controlled Supply Bitcoin Wiki; learnmeabitcoin halving page (source: Learn Me a Bitcoin).

The geometric-series sum across all eras equals not exactly 21,000,000 but 20,999,999.9769 BTC because Bitcoin Core's GetBlockSubsidy uses an integer right-shift to halve the satoshi count each era; any halving producing a fractional satoshi rounds down. After roughly 33 halvings the subsidy rounds to zero, and no further bitcoin is created regardless of subsequent blocks mined.

The 5th era began at the April 2024 halving (block 840,000) when the subsidy dropped from 6.25 BTC to 3.125 BTC. Daily issuance is ~144 blocks × 3.125 = ~450 BTC, or ~164,000 BTC per year. The next halving to 1.5625 BTC is projected for April 2028 at block 1,050,000. For how each halving event reshapes miner economics, the protocol mechanic is simple; the market consequences are where complexity lives.

How Bitcoin Core enforces the cap: GetBlockSubsidy in software

Knowing Bitcoin has a 21M cap is useful. Knowing how it is enforced explains why the cap is credible.

New bitcoin enters circulation through the coinbase transaction at the top of every valid block, the only place in the protocol where new coins are minted. Miners specify a coinbase output paying themselves the subsidy plus transaction fees.

The maximum subsidy is computed by the GetBlockSubsidy function in Bitcoin Core's consensus code GitHub. The function takes block height and consensus parameters and returns the allowed subsidy in satoshis. The core logic is a right-shift that halves the initial 50 BTC (5,000,000,000 satoshis) every 210,000 blocks:

int halvings = nHeight / consensusParams.nSubsidyHalvingInterval; CAmount nSubsidy = 50 * COIN; nSubsidy >>= halvings; return nSubsidy;

In practice the subsidy rounds to zero satoshis around the 33rd halving because each integer right-shift halves the satoshi count.

Distributed enforcement. Every full node runs this same function on every received block. If a coinbase transaction claims more than GetBlockSubsidy(height) + sum_of_fees, every validating node rejects the block as invalid regardless of proof-of-work investment behind it. Bitcoin Core (source: Bitcoin Core) and compatible alternative implementations enforce the same subsidy ceiling, which is why no single node operator or developer team can change the cap unilaterally.

For how UTXO outputs become spendable after coinbase issuance, spend mechanics differ from issuance: spending is signature-constrained, issuance is consensus-rule constrained.

Current state April 2026: 95% mined, ~1.05M remaining

The supply curve is far through its lifetime issuance and now in a long shallow tail. As of April 2026, current trackers put the mined total in a tight range:

Source: iTrustCapital (source: Itrustcapital); Bleap Finance Bitcoin supply tracker (source: Bleap).

Remaining issuance is heavily front-loaded. The 5th era plus 6th era together issue ~984,375 BTC, close to 94% of the remaining ~1.05M; the other ~6% spreads across ~110 years of ever-smaller subsidies. By the early 2030s issuance becomes a rounding effect on total supply.

For how Bitcoin's difficulty adjustment keeps blocks near ten minutes, the "around 2140" projection depends on ten-minute averages; early eras ran slightly faster, shaving the projection.

Why the final supply is 20,999,999.9769 BTC, not exactly 21M

The literal protocol maximum is slightly lower than the 21M shorthand: 20,999,999.9769 BTC, or about 20.99999998 million.

The shortfall is a software artifact. Bitcoin tracks satoshi balances as 64-bit integers (1 BTC = 100,000,000 satoshis). The GetBlockSubsidy halving uses the bitwise right-shift operator >>, which divides by powers of two with integer truncation. Each halving cuts the satoshi count in half; any odd-satoshi remainder is dropped.

The 32nd era pays 1 satoshi per block for 210,000 blocks (2,100 BTC contribution); after that the right-shift produces zero forever. Summing the truncated geometric series yields 20,999,999.9769 BTC as the asymptotic ceiling. The 0.0231 BTC shortfall is rounding error at network scale, but for technical correctness "21 million" is shorthand and "20,999,999.9769" is the literal number.

When will the last satoshi be mined? The "around 2140" estimate

The most-cited answer is "around 2140." Approximately correct, not a calendar date.

The Bitcoin Wiki controlled-supply page projects the final coin around block 6,930,000, with the date pinpointed at May 7, 2140 under one specific assumption: that average network hashrate stays constant at the level observed at block 367,500 (mid-2015). Under any other assumption the date shifts; hashrate has grown several thousand-fold since 2015, and average block intervals run consistently below the ten-minute target during high-hashrate eras.

Why "around 2140" is approximate:

• Halvings trigger by block height, not calendar time. The protocol counts blocks, not years.

• Difficulty adjusts every 2,016 blocks to target ten-minute intervals, but real intervals are Poisson-distributed and often run slightly faster than target.

• Over a multi-decade horizon, small deviations compound. A consistent 9.7-minute average shaves several years off the 2140 projection.

• Final eras issue single satoshis; the terminal date depends on whether anyone is even mining at marginal economics that thin.

The honest answer: the subsidy rounds to zero somewhere in the 2130s-2150s window. "Around 2140" is shorthand; any specific day in a specific year overstates model precision.

The lost-coin question: Accessible supply is below 21M

The protocol cap is 21M BTC. Accessible supply is meaningfully lower, by an estimated 3-4 million BTC, though the figure is inherently uncertain because no on-chain marker distinguishes a lost key from a long-dormant one.

All are estimates with documented assumptions, not verified counts. Chainalysis published the original 2017 report; Glassnode publishes ongoing dormancy heatmaps.

What lost coins do and do not change

Functional supply (coins current holders can actually move) is roughly 16-18M BTC depending on which estimate one accepts. For protocol-monetary-policy claims, 21M is correct. For market-availability claims, accessible supply is the right number.

For how the UTXO model tracks unspent outputs across the chain, outputs older than ~10 years are on-chain but indistinguishable between "lost" and "long-term HODL."

Could the 21M cap ever change? The political reality

In principle yes; in practice almost certainly never.

The 21M ceiling is a consensus rule in GetBlockSubsidy. A change requires a hard fork: nodes running modified software with a higher cap accept blocks that current-software nodes reject. The chain splits. Whichever version retains majority economic activity (exchanges, custodians, large holders, miners) becomes "Bitcoin"; the other is a separate asset.

Why a cap-raising hard fork has effectively zero chance of succeeding:

The cap is credibly committed not because the code is immutable (it is mutable) but because no coalition has both the means and the interest to coordinate a value-destroying change.

For how Bitcoin nodes, miners, and wallets divide enforcement, no single layer can change the cap; consensus requires alignment across layers with no coordinating mechanism for value-destroying changes.

Bitcoin vs fiat: Verifiable scarcity in numbers

Bitcoin's issuance is enforced by software and distributed consensus; fiat supply is governed by central-bank policy. Both are valid governance models; the comparison shows what fixed-supply means in practice.

US M2 versus Bitcoin supply growth, Feb 2020 to Feb 2026:

Source: Federal Reserve H.6 release (source: Federal Reserve) and FRED M2SL (source: FRED); Bitcoin supply from Blockchain.com and Newhedge.

The 2020-21 window was the most extreme: US M2 expanded ~40% in ~18 months during pandemic monetary response, the fastest peacetime expansion on record. Bitcoin supply over the same window grew ~5% as the geometric series proceeded on schedule. The 2022-23 QT period brought the first M2 contraction since the 1930s, but cumulative six-year growth remains several times Bitcoin's.

The structural difference is verifiability. Anyone can read GetBlockSubsidy and project issuance at any future block height with arithmetic certainty. M2 expansion or contraction depends on FOMC decisions not yet made. Central-bank discretion has genuine countercyclical uses; this is a description of two governance models, not an endorsement.

For Bitcoin as an inflation hedge or just another risk asset, supply discipline is half the story; demand response to monetary expansion is the other half and is far less predictable.

Bitcoin vs gold: Stock-to-flow as a description of scarcity

Gold has long been the reference scarcity asset. Stock-to-flow (S2F) is the standard commodity scarcity ratio: total existing supply divided by annual new production.

Source: Bitbo S2F live chart (source: Bitbo); Swissgoldsafe.ch S2F fundamentals (source: Swiss Gold Safe).

Bitcoin's S2F sits ~1.7-2x higher than gold's in 2026 and rises further with every halving. As a descriptive scarcity metric, this is meaningful: Bitcoin's annual supply growth (~0.82%) is below gold's ~1.5% mine output relative to above-ground stock.

As a price-prediction model, S2F is much weaker. Plan B's 2019-2020 model implied BTC price targets ($100K by late 2021, $288K by 2024) that did not materialize on stated timelines. Critiques:

• S2F captures only supply-side scarcity. Price depends on demand, regulation, sentiment, and macro conditions, none in the ratio.

• Back-tests on three prior halving cycles cannot reliably forecast future prices.

• The model conflates scarcity description with demand assertion.

• Plan B's specific targets failed; the analyst community has moved toward treating S2F as a scarcity metric only.

S2F is one lens for relative scarcity, not a forecasting law. For whether Bitcoin functions as a store of value, fixed supply is one input, not the answer.

After ~2140: Fee-only mining and the security budget question

When the subsidy rounds to zero around 2140, miners earn only transaction fees. The security model transitions: miner revenue (and therefore attack cost) becomes a function of fee-paying block-space demand rather than new issuance.

This is the most-debated long-horizon question about Bitcoin's design. Two camps:

• Fee-sufficiency optimists argue that as adoption grows, on-chain block-space competition produces per-block fee revenue large enough to match current subsidy + fee levels. Lightning settlements, custody, and large-value transactions create natural fee demand.

• Security-budget pessimists argue that fee variance is high, fees concentrate in mempool-congestion windows leaving quiet periods with low revenue, and total miner revenue may fall below today's hashrate baseline.

Neither side has a settled answer because the economics play out over 110+ years on adoption trajectories no one can forecast. The protocol does not promise the security budget will remain at 2026 levels post-2140; it promises issuance will stop. Fee-market design determines the rest.

For why proof of work costs real electricity, lower miner revenue means lower hashrate means lower 51%-attack cost.

What fixed supply is and is not: An honest read

Fixed supply means the protocol will not allow more than ~21M BTC to be issued. That is a verifiable supply-side fact.

What it implies:

• Predictable, finite issuance schedule anyone can verify by reading code or running a node

• No discretionary monetary policy; no entity can issue more bitcoin to address economic conditions

• Cumulative supply growth declining toward zero on a predetermined schedule

What it does not imply:

• Demand will remain constant or grow over any time horizon

• Current prices represent fair value

• Prices cannot fall significantly (Bitcoin has had 70-85%+ drawdowns from prior peaks while remaining scarce)

• Fixed supply guarantees any return; it is an input to valuation, not a price floor

• Accessible supply equals protocol supply (lost coins reduce one without changing the other)

• The cap cannot be technically modified (it can; political-economic forces make it functionally permanent)

The argument "only 21 million exist, therefore the price must rise" conflates supply constraint with demand certainty. A credible scarcity narrative attracts demand and supports valuations, which is a legitimate market dynamic; but scarcity itself is not what forces price upward, shifting demand against constrained supply is. For why Bitcoin remains volatile despite fixed supply, the answer lies in demand variance, not supply variability.

From Blofin's exchange-operator vantage point, the supply-cap question shows up most concretely in client conversations about long-horizon allocation reasoning. Our practical observation is that the strongest defensible framings of the 21M cap distinguish three things, protocol-level issuance ceiling (firm), accessible-supply estimate (uncertain by ~3-4M BTC), and price implication (separate question depending on demand), and the weakest conflate all three into "scarcity equals price appreciation." Where we do take a position: any honest read of Bitcoin's monetary properties should treat the 21M cap as one input among several, not a thesis-completing argument; demand-side variables (regulatory clarity, institutional adoption, custody infrastructure, fee-market evolution post-issuance) are equally load-bearing for any multi-decade allocation case.

Common confusions worth flagging

"The whitepaper says 21 million." 

It doesn't. Section 6 talks about issuance and inflation in general terms; the 21M figure is implicit in the geometric series produced by the 50 BTC subsidy + 210,000-block halving cadence in Bitcoin Core, not a stated number in the paper.

"The cap is exactly 21,000,000 BTC." 

Close but slightly off. The asymptotic limit is 20,999,999.9769 BTC due to integer-satoshi truncation in the right-shift halving operation. Round shorthand of "21M" is fine; literal protocol arithmetic produces the longer figure.

"All bitcoin will be mined by 2140." 

Approximate, not exact. The May 7, 2140 figure assumes constant hashrate from mid-2015 levels. Real-world hashrate growth has consistently shaved block intervals below ten minutes, which advances the date; further halving-era variations make the exact terminal date uncertain by years to decades.

"Lost coins lower the 21M cap." 

No. Lost coins reduce accessible supply (estimated 16-18M usable today) but do not change protocol issuance. Both numbers are valid measures of different things.

"Bitcoin can be made unlimited if developers decide." 

Technically possible via hard fork; practically blocked by economic alignment of every existing holder, miner, exchange, and custodian against a value-destroying change. The cap is credibly committed because changing it would catastrophically damage the change's would-be coalition.

"Stock-to-flow proves Bitcoin will hit $X price." 

No. S2F is a scarcity description, not a price model. Plan B's specific price targets failed to materialise on stated timelines; the analyst community treats S2F as a scarcity metric rather than a forecast.

"Fixed supply protects against inflation." 

Protects against protocol-supply-driven inflation, yes. Does not protect against price declines caused by demand-side changes. Bitcoin has had 70-85%+ drawdowns while remaining scarce throughout.

FAQ

Why is Bitcoin capped at 21 million?

The 21 million cap is the mathematical result of Bitcoin's protocol parameters: a 50 BTC starting block subsidy and a halving every 210,000 blocks. The geometric series 50 × 210,000 × (1 + 1/2 + 1/4 + ...) sums to 21,000,000 BTC. Satoshi Nakamoto encoded these values in the original code without leaving a written explanation of why exactly 21 million; the round shorthand is widely treated as a deliberate "predictable, finite" design preference rather than a calibrated economic target. The cap is enforced by distributed consensus, and no single entity can change it without persuading the entire network to adopt new software.

How does Bitcoin actually enforce the 21 million cap?

Every full node runs the GetBlockSubsidy function in Bitcoin Core's consensus code GitHub on every received block. The function computes the maximum allowed subsidy at any block height by halving the original 5,000,000,000 satoshis (50 BTC) for every 210,000-block interval using a bitwise right-shift. If a coinbase transaction claims more than GetBlockSubsidy(height) + sum_of_fees, every validating node rejects the block regardless of the proof-of-work behind it. Distributed verification across thousands of nodes with no central coordinator is what makes the 21M cap a protocol guarantee.

How many bitcoins exist as of 2026 and how many are left?

As of April 2026 approximately 19.95-20.01 million BTC have been mined, ~95% of the 21 million cap, with roughly 1.0-1.05 million BTC remaining Itrustcapital. The current 5th halving era (since April 2024 at block 840,000) issues 3.125 BTC per block, producing ~450 BTC per day or ~164,000 per year. The next halving to 1.5625 BTC is projected for April 2028 at block 1,050,000. The remaining ~1.05M will be issued progressively over ~115 years, with each successive halving cutting the per-block subsidy until the satoshi count rounds to zero around 2140.

When will the last bitcoin be mined?

The standard answer is "around 2140." The Bitcoin Wiki controlled-supply page projects May 7, 2140 specifically Bitcoin Wiki, but that figure assumes constant hashrate from mid-2015 levels, which has not held; hashrate has grown several thousand-fold and block intervals run consistently below the ten-minute target. Halvings trigger by block height, not calendar time, so the terminal date depends on cumulative block-time deviations over more than a century. The protocol rounds the subsidy to zero somewhere in the 2130s-2150s window. "Around 2140" is shorthand; any specific day overstates model precision.

Do lost bitcoins reduce the 21 million cap, and what is accessible supply?

No, lost coins do not reduce the protocol cap. They reduce accessible supply (coins current key-holders can actually move). Chainalysis and Glassnode estimate 2-4 million BTC permanently inaccessible, though figures have material uncertainty since no on-chain marker distinguishes lost from long-dormant. Accessible supply is therefore ~16-18M BTC depending on which estimate one accepts, while the protocol cap of ~20,999,999.9769 BTC is unaffected. Use protocol supply for monetary-policy questions ("can Bitcoin be inflated"), accessible supply for market-availability questions ("how much could realistically come to market").

Could the 21 million cap ever be changed?

Technically yes via hard fork; practically almost certainly never. Changing the cap requires consensus across nodes, miners, exchanges, custodians, and large holders, every group of which loses purchasing-power claim if supply expands. The ~$1-2 trillion in market capitalization aligned to the current cap means a coalition supporting a cap-raising fork has effectively zero chance of forming. Past contentious hard-fork attempts (Bitcoin Cash 2017, Bitcoin SV 2018) split off into separate assets without displacing the original chain. The cap is credibly committed not because the code is immutable but because no coalition has both means and interest to coordinate a value-destroying change.

Does fixed supply guarantee Bitcoin's price will rise?

No. Fixed supply is a supply-side fact: the protocol will not issue more than ~21M BTC. Price reflects supply and demand, and demand depends on adoption, regulation, competing assets, macro conditions, and sentiment, none of which the protocol determines. Bitcoin has experienced 70-85%+ drawdowns from prior peaks while remaining scarce throughout. A credible scarcity narrative can attract demand, but the causal chain runs through demand, not directly from scarcity to price. Stock-to-flow models built on the supply-only assumption have failed to produce reliable forecasts. Treat the 21M cap as one input to valuation reasoning, not a thesis-completing argument.

Researched and written by the BloFin Academy editorial team with AI-assisted drafting. Factual claims independently verified against Bitcoin Core consensus source src/validation.cpp GetBlockSubsidy function GitHub, the Bitcoin whitepaper Section 6 on incentive and inflation Bitcoin.org, the Bitcoin Wiki Controlled Supply page for the 20,999,999.9769 BTC asymptotic limit and May 7, 2140 final-block projection Bitcoin Wiki, the learnmeabitcoin halving page for the 5th era 3.125 BTC subsidy Learn Me a Bitcoin, Blockchain.com circulating-supply chart and the Newhedge / iTrustCapital trackers for the April 2026 ~19.95-20.01M BTC mined figure Itrustcapital, Bitbo and Swissgoldsafe stock-to-flow data for the BTC ~121 versus gold ~62-70 ratios Charts, the Federal Reserve H.6 Money Stock Measures release and FRED M2SL series for the US M2 expansion comparator Federal Reserve, the Nakamoto Institute archive for Satoshi's early correspondence on the design choice Nakamoto Institute, Chainalysis and Glassnode lost-coin research summaries for the 2-4M BTC inaccessible estimate, and Bitcoin Core documentation Bitcoin Core for the distributed full-node enforcement model.

Disclaimer: This content is for educational purposes only and does not constitute financial, investment, legal, or tax advice. Crypto assets are highly volatile and carry significant risk of loss. Always verify local regulations and consult a qualified professional before making financial decisions.