Here’s a question I don’t see asked enough: if Bitcoin’s maximum supply is 21 million coins, but somewhere between 5 and 6 million of those are permanently gone – destroyed wallets, lost keys, Satoshi’s untouched genesis block – what does that actually do to the price math? Because the headline figure of 21 million is doing a lot of heavy lifting in most scarcity arguments, and I think it’s misleading at best.
TLDR
- Bitcoin’s effective liquid supply is roughly 15-16 million BTC, not 21 million – an estimated 5-6 million coins are permanently inaccessible due to lost keys, destroyed wallets, and Satoshi’s dormant holdings.
- The supply shock math is stark: if demand doubles against a 16M coin float rather than 21M, price pressure is roughly 31% more intense than standard scarcity models assume – and the stock-to-flow model doesn’t account for this at all.
- For investors, this makes self-custody using hardware wallets critical – not just for security, but because every coin lost to a bad key is supply permanently removed from the market, tightening the float for everyone who holds correctly.
The framing matters. When people argue about Bitcoin scarcity, they almost always reach for “only 21 million will ever exist.” That’s true. But it’s not the whole picture. The real question for price modeling is: how many Bitcoin are actually available to be bought or sold?
The answer is significantly lower than 21 million. And the gap has real implications for anyone thinking seriously about where Bitcoin’s price can go.
Protect what you hold – don’t add to the lost coin count
Self-custody done right keeps coins in the float.
The Supply Math Nobody Wants to Run – Bitcoin Lost Coins Price Impact
With ~5 million BTC permanently gone and 1.4 million locked in ETFs, Bitcoin’s effective tradeable float is approximately 14.6 million coins – not the 21 million most price models assume. That 31% supply gap directly amplifies every demand event.
Let me do the arithmetic that most Bitcoin content skips.
According to Glassnode’s HODL Wave data and academic research on dormant addresses, somewhere between 5 and 6 million BTC are effectively gone. Not “locked up in cold storage by someone who might sell someday” – actually, permanently inaccessible. Private keys lost. Hard drives in landfills. Wallets with forgotten passphrases. Coins from 2009 and 2010 sitting in addresses that have never, not once, moved a single satoshi.
Strip those out and you get an effective circulating supply of 15 to 16 million BTC.
Now layer in what’s been absorbed by institutional holders. Bitcoin ETFs currently hold approximately 1.4 million BTC – about 6.77% of total supply – and most of that isn’t going anywhere on a day-to-day basis. That’s demand that has already cleared the market and is now sitting in custody.
Add long-term holders who have demonstrated years of conviction and you’re looking at a float – coins that could realistically change hands in a given market cycle – that’s genuinely tight.
Where Did 5 Million Bitcoin Actually Go?
The Chainalysis Lost Coin Report and Glassnode age cohort analysis break down the losses into rough categories. Here’s how I think about them:
Satoshi’s holdings (~1 million BTC). The Bitcoin creator’s wallets – particularly coins mined in the genesis block era – have never moved. Not once in 16+ years. The probability that these coins will ever move is, at this point, academic. Either Satoshi is dead, has lost access, or has made a considered decision to never touch them. For price modeling purposes, treat this 1 million BTC as gone.
Early miner losses (500K – 2M BTC). In 2009 and 2010, Bitcoin was worth fractions of a cent. Miners stored keys on hard drives the same way they stored MP3 files – casually, on equipment they later threw away. The famous story of James Howells, who is still trying to dig up a Welsh landfill for a drive containing 8,000 BTC, is representative of an entire era of casual key management. Estimates here are wide because there’s no clean way to distinguish “lost” from “hodling aggressively,” but the academic consensus points toward 500K to 2M BTC that are simply unreachable.
Addresses with no activity for 15+ years (~3.2 million BTC). This is where the Glassnode data gets interesting. An estimated 3.2 million BTC sits in addresses that haven’t seen a single outgoing transaction in over 15 years. Some of this overlaps with Satoshi’s holdings and early miner losses. The point is that this supply has demonstrated – through 16 years of bull markets, crashes, and multiple halving cycles that would have been obvious selling opportunities – that it is not responding to price signals.
Failed wallets and exchange collapses. Mt. Gox, QuadrigaCX (where the founder died with the only private key, allegedly), various early exchanges that folded without recovering user funds. Some of this gets counted in on-chain analysis; some of it doesn’t.
The total lands somewhere in the 5-6 million BTC range. Let’s use 5 million as a conservative estimate for the math that follows.
The Price Implication – Running the Numbers
Here’s where it gets concrete. If you believe basic supply-and-demand mechanics apply to Bitcoin (and they do – this isn’t a debate), then the effective float matters more than the headline supply.
Consider two scenarios:
| Metric | Headline Supply Model | Effective Supply Model |
|---|---|---|
| Total Supply | 21,000,000 BTC | ~16,000,000 BTC |
| Lost/Unreachable | Ignored / 0 | ~5,000,000 BTC |
| ETF + Institutional Holdings | 1,400,000 BTC | 1,400,000 BTC |
| Tradeable Float (est.) | ~19,600,000 BTC | ~14,600,000 BTC |
| Supply Scarcity Premium | Baseline | +34% price pressure vs. headline model |
If demand doubles – same number of buyers chasing the same investment thesis – but the available supply is 16 million rather than 21 million, the price impact is proportionally more severe. The market has to clear with fewer coins. This isn’t speculative; it’s arithmetic.
The stock-to-flow model, one of the more widely cited Bitcoin price frameworks, uses the 21 million headline figure. It ignores lost coins entirely. If you adjust S2F for effective supply, you get a meaningfully more bullish number – not because the model changed, but because the input was wrong. I’ve written more about this over at Bitcoin Stock-to-Flow Model 2026 if you want to dig into the methodology.
Satoshi’s 1 Million: The Silent Market Maker That Never Acts
I want to spend a moment on Satoshi’s coins specifically because they get treated like a wildcard risk when they should be treated as a fixed supply constraint.
The argument I sometimes hear: “What if Satoshi’s coins suddenly move? That would crash the market.” Sure. But every year that passes without those coins moving makes the scenario less plausible, not more. We have now had 16+ consecutive years of every conceivable incentive to sell – multiple ATHs, regulatory uncertainty, bull markets, bear markets – and those coins have not moved.
At some point, probability is just probability. A dormant wallet for 16 years, across conditions that would have made a rational actor take profits dozens of times, is almost certainly not coming back. For supply modeling purposes, Satoshi’s ~1 million BTC should be treated the same as destroyed coins.
That’s a different frame than most analysts use, but I think it’s the right one for investors trying to understand the real float.
For more on the institutional side of the supply equation, the Bitcoin ETFs Hold 6.77% of BTC Supply piece is worth reading alongside this – ETF flows and lost coins are both compressing the available float, just through different mechanisms.
Lost Coins Beat Halving on Permanent Scarcity
This is the comparison I find most underappreciated.
The halving reduces Bitcoin’s new supply issuance by 50% every four years. It’s baked into the protocol, predictable, and gets enormous attention. But it’s incremental and time-based – each halving affects future supply, not existing supply.
Lost coins are a different kind of scarcity. They’re permanent and retroactive. A coin lost in 2011 removed that supply from the market that year and every subsequent year. The scarcity compounds. Every halving cycle that passes is also another 4 years of normal key attrition – people die, hard drives fail, passphrases are forgotten.
The important nuance from Chainalysis research: loss rates are slowing down. Cold storage practices have improved dramatically since 2017. Hardware wallets have become mainstream. The era of casual key management that cost early Bitcoin holders billions is largely over. This means the existing 5-6 million lost BTC represents a historical peak – future losses will be incremental rather than the mass casualty events of the early mining era.
But that also means the lost coins we already have are essentially permanent. They’re not coming back, and the rate of new losses is low enough that you can model that 5-6 million as a fixed supply reduction going forward.
This connects directly to the supply scarcity thesis for 2028 and beyond. MicroStrategy’s 525K Bitcoin and the 2028 Halving looks at how institutional accumulation stacks on top of this underlying supply compression – worth reading if you’re thinking about the next cycle.
What This Means for the 2028 Cycle
The 2028 halving will cut new supply issuance to approximately 1.6 BTC per block – down from the current 3.125. Pair that with:
- 5-6 million BTC permanently out of circulation
- 1.4 million BTC in ETFs with long-duration mandates
- Improving self-custody practices that reduce future loss rates
- Slow but steady institutional adoption continuing to absorb float
The effective market is operating on a tighter float than headline metrics suggest. If demand growth continues – whether through ETF inflows, retail adoption, or corporate treasury allocations – the price mechanics are working against a 16 million coin float, not a 21 million coin float.
This doesn’t mean Bitcoin goes to infinity on a straight line. Markets are messier than supply-demand models. But it does mean that anyone discounting Bitcoin’s long-term price ceiling because “21 million is a lot of coins” is working with the wrong number.
The accurate number is closer to 15-16 million, and it’s probably shrinking slowly even now.
Track on-chain supply dynamics with a trusted exchange
Get exposure to Bitcoin’s effective scarcity.
Implications for Individual Investors
If you accept the effective supply argument, a few practical things follow:
Your self-custody practices matter to the whole market. Every coin lost to a bad key backup is supply removed permanently. That sounds dramatic, but it’s true – you losing your coins makes everyone else’s slightly more scarce. The flip side: holding properly means you’re participating in the correct way in a market that rewards sound custody.
Wealth concentration math changes. I’ve looked at the numbers for what Bitcoin ownership means at various wealth thresholds in Bitcoin wealth targets by 2032 The analysis changes materially when you run it against 16 million coins rather than 21 million. Owning 1 BTC in a 16M-coin world is more concentrated wealth than owning 1 BTC in a 21M-coin world.
The lost coin estimates have uncertainty bands. Glassnode’s HODL Wave data is the most rigorous on-chain analysis available, and Chainalysis lost coin research is the best independent verification. But neither can definitively distinguish “lost forever” from “cold storage with an owner who hasn’t touched it.” The 5-6 million estimate is the academic consensus; the real number could be somewhat higher or lower. Model with uncertainty.
This doesn’t change short-term price action. Lost coins don’t affect whether Bitcoin trades at $80K or $100K next Tuesday. They’re a structural argument about long-term ceiling and floor dynamics. Don’t use supply arguments to justify short-term trades.
FAQ
Is the “21 million Bitcoin” supply figure accurate?
No, not as a description of available supply. The 21 million hard cap is absolutely real – Bitcoin’s protocol will never issue more. But an estimated 5-6 million of those coins are permanently inaccessible due to lost private keys, destroyed hardware, and provably dormant wallets like Satoshi’s genesis block holdings. The effective liquid supply is approximately 15-16 million BTC, and that’s the figure that matters for price mechanics.
Do stock-to-flow models account for lost Bitcoin?
No, and that’s a meaningful flaw. The standard stock-to-flow model uses the 21 million total supply as its stock figure. It doesn’t subtract lost or permanently dormant coins. Adjusting for the effective supply of 15-16 million BTC produces a more bullish S2F output – not because the model logic changed, but because the input was more accurate. Lost coin estimates are uncertain enough that you’d want to run sensitivity ranges rather than a single adjusted figure.
Could Satoshi’s coins ever move and crash the market?
Theoretically, yes – but the probability decreases with each passing year. Satoshi’s wallets – holding approximately 1 million BTC that have never moved – have been dormant through 16+ years of every conceivable market condition including multiple all-time highs where selling would have been rational. Either Satoshi is no longer alive, has permanently lost access, or has made an ideological decision not to sell. For supply modeling, the rational assumption is to treat those coins as permanently removed from the float. A sudden movement of 1 million BTC would be market-destabilizing, but it becomes less likely with each passing year of inactivity.
How do lost coins compare to the effect of the 2024 halving?
Lost coins dwarf any single halving by pure magnitude. The 2024 halving cut new supply issuance from 6.25 BTC per block to 3.125 BTC per block – reducing annual new supply by roughly 164,000 BTC. Lost coins represent a one-time permanent removal of 5-6 million BTC from circulating supply. The difference is that halvings are predictable, scheduled, and priced in by market participants in advance. Lost coins are a structural background condition that most models simply ignore.
Is the 5-6 million lost coin estimate reliable?
It’s the best available estimate, not a precise measurement. Glassnode’s HODL Wave analysis identifies BTC that hasn’t moved in extended periods; Chainalysis applies additional methodology to separate presumed-lost from long-term cold storage. The challenge is that on-chain data alone can’t distinguish a disciplined long-term holder from a lost wallet – both look identical from the outside. The academic consensus of 5-6 million is a reasonable midpoint, but model it as a range: at 4 million lost, effective supply is ~17 million; at 7 million, it’s ~14 million. The directional argument holds either way.
The supply math is cleaner than most Bitcoin arguments. Fewer coins available to buyers means each coin commands a higher price at any given demand level. Five million coins gone is not a rounding error – it’s a 25%+ reduction in the effective float, and almost no mainstream price model is accounting for it properly.
That’s either a significant analytical gap or a significant opportunity, depending on how you look at it.
