There is no single dataset behind a chart like this. The widely shared versions are
pictures, published quarterly, with no data file and no archive. So this page assembles
it from primary sources, and every slice carries a different weight of evidence. That
difference is worth more than the numbers.
Measured exactly
Coins mined is the only line here with no argument attached. It comes from the
blockchain itself, and every slice is stacked against it rather than against the 21 million
cap, so the chart never pretends coins exist before they were mined.
To be mined is simply 21,000,000 minus that.
Traced to filings and to the chain
- Funds & ETFs is the strongest of the estimated lines. Grayscale filed its
exact bitcoin balance with the SEC every quarter from 2014, so the pre-2024 history is
audited rather than guessed.
- Businesses is anchored on Strategy, which publishes every purchase, and on
the quarterly filings of the listed miners.
- Governments is assembled from court documents, forfeiture filings and
announced sales, so it moves in steps rather than curves. That is real: governments
acquire bitcoin in seizures, not by accumulating.
Assumed, not measured
- Satoshi is the Patoshi-pattern estimate of roughly 968,000 coins. The chart
ramps it through 2009 and early 2010 as those blocks were mined, then holds it flat,
because not one of those coins has ever moved.
- Lost is a model, and the slider exists so you can see that. The page spreads
the loss across coins mined before 2014, because the era of casual mining and thrown-away
drives is where the losses actually are. When and how coins were lost is unknowable;
only the rough total is arguable.
Not measured at all
Individuals is a residual. It is what is left after everything above, which means
it silently absorbs every error in every other slice. It is drawn here as
Individuals & unattributed for that reason. Treat a change in this slice as a
change in what we can attribute, not necessarily as people buying or selling.
Why this shows more for individuals than the chart you have seen
Roughly one percentage point, and it is not a disagreement about individuals. It is the
four exclusions below. Every coin taken out of a government or business slice because the
evidence does not hold up has to land somewhere, and the residual is where it lands.
A wider individuals slice here is the visible cost of a narrower definition of proof
elsewhere, not a different view of how many people own bitcoin.
What was deliberately left out, and why
Four large numbers that appear in most versions of this chart are excluded here:
- China's 190,000 BTC. Seized in the PlusToken case, then ordered converted
to fiat by the court. On-chain analysis places the liquidation in 2019 and 2020.
Trackers still list China as the world's second-largest state holder on the strength
of a seizure that was sold years ago.
- Ukraine's 46,000 BTC. Not a state holding at all. It is the sum of personal
asset declarations filed by 652 individual civil servants, one of whom declared about
18,000 BTC on his own. It has been copied into sovereign tables as if it were treasury.
- Bulgaria's 213,000 BTC. The 2017 announcement said suspects had invested in
the coins. It never said they were seized, and Bulgarian prosecutors later denied
holding them.
- Futures ETFs. They hold contracts, not coins. Counting their notional
exposure as bitcoin would double count the coins backing it.
One inclusion is worth flagging the other way: the US figure contains about 94,600
coins a court has ordered returned to Bitfinex, and about 127,000 still contested in
the Prince Group forfeiture. Officials say roughly 200,000; trackers say 328,000. The
gap is litigation, not accounting.
Why the slices can be added together at all
Most ownership charts quietly double count. These are the traps, and how this page
avoids them:
- Institutions are inside the ETF slice, not beside it. Sovereign wealth
funds, hedge funds and pensions hold ETF shares. Those coins are already counted
once, in the fund. A 13F-derived institutional wedge drawn next to an ETF wedge counts
them twice.
- Feeder funds hold other funds. Several non-US listings marketed as spot
products are feeders into a US ETF, so their coins are already in the US total.
- Exchange balances are not a category. Most spot ETF bitcoin sits with one
custodian, so on-chain exchange-balance metrics already include ETF coins.
- The Grayscale spin-off is not an outflow. In July 2024, about 27,000 coins
moved from GBTC into a new Grayscale trust. Charts that track GBTC alone show a cliff
on that date that never happened. This page counts the two together.
Corporate treasuries and funds are genuinely separate at the large end: the named
treasury companies custody coins directly rather than holding ETF shares. That is what
makes stacking them legitimate.
Sources
Every dated point in this page carries its own source in the data file. Where two
trackers disagree materially, the more conservative figure is used and the disagreement
is noted above rather than averaged away.