Crypto Ownership Statistics: Who Actually Owns It

Two research shops counted crypto owners for the same year, 2024. Triple-A said over 560 million. Crypto.com said 659 million. Same question, same planet, 99 million people apart — about 18%, or the population of Egypt, in the margin.

That gap is not sloppiness. It's what happens when the thing you're counting doesn't exist in the data. Blockchains record addresses. Surveys record what people say when asked a specific question in a specific month. Neither produces a count of owners, and the numbers get published as though they do.

An address is not a person

Start with what the ledger actually is. Glassnode puts it plainly: "A Bitcoin address is not an 'account'. One user can control multiple addresses, and one address can hold the funds from multiple users."

The size of that distortion is measurable. When Glassnode clustered addresses into entities in January 2020, ~28.4 million non-zero addresses collapsed into ~23.1 million entities — 18.5% fewer — and the firm still called its own result a "strict upper limit for the number of Bitcoin entities", because ~96% of clustered entities were single addresses that clustering can't merge without evidence. That analysis is six and a half years old: a calibration for how far addresses overstate people, never a current count.

Clustering also moves the famous concentration line. Glassnode's own correction: "around 2% of network entities control 71.5% of all Bitcoin" — "substantially different from the often propagated '2% control 95% of the supply'."

And some of those addresses belong to nobody. The FCA's July 2026 market description reports that 9% of UK cryptoasset holders "have forgotten their private key and as a result lost access to their wallet." Their coins still sit in every distribution table on the internet.

The global estimates, side by side

Run the headline through a denominator and it gets awkward. 741 million against a world population of 8,140,897,523 is 9.10% of every human alive, infants included. The Federal Reserve puts US adults at 10% — in a country Chainalysis ranks second worldwide. For both to be true, the rest of the planet, newborns and all, would have to own crypto at roughly the American adult rate. That near-parity is the tell.

Crypto.com hands you the evidence itself. The same report notes there "could be approximately 300,000 to 1.2 million people invested in BTC via US spot exchange-traded funds" — a four-fold range, offered without a method, for a category holding tens of billions of dollars.

Chainalysis is the honest one here because it refuses to guess: its index measures transaction activity, not owners, admits web-traffic attribution is imperfect thanks to "VPNs and other similar tools", and excludes "OTC desks, informal markets like hawalas, or cash-based crypto shops."

Same country, four answers

The United States is the most heavily surveyed market in this data set. It still can't tell you how many Americans own crypto.

Three times the spread, and almost none of it is real disagreement about America. "Ever used" will always beat "currently own", which will always beat a Fed question about the prior year. Stack them in one chart and you have manufactured a trend out of grammar.

Security.org is the outlier, and its own history says so: 15% in 2021, then 33% in 2022, then 27–30%. A population doesn't double its crypto ownership in a year and then give a third of it back. With n=992 and self-reporting, that's sentiment, not measurement.

The Fed's series is the most disciplined available: 12% in 2021, 8% in 2024, 10% in 2025. Gallup's 17% figure travels widely without its definition attached — Gallup's "investors" are adults with $10,000 or more invested, not the general public. And Pew says its own use has "remained fairly stable" against 16% in 2021, so anyone quoting 16→19 as growth is quoting past the source.

None of these percentages should be multiplied into a headcount. No agreed adult-population denominator sits next to them.

The one country where ownership fell

The FCA's July 2026 market description is the closest thing to a national statistic in this field: "As of August 2025, an estimated 8% of UK adults own cryptoassets, equivalent to 4.5m people." It comes from YouGov fieldwork run for the FCA — 2,353 quota-controlled interviews plus a 1,053-person boost of current and former owners, 5 August to 2 September 2025.

That 8% is a decline. The Block and CoinDesk reported the wave as roughly 12% falling to 8% — about 7 million people down to 4.5 million — still near double the ~4% of 2021. (The FCA's own PDF states only the level, and that demand "more than doubled since 2020".) Awareness sits at 91%, flat on 2024. People know what it is. Fewer of them are holding it.

Then the FCA contradicts itself. One passage: "7% of cryptoasset owners reported participating in either borrowing or lending, while 22% reported participating in staking." Another, same document: "9% of current cryptoasset users take part in lending, with 2% taking part in borrowing." A union cannot be smaller than one of its parts. If the best regulator-grade source in the field publishes 7% and 9% for overlapping questions in one PDF, the aggregators recycling these numbers third-hand are not doing better.

Demographics, where somebody actually asked

The FCA's UK profile is the most detailed public one: 68% male, 58% under 34, 92% under 54, 62% from higher socio-economic backgrounds (ABC1), and 63% say they're willing to take investment risks against 24% of the wider UK population. Average portfolio £2,250, but skewed — most own under £1,000 while about 16% hold more than £5,000, and 72% leave the coins on the platform they bought them from.

In the US, Pew finds men 18–29 at 38% against women 18–29 at 15%, and men 30–49 at 40% against women 30–49 at 17% — a gender gap wider than any other split it measures. Republicans moved from 16% to 22% while "Democrats' use has held steady at 17%". Gallup's bands read lower — men 18–49 at 25%, women 18–49 at 8% — because it asks about current ownership, not lifetime use. Don't average the two.

One Fed finding deserves more attention: 6% of unbanked adults used crypto to make a financial transaction, against 2% of banked adults.

Where the bitcoin actually sits

Address-level distribution, as of 16 July 2026, against 20,057,293.75 BTC mined (not the 21M cap — trackers that use the cap understate every share by about 4.7%):

Source: BitInfoCharts, snapshot 16 Jul 2026 22:26 UTC. 59,294,945 addresses hold a balance.

Two readings fall out. 98.35% of addresses hold under 1 BTC and share 7.07% of the supply; only 979,740 addresses — 1.65% — hold a whole bitcoin. At the other end, 2,044 addresses (0.0034%) hold 35.93%.

Now the part that refutes the naive reading. The four addresses in the 100,000+ bracket hold 700,779 BTC between them, and every one is an exchange or broker cold wallet: two Binance (248,598 and 181,321), Robinhood (140,850), Bitfinex (130,010). Those balances belong to millions of customers. The fifth belongs to Tether; the sixth is labelled "Bitfinex-Hack-Recovery" — the DOJ seizure documented in our piece on why blockchain isn't anonymous. "The top four addresses control 3.49% of bitcoin" is arithmetically true and tells you nothing about ownership.

Entity-level work lands in the same place from the other direction: Gemini and Glassnode found that "over 30% of Bitcoin's circulating supply is now held by just 216 centralized entities" — ETFs, exchanges, custodians and corporate treasuries — as of June 2025.

The identifiable holders

Do not add these up. ETF coins sit with custodians — largely Coinbase Custody — and providers differ on whether custodial balances count inside "exchange reserves". Any "X% of bitcoin is accounted for" claim built by summing this table is double-counting.

Exchange balances deserve their own warning. Santiment's 6.6% (lowest since 2017; ETH at 4.3%, lowest since 2015) and an aggregated CryptoQuant/Glassnode figure of roughly 1.8 million BTC from December 2025 are seven months and two methodologies apart. They aren't comparable, and neither is wrong. Cite the percentage with the provider and the date; don't publish an absolute count.

And the corporate line hides the most important story on this page. Strategy is 66.8% of the entire tracked public-company total, and trackers render its 843,775 BTC as a tidy number. The 8-K filed 13 July 2026 puts the aggregate purchase price at $63.69 billion — $75,476 per bitcoin, against a spot price of $64,079.61. The position is roughly 15% under water. The 8-K filed a week earlier discloses 1,363 BTC sold at an average of $59,256 and 2,225 BTC at $60,773 — below cost — to fund preferred-stock distributions, plus a "BTC Monetization Program" announced 29 June 2026 under which Strategy may sell bitcoin to raise up to $1.25 billion, capacity still fully untapped. The same filing reports an $8.32 billion loss on digital assets for the June quarter.

The tracker's number is correct, and on its own it misleads. Cite the filing.

Country rankings measure different things

Chainalysis's 2025 index ranks India first for the second year running and first across all four sub-indices, then the United States, Pakistan, Vietnam and Brazil; the UK is 11th. India received $338 billion in on-chain value in the twelve months to June 2025, up 99% year on year.

Triple-A's estimated ownership rate produces a different top: UAE 31.0%, Singapore 24.4%, Turkey 19.3%, Argentina 18.9%, Thailand 17.6%, Brazil 17.5%, against a 6.9% world average — 2024 vintage, no published formula.

Neither list is wrong. One measures transaction throughput weighted by income, the other estimates the share of a population that owns something. Merge them into a single "crypto adoption by country" ranking and you have nonsense.

What these numbers cannot tell you

• An address is not a person. 59.29 million addresses hold BTC; the four largest are exchange cold wallets holding for millions of customers.
• A person is not an address. Clustering cut 28.4M addresses to ~23.1M entities in 2020, and Glassnode called even that a strict upper limit.
• An entity is not a person either. Binance's cold wallet is one entity and millions of owners. Strategy is one entity and thousands of shareholders.
• "Owner" is not a defined term. 10% / 14% / 19% / 30% for the same country in an overlapping window, because four organisations asked four questions.
• On-chain estimation is not a census. 741 million rests on "our on-chain estimates"; the index that does disclose its method measures activity instead.
• Some of the supply is gone, and nobody agrees how much. River's page says both "nearly 4 million" and "roughly 1.57 million" bitcoin are lost — 2.5× apart, no method given, under River's own accurate line: "There is no way to know exactly how many bitcoin are lost forever."

The useful posture isn't picking a winner. It's quoting a number with the name of whoever produced it, the date they produced it, and the question they actually asked. Everything on this page is dated for that reason.