Who are the whales? The map of the 21 million and the holders without names

On July 6, 2026, Strategy — the company formerly known as MicroStrategy — held 843,775 BTC. That is roughly 256,000 coins short of the ~1.1 million that Satoshi Nakamoto mined and never moved. For sixteen years, nobody came anywhere near that untouched first place. Now the gap is visible.

In Bitcoin’s own vocabulary, holders at this scale are called whales. The digital-gold analysis located Bitcoin’s claim to that name in two layers of decentralization: no operator inside the system, and no authority among the people around it. The founder’s absence shows up in the ownership landscape too. So what does that landscape look like in 2026 — who actually holds the 21 million?

The map of the 21 million

The whole picture first, in one table — then the same numbers by area.

The segments cut along ownership — whose coins they are. Coins that exchanges custody for clients answer a different question (who holds the keys) and overlap the rows above, so they get no row of their own. Even the size of that custodial pool is contested: CryptoQuant puts it near 12% of supply, Glassnode near 16%, and the difference is nothing more than two labeling methodologies disagreeing about which addresses belong to exchanges.

The largest cell, “individuals & unclassified,” is what remains after every nameable holding is subtracted. It contains millions of ordinary wallets, large holders no tracker has yet named, and holdings that cannot be named at all (more on those below).

Corporate treasuries — Strategy’s single point of concentration

Of the 1,264,579 BTC on public-company balance sheets, one company holds two-thirds. Second place is an order of magnitude away: as of 2026, Twenty One Capital ranks second at 43,514 BTC, Metaplanet third at roughly 40,000, MARA fourth at nearly 39,000 — and every public company other than Strategy combined holds 420,804 BTC, barely half of Strategy alone.

When Strategy disclosed its first purchase in August 2020, a bitcoin corporate treasury was still an experiment.

Each anchor traces to the company’s SEC disclosures. The curve breaks upward in 2024, when Strategy roughly doubled four years of accumulation in a single year and kept buying.

Spot ETFs — the buyer that arrived in 2024

US spot ETFs were approved in January 2024. Two and a half years later the ETF pool holds about 1.21 million BTC. The largest single product is BlackRock’s IBIT: 551,918 BTC at the end of 2024, 776,475 at the end of 2025, a spring-2026 peak in the 810,000–820,000 range — and then a record outflow stretch in May and June that left it at 733,947 BTC in early July.

This segment owns differently. IBIT’s coins are not BlackRock’s assets; the trust custodies them for ETF shareholders. A corporate treasury sits on its own balance sheet and grows by management decision; an ETF is a wrapper around investors’ money and breathes with their flows, daily. The nearly 80,000-coin decline since spring 2026 is that breathing, printed directly into the number.

Nation-states — from seizure to reserve

The US government’s balance was never bought; it is a genealogy of seizures. Silk Road-linked coins, 69,370 BTC (civil forfeiture, November 2020). James Zhong’s 50,676 BTC (November 2021). The 2016 Bitfinex hack recovery, 94,636 BTC (February 2022). Auctions and victim restitution had trimmed the pile to about 198,000 BTC by March 2025, when an executive order stopped the selling and converted the balance into a Strategic Bitcoin Reserve. In October 2025 the largest forfeiture action in Justice Department history — 127,271 BTC from the Cambodia-based Prince Group case — pushed the attributed balance to 328,372 BTC as of February 2026.

The 649,954 total is the sum of BitcoinTreasuries’ full government list, which includes small sovereign holders beyond the six countries above; the per-country figures also carry different disclosure and attribution dates, so the major rows alone do not add up to the total. The last row carries that remainder explicitly.

Russia does not appear in this table. Its seizure-derived holdings are estimated at 50–200 BTC, and the reserve concept under discussion there extends only to confiscated assets.

The unattributed side — criminal balances and North Korea

Inside “individuals & unclassified” sit holdings that cannot be named. As of July 2025, Chainalysis estimated on-chain balances held directly by illicit entities at about $15 billion — and about $75 billion once wallets downstream of them are included. No BTC-denominated count has been published, and a dollar-denominated balance swells with the bitcoin price alone.

North Korea (Lazarus Group) is counted in flows, not balances. Chainalysis’s lower-bound estimate puts cumulative theft at $6.75 billion through the end of 2025 — $2.02 billion in 2025 alone, most of it the ~$1.5 billion Bybit incident in February — with more than $600 million added in the first half of 2026. The distinction matters: this is not an amount currently held. Most of what was stolen was denominated in Ether and stablecoins, laundering and cash-out are fast, and no published estimate isolates what remains as a bitcoin balance. A cumulative $6.75 billion cannot be placed on a map of the 21 million as a holding.

The whales — the named and the nameless

“Whale” has a working definition in the analytics industry. Glassnode counts whales as clustered entities holding 1,000 BTC or more, exchange balances excluded, at the top of a scale that starts at shrimp (under 1 BTC) and climbs through crab, octopus, fish, dolphin, and shark to whale (1,000–5,000 BTC) and humpback (over 5,000). By its March 2023 estimate, entities above 1,000 BTC held about 6.64 million coins — roughly a third of circulating supply. The unit is the entity, not the address: one holder controls many addresses, so analytics firms bundle addresses with clustering algorithms before counting anything.

Every named holder in the sections above is a whale by that measure, and nearly every name rests on evidence from outside the chain: Strategy’s number comes from its SEC filings, IBIT’s from the fund’s own reports, government balances from forfeiture records. On-chain evidence alone has produced few names. The two best-established are Satoshi, identified not by any disclosure but by the Patoshi mining pattern in the earliest blocks, and the Mt. Gox bankruptcy estate, whose roughly 142,000 BTC were mapped in court-supervised proceedings before repayments began in July 2024.

The rest of the whale pool has no name attached. When a headline reads “a whale bought” or “a whale sold,” the source is usually Whale Alert, which posts large transfers to social media — for Bitcoin, above $50 million between known addresses and $100 million for unknown ones — and names the sender and receiver only when it recognizes them, which in practice means exchanges. Beyond those labels, everything is inference. Clustering rests chiefly on the assumption that all inputs of one transaction share one owner — a linkage the whitepaper itself conceded in its privacy section — and since no ground-truth record of who owns what exists, the accuracy of the clustering cannot itself be measured. The cautionary example is Mt. Gox: the chain’s largest cluster, more than ten million addresses, was initially mis-merged by that very assumption because the exchange once let users import their own private keys.

July 2025 showed the scale of what remains nameless. A holder inactive since April 2011 moved about 80,000 BTC through Galaxy Digital over two weeks; Arkham’s blockchain data traced every step in real time, and the final tranche alone was reported at $4.8 billion — yet no tracker has put a name on the owner. Fourteen years of silence ended in full public view, pseudonymously.

None of this is a defect in the trackers. It is the structure Satoshi warned about in July 2010:

“It’s possible to be pseudonymous, but you have to be careful.”

A forum post the following day stated the transparent half plainly:

“The history of a coin is publicly available. Anyone can see the flow of BitCoins from address to address.”

The map on this page is drawn inside that structure: the history is public, the identities are not, and analysis can sometimes bridge the gap. A whale gets a name when it identifies itself — a filing, a court docket — or when analysis and circumstance close the gap from outside. Until then it sits in the map’s largest cell: visible in every movement, named in none.

The distance from “no controlling entity”

What the digital-gold analysis means by its second layer — decentralization of people and organizations — is that nobody holds authority over the protocol. Concentrated ownership does not directly break that definition. Strategy, IBIT, and the US Treasury cannot change one line of consensus rules, however many hundred thousand coins they hold. A premine allocated to founders at issuance and a stack bought on the open market have entirely different origins.

The ownership landscape still shifts the second layer’s surroundings. ETF flows have become a primary circuit of price formation, and sovereign reserves introduced a political variable — the distance between nation-states and Bitcoin — that the whitepaper never contemplated. And first place has not changed hands. That ~1.1 million coins have sat still for sixteen years — that the founder’s keys never became a market authority — is itself part of the second layer, as documented in Whale Alert’s independent analysis and the survey of converging estimates. Next to the bar that never moves, only the other bars trade places.

Limits

• The numbers move. Each value carries its as-of date in the body and tables; the ETF pool alone shifted nearly 10% within months in 2026.
• Tracker disagreement varies by segment. Government totals differ by more than 2x between River (305,000 BTC, December 2025) and BitcoinTreasuries (649,954 BTC, July 2026); the latter includes the large US forfeiture of October 2025, and this page uses the newer figure.
• Lost-coin estimates span 1.6M–3.7M BTC. The figure uses the lower bound of the widely-cited 2.3M–3.7M range.
• Satoshi’s own count is an estimate with a range of its own, from BitMEX Research’s 600,000–700,000 to Whale Alert’s 1,125,150. This page uses the ~1.1M convergence value, consistent with the rest of this archive.