Volume III · Bitcoin · Money Research
09 — Supply and Control: Who Holds Bitcoin, Who Runs the Network, and Who Benefits
Who controls Bitcoin's rules, transfers and holdings?
Validators enforce their chosen rules, while miners, custodians, key holders, exchanges and law affect different steps in using the asset.
- Validators enforce chosen validity rules; miners propose blocks and influence inclusion.
- Custodians, keys, exchanges and law affect different uses of a coin.
- Dormancy does not prove a key is lost.
Two different kinds of control
Bitcoin's issuance rules and its market infrastructure are different kinds of control. The rules — the issuance schedule and what counts as a valid block — are checked by independently validating nodes; a holder, miner, developer or government cannot make a rule change binding on other nodes merely by announcing it. The market — dollar prices, trading liquidity, custody and block construction — has concentrated intermediaries. Coin ownership gives no vote over protocol validity, while ownership and infrastructure can give influence over flows and access. ETF trust assets, custodian-controlled keys and shareholders' beneficial interests must not be counted as three separate owners of the same coins.
The supply, September 2026
The subsidy has been 3.125 BTC per block since the fourth halving in April 2024; at block 1,050,000 it is scheduled to fall to 1.5625 BTC. At the target ten-minute block interval, 3.125 BTC per block corresponds to about 450 BTC per day, not an exact daily issue rate. A subsidy schedule gives a theoretical cumulative maximum; it is not an independently measured count of spendable coins, and miners need not claim every allowed satoshi. Bitcoin Core's UTXO-set report is the more relevant on-chain measure for unspent output value. Transaction fees are a separate source of miner income, with the long-run security question examined in “The Defects That Have Stopped Bitcoin From Becoming a Global Currency”. Bitcoin Core, `gettxoutsetinfo`.
The scheduled issuance total is not a measured liquid float. Bitcoin Core separately reports the value of currently unspent transaction outputs and identifies some provably unspendable amounts; neither figure reveals which otherwise spendable outputs have inaccessible keys. A coin's failure to move, including coins attributed to early miners, is evidence of inactivity, not proof of permanent loss. Exchange-wallet labels, coin-age bands and fund holdings answer different questions and can overlap. Without a reproducible address/entity method and a demonstrated trading-volume model, a precise “lost coins” or effective-liquid-supply estimate cannot be inferred from those observations. Bitcoin Core, `gettxoutsetinfo`.
| Measure | What it can establish | What it cannot establish |
|---|---|---|
| Issuance schedule | Subsidy allowed under consensus rules | Coins available for sale |
| Unspent outputs | Value currently recorded in the UTXO set | Whether every private key remains accessible |
| Dormant outputs | Time since an output last moved | Permanent loss or the owner's willingness to sell |
| Custodian and fund reports | Assets held for a named legal vehicle at a report date | A separate beneficial owner for each custody label |
Who holds it
Exchange-traded funds. A spot fund's bitcoin belongs to its trust or other legal vehicle; its shareholders hold an interest in that vehicle's net assets, and a custodian controls keys on its behalf. BlackRock sponsors IBIT but should not be described as personally owning every coin in its trust. An ETF sponsor, trust, custodian and shareholder are therefore not additive holder categories. The trust's SEC reports supply dated asset and custody figures; any September 2026 cross-fund total needs a reconciled, same-date issuer-level dataset before publication. iShares Bitcoin Trust ETF, June 2026 Form 10-Q.
Corporate treasuries. Public companies can report Bitcoin as a treasury asset, but company counts, coin totals, financing structures, share premiums and dispositions must be taken from same-date issuer filings and a non-overlapping entity inventory. The former 190-company/1.25-million total, 31 August Strategy balance, named-company league table, preferred-dividend funding and first-sale assertion are withheld here: the 29 June 2026 Strategy 8-K reviewed by this package did not locate the quoted 845,050 figure, and a share-price move cannot prove a coin sale. A custody wallet may also include client property rather than a company's beneficial assets. SEC Strategy filing, 29 June 2026; iShares Bitcoin Trust ETF, June 2026 Form 10-Q.
Governments. No reliable cross-country total follows from adding labelled government wallets. They mix seizures under litigation, finally forfeited property, state-company assets, central-bank holdings and private declarations, often at different dates. The US reserve order admits only eligible, finally forfeited Treasury bitcoin; the October 2025 Prince Group announcement was a civil forfeiture complaint concerning about 127,271 coins, not proof of final forfeiture or reserve inclusion. UK prosecutors report about 61,000 recovered coins subject to civil-recovery proceedings, not Treasury reserve assets. The disposition of coins seized in China's PlusToken case cannot be established from the cited tracker or the public record reviewed here; neither an extant 190,000-coin holding nor a sale should be asserted. Bhutan's state-linked mining is documented, but transfers from labelled wallets do not prove sales or a current balance. Ukraine's reported coins come from officials' private declarations, not public accounts; such declarations require their own verification. Sovereign-fund ETF shares and subnational holdings must also be kept separate from a national central-bank reserve. US reserve order; DOJ Prince Group complaint announcement; UK Crown Prosecution Service; Bhutan DHI; Ukraine NACP.
Stablecoin issuers and exchanges. An issuer balance sheet, affiliates, exchange custody and fund assets are different legal and economic categories. The former Tether June 2026 coin, gold and Treasury totals, affiliate claim and Mt. Gox estate deadline are withheld until the original issuer assurance/account statement and court order are reviewed with dates and consolidation scope. Wallet labels for Coinbase, Binance and other operators can combine proprietary and client assets; Coinbase-custodied trust coins must not be counted twice. iShares Bitcoin Trust ETF, June 2026 Form 10-Q, custodian structure.
Concentration. Address-level statistics are not a distribution of people: one person may control many addresses, while an exchange or ETF vault may hold assets for many people. Entity clustering can improve that picture but depends on attribution methods and incomplete coverage. Fund assets, exchange custody, corporate treasuries and government holdings should be reconciled at the beneficial-owner level before summing them. The proposed 8–9 million “institutional” total is withheld here because it does not establish that its categories are mutually exclusive.
Who runs the network
Mining. Hashrate is estimated from block timing and difficulty rather than counted miner-by-miner. Hashrate Index's 14 September 2026 report gives a 943 EH/s seven-day estimate and 928 EH/s thirty-day estimate under its method; those observations do not establish country shares, AI-related shutdowns, pool-control rights or attack costs. The former September pool percentages, template-producer share, chip-market shares, hashrate geography and Stratum V2 pledge-as-deployment claim are withheld pending dated block-template attribution, manufacturer shipments/installed-fleet denominators and country methodology. A miner-side template option under Stratum V2 is not automatic basic-protocol deployment. Hashrate Index, 14 September 2026 weekly roundup, “Hashrate & Difficulty”; Stratum V2 protocol overview.
Concentration in mining is a transaction-inclusion and reorganisation risk, not authority to issue coins that validating nodes reject. A pool can choose block templates and omit transactions from its own blocks; other miners may still include them. A sustained majority could delay inclusion or reorganise recent blocks, but its capability, cost and incentive depend on access to hardware, energy, existing hashpower, coin price, fees and the target transaction. The earlier $5 billion hardware estimate and assumption that an attacker must lose more than it gains are withheld without a specified attack model. State pressure on pool operators and compromised mining firmware are possible channels to investigate, not a verified account of a particular intervention.
Development. Developers may propose code and merge changes into a repository, but adoption by validating users is separate from repository authority or miner signalling. The former contributor/maintainer counts, donor totals, Core/Knots reachable-node shares, OP_RETURN dispute outcome and one-person-maintainer attribution need dated repository, funding and reachable-node records before being treated as current concentration observations. Miner block templates and user validity checks are different powers; no one software team can change other nodes' rules unilaterally. Bitcoin Developer Guide, “Block Chain”.
The next test. Quantum migration would require technical design and broad agreement among users, miners, wallet makers and custodians. BIP-360 remains a draft soft-fork specification: its proposed P2MR output removes Taproot's key-path spend and can mitigate long exposure for newly created outputs, but does not itself migrate old coins or fully solve short-exposure attacks. BIP-361 remains a draft informational proposal that depends on a future post-quantum signature design; its proposed migration phases would begin only after activation. Neither document establishes a Bitcoin-wide deadline or a settled treatment of unmoved legacy outputs. Trust assets, custodian balances and corporate treasuries overlap in some reports, so their holders' economic stakes also need reconciliation before quoting one aggregate share. BIP-360; BIP-361.
Corporations that benefit from control of supply or flows
The asset's supply cannot be manipulated; its price and its plumbing can be, and have been.
Market-manipulation claims. Historical enforcement orders, academic estimates, exchange-bot histories and reported-volume studies address different markets, periods and standards of proof. The former Tether/Bitfinex percentages, Mt. Gox bot quantity, industry fake-volume rankings and assertion that offshore perpetual futures set today's reference price are withheld here pending each primary order, dataset and venue/price methodology. A past enforcement finding cannot be extended to all current Bitcoin volume or taken as a mechanism proving each asset-price move.
Paper bitcoin. Custodial claims can exceed available assets when intermediaries lend, lose or misstate them, as earlier exchange and lender failures illustrate. A proof-of-reserves report is not automatically a financial-statement audit: depending on its procedures, it may omit liabilities, rights to assets or temporary borrowing, and a point-in-time result does not establish future solvency. A cash-created ETF share is an interest in a trust's net assets, not a promise of one whole bitcoin; creation and redemption mechanics alone do not prove an arbitrary share-to-coin inequality. The trust's reported assets, liabilities and custody arrangements must be assessed from its filings. PCAOB staff investor advisory; iShares Bitcoin Trust ETF Form 10-Q.
Potential beneficiaries and conflicts. Fund sponsors, custodians, exchanges, stablecoin issuers, hardware makers and market makers earn different fees or influence different steps; a reported revenue stream is not a protocol-validity power. The former Strategy premium/dividend, ETF-fee, Coinbase/Tether earnings, market-maker liquidity, Trump-family ownership/profit and conflict-of-interest bundle is withheld until named SEC, issuer, ownership and ethics records are reconciled for the same date. Executive reserve authority creates a public-policy question, but its magnitude cannot be quantified without a legally classified Treasury inventory. U.S. reserve order, §§2–3.
What each actor can control. A validating node rejects blocks that violate the rules it runs. A developer can propose code but cannot make another node install it; a miner selects transactions and valid blocks but cannot force a node to accept invalid issuance. Rule changes are possible through broad adoption or a fork, so “no entity can change the cap” needs the qualification that no actor can change other validators' rules unilaterally. Custodians can restrict client withdrawals and may control keys for trusts, companies and individuals; exchange access, lawful seizure, compromised keys and physical coercion are separate ways to constrain spending. A signature demonstrates key control, not clean beneficial title or complete liabilities. Concentrated markets can affect dollar pricing and liquidity without controlling the validity rules. These channels require separate, dated measurements before one government or company is assigned an absolute share of control.
Key takeaways
Bitcoin has a public issuance schedule, but the number of permanently lost coins and the effective trading float cannot be read from dormant outputs or wallet labels. Custodians, trusts and shareholders are overlapping layers, so their reported holdings cannot be added as independent ownership. Coin ownership does not confer a vote over consensus rules, while concentrated mining, custody, trading and development can affect access and market outcomes. These distinct forms of control need separate, dated measurements before a numerical concentration or liquidity conclusion is warranted.