Who Is Really Buying Bitcoin? The Changing Structure of Institutional Demand

Two institutional channels now hold similar amounts of bitcoin. They run on different engines, and in 2026 both have lost speed.

Last week's ETF figures were small and revealing. U.S. spot bitcoin ETFs closed the week of October 19 with $301 million of net inflows, which reads like a recovery until the daily detail arrives. About $1.07 billion came in from Monday to Wednesday, then $773.5 million left on Thursday and Friday, including $635.6 million on Thursday alone, the largest single day outflow since June. Bitcoin slid toward $80,000. In the same stretch, Strategy, the largest corporate holder, disclosed in its filing for the week to October 18 that it had bought 267 bitcoin for about $22 million and spent $193 million retiring its own preferred stock.

Two filings, two kinds of institution, and no obvious marginal buyer in either.

The usual answer to the question in the title is short: Wall Street, through ETFs. That answer was defensible for most of 2024 and 2025. It is harder to defend now, and the reasons say more about how institutional demand is constructed than any single week of flows could.

Two channels of similar size

Start with the stock of bitcoin rather than the flow. U.S. spot ETFs together hold about 6.4 percent of bitcoin's market capitalization, based on SoSoValue data from October 20, when combined net assets stood at roughly $106 billion. Public companies hold about 1.3 million bitcoin across 173 firms, which BitcoinTreasuries puts at 6.2 percent of the 21 million maximum supply as of mid October. The denominators differ slightly and trackers disagree at the margins, but the conclusion survives the rounding. Funds and corporate balance sheets each own roughly six percent of all the bitcoin there will ever be.

That matters because the common shorthand treats corporate treasuries as a rounding error beside the ETFs. They are far larger than that. Strategy alone held 848,517 bitcoin on October 18, according to its SEC filing, at an average cost of $75,446. That is a little over four percent of the maximum supply and about two thirds of everything public companies hold. The next two largest corporate holders, Metaplanet and Twenty One Capital, hold roughly 44,000 and 43,500 each.

The two channels are similar in size and different in construction. An ETF is a wrapper. Its assets rise when advisors, banks, funds and individuals decide to allocate and fall when they decide to leave, and the sponsor takes no view on price. A treasury company is a financing machine. It buys bitcoin when it can raise capital on acceptable terms, and the terms depend on how the market is pricing its own shares and preferred stock that week.

	U.S. spot ETFs	Public company treasuries
Holdings	About 6.4% of bitcoin's market cap	About 1.3 million BTC, 6.2% of the 21 million cap
What drives buying	Allocation decisions by advisors, banks, funds and individuals	Access to capital markets and the premium on the company's own securities
2026 net change	About +10,800 BTC through October 24	Strategy alone about +176,000 BTC from January 1 to October 18
Concentration	IBIT about 63% of fund assets	Strategy about 65% of holdings

Sources: SoSoValue, Darkfost, BitcoinTreasuries, Strategy SEC filings. Strategy's year to date change and fund concentration shares are DEXENTRAL calculations from those sources.

What the ETF channel did in 2026

Net flows into the ETFs have been close to flat this year. On-chain analyst Darkfost calculated that the 2026 cumulative net inflow had only just returned to positive territory, at 10,784 bitcoin, after most of the year in net outflow. Since launch in January 2024 the funds have taken in about $57.6 billion in cumulative net flows, according to Decrypt's tracker. October as a whole is still down about $314 million.

Flat flows can conceal a great deal of change underneath, and here they do. Analyses of second quarter 13F filings show institutional holdings of the ETFs rising 7.5 percent, from 498,389 to 535,723 bitcoin, in a quarter when the price fell 14.2 percent. The same analysis puts institutions at a record 44.2 percent of ETF ownership, up from 38.4 percent, with retail holders net sellers of roughly 75,800 bitcoin. A second methodology, used by CoinShares, counts professional filers at about 261,000 bitcoin, or 20.8 percent. The gap between 44 and 21 percent is a finding in its own right. Registered investment advisors file 13Fs, and CF Benchmarks data show advisor holdings of IBIT shares rising 145 percent over 2025, so a good share of what the data labels institutional is likely to be clients' money held through an advisor.

The composition shift that matters most sits in the hedge fund column. One compilation of 13F data puts hedge funds at 42.5 percent of reported crypto ETF value at the end of 2024 and 14.4 percent at the end of June, with banks and brokers at 22.7 percent and pension and sovereign funds at only 3.8 percent. Much of the earlier hedge fund position was probably the basis trade, which buys the ETF, sells CME futures and collects the spread. That spread was running around 15 percent annualized in late 2024, according to Velo data, and was about five percent when The Block reported on it in April. CME bitcoin futures open interest had by then fallen to a 14 month low as the trade unwound, and Binance had overtaken CME in open interest for the first time since 2023. Fast money left the wrapper. Slower money, much of it arriving through bank and advisor platforms, took part of its place. Morgan Stanley listed its own spot fund, MSBT, on April 8 at a 0.14 percent fee, the lowest in the category.

The corporate flywheel and its reversal

Strategy is the clearest example of a financing driven buyer, and its 2026 shows how quickly that engine can change gears. In its fourth quarter results the company said it raised $25.3 billion of capital in 2025 and acquired 41,002 bitcoin in January 2026 alone, lifting holdings to 713,502 on February 1. Working back from that filing, it began the year with about 672,500. By May 31 it held 843,706. In five months one company added more than 171,000 bitcoin, roughly sixteen times what the ETFs have absorbed net over the whole year to date.

The mechanism needs a market that pays a premium for the equity. When a company's shares trade above the value of the bitcoin behind them, selling new shares to buy more bitcoin raises the bitcoin per share, and the loop feeds itself. When the shares trade below that value, the same issuance dilutes existing holders. The multiple of market value to bitcoin held, known as mNAV, is the dial. On a basic measure tracked by SaylorTracker it fell to 0.58 on June 26, as Yahoo Finance reported. Strategy's own dashboard, which defines the ratio differently, put it near 1.1 in late October, according to TipRanks.

The company's conduct since June reads like a firm managing a balance sheet. On June 29 it announced a program permitting sales of up to $1.25 billion of bitcoin to fund its dollar reserve, and in the week to July 5 it sold 3,588 bitcoin for $216 million, its largest sale since it began buying in 2020, according to Bloomberg. Its filings for the three weeks ending October 18 show bitcoin purchases of $28.7 million, $21.5 million and $22.1 million, set against preferred stock repurchases of $154.1 million, $192.4 million and $193.1 million. A dollar reserve of $4.83 billion stands behind its dividend obligations. Holdings moved from 843,706 on May 31 to 848,517 on October 18, a net gain of about 4,800 bitcoin in four and a half months.

Retiring preferred stock that trades slightly below its $100 par value is a defensible use of capital, and the preferred shares closed near $99 on October 20. VanEck's Matthew Sigel has argued that trading around par could allow larger bitcoin purchases to resume. What the filings show is a change of priority, and a reminder that demand built on capital market access is only as strong as the market's appetite in a given week. DEXENTRAL examined the leverage side of this structure in an earlier piece on the largest corporate bitcoin position in history.

The dollars behind the demand

Buyers need funding, and two of the usual gauges have stopped helping. Stablecoin supply, often read as dry powder waiting to enter the market, has moved sideways for a year. Artemis data showed it flat near $300 to $305 billion from October 2025 to March 2026, after roughly doubling over the preceding fourteen months. It set a record of about $322 billion in May and has since drifted back to between $303 and $310 billion. Tether's USDT and Circle's USDC make up about 82 percent of the total. A supply measure that has stopped growing is a weak argument for a coming wave of purchases, and stablecoin balances serve many purposes beyond buying bitcoin.

The price of money has moved the other way. On September 16 the Federal Reserve raised its target range by a quarter point to 3.75 to 4 percent in a unanimous vote, its first increase in more than three years, and its statement said inflation remains elevated. The committee meets again on October 27 and 28. A non-yielding asset now competes with a policy rate that is rising instead of falling. No forecast is offered here about what the committee decides or how bitcoin responds. The narrower point is that the funding side and the demand side of this market are being tested at the same time.

Concentration at every layer

Whichever channel one examines, a few names carry most of the weight. On SoSoValue's October 20 figures, IBIT held about 63 percent of spot ETF assets, with Fidelity's FBTC at roughly 14 percent and Grayscale's GBTC at roughly 10 percent. DEXENTRAL's calculation from those shares puts the Herfindahl-Hirschman Index for the fund market above 4,000 on the standard 10,000 scale, which is concentrated by any conventional measure. Price competition has not dislodged the leader. MSBT charges 0.14 percent and Grayscale's Mini Trust 0.15 percent, against 0.25 percent for IBIT and FBTC and 1.50 percent for the original GBTC, according to NerdWallet's October 22 table. By an early September tally of Farside data, MSBT had gathered about $540 million of cumulative flows against roughly $64 billion for IBIT. Buyers in this market appear to value liquidity and familiarity over a few basis points.

Custody concentrates further. Bitwise has estimated that Coinbase holds the bitcoin behind 80 to 84 percent of spot ETF assets, including IBIT's. Fidelity self custodies through its own digital asset affiliate. IBIT's prospectus names Anchorage as an additional available custodian, with no current plan to move assets, as CryptoSlate reported. This concentration brings benefits: fewer moving parts, regulated counterparties, tight trading spreads. It also means that funds which look diversified share operational dependencies, and that a stress event would reach many of them at the same time. Risks of this kind are invisible in calm markets and dominant in the other kind.

The buyer side repeats the pattern. Roughly two thirds of public company holdings sit with one firm, so one company's capital structure shapes the behavior of an entire channel.

What this analysis cannot settle

Several caveats deserve a straight statement. Net flow figures conceal gross flows, and a flat year can contain heavy buying and heavy selling by different holders. 13F filings show quarter end positions roughly six weeks after the fact, cover only managers above a size threshold, and do not identify the final beneficial owner behind an advisor's account. Corporate holdings trackers differ by tens of thousands of bitcoin depending on coverage, so every total in this piece should be read as approximate.

The opposing view deserves its full weight. The ETF channel has rails that treasury companies lack: advisor platforms, retirement accounts, bank distribution, and no dependence on a share premium. If those rails keep widening, slow money could come to matter more than any single company's financing engine, and a quiet year in the flows would then be a pause. The data to date cannot rule that reading out. It cannot rule out the reverse either, in which allocation demand needed a rising price to attract it, and the current flat flows are the first honest look at what remains. Strategy could also resume heavy buying if its preferred securities hold par and its shares keep a premium.

What to watch

Four readings will sharpen the picture in coming weeks. Third quarter 13F filings, due in mid November, will show whether the bank and advisor share kept rising through a quarter in which the Fed began tightening. Strategy's Monday filings will show the ratio of bitcoin bought to preferred stock retired, and a rising ratio would suggest its equity funding is reopening. ETF flows around the October 28 rate decision will show how price sensitive the newer holder base is. And total stablecoin supply against its May record will indicate whether new dollars are arriving or existing ones are circulating.

The question behind the question

The title does not have one answer, and that is the finding. In 2024 the marginal buyer was often a fund harvesting a spread. For stretches of 2025 and early 2026 the most visible buyer was a company with a premium to spend. In recent months it has been a population of advisors and banks adding in modest increments while other holders sell. Each of those buyers had a different funding source, a different sensitivity to rates and a different way of failing.

A market whose demand migrates between sources is best read through those sources, one filing at a time, with less weight on the headline total. Demand that rests on a spread, a premium or a rising price is conspicuous when it arrives and scarce when it is needed. Demand that rests on a client file at an advisory firm is slower and quieter, and tends to still be there next quarter. Which kind is carrying the market today is a question the filings will answer sooner than the commentary does.

---

A reader who holds bitcoin through a fund owns shares in a trust that holds a claim on a custodian. A reader who holds directly owns the keys and carries the operational burden instead. Neither route is free of risk, and the right design depends on size, access needs and tolerance for operational detail. For those deploying serious capital, DEXENTRAL's Private Digital Asset Advisory covers custody setup and access control, operational security and failure mode analysis, and a documented operating framework, delivered as five hours of private consultation. All of this begin with a free 30 minute Welcome Assessment Call.