Who Actually Holds Bitcoin: 13F Filings Map $54 Billion of Institutional ETF and Treasury Exposure

Two and a half years after the SEC approved spot bitcoin ETFs in January 2024 [VERIFY: SEC approval order, Jan 2024], the quarterly 13F record finally answers the question the launch raised: who actually holds this stuff? The Q2 2026 filings show roughly 1,950 institutional managers reporting spot bitcoin ETF positions worth about $54 billion in aggregate [VERIFY: Q2 2026 13F filings, EDGAR aggregation], approximately 26% of the funds’ combined $210 billion in assets [VERIFY: Farside / issuer AUM data]. The other three-quarters belongs to holders the filings cannot see: retail accounts, managers below the reporting threshold, and non-US entities. What the visible quarter-by-quarter record does show is a clean two-act structure. Hedge funds built first and fastest, advisors arrived later and stayed, and alongside both, a parallel ownership channel, the corporate treasury company, accumulated roughly 1.05 million BTC [VERIFY: company treasury disclosures] that never touches a 13F at all.

The 13F lens, and what it cannot see

Form 13F is the only systematic public window into institutional bitcoin ownership, and its blind spots define the analysis. Managers must file only if they run more than $100 million in covered US-listed securities, and filings land up to 45 days after quarter end [VERIFY: SEC Form 13F rules]. The form reports long positions in US-listed instruments only: ETF shares are covered, but short positions, futures, offshore vehicles, and direct coin custody are all out of scope [VERIFY: SEC Form 13F instructions]. That last exclusion matters enormously for bitcoin, because the single largest institutional holder class, the treasury company, holds coin directly and therefore appears in the record only through its own equity. Reading 13F data as "institutional ownership" without those caveats overstates precision; reading the trend across ten consecutive quarters, however, is exactly what the form is good for.

The venue architecture also shapes what gets filed. IBIT trades on Nasdaq, FBTC on Cboe BZX, and GBTC on NYSE Arca [VERIFY: fund prospectuses], and Coinbase serves as custodian for 8 of the 11 funds launched at approval [VERIFY: fund prospectuses / Coinbase Institutional], a dependency that makes COIN a levered play on the entire complex regardless of which ticker an institution buys. The approval itself was the textbook regulatory catalyst: a single SEC order converted an asset most compliance departments could not touch into a line item any 13F filer could hold, and the filings that followed are the measurable record of that conversion.

The build timeline: hedge funds first, advisors later

The cumulative build is best read as two distinct regimes. In the first year, hedge funds dominated: by mid-2024 they accounted for roughly 45% of all 13F-reported spot ETF value [VERIFY: 13F holder-type classification, Q2 2024], and much of that exposure was not a directional bet at all. Funds were running the basis trade, long ETF shares against short CME bitcoin futures, harvesting the spread [VERIFY: CFTC Commitments of Traders, leveraged-funds positioning], and CFTC data shows leveraged-fund net shorts peaking alongside the Q4 2024 filing surge to roughly $27 billion of reported value [VERIFY: CFTC COT data; Q4 2024 13F aggregation]. When bitcoin drew down roughly 25% peak to trough in Q1 2025 [VERIFY: BTC price history], the basis compressed, the trade partially unwound, and reported value fell to about $22 billion [VERIFY: Q1 2025 13F aggregation], the only quarterly decline in the series.

The second regime is slower and stickier. From Q2 2025 onward, reported value climbed every quarter, from about $29 billion to $36 billion, $44 billion, $48 billion, and finally $54 billion in Q2 2026 [VERIFY: Q2 2025 through Q2 2026 13F aggregations], and the composition beneath the line inverted. Hedge funds’ share of reported value fell to roughly 28% by Q2 2026 while investment advisors rose to about 47% [VERIFY: 13F holder-type classification, Q2 2026], with the advisor count among filers rising and the hedge-fund count falling in every quarter since Q1 2025 [VERIFY: 13F holder-type counts]. The distinction matters for flow durability: a basis-trade position exists only while the spread pays, but an advisor allocation is a standing model-portfolio weight, rebalanced rather than liquidated. The $18 billion of net spot ETF inflows in the first half of 2026 [VERIFY: Farside flow data, Jan-Jun 2026] arrived predominantly through that advisor channel.

The ownership map: advisors, funds, sovereigns, treasuries

Mapped by holder type, the visible ownership base sorts into four blocks of very different character. Investment advisors are the largest 13F block, roughly $25 billion of reported value spread across hundreds of platforms [VERIFY: Q2 2026 holder-type aggregation], with wirehouse and RIA names like Morgan Stanley at about $1.9 billion, Wells Fargo at $1.1 billion, LPL Financial at $0.9 billion, and Hightower at $0.7 billion [VERIFY: respective Q2 2026 13F filings]. Hedge funds and proprietary traders hold roughly $15 billion [VERIFY: holder-type aggregation]. The sovereign-and-pension block is small, about $3.1 billion [VERIFY: holder-type aggregation], but carries outsized signal value. And the treasury companies sit apart entirely: Strategy’s roughly 641,000 BTC is worth about $71 billion at market [VERIFY: Strategy disclosures; derived at prevailing price], larger than the entire 13F-reported ETF complex combined.

The sovereign-and-pension block deserves its own paragraph despite its size. The State of Wisconsin Investment Board became the first US state pension to disclose a spot ETF position, in its Q1 2024 filing, and reported roughly $0.6 billion by Q2 2026 [VERIFY: SWIB 13F filings]. Mubadala’s roughly $0.9 billion IBIT position [VERIFY: Mubadala Q2 2026 13F] is the marker most allocators watch: sovereign wealth choosing a US-listed wrapper over direct custody validates the ETF as an institutional instrument, not merely a retail convenience. These positions are small against either institution’s balance sheet, but pension and sovereign filings function as permission structures for the next tier of allocators, which is why each new name in this block moves sentiment more than its dollar size implies.

The largest filers, one by one

Ranking individual filers requires reading each one’s business model, because a dollar of reported value means different things in different hands. Millennium’s roughly $3.1 billion across IBIT and FBTC [VERIFY: Millennium Q2 2026 13F] is substantially arbitrage inventory; Goldman Sachs’ $2.4 billion [VERIFY: Goldman Q2 2026 13F] largely aggregates wealth-management client accounts rather than a house view; Jane Street’s $1.4 billion [VERIFY: Jane Street Q2 2026 13F] is market-maker inventory that can flip sign week to week. The concentration is real but moderate: the top 10 filers account for roughly 30% of all 13F-reported ETF value [VERIFY: Q2 2026 aggregation], while the long tail of nearly 1,950 filers supplies the rest.

Treasury companies: the ownership channel 13Fs cannot see

The corporate treasury channel predates the ETFs by four years and still dwarfs them in coin terms. Michael Saylor initiated the model at MicroStrategy, now Strategy, with its first purchase in August 2020 [VERIFY: MicroStrategy Aug 2020 8-K], and the company has since accumulated roughly 641,000 BTC [VERIFY: Strategy 8-K disclosures], about 3% of the 21 million terminal supply [VERIFY: derived]. Marathon Digital holds roughly 50,000 BTC worth about $5.5 billion [VERIFY: MARA treasury disclosures], and the full cohort now spans roughly 140 public companies holding about 1.05 million BTC combined [VERIFY: treasury disclosure aggregators], near 5% of terminal supply [VERIFY: derived]. None of this appears on a Form 13F, because coin held in corporate treasury is not a covered security; the disclosure channel is the 8-K and the earnings deck, on the company’s own schedule.

For investors, the treasury company is a fundamentally different instrument from the ETF. An IBIT share is a custody receipt; an MSTR share is a leveraged claim on a corporate balance sheet that issues equity and convertible debt to buy more coin, and it trades at a premium or discount to its holdings that has itself become a traded signal. The comparison investors reflexively reach for is gold as a store of value, and Larry Fink’s own arc, from calling bitcoin an index of money laundering in 2017 to describing it as digital gold by the ETF launch [VERIFY: Fink public interviews], is the institutional adoption story compressed into one career. But gold’s ownership base has no equivalent of a single company holding 3% of terminal supply, which is where the analogy, and the risk analysis, must part ways.

Concentration, and how to position around it

The shape of institutional ownership is therefore a barbell with a visible middle. On one end, a long and lengthening tail of advisor allocations, individually small, collectively the largest and stickiest 13F block. On the other, extreme single-entity concentration risk: Strategy alone holds more bitcoin at market value than all 1,950 13F filers report combined [VERIFY: derived comparison], and the top 10 filers hold roughly 30% of the visible ETF value [VERIFY: Q2 2026 aggregation]. The tradeable expressions map cleanly onto that structure: IBIT, FBTC, and GBTC for the custody receipt at three fee and liquidity points; MSTR and MARA for leveraged treasury exposure; COIN for the custody-and-trading toll booth beneath 8 of the 11 funds [VERIFY: fund prospectuses]; HOOD for the retail flow that the 13F record explicitly cannot see [VERIFY: HOOD crypto transaction revenue]; and BLK for the fee stream on the largest fund itself.

The base case from here is continuation of the second regime: advisor share grinding higher each quarter as more platforms complete approval, hedge-fund share oscillating with the futures basis, and one or two new sovereign or pension names per filing season doing more for sentiment than for flows. The bear case is a repeat of Q1 2025 at advisor scale, a drawdown deep enough to test whether model-portfolio allocations rebalance into weakness or get pulled, plus the standing wrapper risk that a treasury-company premium compresses violently. Either way, the 13F record updates on a 45-day lag [VERIFY: SEC Form 13F rules], four times a year, and Ezana’s institutional-holdings tooling tracks the filer-level deltas each season; the quant workbench can backtest whether following the advisor build, rather than the headline flows, has actually paid. Watch the Q3 2026 filings in mid-November: the advisor share, not the price, is the number that tells you whether the institutionalization trade is still on.