Druckenmiller's $88M Crypto Backdoor: Why Q2 13Fs Show Funds Buying Miners, Not Bitcoin

The Q2 2026 13F season produced a headline that reads like a contradiction. Bitcoin fell roughly 14% during the quarter. Hedge funds cut their Bitcoin ETF exposure. And yet Stanley Druckenmiller's Duquesne Family Office — a manager with no history of loud crypto advocacy — showed up in the filings with about $88 million spread across two of the most crypto-levered equities on the tape.

That is not a contradiction. It is a rotation, and it is the most interesting thing buried in this quarter's data.

What Duquesne Actually Bought

Two positions, both new:

• Bitdeer Technologies (BTDR) — roughly 4.1 million shares worth about $64.7 million, at an average cost near $12.26 per share.
• Hyperliquid Strategies (PURR) — roughly 2.9 million shares worth about $23.1 million.

The pairing is instructive. Bitdeer is not a pure-play miner in the 2021 sense; it manufactures mining hardware and operates high-performance computing data centers across the US and internationally. That puts it in the same conceptual bucket as the power-and-compute infrastructure names that hedge funds have been accumulating all year — the picks-and-shovels layer underneath both AI and crypto.

Hyperliquid Strategies is the other kind of bet entirely: a publicly traded vehicle that accumulates HYPE tokens, giving equity investors indirect exposure to a specific protocol. It is a token wrapper with a ticker.

One is infrastructure. One is a levered token proxy. Buying both in the same quarter is not a Bitcoin thesis — it is a bet that crypto-linked equities re-rate faster than the underlying assets do.

The Split Nobody Highlighted

Zoom out and the Q2 filings show two institutional cohorts moving in opposite directions.

On the ETF side, roughly 1,900 institutions disclosed Bitcoin ETF exposure, down from about 2,000 the prior quarter — but aggregate holdings rose about 7.5%, from roughly 498,000 to 536,000 BTC equivalent. Fewer filers, bigger positions. The buying was concentrated in banks and quantitative managers; sovereign wealth funds and endowments largely sat still. Hedge funds, meanwhile, were net reducers.

So the discretionary money left the ETF wrapper. Where did it go? Into equities like Bitdeer, where Jane Street reportedly built a position north of $112 million during the quarter, and Citadel added as well. BlackRock, State Street, and Citadel all showed up in PURR.

This is the pattern worth internalizing: when institutions want beta to a theme but also want the ability to size aggressively, trade liquidity, and avoid the governance headaches of holding a commodity ETF on a 13F, they buy the operating companies instead. You can see the same behavior in how funds expressed the AI trade through power and cooling names rather than through the obvious semiconductor leaders. Our flow tracker makes those substitutions visible quarter over quarter.

Why This Is a Signal and Not Noise

There are three reasons to take this seriously rather than filing it under "hedge funds chase volatility."

First, the cost basis. Bitdeer at roughly $12.26 average was not a momentum entry. That is accumulation into a drawdown, during a quarter when the underlying asset was down double digits. Buying weakness in a levered proxy requires a view, not a chase.

Second, the overlap. When Jane Street, Citadel, and a family office with Druckenmiller's macro pedigree all land in the same mid-cap name in the same quarter, that is not coincidence — it is a shared read on the same setup. You can check how many top-tracked managers hold any given position on our overlap tool.

Third, the macro fit. Druckenmiller's broader Q2 book leaned toward a soft-landing posture with selective AI exposure. Crypto-adjacent compute infrastructure sits at the intersection of both: it benefits from easing financial conditions and from datacenter demand that has nothing to do with token prices. That is a coherent position, not a lottery ticket. Our macro consensus page tracks how these positioning signals aggregate across managers.

The Caveats That Matter

13F data is a snapshot as of June 30, disclosed six weeks later. In a category this volatile, that gap is enormous — these positions may already be halved or doubled. 13Fs also show only long US equity positions, so any hedges, shorts, or direct token holdings are invisible. A fund that looks aggressively long crypto equities on paper could be running a spread trade you will never see.

And Duquesne is a family office. It has no clients, no redemption pressure, and a risk tolerance most institutions cannot replicate. Sizing $88 million into two illiquid, high-beta names is a different decision for them than for a fund with quarterly reporting obligations.

The Takeaway

The interesting question after this filing season is not "are institutions buying crypto?" It is "which wrapper are they using?" Q2 answered that: the discretionary money moved out of the ETF and into the operating companies and token proxies, while the systematic and banking money did the opposite. Those two cohorts rarely disagree this cleanly. Watch which one is right when the Q3 filings land in November — you can follow every manager's changes as they publish on our investor directory.

---

Data sourced from public SEC 13F filings. Educational research only — not investment advice.