Leopold Aschenbrenner: 2,065% Returns, 4x Leverage, and a $16 Billion Forced Sale to Citadel

Leopold Aschenbrenner is no longer just a name attached to a forced sale. In Fernando Sánchez's telling, he is a case study in market hubris: a thesis that may have been directionally right about artificial intelligence, destroyed by a capital structure that could not survive being early and leveraged.

The Prodigy Arc

At 19, Aschenbrenner graduated first in his class at Columbia and joined OpenAI's safety team. In April 2024 he was dismissed over an alleged information leak — an accusation he has always denied.

After leaving, he published a 165-page viral essay, "Situational Awareness," arguing that artificial general intelligence would arrive by 2027 and that compute-infrastructure companies were the most undervalued assets in the market. The essay was not a trading note. It was a fundraising document. It helped him raise $225 million from prominent Silicon Valley backers — including Stripe's founders and GitHub's former CEO — to launch a hedge fund, despite no prior portfolio-management track record.

The fund took the essay's title: Situational Awareness.

Concentration Plus Borrowed Money

Because it was a private US hedge fund, it faced none of the diversification limits that constrain many retail or UCITS vehicles. Fernando highlights the operating conditions that produced the explosive ascent:

Extreme concentration. Up to 40% of the portfolio could sit in a single company.

Massive leverage. The fund ran approximately 4-to-1 leverage, borrowing from large banks including Goldman Sachs, JPMorgan, and Bank of America.

Unreal returns — until they were not. In 2025 the fund reportedly returned 2,065%. In the first half of 2026 it added another 439%. Assets under management ballooned from $225 million to more than $20 billion in under two years.

That is not a track record. That is a volatility machine wearing a track-record costume.

The July 2026 Fracture

The break came in July 2026 on a rumor: Meta Platforms might begin selling its own compute capacity — competing directly with Aschenbrenner's core infrastructure holdings (widely identified with names such as CoreWeave and Nebius). Those positions fell roughly 38% to 46% in a month.

At 4x leverage, a drawdown of that size is not an opinion. It is a margin call.

Between the night of July 29 and July 30, Aschenbrenner was forced into a fire sale of approximately $16 billion in assets at distressed prices to cover debts. The buyer at the bottom, in the irony that completes every leverage story, was Citadel — Ken Griffin's firm.

The same window that marked the SOXX low near 464 and the climax of Korea's leveraged-ETF liquidation was the window that transferred Aschenbrenner's public book to stronger hands. The cleanup Cava described as healthy was not abstract. It had a name, a notional, and a counterparty.

The Investor Letter as Makeup

After the disaster, Aschenbrenner sent limited partners a letter that Fernando's sources call financial cosmetics. It reported the fund still up roughly 80% year to date — while omitting what mattered:

• That performance came almost entirely from a private Anthropic stake, marked on paper and not freely liquid.
• The public equity book — where most participants' real money sat — had collapsed by about 90% in July.
• Many late investors cannot redeem because of a lockup into 2028.

On paper, the future still looked bright. In cash that could be sold, the present had been wiped out.

The Lesson That Survives the Narrative

Aschenbrenner's AGI timeline and compute-infrastructure bias may yet prove directionally correct. Meta's own Q2 commentary — refusing to rent scarce compute because selling intelligence earns higher margins than selling raw capacity — only underscores how contested and valuable that stack remains.

None of that rescued a book running 4x leverage with 40% single-name risk into a one-month rumor shock.

Fernando's moral is the line that belongs on every risk dashboard:

The market does not pay you for knowing the future. It pays you for being solvent when that future arrives.

Excessive leverage turned a potentially correct investment thesis into a mathematical ruin. The buyers who were solvent on July 29–30 — Citadel among them — are the ones who own the other side of that transfer.

For investors watching August's "healthy" bull market after the purge, the Aschenbrenner file is the human face of the cleanup: not a bubble bursting because AI failed, but borrowed money meeting a drawdown it was never designed to survive.

---

This analysis is based on Fernando Sánchez's commentary on Leopold Aschenbrenner's fund collapse, August 2026. For informational purposes only — not financial advice.