Situational Awareness’s reported 67% July loss: what leverage changes in an AI portfolio

A positive year can contain a devastating month

According to Reuters, the portfolio had gained 439% by the end of June and was still up about 80% for the year after July. Those reported returns can coexist with a 67% monthly loss.

An illustrative portfolio starting at 100 would reach 539 after a 439% gain. Keeping 33% of that value after the decline gives 539 × 0.33, or approximately 178. The resulting gain of about 78% is broadly consistent with the rounded reported year-to-date figure. This is an arithmetic illustration, not a reconstruction of subscriptions, withdrawals or the fund’s accounts.

An investor entering near the June peak would have a very different experience from one invested at the start of the year. A positive year-to-date headline does not erase that later investor’s loss. After a 67% decline, the remaining value of 33 needs to rise by about 203% to regain 100.

That distinction matters when judging a spectacular track record. Entry date and drawdown can explain more about an investor’s experience than the return shown from the fund’s chosen starting point.

Leverage can shorten the time available to recover

Borrowing increases exposure relative to an investor’s capital. When asset prices fall, losses consume a larger share of that capital; financing terms can then require more collateral or smaller positions.

A manager may still believe a holding is undervalued. That belief does not provide the cash required by a lender. If financing headroom runs out, a portfolio can have to sell before its longer-term thesis has time to develop.

This is a general mechanism. The reports describe Situational Awareness reducing leverage, but its precise borrowing multiple, collateral arrangements and sale terms have not been disclosed. They do not allow outsiders to identify every margin call or establish the exact order of trades. Treating a plausible financing sequence as a fully documented account would overstate the evidence.

Liquidity also changes the choices available under pressure. A listed security may be easier to sell than a private investment, but a large position can move the market as it is unwound. A portfolio-wide transaction can reduce that execution problem, although its economic outcome depends on the price and contractual terms.

Different sectors can share the same risk

Situational Awareness’s website describes investments across technology, energy and other businesses affected by AI. Semiconductors, electricity suppliers, data centers and software occupy different industries. They can nevertheless depend on a common assumption: substantial AI spending will continue and produce attractive returns.

If investors lower that expectation, several holdings can fall together. Sector variety then offers less protection than it appears to. Expensive valuations and crowded positioning may reinforce the correlation.

This is an explanation of potential shared risk, rather than proof that every holding suffered for the same reason. Reconstructing the portfolio’s actual losses would require positions, hedges and trading records beyond the public information available here.

Why Form 13F cannot reveal the leverage multiple

Quarterly SEC Form 13F filings cover specified reportable securities. They do not include short positions, and written options generally are not reported. They also do not provide a complete account of cash, private holdings, borrowing or transactions during the quarter.

Adding disclosed long positions therefore cannot establish a fund’s net asset value, effective leverage or total loss. It can identify parts of the portfolio at a reporting date, but it is insufficient to validate precise exposure figures circulating on social media. The SEC’s Form 13F guidance explains the reporting limits.

What the Citadel transaction does—and does not—show

Reuters and The Wall Street Journal reported the transfer of the public-equity portfolio to Citadel. Selling the book is different from closing the investment firm; the reports indicated that Situational Awareness retained other assets and continued operating.

Without disclosed transaction prices and financing terms, outsiders cannot determine whether Citadel paid a substantial discount, how much loss the sale crystallized, or whether “rescue” accurately describes the agreement. There is also no basis in the cited information for claims that Citadel deliberately caused the losses.

Large sales can affect individual securities, but that possibility does not show that one fund caused the entire July correction in AI stocks. Changes in growth expectations, valuation, profit-taking and other investors’ positioning would need to be examined alongside its activity.

The questions that survive the headline

Portfolio issueUseful question
Time horizonHow long can the position be held if the thesis takes longer to work?
Common exposureWhich holdings depend on the same spending or valuation assumptions?
FinancingWhat would a hypothetical 20–30% adverse move do to collateral headroom?
LiquidityCan positions be reduced at a realistic price during stressed trading?
Track recordWhat were the drawdowns and returns for investors entering at different dates?

The stress range in the table is an example, not a measure of how much risk is safe. Financing and liquidity need to be considered together with the size and correlation of positions.

For any subsequent rebuilding of the public-equity book, the important disclosures would concern exposure limits, borrowing terms, loss-management rules and redemptions. Private-asset valuation and liquidity would matter too. A renewed strong return would say less about resilience than evidence that the portfolio could withstand another severe reversal.

Sources