The $45 Billion Lesson Inside Situational Awareness's Options Comeback

The market's most watched comeback is being built in the options market. Situational Awareness — the from-nowhere AI hedge fund run by former OpenAI researcher Leopold Aschenbrenner — reached a peak of roughly $45 billion this summer before nearly blowing itself up, and it is now trading its way back through options rather than the borrowed cash that nearly killed it. That distinction is not a footnote about one fund. It is the whole lesson, and it applies to anyone considering a leveraged bet on a hot idea.

The collapse was about structure, not the story

Say what you want about Aschenbrenner's AI thesis — that capable machines would consume semiconductors, memory, data centers, and power by the truckload. It may well be right. But the fund died anyway, because of how the idea was packaged, not whether the idea was true.

Aschenbrenner built Situational Awareness on borrowed money. Reported leverage ran as high as 400%, and the book was concentrated in the names that would benefit most from an AI buildout: SK Hynix, Micron, SanDisk, CoreWeave, Nebius, the power and crypto-mining names around them. It worked spectacularly for a while — the fund was up more than 1,000% since its July 2024 launch and as much as 439% net through the first half of this year. Then, in July, the AI-infrastructure complex cracked. Core holdings fell 27% to 54% in a single month, the Nasdaq 100 dropped more than 10%, and SK Hynix lost about a third of its value.

Here is the mechanical part most people miss: the fund didn't lose because it was wrong. It lost because it was leveraged and correlated. When the longs fell, the margin cushion — the equity the fund had to maintain with its prime brokers — got eaten. Those brokers — Bank of America, Goldman Sachs, JPMorgan — demanded more collateral the fund didn't have. That is the margin call, and a margin call does not care what you believe. Before the open on July 30, the fund sold its entire public-equities book, longs and shorts, in a single block at a discount. Ken Griffin's Citadel bought it. Assets went from $45 billion to roughly $10 billion in a matter of weeks.

The proof is in what happened next

The strongest evidence that this was a structure problem and not a thesis problem came the same day. Once the forced seller was out of the picture, the high-beta names that had just been dumped jumped 20% to 27% intraday — Nebius up 27%, IREN up 26%, Bloom Energy up 25%. The "liquidation discount" simply repriced. The collateral was fine once nobody was being forced to sell it at any price. That is leverage at work: it doesn't just magnify your gains, it converts a bad month into a fire sale, and it lets the market set the price for you at the worst possible moment.

Now the part of this that is genuinely worth staring at: the fund's protective positioning didn't save it either. In the first quarter of this year it had put protection layered over the semiconductor complex — more than $1.5 billion of puts on Nvidia, over $2 billion on the VanEck Semiconductor ETF, over a billion each on Broadcom and Oracle, roughly $969 million on AMD. Those were meant to hedge. They failed, because they were long the same complex they were supposedly protecting. When semiconductors gapped down, the puts and the longs moved together. A hedge only helps if it isn't correlated with the thing you're losing money on; put the hedge on the same names you're already naked in, and you've bought theater.

What the options comeback means

So now the reborn fund — managing roughly $15 billion — is back in the options market. That is exactly the right move mechanically, and it is worth understanding why. With a bought option, the most you can lose is the premium you paid. There is no margin call, no broker demanding more collateral in the middle of a bad week, no forced sale. If you buy a call because you think AI compute is going to explode, and it takes longer than you hoped, your position bleeds out slowly through time decay — painful, but survivable. The mechanism that actually killed the fund cannot reach you through a long option the way it can through a levered cash position.

But don't mistake a different structure for safety. Options still cost you: you pay for the time, and you pay for the volatility, and if everyone expresses the same scarred idea through the same dealers' desks, the marginal buyer of semiconductors starts being an options trader rather than a cash buyer — a shift that changes how the complex behaves. And the deeper point about correlation hasn't gone away. Whether the position is a leveraged stock or a call spread, it is still one big bet on one big theme, and one big theme can still turn into one big drawdown.

The retail takeaway isn't about this fund at all. It's about the difference between holding an idea and holding a position designed to be force-liquidated. The same thesis that blew up a $45 billion portfolio — bought one name at a time with no leverage — would have been merely an uncomfortable year. Leverage is what turns a drawdown into a wipeout, and correlated leverage is what turns a portfolio of many names into a portfolio of one. Options can't fix that. They can only change the way the knife falls.