The alert went out before the candle closed.
July 17, 4:12 PM New York time. A cluster of AI momentum names — the same names every institutional deck has been worshipping since the first quarter — started bleeding in unison. Not a sharp drop. A slow, surgical leak. First the biggest names, then the satellite suppliers, then the options chain. By July 24, the leak had become a structural break. And somewhere in that chaos, a fund called Situational Awareness discovered that borrowed leverage has a memory, and that memory is violent.
The Financial Times reports that Situational Awareness has approached investors and lenders after a period of heavy borrowing amplified its losses during July's AI stock sell-off. That is the polite, boardroom version. The less polite version: the same trade that minted paper millionaires for eighteen months just revealed how much of that wealth was built on someone else's balance sheet. We didn't just watch the chart, we lived it. We watched the funding spreads, the margin desks, the quiet shifts in the options flow. And now the rest of the market gets to see what was hiding in plain sight.
Let's be clear about what Situational Awareness actually is. It is not a crypto fund. It is not a traditional asset manager. It sits in the hybrid zone that my old trading desk used to call 'the place where smart people go to blow up quietly.' A research-heavy operation with an AI-native investment thesis. It borrows capital, concentrates into frontier AI names, and tells itself the narrative is strong enough to survive any drawdown. The FT says the firm is now in talks with investors and lenders. Translation: the margin clerk has already knocked.
This is not a story about one fund. This is the first visible crack in the facade of an 'unleveraged' AI bull market. Because here's what the mainstream headlines will miss: the AI rally has been running on borrowed liquidity since long before July. Not just this fund. The entire complex. From ETFs that use derivatives to hit their exposure targets, to quant shops that lever up the same five names, to the retail trader who bought calls because a YouTuber said so. The difference is that most participants don't see the plumbing. But when a single fund needs to approach lenders after a month of losses, the plumbing just became visible. And once the plumbing is visible, trust starts to leak.
The fund's name is a gift to satirists. Situational awareness means knowing what is actually happening around you. A leveraged fund with that name just discovered that it did not know its own balance sheet. That is not just a bad joke. It is the thesis of this entire piece. The people running the AI trade are not the only ones missing the signal. The lenders are missing it. The prime brokers are missing it. The ETF sponsors are missing it. And the retail crowd, as always, is the last to know.
Let's pull the hood open.
The Mechanics of Borrowed Amplification
The core error isn't that Situational Awareness borrowed money. Borrowing is not a sin. Every serious trader has done it. The error is believing that leverage is a tool for sizing up, when in reality it's a tool for losing faster. That truth is hidden in the math.
Say you have $100 million in equity. You borrow $200 million. Now you control $300 million in AI stocks. The trade feels strong. The trend is your friend. Then the AI complex drops 25% over three weeks. Your $300 million position becomes $225 million. You still owe $200 million. Your equity is now $25 million — a 75% drawdown, not a 25% drawdown. That is the amplification. That is why the FT story matters.
But the math gets worse when the lender starts asking questions. When a fund is down 75%, the lender does not say: 'Hang tight, the narrative will recover.' The lender says: 'Give me more collateral or we close the position.' At that point, the fund has four options: sell assets into a falling tape, raise new capital, negotiate a standstill, or find a lifeline. The FT's reporting says Situational Awareness has chosen the negotiation path and is exploring the capital-raise path. The market sees this as a rescue. I see it as a forced disclosure of the true risk profile underneath an entire category of AI-exposed balance sheets.
This is Archegos with an AI wrapper. I covered the Archegos collapse from a risk desk perspective in 2021. The names were different. The leverage was different. The liquidity was different. But the emotional sequence was identical. First, the fund gets confident because the winners keep winning. Then the drawdown comes in the least expected corner. Then the margin call arrives. Then the lender finds out that the collateral is mostly the same five stocks everyone else is selling. And suddenly, a one-fund problem becomes a market-wide vacuum.
The noise fades, but the pattern remembers.
The same pattern has played out in crypto every cycle. I lived through the 2017 Telegram sprint, when I was manually monitoring more than fifty channels and watching ICOs borrow hype from each other. I saw the same thing in DeFi summer 2020, where protocol treasuries were leveraged against their own native tokens. And I saw the final act in 2022, when the FTX collapse vaporized capital that was, in reality, just a promissory note from one entity to another. The instruments change. The math doesn't.
What July's Tape Actually Showed
Let's get specific about the sell-off. The July AI drawdown wasn't a normal correction. It wasn't a news-driven crash, like a weak earnings report or a Fed surprise. It was a liquidity-driven cascade. That's a crucial distinction.
In a normal correction, price declines because participants reassess value. In a liquidity cascade, price declines because participants are forced to sell, regardless of value. The tape doesn't care about your thesis. It cares about your margin ratio.
Look at the daily price action in the AI complex in mid-July. There were days when the biggest name would open flat, rally two percent, and then, without any apparent catalyst, reverse into the red. Spot volumes were elevated, but not panic-level. Options implied volatility stayed too calm. That's the signature of a leveraged book quietly deleveraging in the background. The alert went out before the candle closed. But only the people watching the funding channels — the borrowing desks, the repo desks, the prime brokers — could see it.
I was in Dubai during that week. I had no direct position in the AI trade. But I was watching the same liquidity signals I've watched in crypto for years: the basis between spot and futures, the utilization of margin lending, the sudden quiet in high-touch execution. What I saw did not look like an ordinary reallocation. It looked like someone was selling into strength because they had to, not because they wanted to.
The FT story confirms that suspicion. Situational Awareness was one of the entities selling because it had to. And the fact that this fund was caught this unprepared is the irony that should make every investor pause.
The Hidden Debt Snowball
Here's the part that is even more uncomfortable. The AI trade is not a single position. It's a stack of debt layers. Layer one is the direct borrowing of a fund like Situational Awareness. Layer two is the margin lending done by prime brokers to dozens of smaller funds with similar concentrations. Layer three is the synthetic exposure embedded in ETFs and structured products. Layer four is the options market, where dealers are forced to hedge delta by buying or selling the same underlying stocks. Each layer looks safe on its own. Each layer is transparent in isolation. But together, they create a debt snowball that rolls faster as the price falls.
This is where I want to use a term from my cybersecurity training: horizontal privilege escalation. In an attack, you don't need to break into the main vault. You compromise one user, then use their access to move sideways across the network. The leveraged AI trade is the same. You don't need the central lender to fail. You just need one fund's margin call to be big enough that the lender needs to reassess its other clients. Once that happens, the lender tightens terms for everyone. And everyone is forced to sell a little bit more. That is horizontal leverage escalation. The leak moves sideways before it moves down.
The FT report is the first public notification of that escalation. The fund approached its investors and lenders. That is not just an event. It's a protocol change in the relationship between every borrower and every lender in the AI complex. The terms are about to get worse for everyone.
The Math Nobody Runs Until It's Too Late
Let's do a more detailed scenario. Suppose Situational Awareness controlled a $500 million book built on $150 million of equity and $350 million of borrow. The AI complex drops 20%. The book loses $100 million. Equity falls from $150 million to $50 million. That's a 67% equity drawdown. The leverage ratio, which was 3.3x at the start, is now effectively 9x because the debt hasn't changed while the equity has collapsed. The lender doesn't care about the initial ratio. The lender cares about the current ratio. And a 9x ratio is unacceptable. That's why a fund that was 'fine' in June becomes a distressed borrower in July.
The math gets uglier when you consider the options market. If the fund used calls or put spreads to gain exposure, the dealer on the other side is hedged in the underlying stock. When the stock falls, the dealer sells stock to reduce delta. That selling pushes the stock lower. That lower stock forces other leveraged participants to sell. The cycle feeds itself. It is the exact same loop you see in crypto when a liquidation engine starts cascading through a decentralized protocol. The code is different. The behavior is identical.
In my crypto audits, I've watched this loop play out in real time on lending protocols like Compound and Aave. The liquidation starts when the health factor drops below one. The liquidator buys the collateral at a discount. The sale pushes the price down. The next account gets liquidated. The cascade continues until there are no more undercollateralized positions. The market finds the floor only after the forced sellers are done. The AI market has a similar liquidation engine, but it's not on-chain. It's in the prime brokerage agreements. It's in the ISDA contracts. It's in the quiet conversations between lenders and borrowers. The engine is slower, but the output is the same.
This is why the FT story is so important. It's the first time we're seeing the AI market's liquidation engine in action. The engine is not located in a smart contract. It's located in a boardroom. But it's still moving the price.
The Lending Side Is the Real Story
Here is the part that nobody wants to talk about. The firms that lent Situational Awareness the money were not passive victims. They were active participants in the leverage game. They underwrote the concentration. They priced the collateral as if AI stocks could only go up. And now, instead of taking the loss, they're reportedly at the table negotiating.
Why? Because if they force a full liquidation, they have to mark the loans to market. And if they mark the loans to market, they have to admit that the collateral was overvalued. That's a problem for their own balance sheets, their own investors, and their own regulatory capital. So the rational move is to kick the can. Restructure. Extend. Pretend.
But here's the contrarian angle: the lenders might actually be the smarter side of this trade.
Think about it. They know the fund's positions. They know the collateral. They know exactly what a continued drawdown would do. If they're willing to negotiate rather than force liquidation, they're signaling that they believe the underlying assets are still worth more than the loan amount over a longer time horizon. In other words, the lenders are telling you that the AI trade is not dead — it's just overleveraged. From static streams to living liquidity. The capital that was moving through the system is now, suddenly, very alive to the dangers around it.
The market will read the FT headline as bearish. I read the negotiation as a bullish signal for the underlying assets, and a bearish signal for the leverage structure around those assets. The assets survive. The leverage gets repriced. And anyone who is stuck on the wrong side of that repricing gets wiped out.
This is the same dynamic that played out in crypto in 2022. The underlying technology — the chains, the protocols, the code — didn't die. But the leverage built on top of it did. Celsius, BlockFi, FTX. Those weren't technology failures. They were capital structure failures. The same is happening in AI. The technology is fine. The capital structure is fragile. And the first visible fracture has just been reported.
Why This Matters for Crypto Traders
Now, you might be asking: what does an AI fund's margin call have to do with a crypto audience? The answer, from my seat, is everything.
The AI trade and the crypto trade are now deeply entangled. Not because Bitcoin trades in sync with Nvidia, although it often does. Not because the same macro flows move both. But because the same levered, narrative-driven, momentum-chasing capital is rotating between them. When AI stocks start to bleed, that capital doesn't go to cash. It goes to the next high-beta narrative. And lately, that has meant crypto.
I saw this in real time in July. As the AI complex sold off, certain crypto sectors — the ones with their own AI narratives, like decentralized compute, data availability, and GPU-tokenized schemes — suddenly saw volume spikes that had nothing to do with fundamentals. The narrative rotated. The leverage didn't disappear. It migrated.
This is the liquidity fragmentation problem resurfacing. In crypto, we're supposed to believe that liquidity fragmentation is a technical issue that requires new cross-chain protocols, new bridge standards, new infrastructure to unify it. But from where I sit, liquidity fragmentation isn't a technical problem. It's a behavioral one. Capital doesn't get fragmented by technology. It gets fragmented by fear and greed in identical measure. The same hundred million that was buying AI calls in July is now buying something else in August. The patterns look different. The balance sheets are the same.
When a leveraged AI fund like Situational Awareness has to approach lenders, it sends a signal to every leveraged participant in every market: the era of cheap, passive, confidence-based borrowing is ending. That signal is not confined to the AI complex. It moves into crypto, into real estate, into every asset that was priced off rising confidence. And in a bear market, that's the kind of signal that starts avalanches.
Spot-Check: Reading the Next Distress Before the Headline
Let me give you a few operational flags I'm watching. These are the kinds of things that don't make the FT front page but tell you more than any headline ever will.
First, the yield on secured funding. If the cost of borrowing against AI equities jumps relative to the cost of borrowing against Treasuries, the leverage in the system is getting expensive. That's a leading indicator. If you're not watching funding spreads, you're trading blind.
Second, the balance sheet of the lenders themselves. If a major prime broker or specialty finance company reports an unexpected increase in 'unrealized losses on lending collateral,' you should parse that immediately. It means the loans are in trouble but haven't been marked down yet. The accounting can hide the flow for a quarter, but the flow is already there.
Third, the correlation of drawdowns. Watch whether AI stocks and crypto start falling on the same days, in similar percentages. That's the smoking gun of shared leverage. When two unrelated assets move in lockstep under pressure, it means one margin desk is selling both to cover its own positions. We saw this in March 2020. We saw it again in late 2022. If the correlation spikes now, the market is telling you that leverage is hitting the same exit door.
Fourth, the 'rescue' narrative. Whenever a fund approaches investors and lenders for a lifeline, pay attention to the terms. If the rescue comes with harsh covenants — forced deleveraging, restricted redemptions, collateral haircuts — the pain is far from over. If the rescue comes with vague language about 'strategic options,' the pain is worse than it appears. Situational Awareness's approach, per the FT, is in its early stages. That means the terms haven't been set. That means the uncertainty is still elevated.
I've run this checklist in crypto for years. It kept me out of some bad positions and, honestly, it got me into some good ones. The key is not to predict the future. The key is to know what to watch in the present. The noise fades, but the pattern remembers.
The Contrarian Angle: What Everyone Gets Wrong
The conventional take on this FT story is straightforward: leveraged AI fund blowup, market risk, everyone should be careful. But the contrarian take is more nuanced, and I think it's more useful.
First, the fact that Situational Awareness is approaching investors and lenders — not just quietly liquidating — suggests that its positions are not as toxic as they could be. If this were a true collapse, we'd see a forced sale of hard assets and a cascading markdown. Instead, we're hearing about negotiations. The management team believes they have time. The lenders seem to agree, at least for now. That tells me the underlying assets — the AI names themselves — still have buyers at some price. The problem is isolatable, at least for the moment.
Second, the narrative framing is wrong. Everyone will say 'AI stock sell-off causes leveraged losses.' The better framing is 'borrowed confidence is the culprit.' The AI stocks didn't fail. The leverage failed. In a market where every major rally is amplified by borrow, the correction was always going to be amplified too. This is not a problem with AI. It's a problem with the financial infrastructure around it.
Third, and this is the one that gets me called a contrarian at every dinner, there is a decent chance this ends up being a buying signal for the very same AI names that got sold off. Here's why. Once the leverage is restructured or removed, the remaining holders of those assets have stronger hands. The forced sellers leave the book. The marginal buyer stops being a leveraged chaser and becomes an actual allocator. That's how bottoms are made. Not when the news is good. When the weak hands have been shaken out.
I watched this exact sequence in crypto through the collapse of the Terra ecosystem. The immediate reaction to Terra's death was to sell everything. Five months later, a lot of the infrastructure that survived the death spiral was the foundation of a genuine market recovery. The leverage died. The assets didn't.
So when the FT reports that Situational Awareness has approached investors and lenders, don't just read it as a warning. Ask yourself: what if this is the moment when the AI trade's weakest leverage is finally being removed? What if this is the signal that the last forced seller is at the table? In a bear market, the best contrarian positions are built when the margin calls are loud, not when the recovery is obvious.
The Hidden Risk: What Comes After the Rescue
Now, before you go out and buy the dip, let me put the danger on the table.
The rescue of a leveraged fund is a two-stage process. Stage one is the negotiation. Stage two is the post-rescue deleveraging. Most market participants only price stage one. They see 'the fund is being rescued' and think the selling is over. But the worst selling often comes in stage two, when the fund has accepted that it must reduce its footprint. That's when the true portfolio liquidation happens — quietly, methodically, into any bounce.
The 2022 crypto bear market was full of these deferred liquidations. A lender would announce a 'constructive engagement' with a borrower, the market would rally, then two weeks later the borrower would file for bankruptcy after quietly selling everything in the interim. The timeline matters more than the headline. The announcement is not the end. It's the beginning of the end.
I would also watch the second- and third-order effects that this story will have on other leveraged participants. When one fund is seen approaching lenders, every prime broker starts auditing its other clients more aggressively. Margin requirements get tightened. Lending spreads get wider. Collateral haircuts get bigger. This is the 'risk-off in the background' effect. The FT story may be about one entity, but the consequences are about everyone.
This is the point where I remind myself of a lesson from my cybersecurity background. The best hacks don't target the system directly. They target the trust between the system's components. A margin call isn't a hack. But it follows the same logic: compromise one node, and the entire network begins to question every other node. The result is not just a single failure. It's a cascading reduction in trust, which in markets is equivalent to a cascading reduction in liquidity.
From static streams to living liquidity — and then, suddenly, to dried-up liquidity. That's how a story like this ends if the restructuring fails. Not with a whimper. With a funding vacuum.
My Personal Read
I've been in this industry long enough to stop predicting and start preparing. The Situational Awareness situation is not a crash prediction. It's a warning light. And warning lights are useful only if you're already paying attention to the dashboard.
Based on my audit experience across DeFi protocols and my time monitoring market structure across asset classes, I'll tell you exactly what I'm doing. I'm reducing exposure to assets that are dependent on borrowed money finding a home. I'm watching funding spreads like a hawk. I'm keeping more dry powder than my ego would prefer. Shiny objects distract, but dry powder preserves. The AI narrative will have more shiny objects by the time this cycle is over. The question is whether you'll have the capital to buy them.
I'm also checking the 'trust the code, verify the art, ignore the hype' line. In crypto, the code is the contract. In AI stocks, the 'code' is the balance sheet. Verify the balance sheet. Ignore the hype. The hype says AI is the future. Fine. The balance sheet says the future was borrowed. Also fine. Both things can be true simultaneously. The problem is when you confuse the hype for the balance sheet.
The FT's reporting on Situational Awareness is exactly the kind of story that gets dismissed as 'one bad actor' or 'idiosyncratic risk.' It's not. It's the tell-tale heart of a market that has been trading on borrowed time. The noise will fade. The pattern will remember. And in six months, when you look back at this moment, you'll either have used it to reposition or you'll be part of the noise.
What I'd Do if I Were on the Other Side
Let me step into the shoes of the lenders for a moment. If I were the counterparty holding the loan to Situational Awareness, what would I be asking for right now?
First, I'd want a complete list of every position, including the ones that haven't been marked to market yet. I'd want to see the collateral in the context of the entire market's cross-correlations. I'd ask what happens if Nvidia drops another 10% while Microsoft drops 8% and the crypto AI narrative rotates back into a drawdown. I'd stress-test the fund's book the way I stress-tested smart contracts in DeFi audits. The math doesn't care about the story. The math cares about the liquidation price.
Second, I'd ask for a funded deleveraging plan before I put in a single new dollar. Not a plan that says 'we'll wait for the recovery.' A plan that says 'these are the positions we can exit without moving the market too much, and this is the order in which we will exit them.' That gives the lender control over the path of the liquidation. Control is the only thing that matters when the market turns against you.
Third, I'd ask for a personal guarantee from the principals. That might sound harsh, but it's the only mechanism that aligns incentives. If the managers believe their own thesis, they should be willing to put their own net worth on the line. If they're not, then their thesis is just a pitch. Trust the code, verify the art, ignore the hype. The code of finance is the collateral and the guarantee. The art is the PowerPoint.
I'm not saying that this specific rescue will follow that path. I'm saying that if it doesn't, the signal is even more bearish than the headline suggests. The terms of the rescue will tell you everything. Most market participants will only see the announcement. I'll be reading the footnotes.
The Deeper Pattern: Every Bull Market's Favorite Trade Dies the Same Way
Let me zoom out for a second. Every major bull market produces a favorite trade. In 1999, it was any stock with a dot-com suffix. In 2005, it was housing. In 2017, it was ICOs. In 2021, it was DAO treasuries and NFT collections. In 2024, it's AI. The assets change. The leverage mechanics don't. And the endgame is always the same: the crowd discovers that the trade was never really about the asset. It was about the borrow.
The AI trade was always a borrowing trade. The companies are, in many cases, genuinely transformative. But the institutional structures around them — the leveraged funds, the structured products, the collateralized lending agreements — are just modern versions of the same debt machines that have been blowing up for centuries. When the crowd believes the story will never end, the debt grows quietly. When the story stalls, even briefly, the debt becomes the story.
Situational Awareness is just the first name on the list. There will be others. Some will be in AI. Some will be in crypto. Some will be in private credit. The reason I'm writing this now is because the FT report is the first time a leveraged AI fund has publicly approached lenders for a lifeline. That is the kind of moment that marks the beginning of a cycle, not the end. The alert went out before the candle closed. It will go out again.
The Takeaway: Watch the Next Margin Call, Not This One
Here's my honest, forward-looking read.
The Situational Awareness situation is the first public crack, but it won't be the last. The AI trade has been the world's favorite crowded trade for two years. Crowded trades don't end with one fund approaching lenders. They end when the leverage underneath them has been completely repriced, or completely destroyed. We are at the very beginning of that process. That means the next margin call will be more important than this one. The next name that approaches investors and lenders will tell you whether this is an isolated incident or a systemic shift.
Watch the lending desks. Watch the funding markets. Watch the correlation between AI stocks and crypto on down days. And most importantly, watch the language of the rescue. If the coming restructurings are fast and decisive, the bottom will come quickly. If they're slow and negotiated, the pain will drag on. The market's memory is long. The leverage's appetite is short. The pattern remembers, even when the narrative doesn't.
The alert went out before the candle closed in July. It will go out again. The only question is whether you'll hear it before your own margin ratio is the one being tested.
