The subpoenas landed on the desks of America's four largest banks last week. The SEC wants trading timestamps and loan communications. Not from the fund that lost 67% of its value. From the banks that funded it.
This is the tell. The SEC isn't investigating a failed investment strategy. It's investigating the plumbing. And the plumbing is where the structural truth always lives.
Code does not lie, but it does leave traces. The traces here point to a systemic question: when a 24-year-old former OpenAI researcher borrows tens of billions of dollars to bet on AI concentration, who is responsible for the failure? The trader who placed the bet, or the institutions that handed him the chips?
The Context: A Collapse in Three Acts
Situational Awareness AI Fund launched with a simple thesis: AI infrastructure is the new oil, and concentrated exposure is the only rational position. The fund borrowed hundreds of billions of dollars from Bank of America, Citigroup, Goldman Sachs, and JPMorgan. It piled into Anthropic shares, bitcoin miners, and a handful of AI infrastructure names.
For a while, the thesis worked. Then the market turned. Margin calls came. The fund lost 67% of its value in weeks. Citadel stepped in and bought the fund's book at a discount. The founders walked away with their reputations bruised but their personal wealth largely intact.
The banks walked away with subpoenas.
Based on my experience auditing smart contracts in 2017, I recognize this pattern. When a system fails, the first instinct is to blame the most visible actor. But the structural fault is usually in the infrastructure. The SEC knows this. That's why they're asking about loan communications, not trading strategies.
The Core: What the Subpoenas Actually Reveal
The SEC's request for trading timestamps and loan communications points to two specific investigative threads. First, market manipulation. Timestamps reveal whether the fund was engaging in layering, spoofing, or coordinated selling. Second, credit fraud. Loan communications reveal whether the banks knew about the fund's leverage concentration and continued lending anyway.
The second thread is the dangerous one. If the banks knew the fund was over-leveraged and kept extending credit, they're not just counterparties. They're enablers. The SEC can pursue aiding and abetting charges under Section 20(e) of the Securities Exchange Act. The precedent is Archegos.
In 2021, Bill Hwang's family office collapsed with over $20 billion in losses. The banks that financed him — Credit Suisse, Nomura, Morgan Stanley — paid billions in fines and settlements. Credit Suisse alone paid approximately $500 million to US and UK regulators. The SEC's playbook is already written.
But there's a critical difference. Archegos used total return swaps to hide its positions. Situational Awareness appears to have used direct borrowing. That's more transparent, but it also means the banks had clearer visibility into the fund's leverage. If they saw the risk and didn't act, their liability is harder to escape.
The SEC's choice to investigate the banks rather than the fund is strategic. The fund is already dead. The banks are alive and regulated. By going after the banks, the SEC sends a message to every institution that finances leveraged AI bets: you are the first line of defense, and you will be held accountable.
This is the regulatory equivalent of auditing the smart contract instead of blaming the user who clicked the malicious link. The user made a mistake, but the contract had the vulnerability. The banks are the contract.
The Contrarian Angle: The Banks Are the Real Story
Here's the counter-intuitive part. The fund's collapse is not the story. The fund was a 24-year-old's concentrated bet that went wrong. That happens every day in markets. The story is that four of the world's largest banks — institutions with sophisticated risk management departments and decades of regulatory experience — collectively lent hundreds of billions of dollars to a single fund with a concentrated AI thesis.
Either the banks' risk models failed, or they were overridden. Both possibilities are troubling.
If the risk models failed, then the banks' AI-driven risk management systems are not fit for purpose. If the models were overridden, then there's a cultural problem where revenue generation trumps risk management. The SEC's investigation will determine which scenario applies.
There's also a third possibility, one that's more uncomfortable. The banks may have understood the risk perfectly well and decided to lend anyway because the fees were too attractive. This is the "yield is a symptom, not the cure" problem. The banks were chasing yield on AI-themed lending, and they ignored the structural risks.
In the red, we find the structural truth. The red here is the margin calls, the forced selling, and the 67% loss. The structural truth is that the banks' lending practices to AI-focused funds are dangerously loose.
The Takeaway: The AI Leverage Cycle Is Just Beginning
The SEC's investigation is not the end of this story. It's the beginning of a new regulatory chapter. The AI narrative is too powerful, and the leverage that's been built on it is too large, for this to be a one-off event.
Expect to see new disclosure requirements for AI-themed funds. Expect banks to face stricter scrutiny on their lending to concentrated AI positions. Expect the cost of leverage for AI funds to rise significantly.
Governance is the art of managing disagreement. The disagreement here is between the AI narrative's promise of transformative returns and the reality of concentrated leverage. The SEC is stepping in to manage that disagreement.
The question is whether the banks will learn from this or repeat the pattern. Based on my experience watching the 2022 bear market collapse, I'm not optimistic. The incentives haven't changed. The fees are still too attractive. The risk models are still too optimistic.
We build frameworks, not just tokens. The framework here is regulatory oversight of AI-themed leverage. It's being built now, in the wake of this collapse. The question is whether it will be strong enough to prevent the next one.
Trust is verified, never assumed. The SEC is verifying. The banks are being tested. And the AI leverage cycle is just beginning.
