Wall Street is chock full of remarkably smart people, impressively credentialed with advanced degrees from universities reputed to be the finest in the land. But as my late father repeatedly instilled in me, there is no substitute for experience. My father frequently railed about brilliant theorists who developed and promoted products based on grand ideological assumptions rather than well-worn reality.
His wisdom was validated once again this morning in The Wall Street Journal’s exclusive report that Ken Griffin’s Miami-based mega-fund Citadel has acquired the bulk of the public stock portfolio of Situational Awareness LP.
If you aren’t following the drama in Silicon Valley and financial circles, allow me to explain why this is yet another quintessential “you can’t make this up” Wall Street story.
Situational Awareness is an AI-focused hedge fund founded by Leopold Aschenbrenner—a former OpenAI researcher in his mid-20s who two years ago became one of Silicon Valley’s most celebrated artificial intelligence visionaries. Aschenbrenner made a massive splash with his viral, 165-page manifesto detailing the parabolic rise of AGI, positioning himself as the ultimate prophet of the multi-trillion-dollar AI build-out.
The investment world didn’t merely embrace Aschenbrenner’s AI thesis. It embraced Aschenbrenner. Investors wagered that extraordinary insight into artificial intelligence would naturally translate into extraordinary skill managing billions of dollars. Those are two very different disciplines.
Situational Awareness attracted massive pools of capital from high-profile backers, including the founders of Stripe, tech executives Nat Friedman and Daniel Gross, and Jane Street—the notoriously secretive quantitative trading juggernaut. Driven by eye-popping first-half returns and fueled by massive leverage provided by prime brokers like Goldman Sachs and JPMorgan, the fund’s total footprint rapidly swelled past $20 billion in assets under management.
Before becoming an AI prophet, Aschenbrenner was a prominent figure in the Effective Altruism movement. In 2022, he served on the five-person grant-making team for the FTX Future Fund—the philanthropic arm funded by Sam Bankman-Fried before FTX collapsed into bankruptcy and Bankman-Fried was sentenced to 25 years in prison.
The web of connections gets even tighter. Bankman-Fried and his convicted associate, Caroline Ellison, both began their careers as traders at Jane Street. While neither Aschenbrenner nor Jane Street have been accused of wrongdoing in the FTX affair, the overlap illustrates how remarkably small—and self-reinforcing—the AI ecosystem has become. The same circles of elite technologists, investors and intellectual influencers repeatedly validate one another’s ideas, often before the marketplace has a chance to do so.
Reality arrived in July.
Aschenbrenner’s fund placed an extreme, highly levered bet on the AI build-out—going long on memory chipmakers and AI suppliers like SK Hynix, Micron and Nebius while shorting software companies. When the market shifted and technology shares suffered a brutal drawdown, leverage transformed a bad month into a catastrophic margin call.
That’s when theory met experience.
Enter Ken Griffin.
Griffin began trading from his Harvard dorm room in the late 1980s before launching Citadel in 1990 with $4.6 million in capital. Over the intervening decades, he built the firm into a global financial powerhouse managing approximately $70 billion in assets and posting average annual returns of 19.2 percent after fees since inception.
The irony is delicious. The portfolio assembled by one of Silicon Valley’s celebrated AI visionaries ultimately landed in the hands of one of Wall Street’s most accomplished risk managers.
Citadel relies heavily on real human intellect rather than Silicon Valley’s artificial kind, employing more than 260 PhDs who process over 100 petabytes of data to make daily trading decisions.
Griffin recently opined that AI is a technology most business executives don’t actually understand. He shared a story of sitting around a table with corporate leaders, asking them to share specific examples of how AI was transforming their operations.
“I couldn’t help myself. I’m like, ‘Let’s go around the table and share stories about how AI is transforming your business,’” Griffin recalled. What he got back were four or five “incredible stories” touting productivity gains. But after digging below the surface, Griffin realized that “not one involved AI”—they were simply standard digitization and basic data optimization rebranded to catch the AI wave.
Griffin has been equally dismissive of claims that generative AI could consistently outsmart financial markets or produce outsized market returns. He recounted an instance where a colleague on Citadel’s commodities desk generated a market report using an AI engine.
“The first few sentences—like, wow, that’s really insightful,” Griffin said. “And then you go down below that, and it’s all garbage.”
While Morgan Stanley recently cut three percent of its global workforce to let AI absorb routine functions, Citadel’s market-making subsidiary took a very different path earlier this year by poaching Elan Luger, JPMorgan’s global head of high-touch equity sales trading.
The contrast is telling. Major institutional clients don’t want chatbots handling massive block trades. They demand execution overseen by living, breathing market professionals who understand risk, counterparty trust and human nuance—not algorithms.
High-touch equity trading remains such prized, lucrative territory that when Citadel Securities poached Luger to run its new division, JPMorgan retaliated. The bank promptly notified Citadel Securities that it was shutting off access to high-touch trade ideas and execution, unwilling to subsidize a direct competitor hunting on its home turf.
Aschenbrenner wasn’t unique. Corporate America has spent the past two years creating its own generation of AI oracles whose reputations often race ahead of demonstrated results.
My recent blog post, Misguided CEOs with Egg on Their Faces, highlighted how Ford’s Jim Farley, Starbucks’ Brian Niccol, Uber’s Dara Khosrowshahi, Meta’s Mark Zuckerberg and Accenture’s Julie Sweet all mistakenly assessed AI’s near-term capabilities. Savvy reporters might want to ask more executives to validate how AI is transforming their companies rather than simply accepting sweeping claims at face value.
GM’s Mary Barra is among the most aggressive, boasting that “nearly 90 percent” of the code written by GM’s autonomy team is generated by AI. That’s worth remembering because Barra previously positioned herself as one of Corporate America’s foremost EV visionaries, confidently predicting GM would overtake Tesla in North American EV sales, a feat she missed by a country mile.
Meanwhile, Dr. Sterling Anderson—the autonomy guru GM poached from Aurora with a compensation package worth up to $40 million—has been making the tech trade media rounds, boasting to Fast Company and other publications that GM’s AI tools could cut the automaker’s vehicle development timeline in half.
That sounds remarkably impressive, save for one conveniently omitted detail: Anderson has no background developing high-volume, internal combustion engine trucks or SUVs—the very gas-powered platforms that generate virtually all of GM’s operating profits and finance its massive stock buyback machine.
Rest in peace, Pops. You understood long before Silicon Valley’s AI visionaries captured Wall Street’s imagination that credibility should be earned—not assumed.