You know that feeling when a story is so perfectly absurd that it almost feels fictional? Well, let me walk you through one of the strangest chapters in modern tech history, and why it still matters today.

Right now, in a federal prison cell, Sam Bankman-Fried is serving a 25-year sentence for one of crypto's biggest fraud cases. Meanwhile, the company he invested in with stolen customer deposits is now valued at over $380 billion and literally shaping Pentagon policy on AI weapons. That's not hyperbole—we're talking about Anthropic, which just closed a $30 billion Series G round in February 2026. The company's Claude model is deployed across U.S. intelligence agencies and military operations. And somewhere in the bankruptcy proceedings, there's an 8% stake that SBF acquired back in April 2022 for $500 million that would theoretically be worth over $30 billion today if it weren't frozen in liquidation.

That's a 60x return. In venture capital history, that's absolutely elite tier.

But here's what actually gets me about this story: it's not really about investment genius or even luck. It's about a very specific ecosystem, a philosophy, and how the same underlying logic can lead one person to build a $380 billion AI company and another to federal prison.

Let me take you back to mid-2010s San Francisco. There was this tight-knit circle of people—same shared housing, same parties, same reading list. They all believed in something called Effective Altruism. The basic pitch is elegant: charity shouldn't be based on feelings; it should be based on math. Every dollar should flow toward whatever mathematically maximizes good in the world. And a significant chunk of this movement became obsessed with one particular existential risk: uncontrolled artificial intelligence.

Dario Amodei, who would become Anthropic's CEO, was deeply embedded in this world. He was the 43rd person to sign the Giving What We Can pledge, committing to donate at least 10% of his income. He lived in a shared apartment with Holden Karnofsky, who co-founded GiveWell and Open Philanthropy—basically one of the most influential fundraisers in the entire EA movement. The third roommate was Paul Christiano, a key AI alignment researcher. Both Dario and Paul were technical advisors to Open Philanthropy at the time. Later, Karnofsky married Dario's sister and they all lived together for a while.

This wasn't just a friend group. This was the core network of a movement.

Anthromic's governance structure tells you everything about how deep this went. The Long-Term Benefit Trust—the company's most important governance body—has four members. Three of them come directly from the EA ecosystem: Neil Buddy Shah (former GiveWell managing director), Zach Robinson (CEO of the Center for Effective Altruism), and Kanika Bahl (CEO of Evidence Action). The early funding? Almost entirely from EA-aligned money. Dustin Moskovitz (Facebook co-founder), Jaan Tallinn (Skype co-founder), and yes, SBF—the three biggest financial backers in EA history all became early Anthropic investors.

So how did SBF actually find Anthropic? It wasn't some brilliant market insight. It was literally money flowing within a network toward projects that network had defined as important. EA's money going to EA's problems, solved by EA's people.

Now, SBF was operating under a specific EA philosophy called "earning to give." He left Jane Street, a Wall Street quant firm, to get into crypto. His public pitch was never about getting rich—it was about making as much money as possible so he could donate it to causes that would have maximum positive impact. The EA movement had decided AI safety was the ultimate existential risk. Anthropic's entire mission—"safely developing powerful AI"—was basically EA's standard solution to AI risk.

In May 2021, Jaan Tallinn led Anthropic's Series A with $124 million, and Moskovitz joined in. Then in April 2022, SBF stepped up and led the Series B round. He wrote a check for $500 million, which represented 86% of the entire $580 million round. The other investors? Caroline Ellison (CEO of Alameda), Nishad Singh (FTX Engineering Director), and people from Jane Street. This wasn't a diverse investor base—it was basically SBF and his network writing the check.

Here's where it gets interesting though. Dario Amodei is not stupid. He later said that SBF seemed like someone who genuinely cared about AI and AI safety, which aligned perfectly with Anthropic's mission. But then Dario said something crucial: he noticed "enough red flags." And this is the decision that probably saved Anthropic. They took the money—they needed it for computing infrastructure—but they structured it so SBF got non-voting shares and no board seat. Complete governance isolation.

Years later, after everything collapsed, Dario described SBF's actions as "far, far, far more extreme and egregious than I had imagined." Three "fars" stacked together.

But here's the uncomfortable question: if the red flags were so obvious they needed governance isolation, why take the money at all? You could argue that in early 2022, the AI funding environment was way less vibrant than today. Finding someone willing to write a $500 million check, regardless of concerns, was genuinely difficult. But there's something deeper here about how the EA ecosystem operates. Within that circle, the "cleanliness" of funding sources was never really the priority. What mattered was "effectiveness"—could the money help you do more good? The underlying logic was: making money is the means, doing good is the end, so how you make the money doesn't have to be so particular as long as the final output of good is large enough.

SBF took that logic to its criminal extreme. But at the time he invested in Anthropic, it seemed like just a radical philosophical choice, not an illegal one.

Then November 2022 happened. CoinDesk exposed Alameda's balance sheet. Changpeng Zhao announced he was selling FTT. FTX had a run on it. Within nine days, the entire empire collapsed. SBF got arrested, extradited, tried, and sentenced to 25 years in March 2024. All his assets, including that Anthropic stake, got frozen in bankruptcy proceedings.

During the trial, there was this one moment that the court actually excluded from the record. SBF's defense team tried to argue that his Anthropic investment proved he had foresight and wasn't just stealing—he was making investment decisions that multiplied in value. The prosecutor, Damian Williams, shut that down immediately: whether the investments were profitable was completely irrelevant to the fraud charge. You can steal someone else's money, invest it brilliantly, and still be guilty of theft. The judge agreed. Anthropic's name got excluded from the trial entirely.

Then came the liquidation. In March 2024, Anthropic did a financing round that valued the company at $884 million. Abu Dhabi's Mubadala sovereign wealth fund invested $500 million—exactly what SBF had put in years earlier. The second-largest buyer? Jane Street, SBF's former employer. Craig Falls, their head of quantitative research, personally invested $20 million. So SBF's old firm ended up buying back shares that a former employee had purchased with embezzled funds. The liquidation team recovered $1.34 billion across two rounds, which went into the FTX creditor compensation pool.

Now here's the kicker. What if they'd held? In February 2026, Anthropic just closed that $30 billion Series G round, bringing the post-money valuation to $380 billion. Without dilution, that 8% stake would theoretically be worth over $30 billion. The difference between $1.34 billion and a potential $30 billion is the biggest regret in the entire FTX bankruptcy.

But the liquidation team had no choice. Their job was to liquidate assets fast and repay creditors. SBF will be in prison until at least 2049.

Here's what's fascinating though: Anthropic is now systematically distancing itself from the EA movement, even though it was literally built on EA's money, EA's people, and EA's philosophical framework. The seven co-founders pledged to donate 80% of their personal wealth—that's roughly $38 billion based on current valuations. Nearly 30 Anthropic employees registered for the EA conference in San Francisco, more than double the combined attendance from OpenAI, Google DeepMind, xAI, and Meta's superintelligence labs.

But when Daniela Amodei (Dario's sister, Anthropic's President) was interviewed by Wired, she said, "I'm not an expert on effective altruism. I don't agree with that term. My impression is that it's a bit outdated." This is the same woman whose company is governed by EA people and funded by EA networks, but now EA is suddenly outdated?

It makes sense though. After the SBF collapse, the EA movement's reputation got absolutely destroyed. Any smart company would distance itself from that brand association. But the facts remain: Anthropic's founding logic came directly from EA's arguments about AI risk. Its early funding came almost entirely from EA networks. Its governance is controlled by EA people. They're taking EA's money, using EA's people, living in EA's shared housing, but now pretending they're not EA.

The parallel is almost too perfect to be real. Both SBF and Anthropic's founders believed in the same underlying operating system: you should be willing to take unusual means and bear unusual risks if the final good output is large enough. SBF pushed that logic past the boundary into crime. Anthropic operates on the legal side of that line, but their core proposition—"we must build the most powerful AI ourselves to ensure AI safety"—is itself a grand, almost self-justifying gamble.

They grew in the same soil. Dario and SBF attended the same parties, believed in the same philosophy, lived on different nodes of the same social network. One built a $380 billion AI empire that's now shaping Pentagon policy. The other is in federal prison, watching from a cell as his $500 million investment would theoretically be worth $30 billion if circumstances had been different.

The $500 million check that connected them remains one of the most bizarre chapters in Anthropic's history. It's a story about how the same logic can lead to either a $380 billion company or federal prison, depending on where you draw the line.