Author: Deep Tide TechFlow

Anthropic is arguably the most important AI company on this planet today, if not the only one.

Its Claude large model is deployed at the Pentagon, U.S. intelligence agencies, and national laboratories, used by the U.S. military for intelligence analysis and target selection in military strikes against Iran.

Its annual revenue skyrocketed from zero to $14 billion in less than three years. In February 2026, Anthropic completed $30 billion in Series G financing, with a post-investment valuation exceeding $380 billion. Tech giants like Amazon, Google, NVIDIA, and Microsoft are lining up to invest money.

In the past few weeks, it has been engaged in a high-stakes game with the Pentagon over the weaponization of AI that the whole world is paying attention to.

In the early financing history of this company, one name still captivates many: Sam Bankman-Fried.

In April 2022, before ChatGPT existed and the AI field was nowhere near as hot as it is today, SBF poured $500 million into Anthropic's Series B through his controlled hedge fund Alameda Research, taking 86% of the entire round and acquiring about 8% equity. Seven months later, the FTX empire collapsed, with SBF becoming the protagonist of the largest scam in cryptocurrency history, sentenced to 25 years in prison, and that $500 million was customer deposits from FTX.

But had SBF not been arrested, and if that money had a legitimate source, that 8% stake would theoretically be worth over $30 billion at today's $380 billion valuation. Turning $500 million into $30 billion yields a return ratio of over 60 times; such absolute profit would rank highly in the entire history of venture capital.

A crypto fraudster serving time in a federal prison almost made the most insane bet in AI investment history.

How did SBF find Anthropic in 2022? Why was he bold enough to throw down $500 million? And why did Anthropic accept that money?

The answer is hidden in a circle known as "effective altruism."

A Shared Rental, A Movement, A Check

In mid-2010s San Francisco, there was a group of people living in the same type of shared housing, attending the same type of parties, reading the same type of papers, and adhering to the same philosophy.

This philosophy is called effective altruism (EA). The core proposition is straightforward: charity should not be based on feelings but on calculations. Every dollar should flow towards the direction that mathematically “maximizes good outcomes.” In one important branch of EA, the primary existential risk faced by humanity is neither nuclear war nor pandemics, but runaway artificial intelligence.

Dario Amodei is deeply embedded in this circle.

He is the 43rd signatory of the Giving What We Can Pledge, committing to donate at least 10% of his income, and he became a fan of GiveWell as early as 2007 or 2008.

He lived in the same shared house with two people: one named Holden Karnofsky, co-founder of GiveWell and Open Philanthropy, one of the most influential fund allocators in the EA movement; the other named Paul Christiano, a core researcher in AI alignment. At the time, Dario and Paul were both serving as technical advisors at Open Philanthropy.

Later, Karnofsky married Dario's sister Daniela. After getting engaged, the couple briefly lived with Dario. In January 2025, Karnofsky quietly joined Anthropic as a "technical staff", responsible for security policy. When a reporter from Fortune discovered this, Anthropic had not even publicly announced the appointment.

This is a close-knit social network.

Amanda Askell, an early employee at Anthropic, is the ex-wife of William MacAskill, one of the founders of the EA movement. She is the 67th signatory of GWWC, with her doctoral thesis focusing on a core issue in EA philosophy: how to handle infinity in ethics.

The most important governance body of Anthropic, the "Long-Term Benefit Trust," theoretically holds substantial control over the company, with three of its four members directly coming from the EA system: former GiveWell executive director Neil Buddy Shah, CEO of the Center for Effective Altruism Zach Robinson, and CEO of the long-term funded GiveWell partner Evidence Action Kanika Bahl.

All three of the largest financial backers in EA movement history are early investors in Anthropic: Facebook co-founder Dustin Moskovitz, Skype co-founder Jaan Tallinn, and Sam Bankman-Fried.

This is the real pathway that allowed SBF to find Anthropic—not some genius investment insight or foresight into the AI field, but a financial cycle within the inner circle: EA money flows to EA projects, solving problems defined by EA.

SBF subscribes to a more radical branch of EA, "earning to give." He resigned from the Wall Street quantitative firm Jane Street to dive into cryptocurrency, explicitly stating that his goal is not personal wealth but "altruism," aiming to make as much money as possible before directing it towards causes that can generate the greatest positive impact. Anthropic’s mission, "to develop powerful AI safely," is almost the standard prescription from EA regarding existential risks from AI.

In May 2021, Jaan Tallinn led a $124 million Series A investment in Anthropic, with Moskovitz co-investing. In April 2022, SBF took over and led a $500 million Series B investment, accounting for 86% of the total $580 million raised. Co-investors in the round included Caroline Ellison, Nishad Singh, and James McClave from Jane Street.

The co-investor list itself speaks volumes. Caroline Ellison is the CEO of Alameda, Nishad Singh is the engineering director of FTX, and Jane Street is SBF's former employer.

This $580 million Series B was, in fact, almost entirely funded by SBF and the funds he controlled.

Red Flags and Compromises

Dario Amodei is not foolish.

Later, in a deep interview reflecting on this matter, he recalled that SBF appeared at the time to be “someone who was optimistic about AI and concerned about safety,” aligning well with Anthropic's direction. But then Dario said a key phrase: he detected “enough red flags.”

So he made a decision: take the money but isolate SBF in governance structure. SBF received non-voting shares and was excluded from the board. Dario later appraised SBF's behavior as “much more extreme and egregious than I imagined,” with three "much more" layered together.

This decision later proved to be extremely wise. But it raised a sharp question: if the danger signals were sufficient to warrant isolation in governance, why accept the money at all?

You could argue that the AI financing environment in early 2022 was nowhere near as hot as it is today, and that Anthropic needed significant funds to build computational power; an investor willing to shell out $500 million, regardless of how many "red flags" existed, is hard to come by.

But there is an even more nuanced reason: in the operational logic of the EA circle, the "cleanliness" of funding sources has never been a top priority. What matters is the “effectiveness” of funds; whether they can help you do more. SBF's entire wealth narrative is built on this foundation: making money is a means, doing good is the purpose, so the means of making money need not be overly scrupulous, as long as the eventual output of "good" is large enough.

This logic reached a criminal extreme in SBF’s hands, but at the moment he invested in Anthropic, it still appeared to be a radical yet not illegal philosophical choice.

After the Collapse: A Dark Comedy

The subsequent events are well-known in the crypto circle.

In November 2022, CoinDesk exposed Alameda’s balance sheet, Zhao Changpeng announced the liquidation of FTT, a bank run swept across FTX, and within nine days the empire collapsed. SBF was arrested, extradited, and tried, receiving a 25-year sentence in March 2024. The 8% stake in Anthropic, along with all assets, was frozen in bankruptcy proceedings.

One notable episode during the trial that was excluded by the court is worth mentioning.

SBF’s defense attorney attempted to present the investment in Anthropic as evidence of “foresight,” saying, “Look, he didn’t just squander; he made an investment decision that multiplied in value.”

Prosecutor Damian Williams responded firmly: the profitability of these investments is completely unrelated to the fraud charges. If you stole someone else's money to invest, earning profits still amounts to theft. The judge accepted the prosecution's viewpoint, and the name Anthropic was excluded from the court proceedings.

The prosecution also added a sting: isn't FTX itself the best counterexample? Valued at $18 billion in 2021, $32 billion in 2022, and worth nothing today.

Then came the liquidation auction.

In March 2024, the first round valued at $884 million.

The largest buyer, the Abu Dhabi sovereign fund Mubadala, invested $500 million, precisely the amount SBF invested that year. The second-largest buyer was Jane Street, the former employer of SBF and Caroline Ellison; Jane Street's quantitative research director Craig Falls even personally contributed $20 million to participate. SBF's first job after graduating from MIT was as a trader at Jane Street, and now this old employer spent money to buy back the shares acquired by their former employee with stolen funds.

The two rounds collectively recovered $1.34 billion. This money flowed into FTX's creditor repayment pool, becoming an important source of funds for victims to reclaim their deposits.

What if the liquidation team had not sold?

In February 2026, Anthropic completed the $30 billion Series G financing, reaching a post-investment valuation of $380 billion. If we don’t consider dilution, that 8% theoretically transformed from $1.34 billion to $30 billion. The liquidation team naturally did not have that choice; their duty was to liquidate and repay creditors as quickly as possible, but the numerical difference, $1.34 billion vs a potential over $30 billion, is key to understanding why this story continues to be discussed today.

It is the largest regret in the entire FTX bankruptcy case.

Collective Forgetting of EA

Antrhopic's size and influence today need no elaboration, but an interesting phenomenon is that the company is systematically distancing itself from the EA movement.

All seven co-founders have pledged to donate 80% of their personal wealth, which is worth about $38 billion based on current valuations. Nearly 30 Anthropic employees signed up for the EA conference in San Francisco, more than double the total of OpenAI, Google DeepMind, xAI, and Meta combined.

But Daniela Amodei said in an interview with Wired, “I am not an expert on effective altruism. I don’t agree with that term. My impression is that it's a bit of an outdated way of saying things.” The person saying this is married to one of the most influential fund allocators in the EA movement, who just joined her company.

This attitude of “taking EA’s money, using EA’s people, living in EA’s shared housing, yet not acknowledging being EA” has become understandable after the SBF case. The collapse of FTX plunged the reputation of the EA movement to rock bottom. Anthropic needs to maintain distance from this label, just as any smart company would cut ties when facing negative associations with its brand.

But the facts are there: the founding logic of Anthropic comes from the core discourse of the EA circle regarding existential risks from AI; its early financing was almost entirely sourced from the EA network; its governance structure is held by personnel from the EA system.

A Parallel Universe in Prison

Sam Bankman-Fried is now in federal prison, with the earliest release date set for 2049. By then, he will be 57 years old.

During his time in prison, the AI company he invested in with stolen funds has seen its valuation exceed $38 billion and is engaged in a high-profile game with the Pentagon regarding AI weaponization issues, with its founders becoming frequent guests in The New York Times and on Capitol Hill. If everything were legal, that $500 million bet would have been enough to make SBF one of the highest-return investors of this era.

SBF's “earning to give” and Anthropic’s “safely developing AI” share the same underlying operating system, allowing unusual means and risks in pursuit of sufficiently large positive outcomes.

SBF pushed this logic beyond the boundaries of legality, while Anthropic operates on the safe side of that line. Yet its core proposition—“we must build the most powerful AI to ensure AI's safety”—is, in itself, an enormous bet that is almost self-evident.

They grew from the same soil.

In that soil, Dario and SBF once attended the same parties, adhered to the same philosophy, and lived at different nodes of the same social network. One moved towards a $38 billion valued AI empire; the other walked into federal prison.

And the $500 million check that connects them still remains the most bizarre page in Anthropic's history.