Anthropic may have become the most important AI company on this planet today.

Its Claude large model has been deployed into the Pentagon, U.S. intelligence systems, and national laboratories for intelligence analysis and target selection, even participating in real-world military decision support. On the commercial side, this company took less than three years to grow its annual revenue from 0 to 14 billion dollars. In February 2026, Anthropic completed a 30 billion dollar Series G financing, with a post-money valuation exceeding 380 billion dollars, with tech giants like Amazon, Google, NVIDIA, and Microsoft almost all participating.

At the same time, it is also engaged in a globally watched game with the Pentagon over the issue of AI weaponization.

However, in the early history of this company, there is an investment that still seems extremely bizarre.

The investor is Sam Bankman-Fried.

In April 2022, when ChatGPT had not yet been born, the AI track was nowhere near the popularity it has today. SBF invested $500 million in Anthropic's B round through Alameda Research, accounting for 86% of the entire round of financing, acquiring about 8% of the equity.

Seven months later, FTX collapsed.

This $500 million was proven to come from user deposits. SBF became the protagonist of the largest fraud case in cryptocurrency history and was ultimately sentenced to 25 years in prison.

But if we view the timeline from another perspective, this investment almost rewrote history.

According to the valuation of $380 billion in 2026, the theoretical value of that 8% equity exceeds $30 billion. $500 million turns into $30 billion, a return exceeding 60 times. This is an extremely rare absolute return in the entire history of venture capital.

A person serving time in federal prison almost completed the craziest bet in the history of AI investments.

The question is, how did he find Anthropic in 2022?

The answer lies not in 'vision,' but in a circle.

This circle is called Effective Altruism (EA).

In mid-2010s San Francisco, there was a group of people living in highly overlapping social structures: they lived in similar shared houses, attended the same parties, read the same papers, and adhered to a common philosophy—that goodness should be calculated, and every dollar should flow to where it can 'maximize good outcomes.'

In this system, an important branch believes that humanity's greatest long-term risk is not war or disease, but uncontrolled artificial intelligence.

Anthropic's founder Dario Amodei is deeply embedded in this network. He has long collaborated with key figures from GiveWell and Open Philanthropy, maintained close relationships with researchers in the AI alignment field, and even rented houses, worked together, and established family relationships with these people in real life.

Anthropic's early employees, governance structure, and even trust system largely stem from this network.

More importantly, several of the most important funders in this circle—Facebook co-founder Dustin Moskovitz, Skype co-founder Jaan Tallinn, and Sam Bankman-Fried—are all early investors in Anthropic.

This made SBF's investment path very clear.

This is not a story about 'seeing AI in advance,' but a typical case of 'capital flow within a circle': EA's money flows to EA projects to solve the problems defined by EA.

SBF adheres to a more radical branch within EA—'earning to give.' His logic is simple: first make as much money as possible, then invest that money in fields that can generate the greatest positive impact. The mission of Anthropic—'safely developing powerful AI'—happens to be one of the most core issues within this system.

Thus, investment took place.

In 2021, Jaan Tallinn led the A round for Anthropic; in 2022, SBF took over, writing a $500 million check in the B round. The vast majority of funding for this round actually came from him and the funding system he controlled.

But this money is not without its doubts.

Dario Amodei later recalled that SBF had already shown 'enough red flags' at that time. Thus, he made a key decision: to accept the funding but isolate it in governance structure. What SBF received were non-voting shares, excluded from the board.

This decision has proven to be extremely critical in hindsight.

But the question remains: since the risks have been perceived, why still choose to take the money?

Deeper answers are still related to EA.

In this philosophical system, the evaluation criteria for funds do not entirely depend on their source, but rather on their 'effectiveness'—whether they can bring about greater good. SBF's entire wealth narrative is built on this logic: making money is a means, while doing good is the end.

This logic ultimately crossed legal boundaries in his case, but in 2022, it still seemed merely radical rather than criminal.

Everything that happened afterward constitutes a reality that is almost darkly humorous.

FTX collapsed in November 2022, SBF was arrested, tried, and convicted. The portion of equity in Anthropic was included in the bankruptcy liquidation. In court, the defense tried to present this investment as evidence of 'foresight,' but the prosecution responded very directly: investing with stolen money, even if it makes a profit, is still stealing. The judge ultimately adopted this viewpoint.

The liquidation team then sold these shares, recovering about $1.34 billion in total, which were used to compensate the affected users.

But if these shares had not been sold, the story would be completely different.

According to the valuation in 2026, this portion of equity theoretically exceeds $30 billion. The gap between $1.34 billion and $30 billion has become the most dramatic 'unrealized gain' in the entire FTX incident and is also the reason this history is still frequently mentioned.

At the same time, Anthropic is also undergoing changes.

Although its founder still promised to donate most of his wealth, employees still maintained contact with the EA community, and the company's structure was still deeply influenced by it, in public statements, it is gradually distancing itself from the label of 'effective altruism.' This separation became particularly reasonable after the FTX incident—any company would actively distance itself from a tarnished label.

But in any case, the path to the birth of Anthropic cannot be erased.

Its starting point of thought, funding sources, and organizational structure are deeply embedded in the EA network.

And this also creates a stark contrast in the entire situation.

Sam Bankman-Fried is currently serving time in federal prison and will not be eligible for release until 2049 at the earliest.

And during the time he is serving his sentence, the company he once bet on has grown into one of the core forces of the AI era.

If everything is legal, this $500 million investment would be enough to make him one of the most successful venture capitalists in history.

SBF's 'make money and donate' philosophy shares the same underlying logic as Anthropic's 'safe development of AI': to achieve a sufficiently large good, one can endure extremely high risks and unusual paths.

The difference is that SBF crossed the legal boundary, while Anthropic is still operating within it.

They once lived in the same social network, participated in the same discussions, and adhered to the same ideas.

In the end, one turned into a $380 billion AI empire, and the other walked into federal prison.

And what connects them is that $500 million check.

To this day, it remains one of the most intriguing pages in Anthropic's history.#SBF #AI $BTC $ETH