Well it's not a random 25 year old. He became valedictorian at Columbia and graduated at 19. He was then briefly a part of OpenAI's superalignment team.
After he got fired, he wrote a 165 page thesis on AI in early 2024 that all ended up becoming true in 2026; you can read it for yourself and be the judge: https://situational-awareness.ai
That's what he based the fund on.
It's generational foresight not available to 99.999999999999999999% of 25 year olds. I want to get ahead of the default cynicism and "oh it's just dumb luck woe is me" aspect of HN.
> generational foresight not available to 99.999999999999999999% of 25 year olds. I want to get ahead of the default cynicism and "oh it's just dumb luck woe is me" aspect of HN.
And I want to get ahead of this absurd romanticism. I have plenty of counterpoints in all types of technical domains (3D printing, blockchain, you name it). Oracles don't exist, just stories of right place right time survivors.
> That's what he based the fund on.
His thesis was so true that his fund folded. Fancy that.
This is only true if newer investors get pro-rata ownership of the Anthropic stake. If that position is side-pocketed, then any investors who were not there at the time of the Anthropic allocation are looking at around 90% loss on their investment.
You can argue the semantics all you want and ignore the original premise of my comment - which is romanticising people who have perceived "foresight" is super weird.
His fund didn’t fold btw. They sold / were forced to sell a position that was several hundred percent up at a very very bad price relative to peak. AFAIK the fund is still up on the year though.
AIUI they were forced to sell all their positions in public companies at a net loss. The fund is still up for the year only because of a prior investment in Anthropic.
It's not part of the strategy or operating agreement of a GP/LP structure to sell a specific fund to another investment firm at a discount. The fund may buy and sell assets to create returns for LPs but selling the fund itself at a discount is not a successful exit, e.g. they folded.
I feel like I’m arguing against the grain here, and it seems to me like there’s a lot of emotion around this topic. I’m not even a defender of AI billionaires. I just posted because I see this type of discussion around SA recently and it’s mistaken. So I’ll just leave this comment, and it goes for the other replies too.
What happened to SA is not what anyone would call “a fund folding”. Normally you’d assume folding means the fund went bankrupt, or closed at a loss.
This fund is not closing and not on a loss. It’s not bankrupt, and apparently it is opening up new positions and continuing to invest in its main theme last I saw it in the news.
It is on a major drawdown from its peak a few months ago, but it’s still up for the year, most of its investors are likely well in the positive. If you make a huge investment and it goes up 5x before trimming its gains and ending up 100%, you’ll be annoyed, you’ll wish you sold at the top, but you’ve still doubled your money.
Getting margin called and closed this way is not the same as going bankrupt.
> Getting margin called and closed this way is not the same as going bankrupt.
Technically we don't actually know the details here since no one here is privy to the specific fund structure - which is very likely WAY more complex than reported.
So, realistically we're arguing semantics. I said it folded and for intents and purposes if you're getting margin called and have to sell, then that's folding.
Just like in poker I can fold and not play anymore and still have a stack of chips.
That is called folding the fund. It did in fact fold. "My fund would not fold if I made different hypothetical decisions" argument means your fund did in fact fold.
Yes, being unable to pay margin call means your fund is folding.
Wait, is this the group who predicted that we'd have expontential AI progress and AGI in 2027 because we'd have AIs training AIs? Predicated on the preposterous assumption that R&DEffort == RateOfProgress. The more sober and realistic assessment is that R&DEffort >= RateOfProgress, with only short bursts of achieving the upper bound. Leave it to a brilliant 25 year old to overfit to one of those short bursts of progress and then assume it will continue forever just as it always has.
What actually happened is that AI-assisted training of LLMs (mind you -- stuck on that same old transformer architecture, with nothing resembling a medium-term memory, and fine-tuning still doesn't really work as a form of "learning") is right now giving us ~linear-ish progress... beecause we already hit the slowdown in the S curve of what human researchers can achieve.
We are also energy-constrained in ways that the exponential forecasts have no answer for other. Compute is not getting more efficient fast enough to achieve anything close to exponential growth driven by growth in computing power. There are social, moral, and political limits on the amount of energy we can dedicate to AI training in any given time period. And it's not just energy, we're short on RAM wafers and water for cooling and eventually we're going to hit other limits on various minerals and elements of the supply chain etc. etc. So even if it were true that R&DEffort == RateOfProgress without resource constraints, we're likely not going to see that exponential progress because we also need corresponding exponential scifi-scale progress in computing efficiency, cooling, and, energy delivery.
Basically this kid made a bunch of very clever and grand predictions which were always kind of ridiculous and no, they have not come to pass in 2026 and we are not at all on track to achieve AGI in the next 12 months, despite what Altman keeps trying to make the public think.
I'm not saying we won't have breakthroughs or significant capability changes. I'm saying that any reasonable expected rate of breakthroughs is not enough to maintain exponential R&D effort and to have it translate into exponential progress towards AGI, unless it's an absolutely monumental discovery on par with the GPT LLM itself. It would be impossible to predict, and it would be totally fallacious to credit this author with being a visionary if such a discovery does in fact arise.
It’s pretty standard insight among the machine learning crowd. This kid founded the effective altruist club at his school. He’s clearly very smart and ambitious, and has leadership qualities, but you don’t have to exaggerate.
Is "valedictorian" a good signal? I'm from Europe so we don't have the concept, but it seems to me that being the exact single topmost student in a cohort would set a perverse incentive to game the rules and hyperfocus on getting this label. To me being somewhere in the top 5%-10% while ignoring some bullshit and not caring about maximizing all grades would indicate a more robust mind that is focused on better priorities. (I'm not claiming it is easy to become valedictorian, I'm just wondering what precisely it indicates about a person, besides obvious talent and conscientiousness etc)
This is anecdotal, however I remember hearing about farmers trying to maximize for egg production. Multiple hens would share a cage. The initial criteria was to only breed the hen that produced the most eggs. Overall egg production declined though. It turns out that the best performing hen was doing so at the cost of the others, ie. eating more than its fair share, being combative. The strategy that worked in raising overall egg production was breeding all the hens from the best performing cage.
I'm not sure failing to understand the concept of hedging and the difference between mean and median returns under heavily leveraged trading counts as generational foresight. Most 25 year olds I know in the tech world have been calling for AGI since about the same time. I do respect his ability to just dive headfirst into something even though he basically took wallstreetbets literally but with billions of dollars under management rather than thousands. I'm not sure why we revere having enough confidence to do something badly so much.
You forgot to add that he was in a relationship and later engaged to the chief of staff at Anthropic while raising and running his fund, but I’m sure he had no access to material non-public information.
Early 2024 is a bit late, isn't it? He jumped on a bandwagon and his AGI fever ramblings quoted in the submission show that he is just a marketing drone.
You sure that's the right number of nines? We'd probably need to look at over a billion earths to find another such 25 year old in that case. Further, if we look at the rest of his body of work, it may temper our estimation of his foresight.
Did you make big bets on specific things like memory? He had insight, he made money, didn't have enough experience to handle it well, but still made great returns after his "downfall"
Well that “foresight” left him to walk right into a total amateur hour cash crunch that would have gotten a junior Wall Street analyst fired… so the evidence of amazing foresight here isn’t great.
Calling the fund “situational awareness” and then demonstrating a complete lack of situational awareness was rather amusing.
There are not enough human 25 years old in history for this many 9s. You had 21 nines, which would only make sense if there were 10 billion billion humans ever and that he is the only one. However, the current estimate is there were 107 billion humans ever alive in history, so your statement is impossible even if the estimate is off by 10000 fold.
Your failure here is that you are using normal math, not Effective Altriusm math. That number should not only include past/current 25 year olds, but also all possible future 25 year olds, including all 25 year olds simulatable by a future powerful AI. In that context, the estimation is quite modest.
It's a rhetorical device that I have yet to see used to good effect. The kind of rhetorical device that undercuts the argument rather than strengthens it.
You're so caught up in irrelevant details like a rhetorical device that you've missed that this guy's fund is up 80% for the year and his investors are having a GREAT year. Worth reflecting on.
However, I think this is a classic example of the category error the OP is talking about. Making accurate predictions isn't necessarily the same thing as asset management. Once you reach that level of fame, people will 100% target and trade against you, and when someone is that well-known, their market positions inevitably get exposed—how could they possibly handle that?
There are large and famous funds that have lasted despite all that you say. So it is very possible to achieve fame, have your positions known, and have people trade against you, and still not be forced into selling your entire public portfolio to satisfy a margin call.
This is why hedging and neurotic levels of risk-management are necessary. Running a highly levered and highly correlated portfolio is a disaster waiting to happen regardless of whether your overall thesis is correct or not.
That's fair and I guess we're about to find out! The thing to appreciate is there's a lot that's unprecedented about all of this. Some of it has to do with Leopold but then there's a whole lot that doesn't. I'll definitely be following along as the story develops over the years...
His strategy was to trade on MNPI available to him via relationships that no larger fund/bank compliance department would sign off on, in a portfolio no fund/bank risk department would allow.
I don't understand how you can read a thesis that predicts AGI in 2027 (and the subsequent explosion to ASI) and say that it "all ended up becoming true in 2026".
> he wrote a 300 page thesis on AI in 2024 that all ended up becoming true in 2026
...no? The essay is literally titled "The Decade Ahead", its predictions have not come true in 2026.
> The AGI race has begun. We are building machines that can think and reason. By 2025/26, these machines will outpace many college graduates. By the end of the decade, they will be smarter than you or I; we will have superintelligence, in the true sense of the word
I'd argue even the 2025/26 part of that isn't true, though the "many" in "many college graduates" makes the claim very difficult to verify.
But even putting all that aside I think it's still a valuable evidence point when discussing "intellectual arrogance". No doubt this 25 year old is very smart when it comes to AI. But there's this belief that if someone is smart at one thing then they're surely smart at everything. Turns out "smart at AI" doesn't automatically translate to "can sensibly manage a $45bn fund".
I'm not paying for the food. I'm paying for them to bring me the food. Those are two different products.