Dive into the shocking corporate collapse of Situational Awareness, a $45 billion tech-centric hedge fund managed by 24-year-old former OpenAI researcher Leopold Aschenbrenner [0:00:10, 0:00:21, 0:00:38]. This finance documentary breaks down how a young trader with zero professional experience raised hundreds of millions from Silicon Valley elites, only to lose two-thirds of his total fund in a matter of weeks due to extreme leverage and a fake "long-short" strategy [0:00:21, 0:00:54, 0:01:03, 0:02:44]. Learn the mathematics behind volatility drag, why Ken Griffin's Citadel bought out the collapsing portfolio carcass during Aschenbrenner's own wedding weekend, and the stark cultural divide between Wall Street risk management and Silicon Valley speculation [0:01:44, 0:02:53, 0:03:00, 0:30:23].#LeopoldAschenbrenner#HedgeFundCollapse#SituationalAwareness#PatrickBoyle#VolatilityDrag#Citadel#LeverageTrap#RiskManagement#SiliconValleyHype#CorporateFinance
AI Generated·May contain errors
Come on guys, stop, it's not funny.
Last week, a 24-year-old named Leopold Aschenbrenner lost about two-thirds of a $45 billion hedge
fund in a matter of weeks.
Before this, his main qualifications for managing $45 billion were a viral essay, a period working
for Sandbank and Freed, and being fired from OpenAI.
A lot of this is stuff you'd normally leave off your CV.
The fund was called Situational Awareness, and the Wall Street Journal reports that this
was a crash he never saw coming, which does raise a small question about the name of the
firm.
He'd raised the money mostly from Silicon Valley tech executives.
The Stripe founders, Nat Freeman and Daniel Gross, then borrowed a great deal more from
Wall Street prime brokers to bet heavily and in one direction on the rapid advancement
of artificial intelligence.
Borrowing billions to make a concentrated bet on a single volatile sector is generally considered
a high-risk strategy.
When the market recently reconsidered whether AI would be immediately profitable, his lenders
sent a wave of margin calls, and the fund lost about 67% of its value in a single month,
which is generally considered a poor month.
Losing this much money is stressful under any circumstances, but for Leopold the timing
was a little inconvenient because the margin calls arrived during his wedding weekend.
The wedding has been described as a multi-day celebration in Carmel, California with the
ceremony at a Tuscan-style villa and the send-off at a spa in the forest.
There was also a pre-wedding colloquium with panels and breakout sessions.
I've actually stopped attending weddings myself because if there's one thing I can't
stand, it's a colloquium.
The couple's only stated request was no gifts, which in hindsight they may have wanted
to revisit.
You have to imagine some guests arrived dreading a four-hour breakout session on the future
of humanity and then cheered up considerably upon hearing that they'd instead get a
front-row seat to a catastrophic financial meltdown.
So how does a 24-year-old with no trading experience borrow billions to blow up his
own fund?
And why was Ken Griffin so willing to rescue him?
But first, if the stress of a $30 billion margin call during your wedding breakout sessions
is causing you to feel unwell, you might want to consider today's sponsor, ZocDoc.
There's a specific kind of healthcare admin that shows up when you move somewhere new
or switch jobs and end up with different insurance.
The doctor you'd finally gotten comfortable with is suddenly out of network or a five-hour
drive away, and you're starting the whole search over from nothing.
Normally that means asking around for recommendations or picking a name off an insurance directory
and hoping.
This week's sponsor, ZocDoc, replaces that guesswork.
It's a free platform where you search local in-network doctors across more than 200-plus
specialties, read reviews from actual patients, and see real appointment openings you can
book instantly online.
No phone call involved.
There are over 150,000 providers listed across all 50 states, and most appointments land
within 24 to 72 hours.
Same-day slots show up more often than you'd expect.
So rather than guessing which of a dozen similarly-rated doctors near you is worth your time, you
can actually compare them and choose based on what matters to you – location, reviews,
whether they take your insurance – before you've committed to anything.
Head to ZocDoc.com forward slash Patrick Boyle or click the link in the description.
Moving house is disruptive enough without also losing track of your own health care.
Thank you so much to ZocDoc for sponsoring our video.
If you want to manage $45 billion on Wall Street, the traditional route involves decades
of staring at spreadsheets, building risk systems, and surviving a few market crashes.
It takes a long time, and in truth, it's a bit tedious.
Silicon Valley has fortunately streamlined this process.
To raise a large hedge fund in California, your CV requires a big following on Twitter,
a one-year stint at a startup, a viral PDF, and a four-and-a-half-hour podcast appearance.
I think there's going to be a lot of unfortunate things that happen.
Look, Leopold Aschenbrenner is a smart kid.
He graduated from Columbia at the age of 19, was an early convert to effective altruism,
and got his first real job at the philanthropic arm of Sam Bankman Freed's crypto exchange
FTX.
He stayed there until the exchange collapsed in late 2022, which is one way to learn about
risk.
It might surprise you to learn that I'm actually one of the few people who hopes that
Sam Bankman Freed gets a pardon.
Not because I think he's innocent, I don't think that at all.
It's that I find him considerably funnier than all of these new guys.
One of the core tenets of Bankman Freed's flavor of effective altruism was that you
should take any bet with a positive expected return, even if repeated plays guarantee eventual
ruin.
That's a risk management philosophy that gets you onto the cover of Forbes right up
until the morning you have to borrow your lawyer's suit for the court appearance.
After FTX imploded, Leopold spent a bit under a year at OpenAI before being fired.
He says that he was let go in retaliation for raising security concerns.
OpenAI called it an improper disclosure of sensitive information.
He reportedly refused to sign a non-disparagement agreement on the way out, walking away from
about a million dollars in equity and declaring that freedom is priceless.
I think that story is a big part of how he got such a big following on Twitter, because
I've been through his tweets and it definitely wasn't the tweets.
I would say, though, that if I left a million dollars on the table over a non-disparagement
clause, I'd bang out a million dollars worth of disparagement that same afternoon.
But that's just me.
His actual trading experience before starting the fund appears to have been non-existent.
He implied on a podcast that he shorted the market in 2020, which would have been impressive
because he would have been 17 years old at the time and U.S. brokerages tend not to allow
miners to open the margin accounts you need to short stocks.
There's that problem of them not being legally able to sign a contract.
The cornerstone of the whole empire, however, was a 165-page essay he self-published in June 2024
called Situational Awareness.
I read all 165 pages for this video and I can't tell you how badly I wanted to read
an AI summary instead, which, given the subject matter, might have been the intended way to
consume it.
You'll notice that everyone who mentions this essay brings up the page count.
Nobody ever tells you that George Orwell wrote a 224-page book called The Road to Wigan Pier,
which possibly tells you something about how gripping Leopold's essay is by comparison.
I'd also be curious how many of the people who cite this essay as scripture actually
reached page 165.
My suspicion is that Situational Awareness is the most referenced and least finished
document in Silicon Valley since the Terms and Conditions.
And look, I'm aware that I'm not in a strong position to complain about people not finishing
long things.
I know that right now, in a number of households, one member of the family has put this video
on and the others are sitting nearby wondering how much longer it is and when normal life
will resume.
To those people, I would only say this.
Your partner is watching a 30-minute plus video about a guy in a quarter zip who has
a funny laugh and lost a lot of money.
On some level, you already knew who they were.
None of your irritation or dissatisfaction is the video's fault.
This is a compatibility issue and you should take that up directly with your partner.
On top of the 165-page essay, there was the four-and-a-half-hour podcast with Dwarkish
Patel, which even at double speed is a lot of Leopold.
Four-and-a-half hours of AI chat and the laughing.
For scale, Watson and Crick's paper setting out the double helix structure of DNA, arguably
the most important scientific paper of the last 100 years, was two pages long.
Now, having read the whole thing, I can report that it is less of a prophecy and more of
a very long summary of what everyone in San Francisco had already been saying at dinner
for about three years.
He says AGI arrives in 2027, that the world will change faster than anyone believes because
of recursive self-improvement, that the free world's survival is at stake, and there's
a great deal about China and robots.
That sort of thing.
There's nothing about the essay that struck me as innovative or even interesting.
It's just that describing the tech industry's group chat back to itself in 165 pages is
not usually what we mean by seeing the future.
It also contradicts itself in ways that are more entertaining than any of the predictions.
In one passage, he describes himself as a firm believer in the American private sector,
who would almost never support government involvement in industry.
A few pages later, he calls for the US government to nationalize the AI labs and run them as
a wartime project, which would be one of the largest peacetime nationalizations imaginable,
proposed by a man who has just told you he opposes government involvement in business.
He believes both of these things apparently at the same time.
It opens, by the way, with the line, you can see the future first in San Francisco,
which struck me as a slightly alarming thing to promise, because it implies that in the
near future, most of us will be homeless and on drugs.
The essay went viral among tech executives when it was released, and Tim Ferriss crowned
Leopold the Nostradamus of AI, which is more fitting than Ferriss probably intended, given
that Nostradamus was a man whose famous predictions only look like predictions once you already
know what happened.
Along with the manifesto came the tech visionary costume.
Elizabeth Holmes had her black turtleneck.
Steve Jobs had his black turtleneck.
Leopold went with a brown turtleneck and then wore a beige quarter zip sweater over it.
It's all so dull.
When you're in California and you want a costume to raise billions of dollars, don't
you realize that Hollywood is right down the road?
Hire a professional costume designer.
Up your game.
Anyway, it all worked out for Leopold.
He launched the fund with a reported $225 million from tech insiders, and after a year
of high returns and a lot of borrowed money, he was trading a $45 billion book.
All of this out of a small office with eight employees.
Ken Griffin and Izzy Englander, on the other hand, have wasted their whole careers building
trading infrastructure and risk teams.
They just don't seem to know how it's done.
The way I'd describe it is that they lack situational awareness.
There are real cultural divides in America, not one divide quite a few, and Leopold,
who we've by now worked out was lacking in situational awareness, was about to collide
with a few of them.
I was listening to an Ezra Klein podcast a few days ago where he interviewed a sociologist,
Jasmine San, about why most Americans don't want the enormous AI data center built out
anywhere near them, and Klein made a really interesting point that struck me.
He pointed out that people in Silicon Valley have spent their entire careers watching impossible
problems cave as soon as enough smart engineers throw code at them.
To that culture, every obstacle is temporary and every wall is just a bug that hasn't
been patched yet.
Washington, he explained, runs on the opposite energy.
In Washington, things that ought to be simple to fix become permanently impossible because
everyone involved wants a different outcome and nobody will budge.
And the one thing Silicon Valley desperately wants to build right now, gigawatt power plants
and vast data centers out in rural counties, turns out to be a Washington problem, not
a coding problem.
Building data centers is all about permits, land rights, utility boards, and angry local
residents.
You can't resolve a room full of people who want opposite things by writing better
code.
Leopold's 165-page essay was, to a large extent, about this infrastructure build-out,
and Silicon Valley loved it because in California, if you write a science fiction manifesto about
how the world is going to change, someone will hand you a $25 million check over coffee.
Tech investors who understood software were perfectly happy to back a 24-year-old whose
portfolio was, when you closely looked at it, a Wall Street Bets account, explained
using longer words.
It's the same credulity that led Elizabeth Holmes raise hundreds of millions for Theranos
on the strength of a board of distinguished elderly statesmen, none of whom knew anything
about blood work.
But California money has a ceiling.
The really large pools of institutional cash, the pension funds, the big allocators, the
people who write nine-figure checks without checking with the group chat first, mostly
sit on the East Coast.
So, with billions already under management and an appetite for a great deal more, Leopold
packed up his quarter zip and took the fundraising tour to New York.
And New York is the third culture, the one Leopold hadn't gotten involved with yet.
Silicon Valley asks whether a thing can be built.
Washington asks whose permission you need.
Wall Street asks a colder question.
What happens when you're wrong?
In New York, investors are less interested in your Twitter following, your podcast appearances,
and your feelings about the future.
They want to know whether you have a sensible portfolio and the faintest understanding of
risk.
According to Rob Copeland's reporting in the New York Times, downtown Manhattan looked
at Leopold and saw a light weight, a one-hit wonder.
Blackstone, the largest investor in hedge funds on the planet, took the meeting and
passed.
Copeland reports that one wealthy New York investor welcomed Leopold into his office
and gave him a proper grilling on risk.
He asked a completely standard question, the sort anyone allocating billions is obliged
to ask.
What was the plan if the AI revolution didn't arrive on schedule?
If adoption slowed?
If the thesis was early or simply wrong?
Leopold apparently had no detailed answer.
He told the investor that he truly believed it would all work out.
Now that is a lovely way to go through life.
It's not, however, a risk management framework.
Institutional investment committees are rarely won over by a young man asking for billions
on the strength of a PDF and a strong feeling.
The New York money passed, deciding to hold on to its cash, which in hindsight turned
out to be an excellent piece of risk management.
To understand how situational awareness lost so much money so fast, we need to look at
how hedge funds are supposed to work.
Matt Levine at Bloomberg laid this out really well.
The basic idea of a long-short equity hedge fund is to strip out the market risk.
If you find 10 good stocks and buy them and the whole market then crashes, your 10 good
stocks will most likely fall with everything else.
They might fall a little bit less, but you still lose money.
That's why you hedge.
You find 10 bad stocks and you short them.
Now if the market crashes, your good stocks fall, but your bad stocks fall too, and because
you short them, you make money there, which cancels out the losses on your longs.
You've insulated yourself against the market as a whole and left yourself exposed to the
thing you're supposedly good at, telling the winners from the losers.
That is a perfectly sound approach if your longs and your shorts are chosen through good
research.
But it's not what Leopold did.
Leopold had one idea, which he'd helpfully written down in a 165-page document.
The idea was that AI was going to change everything.
So he bought the companies he thought would win the AI revolution, the South Korean chip
maker SK Hynix, SanDisk, Bloom Energy, and he shorted the companies he thought would
lose it, mostly software firms like Adobe.
The trouble is that this doesn't hedge your portfolio at all.
As Matt Levine put it, a book like this hasn't reduced its risk, it's rotated it.
You're insulated against the market, but you've intensified your exposure to your
one big idea.
It's a bit like betting $100 that one football team will win a match and then as a hedge
betting another $100 that their opponents will lose.
It feels like two bets, but it's the same bet twice.
Both sides of Leopold's portfolio depended on the identical team.
The moment the market started to worry that the AI revolution might take a little longer
than expected, his long bets on the chip makers would fall and his short bets on the software
companies would rise.
So he'd lose money on both legs of his hedge at precisely the same moment.
He didn't have a hedged book, he had one enormous unhedged bet on a single idea, dressed
up as a hedge fund.
Now that's fine if you're trading your own money on Robinhood, but Leopold was running
a hedge fund and he was four times levered.
He took his investors' money and reportedly borrowed a further three to four times that
amount from prime brokers like Goldman Sachs and JP Morgan and poured the whole lot into
the same thematic bet.
As Warren Buffett once put it, if you're smart you don't need leverage and if you're
dumb you shouldn't be using it.
Now it's worth pausing on the leverage here because normally there's a logic to it.
Funds often use leverage when the thing they're trading is boring and barely moves.
You borrow to turn a tiny stable return into a decent return.
Leopold was trading the most volatile stocks on the planet and then levered up on top of
that to get even more exposure to them.
That's the exact mistake we went through a few weeks ago in the video on Korean retail
traders.
And I'd suggest that if your investment strategy is functionally the same as a retail trader
leveraged to the eyeballs, just with more zeros, it may be time for a rethink.
And Leopold wasn't alone with this brilliant idea.
As Paul Davies pointed out in Bloomberg, the entire market was and still is leveraged long
on AI.
In South Korea, retail investors known locally as ants have borrowed heavily on margin to
buy the same sorts of names like Samsung and SK hynix.
So when the tech rally wobbled in July, the ants in Korea got their margin calls and had
to sell.
SK hynix dropped hard.
At the same time, the software stocks that Leopold was short rallied.
His portfolio was now taking on water from both sides at once and his prime brokers began
demanding more collateral because a bank's enthusiasm for your AI thesis ends precisely
at the point where it might have to hold the bag for it.
With the fund coming apart, Leopold's team reportedly started phoning rival hedge funds
to offload positions and raise cash.
According to the New York Times, his representatives described what they were selling as a portfolio
of hedges that were simply in distress.
And that's a creative way to describe a giant wrong way bet on Adobe.
The rival funds opened the file and saw immediately that these weren't hedges.
They were just large directional bets blowing up in real time.
And once a rival knows that you have losing positions you're desperate to get out of,
the dynamic turns nasty.
They can smell the blood in the water.
If they know you'll shortly be dumping a stock to repay your loans, they short it ahead
of you and make money from a move that they can see coming a mile off.
With prime brokers like Goldman Sachs and JP Morgan issuing margin calls and rival funds
actively betting against his book, Leopold was in a bit of a bind.
He sent a letter to his investors on July 24, acknowledging that the fund had not been
immune to the market ructions.
Now that's a remarkable amount of freight for the word ructions to carry.
A ruction is the kind of thing you have with your neighbor about how they never clean up
after their dog.
It's not traditionally the word used for losing $30 billion during your own wedding.
Though, to be fair, a letter about ructions is a step up from fleeing to the Bahamas and
leaving a post-it note on your desk that says, I'm sorry.
He even added a PS, noting that this might be an unusually good opportunity to add money
to the fund and offering investors the chance to put fresh cash in on August 1st.
Now, you have to admire the confidence.
Your fund is in free fall, you're phoning your competitors to take positions off your
hands, and you find room for a postscript inviting your clients to double down.
He later wrote to investors that he took full responsibility for what had happened and then
reportedly offered them one-on-one phone calls to talk it through over the following week,
which was also his honeymoon.
Nothing quite says, till death do us part, quite like taking angry investor calls from
the forest spa.
With the banks demanding collateral, situational awareness started calling round the big hedge
funds, pitching them the opportunity to buy chunks of the portfolio.
The Financial Times described these talks as an ad hoc effort, which yes, ringing your
competitors in the middle of the night to ask for a multi-billion dollar bailout, while
the wedding planner is on hold on line too, is always going to be fairly ad hoc.
I can't think of many times in financial history that pitching your own margin call
as an investment opportunity has ever worked out, but you never know.
It does sometimes work if you call Ken Griffin.
Citadel, Ken Griffin's $71 billion firm, agreed to take the whole portfolio, though
the word rescue is doing a lot of work in most of the accounts of this, because Citadel
didn't write to anyone's aid.
It won an auction.
Griffin's firm beat out Jane Street and Millennium for the right to buy Leopold's
collapsing book.
This was no bailout.
It was three of the most patient, carefully built firms on Wall Street bidding against
each other for the carcass and Griffin bidding best.
There is a small detail in that list worth noting.
Jane Street, the quantitative trading firm, was one of the bidders.
Jane Street was also one of Leopold's own investors, so one of the firms that had put
money into situational awareness turned up at the fire sale to bid on its remains.
Which tells you something about how Wall Street handles sentiment.
If your own backers are circling the wreckage, kicking the tires, you were never really amongst
friends.
It's a bit of a pattern with these firms.
When Enron collapsed in 2001, Griffin reportedly put a team on a charter jet to Houston more
or less the same day to pick through the wreckage and hire away Enron's best energy traders.
He did much the same when Amaranth Advisors detonated on natural gas bets and again when
Sowood Capital went down.
Demand moves quickly.
If your hedge fund ever catches fire, Ken Griffin is somehow both the first person to
smell the smoke and the last person you want to see walking through the door.
This is one of the key lessons.
The people who ended up bidding to buy Leopold's portfolio at a discount, Griffin,
the folks at Jane Street and Millennium, are exactly the people who built their firm slowly,
over decades, with risk teams and infrastructure and all of the boring machinery Leopold decided
he didn't need.
They weren't faster than him on the way up, but when the fire started they were the ones
standing there with cash, calm and a signed contract by the morning.
Slow and careful, it turns out, is what you are able to be when the fast people need a buyer.
The only reason situational awareness wasn't wiped out entirely is that a large slice of
the portfolio sat in private illiquid startup stakes, most notably a multi-billion dollar
position in the AI company Anthropic, which is where the world gets a little small,
because Anthropic is the company where Leopold's bride serves as chief of staff to the chief
executive.
So, the single asset that saved his fund was a stake in the company his wife helps run.
The two of them met, naturally, at Sam Bankman Freed's charity.
Nearly everyone in this story met at Sam Bankman Freed's charity.
You can't be margin called on an illiquid private asset because there is no daily price
to market against and no way to sell it overnight.
Leopold did try to sell the Anthropic stake to Sequoia and Green Oaks to raise cash,
but the Citadel deal closed first.
So, Leopold was, in the end, saved by the fact that some of his assets were too difficult to
sell in a panic.
He came out of the fire sale with a slimmed down fund holding somewhere between 8 and
10 billion dollars.
Cliff Asness, who runs AQR, manages a couple of hundred billion dollars and has been doing
this for 30 odd years, was asked by the journalist Joe Nocera what he made of the whole affair.
He replied that we should file the phrase,
you never go full Kelly, under the heading,
things of which you should be situationally aware.
Going full Kelly on Wall Street means building a portfolio for the maximum possible return
with no regard whatsoever for the risk of losing everything,
which is a fairly precise description of what Leopold did.
And when a man who has run money for three decades reaches for that particular pun
to describe your fund, it's not a compliment.
So, why exactly does going full Kelly with maximum leverage destroy a fund,
even when the underlying investments do go up over time?
To see how a portfolio with a high expected return can still march reliably towards zero,
we do need to look at the math of something called volatility drag.
Victor Hagani, the long-standing hedge fund manager from Elm Capital Partners,
discussed this exact idea on my interviews channel last year,
and he goes through it in detail with James White in their book, The Missing Billionaires.
I'll keep the math here as painless as I can.
The core problem is that gains and losses aren't symmetric.
If your portfolio falls by 50%, you don't need a 50% gain to get back to where you started.
You need a 100% gain.
A hundred dollars falls to 50, and now it has to double just to break even.
A loss always digs a deeper hole than the same size gain can fill.
This is also incidentally why down 67% but up 80% on the year
is a less reassuring sentence than Leopold seemed to think.
Here's how high volatility investments can eat your money.
Imagine that if I offered you a coin flip.
Heads you win 50%, tails you lose 40%.
If you look at the arithmetic, the average return is a 5% gain per flip.
It sounds like free money.
I appear to be an idiot for offering it.
But watch the path you actually walk.
You start with $100.
You flip heads, win 50%, and you're on 150.
You flip tails, lose 40%, and you're down to 90.
One win, one loss, and you're already down 10%.
Do it again.
Another win, another loss, and you're at $81.
You're slowly going broke playing a game with a positive average return.
And I'm slowly getting rich offering it to you.
The reason is that the flattering average or mean
is propped up by a handful of spectacularly lucky runs.
Long streaks of heads that can happen but that almost never do.
While the path most people actually live through
gets dragged downward in high volatility scenarios.
The wider the swings, the bigger the gap between the mean return and the median return
or the advertised average and the return you're actually most likely to get.
There's a rough formula for this.
Your real compounding growth rate is approximately
your average return minus one half of your volatility squared.
That second time, half your volatility squared is the volatility drag.
It's the tax that the swings themselves charge you.
And it's subtracted from your average to give the rate your money truly compounds at.
Which brings us to the leverage trap.
And this is the part that actually killed the fund.
When you borrow to amplify your trades, two things happen.
And crucially, they don't grow at the same speed.
Your average or expected return scales in a straight line with leverage.
Double the leverage, double the expected return.
But the drag depends on volatility squared.
So it scales with the square of your leverage.
Double your leverage and you double your return.
But you quadruple your drag.
Let's put Leopold's numbers in.
Suppose a concentrated basket of AI stocks has an expected return of 15%
and a volatility of 40%.
Unlevered, the drag is 8%.
And the fund compounds at a perfectly healthy 7% a year.
This is a good business.
You could do this for decades and retire happy.
Now, lever it four times as Leopold reportedly did.
Your average return quadruples to a mouthwatering 60%.
But your drag doesn't quadruple.
It grows by four squared, 16 times.
From 8% to 128%.
Your typical compounding return goes from plus 7% to minus 68% per year.
The exact same leverage that quadrupled the return
in the brochure is the mathematical force
marching the typical outcome straight into the ground.
It's not that the strategy has good years and then one bad one.
The bad outcome was baked into the arithmetic the whole time.
Now, I want to be fair to the expected return here
because it is a real thing.
That 60% average isn't a lie.
There are versions of the world where this bet pays off spectacularly.
Where AI compounds exactly as Leopold expects
and he ends up with a fortune so large it embarrasses everyone who doubted him.
The problem is how little of the probability lives there.
The expected return is propped up by a thin sliver of wonderful outcomes
way out in the right tail.
The trillion dollar timelines.
While the great bulk of the probability sits over on the left in the land of losses.
The average is dragged upward by outcomes that almost never happen.
So the mean stays high while the median,
the outcome sitting right in the middle of everything that could happen to you,
is a heavy loss.
And no individual investor actually receives the average.
Each one gets the path they happen to land on.
It's just that with a distribution this skewed,
the overwhelming majority of those paths cluster down near the median in loss territory.
While the handful that would deliver the advertised average
are off in a tail that almost nobody reaches.
Leopold isn't betting on a bad expected value.
He's betting that the math applies to other people.
And here's the cruel part.
Everything I've just described,
the thin sliver of wonderful outcomes holding the average up,
assumes that you actually get to stay in the game long enough to reach them.
Leverage takes that away.
A big enough downswing triggers a margin call.
Your prime broker sells you out at the bottom
and the position is closed before any of those lucky average rescuing outcomes
ever get the chance to arrive.
You're left with all of the drag and none of the jackpot.
And there's one more problem.
And it's about who his investors were.
Leopold's investors were mostly Silicon Valley insiders.
People whose wealth is already leveraged to the AI boom.
Their startup equity, their stock options, their salaries,
all of it already rides on the same wave.
So putting money into a leveraged concentrated AI fund
does nothing to spread their risk.
It just piles more of the exposure they already have
on top of the exposure they already have.
The whole point of an outside investment
is to own something that zigs when the rest of your life sags.
This is an investment that zigs in perfect unison
with everything else they own.
If the AI trade ever truly breaks,
these are people whose fund, whose portfolio,
whose job and whose Silicon Valley real estate
all go down on the same afternoon.
A huge short-term return like 400% isn't the good news it looks like.
It's usually the last thing you see
before the whole thing comes apart.
So after vaporizing $30 billion in a matter of weeks,
you might assume that Leopold's career in finance
had hit a bit of a roadblock.
But that would be to underestimate the resilience
of a 24-year-old visionary backed by Silicon Valley's tech elites.
In his letter to investors,
he said that he had removed all of the leverage from the fund,
promised to learn the necessary lessons
and described the whole episode as a set of very expensive scars
that would become invaluable lessons.
And he pointed out that the fund had been structured
so that he could take a loss like this and fight another day.
And to be fair to him,
it is remarkably easy to fight another day
when you're fighting with other people's money.
And those people have signed lock-up agreements
that stop them from asking for what's left of it back until September.
Here's the part that tells you everything about the two coasts.
One week after nearly blowing up his fund,
Leopold isn't fending off furious investors demanding their money back.
He's fending off new investors trying to give him more.
According to Bloomberg,
Silicon Valley money has been clamoring to get in since the collapse
and the fund is, for the moment, politely turning them away.
Think about what that means.
On Wall Street, losing two-thirds of your fund in a month
is a career-defining catastrophe.
In Silicon Valley, it's a buying opportunity.
Pat Grady, a partner at Sequoia,
was asked about the turmoil by Bloomberg and said,
Our suspicion is that he's going to be a fixture in Silicon Valley
for a long time to come.
And he wasted no time getting back to it.
Days after Citadel bought his collapsing public book on a midnight fire sale,
Leopold wired $400 million into a single privately-held startup backed by Sequoia.
On top of $100 million, he had put into the same company a month earlier.
So, the man who had been very nearly destroyed by one enormous concentrated bet
marked the occasion by making another one.
Now, the wedding went ahead as planned,
the guests presumably enjoyed the breakout sessions,
and Leopold rides off into the Silicon Valley sunset
to invest his remaining billions.
A little wiser, a little more married,
and forever haunted by the mathematical gap between the mean and the median.
If you found this video interesting,
you should watch my video on the South Korean retail traders next.
They made a very similar mistake with a lot less money.
Don't forget to check out our sponsor ZocDoc using the link in the description.
See you in the next video.
Bye.