Latest / In The Loop / The Man Who Predicted The 2008 Market Crash Just Bet $1 Billion Against AI

Transcript

The Man Who Predicted The 2008 Market Crash Just Bet $1 Billion Against AI · In The Loop

• 0:00
  Michael Bury, the man that predicted the 2008 financial
• 0:03
  crisis and the housing crisis, who they had Christian Bale play
• 0:07
  him in the film The Big Shaw just revealed that he's betting
• 0:10
  against NVIDIA and Palantir, the two biggest companies in the AI
• 0:15
  ecosystem. He bet 80% of his total net
• 0:17
  worth of a $1.1 billion against those companies doing well in
• 0:22
  the next few months. Markets panicked when they found
• 0:25
  out that he'd done this trade. CNBC reported that it was one of
• 0:28
  the worst days on the stock market in recent history.
• 0:32
  Over a trillion dollars was lost in single day just because of
• 0:35
  this news. I'm Jack, and this is in the
• 0:36
  loop. And today we're going to discuss
• 0:38
  why Michael Burry did this, what he saw, and as a result, we're
• 0:43
  going to follow the money and in doing so, explain why I think
• 0:46
  many people's reaction is just wrong.
• 1:01
  So I think let's start with the circular financing that's going
• 1:05
  on right now. So if you know anything about
• 1:08
  Michael Berry and his housing trade, if you've watched The Big
• 1:10
  Short, you know he held the short.
• 1:12
  So which is a short is when you bet against specific companies
• 1:15
  or a market and if they do well, you lose your money.
• 1:19
  Now he bet against all of these companies in 2008 for over 2
• 1:25
  years. So between 2005 and 2007,
• 1:28
  everyone thought it was absolutely crazy because
• 1:29
  everyone's printing money. His investors, the people were
• 1:33
  giving his company money obviously were furious.
• 1:37
  And he held it regardless because he was 100% convicted in
• 1:43
  the problem that he saw. And that was going to eventually
• 1:45
  explode. And as we all know, with the
• 1:47
  2008 financial crisis, it, well, it really did explode.
• 1:52
  Now here's the really interesting part about this
• 1:54
  particular bet against AI. By the time this information
• 1:58
  became public, Bury had already closed both his positions.
• 2:02
  So he was already out. He'd made his bet, he'd made his
• 2:04
  money and already got out. It was just a three month bet,
• 2:09
  Wasn't a two year bet like the housing crisis last time.
• 2:12
  So this is what sent me down a rabbit hole.
• 2:14
  Now, why I wanted to do this episode on that topic because I
• 2:17
  think the market reaction's been, well, you could argue
• 2:21
  ridiculous, but it's as a result of things that have been brewing
• 2:24
  for some time because you have a trillion dollars lost in market
• 2:28
  value and basically a week last week speaks volumes.
• 2:32
  So I guess why did Barry pick these companies, Palantir and
• 2:34
  NVIDIA? Well, First off, Palantir trades
• 2:38
  at 143 times revenue, so they are essentially being valued 143
• 2:44
  times more than they make. That makes it the most expensive
• 2:47
  stock in the S&P 500, and it's more than double what peak 2021
• 2:52
  SAS companies were trading at during the bubble mania.
• 2:56
  And NVIDIA is valued at 30 times sales, which makes the Palantir
• 3:02
  stock actually look reasonable in comparison.
• 3:04
  But when you dig into the SEC filings, you find something
• 3:07
  quite interesting. And many people argue concerning
• 3:11
  because 46% of Nvidia's revenue comes from just four companies
• 3:15
  and all four of them are building their own chips to
• 3:17
  compete with NVIDIA, which is obviously a bit of a red flag.
• 3:20
  And there's also a massive vendor financing issue, which
• 3:23
  I'll get into just now. So 85% of Nvidia's revenue is
• 3:27
  under Fender financing, which is basically when one vendor says
• 3:31
  you will pay me this amount of money over this period, and they
• 3:34
  both kind of commit to that. So the money's not actually real
• 3:36
  yet. It's not landed in that
• 3:38
  company's bank account, but it will do over time.
• 3:40
  It's like the vendor, the person selling a good financing, others
• 3:44
  to purchase that good because everybody really wants it and
• 3:47
  we'll pay for it. And really it's not a problem
• 3:51
  until the money isn't actually real in the ecosystem and it's
• 3:55
  just numbers that are being thrown about.
• 3:56
  And this is where a lot of people have problems with the AI
• 3:59
  sector right now, because money is actually just going in big
• 4:01
  circles. So NVIDIA agrees to invest $100
• 4:16
  billion in Open AI, which is great.
• 4:18
  And then Open AI commits to spend that money on NVIDIA chips
• 4:22
  over a certain period. Now, not all of the money, but a
• 4:24
  big chunk of it. AMD, another producer of chips
• 4:27
  and technology, commits to Open AI buying $90 billion of AM
• 4:34
  DGPUS and in exchange AMD gives Open AI warrants for 160 million
• 4:41
  shares of their company at a single penny each, which is
• 4:45
  basically giving away 10% of their entire company to Open AI
• 4:49
  for purchasing these goods and services from them.
• 4:51
  Open AI essentially pledged $300 billion / 5 years, which is $60
• 4:56
  billion a year and it's a non binding agreement with a party
• 5:01
  that doesn't have the money to pay.
• 5:03
  So Open AI can't afford that. They don't have that money.
• 5:06
  And Microsoft wants Open AI to commit and buy $250 billion of
• 5:11
  Azure services. So that's like their developer
• 5:14
  platform where they provide language models and other
• 5:16
  services. So they want them Open AI to
• 5:18
  commit to $250 billion of those services over the next six
• 5:23
  years. And then Amazon just signed up
• 5:25
  Open AI on AWS for $38 billion a year.
• 5:29
  One company's investment becomes another's guaranteed revenue,
• 5:32
  and it's all within a, you know, a handful of maybe 20 companies
• 5:34
  that this money is just going in circles around.
• 5:36
  And so obviously Barry and other people are looking at this and
• 5:40
  going that's potentially insane. Now I'll get onto my view on
• 5:43
  this shortly, but if you add up all the company's commitments
• 5:47
  and then work backwards, open a eyes revenue would have to scale
• 5:51
  from 20 billion today to $577 billion by 2029 just to support
• 5:59
  the commitments in spending they've given out.
• 6:02
  That's Google size revenue in four years and Google took 20
• 6:07
  years to get there. Right now Open AI lost last year
• 6:10
  $5 billion and they made $3.7 billion in revenue.
• 6:14
  So they're burning $2.25 for every dollar they earn and their
• 6:18
  infrastructure investments that have committed to is 6 billion
• 6:22
  in 2025, then 50 billion in 2027 and 112 billion in 2028 and 173
• 6:28
  billion in 2029. Now that means they have to grow
• 6:33
  by 104% every single year, which is faster than Google ever grew,
• 6:37
  grew. Google grew at about 70% every
• 6:40
  single year. And they were literally
• 6:41
  inventing digital advertising and modern day technology
• 6:45
  platforms at the at the same time.
• 6:47
  And I guess this is what Barry probably saw, lots of circular
• 6:51
  financing with math that doesn't really work out.
• 6:54
  And now maybe it's a stretch because maybe Bury also looked
• 6:56
  at historical parallels where maybe this hasn't worked out.
• 7:00
  You know, we've seen this part before.
• 7:04
  So in 1999 to 2000, a company called Lucent Technologies
• 7:08
  basically had $8.1 billion in vendor financing out there,
• 7:13
  which is 20% of their revenue. They reported $41 billion in
• 7:16
  revenue to investors and to the world because technically
• 7:20
  because of that vendor financing them agreeing with other
• 7:23
  companies that those companies will send them the money, that
• 7:26
  they had $41 billion in revenue as a result.
• 7:29
  But they only actually had $304,000,000 in revenue, actual
• 7:32
  cash coming in. So when the telecom industry
• 7:35
  back then collapsed, obviously that company collapsed with it.
• 7:40
  The SEC ended up charging ten members of their exec team with
• 7:43
  fraud for manipulating essentially billions of dollars
• 7:46
  in revenue. And to put that into context for
• 7:48
  NVIDIA, Nvidia's exposure is 85% of revenues. 85% of revenue is
• 7:54
  potential vendor financing. So that's four times more than
• 7:56
  that. Lucent Technologies that
• 7:58
  obviously collapse when the telecom industry collapsed.
• 8:00
  So obviously that's the mechanics the Barry saw in the
• 8:03
  market. It's what people call a bubble.
• 8:05
  People say it's all going to collapse news with circular
• 8:07
  financing, but here's what I think people get wrong.
• 8:21
  So everyone's obviously freaking out the this is like 2008 and
• 8:25
  it's following very similar patterns as 2008 and the.com
• 8:28
  bubble and everyone's going to lose all their money.
• 8:31
  And I think there's elements of truth in that.
• 8:34
  I'm not saying that there's no comparison and parallels here.
• 8:38
  There is a lot of danger in everyone getting very excited
• 8:40
  and committing a lot of money and including individual
• 8:43
  investors as well as big institutional investors.
• 8:47
  But I think they are overall a little bit wrong because there
• 8:51
  is big differences between now and then.
• 8:53
  The tech companies building this AI infrastructure aren't
• 8:56
  start-ups over than open AI. Open AI, if it doesn't work out,
• 9:00
  is going to cause a big ripple effect.
• 9:04
  Let's just say, let's put it that way, because they've
• 9:06
  committed so much money to so many different vendors who have
• 9:08
  used that commitment to fund other things.
• 9:11
  But anyway, the companies that are predominantly funding this,
• 9:13
  the Microsofts, the Amazons, the Googles are sitting on $500
• 9:17
  billion in cash reserves and they generated over $340 billion
• 9:24
  in this past year. So to put that into context,
• 9:27
  that Lucent technology that collapsed because they were
• 9:30
  financing out too much, well, they had nowhere else to go.
• 9:35
  They were only serving telecom customers who when the telecom
• 9:38
  companies collapsed, that company collapsed as a result.
• 9:41
  Whereas the companies were talking about the Googles, the
• 9:43
  Microsofts, the Amazons. Their revenue is highly
• 9:48
  diversified and Microsoft Azure is already making $86 billion in
• 9:52
  ARR from their AI services. Google has a search advertising
• 9:56
  platform that basically just prints money.
• 9:59
  And AWS is not relying on AI to grow.
• 10:02
  Although it needs to offer competitive services to grow in
• 10:06
  the AI space, it doesn't rely on AI to be successful and grow
• 10:10
  revenues. So although open AI is a
• 10:13
  critical factor and potential ripple effects that are going to
• 10:17
  cause absolute mayhem if they do collapse, nobody's completely
• 10:21
  completely reliant on open AI commitments.
• 10:24
  And the same with the.com bubble era.
• 10:25
  I know I've said this before and I'm not going to spend too much
• 10:27
  time on it as a result, but many of these companies were
• 10:30
  unprofitable startups with no path to revenue and they were
• 10:35
  all investing in the infrastructure they couldn't
• 10:37
  afford but weren't making any money.
• 10:39
  And we're relying on investors to pay for this is different.
• 10:43
  And if you tried to take, say, Claude code away or other
• 10:46
  services or AI platforms away from people, I think they would
• 10:50
  flip the desk. I wouldn't be able to cope.
• 10:52
  So yes, the vendor financing to open AI creates a lot of risk
• 10:55
  and exposure. Yes, if open AI collapses,
• 10:59
  someone is going to lose a lot of money and take a lot of write
• 11:02
  downs. But a write down isn't
• 11:04
  bankruptcy. It's not the systemic crisis
• 11:06
  that the magic 7 and a collapse all of a sudden.
• 11:08
  So the Facebook's matters, the Googles, it's just going to
• 11:12
  delay AI roll outs probably for a few years.
• 11:16
  It's not going to stop anything. And I think that's why Burry
• 11:19
  exited his position after three months instead of trying to hold
• 11:22
  on to that for years like you did with the housing crisis.
• 11:25
  With housing, he saw the systemic crisis, over leveraged
• 11:28
  banks, fraudulent mortgages, a financial system that would
• 11:32
  completely collapse if that domino fell.
• 11:36
  And I think he thought that that was worth holding out throughout
• 11:39
  the pain. But with AI, all he's probably
• 11:41
  seeing is an overvaluation of major companies in the market,
• 11:45
  too much expectation, and as a result, would expect the share
• 11:50
  price of all those companies and the valuations of them to come
• 11:53
  down a bit. So yeah, to conclude, what does
• 12:06
  this all mean? Well, what I think is actually
• 12:09
  happening is that they reinvested against the mechanics
• 12:11
  of the market right now that circular financing is real.
• 12:15
  The maths definitely doesn't work out fully right now, but
• 12:18
  Open AI is betting big that if it goes to an IPO, it's going to
• 12:21
  raise a lot of money and people all want to get and buy in.
• 12:24
  I think that's probably true. And I think the panic about the
• 12:28
  market is, as always, a bit of an overreaction.
• 12:32
  But I do think that all these companies like the Nvidia's, the
• 12:35
  others are well overvalued. I'm an individual investor for
• 12:38
  many of these companies and they are so expensive right now.
• 12:42
  Their their valuations compared to how much they're actually
• 12:46
  making is ridiculous. And everyone wants to get on the
• 12:49
  train because they see how much money is being invested.
• 12:52
  But everyone's a bit worried because it feels too good to be
• 12:55
  true. So there'll be lots of these
• 12:57
  drops over the next coming period.
• 12:59
  I don't think it impacts how much value AI is actually
• 13:01
  delivering week to week. So people are raising this, I
• 13:05
  guess question of a bubble or a future market collapse.
• 13:08
  I hope you use this episode as a guide to have a good
• 13:12
  conversation about it. Anyway, I hope you enjoyed that
• 13:15
  episode. I certainly did.
• 13:17
  It was a really interesting one. I promise next week we'll
• 13:20
  probably get away from bubble conversations into an extra
• 13:23
  interesting topic. But anyway, I'll see you next
• 13:26
  week.