0:04
So, by now you've probably seen a dozen
0:06
of these AI bubble videos. How Nvidia
0:08
and Open AI fuel the AI money machine.
0:11
>> I believe the AI industry is in a very
0:13
>> We are in an AI bubble bubble. Is it a
0:15
bubble? But what they've all missed is
0:17
what's truly behind it all.
0:19
>> And I don't blame them because the
0:21
bubble is confusing as it is. And that's
0:23
all by design. While this tech company
0:25
funds this and in this one back around
0:27
in a loop, Wall Street has been
0:28
bankrolling the entire AI infrastructure
0:31
boom using the very same tactics that
0:33
crashed housing in 2008. Only this time,
0:36
I don't think it's a bubble. It's far
0:38
worse. It's a $1.8 trillion black hole
0:41
where money goes in and it never comes
0:44
back out. Just like before, it will only
0:46
be obvious once it's too late. But if
0:49
you think I'm just fear-mongering,
0:50
Michael Barry, who predicted 2008, is
0:53
already making the same bet. Look, it
0:55
doesn't matter at all whether you're for
0:57
or against AI. It doesn't, cuz I'll show
1:00
you how you're already paying for it in
1:01
a way that's actually easy to understand
1:04
in what I think is the biggest bet in
1:06
human history. [music] And even if they
1:08
get it all wrong, guess who will end up
1:11
paying for it. AI is real and it is
1:14
going to change every industry. The idea
1:16
that chips and is what you want to short
1:34
So to understand this whole thing, the
1:36
best way is to build out the bubble in a
1:38
way where it's actually simple. Because
1:40
once you see the system, you'll begin to
1:42
understand all the players that are
1:43
involved that's out in [music] the open
1:45
and hidden and why this is a black hole
1:48
rather than a bubble. So, let's break it
1:49
down starting with showing you the most
1:51
important players that you need to know.
1:53
First, the AI lab companies. You already
1:55
know this, like OpenAI, Chat GBT,
1:57
Anthropics Claude, and Musk's XAI
1:59
[music] and Grock. Second, the chip
2:01
makers that power AI like Nvidia and
2:04
AMD. And third, data center providers
2:06
like Oracle and Cororeweave that buy
2:08
billions of dollars of these chips,
2:09
store them in massive facilities, and
2:11
then rent out that compute to the AI
2:13
labs. And finally, these guys, which
2:16
[music] I'll reveal who a bit later on,
2:18
but for now, remember those three to
2:20
understand how the system is
2:22
artificially propped up in a circular
2:24
loop. And the loop begins with Nvidia,
2:26
the most valuable company in the world.
2:28
And as you'll see where all the loops
2:30
lead back to. And recently Nvidia
2:32
invests up to hund00 billion in OpenAI.
2:35
Sounds generous, right? Well, there's a
2:37
catch. OpenAI must use that money to buy
2:40
millions of NVIDIA chips. So Nvidia
2:42
gives OpenAI money to buy Nvidia
2:44
products. So essentially, a discount
2:46
disguises an investment, but that
2:48
doesn't matter on paper or to the
2:50
public. As soon as the deal is
2:51
announced, Nvidia's revenue increases,
2:53
stock goes up, and now they have even
2:54
more money to invest back into Open AI.
2:57
And now the loop is in motion. Open AAI
2:59
then signs a $300 billion deal with
3:02
Oracle to build massive data centers.
3:04
And just like that, Oracle stock spikes
3:06
36% in one day. Oh, but I almost forgot
3:09
to mention, Open AI only makes $12
3:11
billion in revenue. So, how the hell are
3:14
they signing $300 billion deals? Well,
3:16
that's the beautiful thing about this
3:18
circle. Oracle takes that $300 billion
3:20
and uses it to buy tens of billions in
3:23
Nvidia chips for their data centers. And
3:25
Nvidia now has more money to reinvest
3:27
into an open AI. And just like that, the
3:29
super loop restarts. But that's just the
3:32
beginning because in order to keep this
3:34
going, you have to remove all the
3:36
perceived risk. [music] So remember
3:38
Cororeweave? Well, they're a data center
3:40
provider that needs chips to rent out to
3:42
an AI lab like OpenAI. And so to make
3:45
this happen, Cororeweave buys huge
3:48
quantities of NVIDIA chips to install in
3:50
their data centers, [music] but with a
3:52
trick. The deal Nvidia generously gave
3:55
to Cororeweave is that if they can't
3:57
find enough customers, Nvidia will buy
3:59
any unused chips back risk-free. But
4:02
hold on, there's more. When Cororeweave
4:05
filed for IPO recently, it turns out
4:07
that Nvidia owned 5% of the company. And
4:10
Nvidia agreed to anchor the IPO with a
4:13
$250 million order to boost up investor
4:16
confidence. So you see the pattern here.
4:18
Revenue gets recorded, stock prices go
4:19
up, and rinse and repeat. So, as I laid
4:22
that out now, it's very easy to blame
4:24
these few companies. But the bubble
4:26
didn't grow to this size on its own. The
4:28
thing is, everyone is playing the same
4:30
game, but in slightly different
4:32
financial [music] engineering
4:34
strategies. Because Nvidia isn't the
4:35
only chipmaker in town. And since AI
4:37
labs like OpenAI require massive amounts
4:40
of compute power, they're also buying
4:42
billions in AMD chips. But in this seal,
4:45
it's extremely unusual. AMD gives OpenAI
4:48
the right to buy 10% of AMD stock for 1
4:51
cent each. 1 cent. But the stipulation
4:54
is is that the stock only vests if AMD's
4:57
stock price rises after the partnership,
5:00
which of course [music] it does. Because
5:02
the very second that OpenAI publicly
5:04
announces the massive AMD deal, AMD's
5:07
stock spikes. And today, AMD's stock
5:10
price is actually higher than Nvidia's.
5:12
And this is exactly where our fourth and
5:14
not so secret player comes in. Big tech.
5:17
Oh yeah, you really thought that they
5:19
wouldn't be here. Amazon has invested $8
5:21
billion in Anthropic, the AI lab that
5:23
makes Cloud. And in return, Anthropic
5:25
has to use AWS cloud and chips. Google
5:28
does the same thing, but $3 billion into
5:30
Anthropic. And don't think OpenAI isn't
5:32
taking similar deals. Microsoft has
5:34
already pumped $13 billion into them to
5:37
keep them locked into their
5:38
infrastructure. So what does this all
5:40
create? It removes actual competition
5:43
and replaces it with this multi-headed
5:46
dependency where everybody's betting on
5:49
everyone else in this small little
5:51
circle. And within those bigger circles,
5:53
smaller bubbles are growing by the day.
5:56
Like when Nvidia invests in XAI, Elon
5:58
Musk's AI lab. And as you've seen the
6:00
pattern by now, XAI buys Nvidia chips.
6:03
But what you may not have heard about is
6:05
that XAI absorbed Twitter or X earlier
6:08
this year. So X AI can pull data from X
6:11
to train Grock its AI model. Tesla then
6:14
uses Grock and its cars and robots. And
6:17
Musk wants Tesla shareholders to fund X
6:20
AI, creating a self-contained ecosystem
6:23
where money, data, and compute all loop
6:26
across Musk's [music] assets. And as
6:28
I've mentioned throughout, the big
6:29
bubble continues to grow at scale
6:31
because the market rewards this sort of
6:33
behavior. Because remember how I said
6:35
that all loops lead back to Nvidia?
6:38
Well, all those stock jumps mean that
6:40
Nvidia is now worth $5 trillion or more
6:43
than Japan and Germany's entire GDP. And
6:46
if they want to keep that growing, they
6:48
need to continue to convince the public
6:50
that the demand will keep rising
6:52
forever. Which means that they have
6:55
every incentive to manufacture demand,
6:58
not just meet it. So hopefully you see
7:00
the full picture now. And it's exactly
7:02
why economists are sounding the alarm of
7:05
roundtpping. The same trick that Enron
7:07
used in the 2000s when company A sells
7:10
an asset to company B with a secret
7:12
agreement to buy back a similar asset
7:14
later. And the result of that looks like
7:17
companies making money when they're
7:20
really not. So I think you don't need to
7:22
be a genius to know why these are red
7:25
flags. But even with all that said, is
7:28
this really a bubble though? Well, to
7:30
answer that, I need to tell you about
7:32
the hidden players quietly making all of
7:39
Because while everyone's been attracted
7:41
by these headlines and stock prices,
7:43
Wall Street has been quietly buying up
7:45
the actual infrastructure that is needed
7:47
for all this to function from land, data
7:50
centers, and power. And along with big
7:52
tech, they're making the biggest bet in
7:54
human history. And as you'll see,
7:56
they're willing to pay any premium to
7:58
make that happen. And the thing is,
8:00
we've only seen the beginning. By 2030,
8:02
$7 trillion is said to be spent on data
8:04
center infrastructure. Since 2022,
8:07
private equity firms have acquired over
8:09
$450 data center companies. That's 80 to
8:12
90% of all mergers in the sector. And in
8:14
2024 alone, they announced $115 billion
8:17
in deals, nearly double the prior two
8:20
years combined. And what private equity
8:22
is doing is that they're buying these
8:23
buildings to lease them to big tech. And
8:25
it's a perfect arrangement. Private
8:27
equity builds a shell to collect rent
8:29
payments and big tech invests their
8:31
money into faster and better AI tech
8:33
while keeping the liability off their
8:35
balance sheets. But you might be
8:36
thinking, why is no one talking about
8:39
this? Well, simple. What do you think
8:41
gets more clicks? Chad GBT passing the
8:43
bar exam or a company most people have
8:45
never heard of like Blackstone acquiring
8:48
multibillion dollar data centers?
8:52
>> Becoming an AI player.
8:53
>> Okay. And now you might be thinking,
8:55
well, so what? Well, because this is how
8:57
the risks multiply. First, if this AI
8:59
boom ends up cooling off and
9:01
hyperscalers need less space, private
9:03
equity is about to be stuck with empty
9:05
warehouses full of unused or outdated
9:07
servers. And if you've seen any of my
9:09
private equity videos, these deals are
9:11
usually funded with massive debt in a
9:13
leverage buyout. So, now you can deduce
9:15
that if they can't pay it back, these
9:17
losses ripple through these banks and
9:19
credit markets. But the second and
9:21
bigger reason why you should care is
9:23
because you're already paying for it.
9:25
Because the infrastructure that private
9:26
equity is buying isn't being built in
9:28
Silicon Valley. It's happening in your
9:31
neighborhood. Private equitybacked
9:32
developers are buying up farmland and
9:35
industrial parks in suburbs across
9:37
America to build these massive data
9:39
center campuses. So places that used to
9:41
be quiet rural counties are turning into
9:43
server farms so big they make Walmart
9:46
look small. But I guess the good news is
9:48
that people are fighting back. In
9:50
Virginia, where a lot of these data
9:51
centers are, a project with 84 data
9:54
centers, where one data server is the
9:56
size of two Walmarts, actually ended up
9:58
being stalled, along with $46 billion in
10:01
other developments. But this community
10:03
backlash isn't just about land. And what
10:05
all this investment from private equity
10:07
fails to mention is that it's actually
10:10
about power. And a perfect example is
10:12
OpenAI's $500 billion Stargate project
10:14
that will require 10 gawatt. So power
10:17
enough for 26 million homes or the
10:20
entire state of Texas where I live in.
10:22
And that's just one project. With amount
10:23
of investment into US data centers that
10:25
are being built, it's estimated that
10:27
combined it will now draw as much power
10:30
as 10 to 15 major cities, not to mention
10:33
the millions of gallons of water a day.
10:35
And the thing is the US grid can't
10:38
handle this. One nuclear plant is around
10:40
1 gawatt of energy and we built one in
10:42
the US in the last 30 years. Renewable
10:44
energy is currently limited by tariffs
10:47
and data centers take around 2 to three
10:49
years to build while power plants take 5
10:52
to 10. So what are these companies
10:54
doing? Well, you really think that with
10:55
all that money from big tech and private
10:57
equity, they haven't found workarounds.
10:59
The solution has been installing gas
11:01
turbines directly at these data centers.
11:03
And they're strategically choosing
11:04
states like Tennessee that lets them
11:06
fasttrack environmental review. All the
11:08
while going against their own green
11:11
pledges that they made to show how much
11:13
they care about the environment. And
11:14
that's exactly why nearly one in five US
11:17
data centers are now concentrated in
11:19
communities already dealing with a bunch
11:21
of pollution. So with all that said, I
11:23
think that you and I can at least cut
11:25
some slack if they're at least paying
11:27
for the increased power usage. But
11:29
lobbying is also in there. They're
11:31
blocking laws that would make tech
11:32
companies pay for the grid upgrades that
11:34
they cause. So instead, regulation
11:37
allows utility companies to spread the
11:39
cost across everyone. Meaning you could
11:41
soon see an extra $10 to $20 on your
11:43
monthly bill. And even if that sounds
11:45
like peanuts, if 50% of Americans live
11:48
paycheck to paycheck, every dollar is
11:50
going to count. Now look, most of you
11:52
watching are Americans who care about a
11:55
strong economy. So maybe I'm just being
11:56
a devil's advocate here, but could this
11:58
all be forgiven if private equitybacked
12:00
data centers create tons of jobs, right?
12:04
Well, during construction, it does. A
12:06
new data center can employ over a
12:07
thousand workers for construction, but
12:10
once it's built, a typical data center
12:12
actually only employs around 50
12:14
full-time workers. That's it. That's if
12:16
the data center even ends up being used.
12:19
Because the thing is, there are already
12:20
signs that we're building far more than
12:22
the market will ever need. Because
12:24
remember, Cororeweave, their entire
12:26
business model depends on renting out
12:27
chips to AI labs from their private
12:30
equity Binance data centers. But the
12:32
thing is, chip rental prices are
12:34
crashing. Nvidia B200 chips went from
12:37
$3.20 per hour per chip to now $2.80 per
12:42
hour, which is below break even for many
12:45
of these sort of operators. So, that's
12:47
sort of a red flag of over supply, but
12:49
it introduces the risk. Let's say if the
12:52
demand that we're predicting never
12:53
catches up, entire data centers could
12:56
become transited assets. And it's
12:58
happened before, just like the 19th
13:00
century railroad lines that lead to
13:02
nowhere. And so if data centers become
13:04
underutilized, the lenders who finance
13:06
all this are going to take some massive
13:08
losses. And with all these things
13:10
connected, huge parts of the AI supply
13:13
chains could become financially [music]
13:15
distressed. So you might be thinking
13:17
that everything that I'm saying so far,
13:19
doesn't that point to a bubble? from the
13:21
circular funding to private equity
13:23
building infrastructure that no one
13:24
might end up using. Not to mention the
13:26
cost that communities pay while getting
13:28
their jobs potentially replaced.
13:35
But here's the thing. When the bet is as
13:38
big as potentially replacing human
13:40
labor, it's not going to be like other
13:42
bubbles where if it pops, the market
13:44
resets and life goes on. This bet is
13:47
different. And so obviously I can't
13:49
predict the future, but to show you why,
13:51
let me walk you through the three
13:53
scenarios of what will happen next to
13:55
understand why this is a black hole. So
13:57
since it's been a little bit too doomer
13:59
for your sake, let's start with the best
14:01
case scenario. The hype actually catches
14:03
up to reality and AI grows into its
14:05
insane valuations [music] and actually
14:08
delivers on what the tech bros promise.
14:10
And it's not that far-fetched. Think of
14:11
like the dot bubble. Yes, a ton of
14:13
companies went bankrupt and markets
14:15
crashed, but from it came Amazon,
14:17
Google, or even YouTube that you're
14:19
using right now. And the same could
14:21
happen here where from the fallout, we
14:23
get real technological breakthroughs and
14:26
AI really does become a force multiplier
14:28
that democratizes opportunity instead of
14:31
deepening inequality. Because what if
14:33
the early research stays true? Where AI
14:35
tools have been found to help
14:36
lowerkilled workers earn more, work
14:38
faster, and compete better. And although
14:40
without a doubt some jobs will be
14:42
replaced, but also millions of people
14:44
end up upgrading their careers. So
14:46
that's the best case scenario, but you
14:48
and I both know that every advancement
14:50
comes with a price. And the real
14:51
question is at what cost? Which brings
14:54
us to the worst case scenario, the
14:56
singularity. Not the sci-fi version
14:58
where we upload our brains and transcend
15:00
biology. I mean the real world version
15:03
where government, big tech and capital
15:05
merge into one singular system of
15:07
totalitarian control and where this guy
15:10
is calling all the shots.
15:12
>> You would prefer the human race to
15:15
>> Uh you're hesitating.
15:18
>> I don't know. I I would This is a long
15:20
hesitation. So many longesitation.
15:22
>> There's so many questions and
15:23
>> should the human race survive? Yeah, it
15:26
sounds insane, but is it that
15:29
far-fetched when it's already happened
15:31
before? After what happened in 2001, the
15:33
US quietly rolled out total information
15:36
awareness where it justified
15:38
wiretapping, mass data collection,
15:40
government agencies spying on citizens
15:42
all because of fear. So, when the
15:44
playbook is all the same, is it crazy
15:46
that in the worst case scenario that
15:47
this happens in the name of security?
15:49
Again, I don't know, but hopefully we
15:51
don't end up in this dystopian
15:53
nightmare. So, what is the most likely
15:56
scenario? Because some creators say it's
15:58
a bubble, others say it's not, or maybe
16:00
a combination of both. But I really do
16:03
think that they're all looking at it
16:04
wrong. Because everything I've described
16:06
so far isn't a bubble. It's more of a
16:09
black hole. And there are three primary
16:11
reasons why. The first is there's no
16:13
exit. Big tech and Wall Street and the
16:16
world at large are in an arms race to
16:17
meet AI expectations. But AI
16:20
infrastructure isn't capital that you
16:22
can just pull back from. It's capital
16:23
that gets sucked in the more you feed
16:25
it. Every new chip requires more power,
16:28
more servers, more cooling, more debt.
16:30
And once the buildout starts, the only
16:32
direction is forward, even if demand
16:34
never materializes. Not to mention,
16:37
because so much of this is financed with
16:39
private credit at double-digit interest
16:41
rates, these companies can't slow down.
16:44
They have to expand, not because demand
16:46
is real, but because the debt also
16:48
requires it. So, this isn't like a
16:50
normal bubble where you're buying assets
16:52
hoping to flip them later. In this case,
16:54
once a data center is built, who the
16:56
hell are you going to sell a giant
16:58
server farm to, which leads to the
17:00
second point that there's no truth?
17:02
These assets stay on the books at full
17:04
value because they barely trade. And
17:06
there's no price discovery until a
17:08
bankruptcy forces a fire sale. So, on
17:10
paper, everything can look healthy even
17:12
if the real value is collapsing
17:14
underneath. And because most of this
17:16
sits inside private equity, private
17:18
credit, and corporate balance sheets,
17:19
the public might not ever end up seeing
17:22
what's actually happening. Just like
17:23
Enron was able to get away with it,
17:25
clever accounting lets you depreciate
17:27
slowly over 20 years when in reality, a
17:30
drop in utilization can wipe out the
17:32
value overnight. So investors,
17:33
regulators, and the public all operate
17:36
under an illusion, not because anyone's
17:38
hiding it, but because a system doesn't
17:40
require the truth. Which leads me
17:42
directly to the third and most important
17:44
reason why this is a black hole. Because
17:46
if it breaks, there will be no alarms.
17:49
Traditional bubbles usually burst when
17:51
sentiment flips and then everybody
17:53
panics at once. But I don't see AI
17:56
collapsing that way. As long as big tech
17:58
keeps announcing breakthroughs and
18:00
trillion dollar deals, confidence is
18:02
going to stay artificially high because
18:04
that bet to replace human labor is a bet
18:07
that we've never even seen before. And
18:09
not to mention with stock prices
18:10
continuing to hit record highs, no one's
18:13
pulling the plug anytime soon. But even
18:15
if like confidence really does slip to
18:17
like record lows, I don't think it will
18:20
even be a sudden pop. It will be more so
18:21
a slow bleed where server farm
18:23
construction pauses, AI startups quietly
18:25
start shutting down, and utilization
18:27
starts dropping. And again, because
18:29
private credit is financing a lot of
18:31
this, the public might not really notice
18:34
for months. So to put it simply, this
18:36
won't be like a Lehman Brothers moment
18:38
like in 2008. It'll be more so like
18:40
thousand micro failures all happening
18:43
slowly behind closed doors. And this is
18:46
exactly what Michael Barry started
18:48
warning about in [music] this AI black
18:50
hole. Turns out Google's actually
18:52
quietly extended the useful life of its
18:54
servers and network gear from 4 years to
18:56
6 years. And that alone cut its 2023
18:59
depreciation cost by about $3.4 $4
19:02
billion and magically boosted reported
19:04
profits by nearly $3 billion without
19:07
doing anything. And they're not alone.
19:09
Microsoft, Meta, Amazon, they've all
19:11
done the same. And in total, big tech
19:13
has added almost $10 billion to profits
19:16
over 2 years just by declaring that
19:18
their hardware now lasts 6 years instead
19:20
of four. And it's this red flag on paper
19:22
that Bur is pointing out is that if
19:24
everybody is treating AI servers and
19:26
chips as if they'll generate value for 6
19:28
years, there's going to be some problems
19:30
when the hardware might actually become
19:32
obsolete in 2 to 3 years. So that's the
19:36
bottomless black hole while also having
19:38
every ingredient of a bubble. A world
19:40
changing story, concentrated debt,
19:42
steady financing, and exposure spread
19:44
across banks and private credit. But
19:46
instead of popping, the capital just
19:48
gets swallowed, disappearing into
19:50
depreciation, energy bills, and endless
19:53
debt payments. And so, obviously, no one
19:56
knows what's actually going to happen.
19:57
But what's clear is that the pace of
19:59
change in investment into AI isn't
20:01
slowing down, it's accelerating. Just as
20:04
I was finishing this video, the White
20:06
House signed an executive order called
20:07
the Genesis Mission, which is
20:09
essentially a federally funded push to
20:11
win the race for AGI or artificial
20:13
general intelligence. And it's literally
20:15
like the Manhattan project of artificial
20:18
intelligence. So the details are still
20:20
coming out, but what this means is that
20:22
the AI bubble or black hole isn't just
20:24
powered by big tech and Wall Street
20:26
anymore. It's now backed by the full
20:28
weight of the US government as every
20:30
country competes in [music] this era's
20:33
cold war arms race. And again, it's this
20:35
big bet that the world is competing on
20:37
that makes it foolish to call this an AI
20:40
bubble instead of what it is, a black
20:42
hole. So whether you're for AI or
20:44
against it, [snorts] here's what you can
20:46
do. Learn how power moves and don't be
20:48
the guy in the common refusing to
20:50
embrace change. Because here's the
20:52
thing, even if you can't change the
20:54
system, the next best move is to learn
20:56
to understand it. Because once you do,
20:58
you can use that knowledge to protect
21:00
yourself or even profit like Michael
21:03
Barry did in 2008 when everybody else
21:05
chose ignorance instead. Because in a
21:08
world of money and power fueled by AI,
21:11
ignorance is only going to be the most
21:14
dangerous position of them all. And so,
21:16
if you're watching this channel, you're
21:17
in the right place. I'll be diving even
21:19
deeper into this next week in my free
21:21
newsletter in the video description. But
21:23
if you haven't seen my last video on
21:25
AI's impact on the job market, go and
21:27
watch our video on it. And don't forget
21:29
to like and subscribe to learn how money
