August 21, 2025 • 41 mins
In part two of this week's two-part Better Offline, Ed Zitron walks you through how America’s economic growth has become dependent on Amazon, Google, Microsoft and Meta’s AI capital expenditures - and how this inevitably harms both the economy and our markets.
Better Offline listener deal: Get $15 Off Where's Your Ed At Premium! Deal goes until the end of August.
https://edzitronswheresyouredatghostio.outpost.pub/public/promo-subscription/better-offline-discount
YOU CAN NOW BUY BETTER OFFLINE MERCH! Go to https://cottonbureau.com/people/better-offline and use code FREE99 for free shipping on orders of $99 or more.
BUY A LIMITED EDITION BETTER OFFLINE CHALLENGE COIN! https://cottonbureau.com/p/XSH74N/challenge-coin/better-offline-challenge-coin#/29269226/gold-metal-1.75in
---
LINKS: https://www.tinyurl.com/betterofflinelinks
Newsletter: https://www.wheresyoured.at/
Reddit: https://www.reddit.com/r/BetterOffline/
Discord: chat.wheresyoured.at
Ed's Socials:
https://www.instagram.com/edzitron
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:03):
Media.
Speaker 2 (00:05):
Hi, I'm Edzetron, and this is the second part of
this week's better offline. As ever, go to the episode
notes by some merch get the challenge coin, subscribe to
(00:25):
my newsletter if you like it, go premium. Either way,
you've got this episode for free if you can stand
the ads, which many of you can.
Speaker 3 (00:32):
Anyway.
Speaker 2 (00:33):
The thrust of this two part is that the AI
trade is going to, as you've probably guessed by now,
end badly, and as I'll also explain, I believe we're
now set up for potentially economic and market wrecking consequences
of this hysterical investment bubble. Today we're going to go
beyond just startups and into how the poison of the
bubble is crept into our economy and how dangerous things
(00:53):
are getting as a result. In the first installment, we
talked about Cursor, the AI coding startup that, despite having
no mote and zero sustainability, has somehow earned a valuation
of ten billion dollars, and then asked why none of
these generative AI companies are being acquired, at least not
in the traditional sense. And no, I'm not including the
acquisitions where a big company snags the talent and leaves
(01:14):
barely breathing bodies of startups in their wake. This matters
because if these AI companies can't get an exit, go public,
or get bought, it raises serious questions about how any
of this ends. Cursor is the best example, and the
fact that Anthropic is hopelessly dependent on Cursor for revenue
despite doing everything they can to kill it, makes the
entire situation feel a little bit weird. Cursor's paths to
(01:36):
viability seem frankly to be non existent, and the collapse
of Cursor will inevitably result in a vast chunk of
anthropics revenue is going up in smoke, which again will
also make its collapse seem inevitable, as how can it
fundraise that evaluation over its previous one when its biggest
customer just died? So much of what I'm describing as
structural and few companies have bigger structural obstacles to survive
(01:57):
all than open AI. As a reminder, open aia I
appears to have burned at least ten billion dollars in
the last two months. It has just raised another eight
point three billion dollars after raising ten billion dollars in
June according to The New York Times, and intends to
receive around twenty two point five billion dollars by the
end of the year from SoftBank, and that is assuming
it becomes a for profit entity by the end of
the year, and if that doesn't happen, the route gets
(02:18):
cut to twenty billion dollars total, meaning that SoftBank would
only be on the hook for a further one point
seven billion dollars. I am repeating myself, and I need
you to really get this. Open Ai just got ten
billion dollars in June, just in June, and had to
raise another eight point three billion dollars in August. And
that is an unbelievable cash burn, one dwarfing any startup
in history, raveled only by Xai, makers of Grok the
(02:40):
races dollar m which loses over a billion dollars a month.
I should also be clear that if open ai does
not convert to a for profit there is no path
forward to continue raising capital. Open Ai must have the
promise of an IPO. It must go public because a
valuation of three hundred billion dollars or more. Open Ai
can no longer be acquired because nobody has that much money.
And if Let's be real, nobody actually believes open ai
(03:03):
is worth that much. The only way to prove that
anybody does is to take open ai public, and that
will be impossible if it cannot convert. And ironically, SoftBank's
large and late stage participation makes exits actually harder, as
early investors will see their holdings diluted as a percentage
of total equity or whatever the hell we're calling it.
Because these aren't real equity shares, their profit participation units.
Speaker 3 (03:25):
They're a nonprofit. You can't do stock in them.
Speaker 2 (03:28):
While a normal company could just issue equity and deal
with the dilution that way, open AI's structure necess states
the negotiation, where companies can obstruct the entire process if
they see fit. Speaking of companies doing that, let's talk
about Microsoft. As I asked in my premium newsletter a
few weeks ago, what if Microsoft does not want open
ai to convert. They own all the IP, they own
(03:49):
all the access to open AI's research, and already run
most of their infrastructure, while assuming the best case scenario
that it would end up owning a massive junk of
the biggest tech stile up of all time. And I'm
talking about equity not open AI's current profit sharing units.
Microsoft might also believe that it stands to gain more
by letting open ai die and assuming its role in
the AI ecosystem. But let's assume that open ai converts
(04:12):
and now has to continue raising money at a rate
that will require it allegedly to only need to raise
seventeen billion dollars in twenty twenty seven. That number doesn't
make any sense considering the open ai already had to
bring forward its eight point three billion dollar fund raised
by at least three months. But let's stick with that idea.
Open ai believes it will be profitable somehow by twenty thirty,
(04:32):
and even if we assume that that means it intends
to burn over one hundred billion dollars total to get there,
is the plan to take open ai public and then
dumper toxic acx onto the public markets. Then it will
just flounder and convulse and die for everybody to seek.
Kind of sucks. Can you imagine open aiss one? How
old do you think this company would do when they
(04:54):
had a true financial audit from a major accounting firm,
and you can claim they already have one. I simply
don't belie leave you, this burn does not seem like
an accountancy firm would be good with them. And if
you want to know what all that looks like Google
we Work, which went from tech industry darling to joke
in a matter of days in part because it was
forced to disclose how bad things actually were when they
(05:14):
filed to go public. No really are linked to an
article I've linked in the spreadsheet called the Strangest and
most Alarming things in we Work's IPO filing. It's a
bad sign if your IPO filing is described as strange
and alarming. Those are not even terms of art, they're
just worrisome. With that in mind, I feel the same
way about Anthropic. Nobody is buying this company a one
hundred and seventy billion dollars, which is how much it's
(05:35):
raising it right now, and thus the only way to
access liquidity would be to take Anthropic public and show
the world how a company that made seventy two million
dollars in January and more than four hundred million dollars
in July twenty twenty five and also plans to lose
three billion dollars or more, and then you show them
that and let the market sign a fair price, especially
if the S one includes a bunch of strange and
(05:56):
alarming stuff that raises questions about whether anthropics value is
actually justified. I'm I'm stuttering here because everyone's acting like
these irrational things like yeah, our open ai, which burns
five billion dollars in twenty twenty four, probably ten or
more in twenty twenty five. If if not fifteen or twenty,
like that's fine, Like this is all good.
Speaker 3 (06:17):
They'll just go public one day, will they really?
Speaker 2 (06:20):
Really?
Speaker 3 (06:21):
Will they? No?
Speaker 2 (06:22):
Really, what's the plan? What's the plan here? Because the
arguments against my work always come down to costs will
go down and these products will become essential. Outside of
chat GPT, there's no real proof that these products are
anything remotely essential, And I'd argue there's very little about
chat GPT that Microsoft can't provide with rate limits via
co pilot. I'd also argue that essential is a very
(06:42):
subjective term, essentially in the sense that some people use
it as search doesn't mean it's useful for enterprises or
the majority of people, or could not be replaced by
anything else. And I guess chat GPT somehow makes one
billion dollars a month in revenue selling access to premium
versions of chat GPT, though I'm not one hundred percent
sure how Assuming it has twenty million customers paying twenty
(07:04):
bucks a month, that's four hundred million dollars a month.
Then five million business customers paying an average of one
hundred dollars a month each, that's nine hundred million dollars.
And is that is that really indicative? Is the average
really that good? Are there that many people paying thirty
five or fifty or two hundred bucks a month? How
many of those versus one hundred bucks? I mean, most
don't pay one hundred bucks. It's either thirty five, fifty
(07:27):
or two hundred. Open ai doesn't break out the actual
revenues behind these numbers for a reason. I believe that
reason is they don't look as good. I also, by
the way, when I was fucking around trying to play
with GPT five, I got offered a team subscription for
a dollar for one month. How how's how many people?
How many of those business customers got a month for
(07:47):
a dollar? How long does open ai counter customer? What
is open AI's churn like? And does it really as
I wrote in my news there, how much money do
open AI and Anthropic make in the year making more
than Spotify at one point five billion dollars a month.
We don't know an open AI, much like Anthropic, has
never shared actual revenues other than in June when they
said ten billion annualized revenue, choosing instead to leak to
(08:10):
the media and hope to wfiscape the actual amounts of
money being spent on its services. Anyway, long story short,
these companies are unprofitable with no end in sight, don't
even make that much money in most cases, and have
valued more than anybody would ever buy them for. And
they don't even have that much in the way of
valuable IP. And on top of that, the two biggest
players burn billions of dollars more than they make serving
(08:31):
these companies that also burn billions of dollars. I wish
someone would make this make sense because it doesn't. But
ed the government will give them more money forever, They'll
bail them out. I'm tired of this fucking point. I'm
so tired of hearing about bailouce. I'm so tired of
people saying they'll bail them out. I get that you're scared,
I get that things are grim, things are absolutely fucking grim,
(08:53):
But engage with reality here. You can't bail this out.
You can't bail it out even if this were going
to happen, will not. Who would they give the money
to and.
Speaker 3 (09:02):
For how long would they give it to? All AI startups?
Speaker 2 (09:06):
Is every startup going to get a paycheck protection program
for generative AI? How would that play out in rural,
red districts where big tech has never been popular, which
are being hit with both massive cuts to welfare as
well as the shock waves of betrayed war that has
made American agricultural exports like feedstocks, which previously went to
China by shiploads, less appealing worldwide. So they just, in
(09:26):
this scenario, by the way, bail out open AI then
stuff it full of government contracts to the tune of
what fifteen billion dollars a year? Right just in this example,
just to be clear, that's the low end of what
this would take to do. And they'll have to keep
doing it forever until Sam Wultman can build enough data
centers to what keep burning billions of dollars. There's no
plan to make this profitable. Say this happens, it won't
(09:49):
be Sam, Now what America has a bullshit generative AI
company attached to it, attached to the state that doesn't
really innovate. It doesn't really matter in any meaningful way
except that it owns a bunch of data centers, which
it doesn't yet. By the way, Microsoft owns all their infrastructure,
the core stuff is barely happening. Fucking hell, I just
don't think this happens. I think it's a silly idea,
(10:09):
and the most likely situation would be that Microsoft would
unhinge its daw and swallow open AI in its customer's whole. Hey, hey, hey,
did you know, by the way, the Microsoft's data center
construction is down year every year, and it's basically signed
no new data center leases.
Speaker 3 (10:22):
I wonder why it isn't building these new.
Speaker 2 (10:23):
Data centers or open AI could be anything. By the way,
Stargate isn't saving them either. As I've written recently, Stargate
doesn't actually exist beyond the media hype it generated, and
every single person who wrote big articles about Stargate being
real and Stargate being connected to Abilene Texas should be
fucking ashamed of themselves. It's deeply unprofessional anyway. That also
(10:45):
counts the hint of the involvement. There are several articles
that hint without directly saying that Abilene Texas was part
of the soft bank stargate thing. You can't wriggle out
of this one when the time comes or will be
ad by which I mean, I'm going to be pointing
at the articles aggressively. We have other stuff to do anyway.
(11:05):
By the way, does the government do this for everybody?
This very important question is everyone getting bailed out? Is
it just anthropical out and open AI? What about Google
and Microsoft? They're going to be burning and come on
a bunch of money. Why do they get the money
no one else? While their industries are going to turn
around and go wait wait, wait, wait, wait wait. Why
do I have to fucking make money when what's the
point is? Because AI is important, You're going to get
(11:25):
a bunch of fraud if this happens, it's not going
to Just to be clear, and what is the limit?
By the way, do they subsidize or compute for companies
like Cursor to what end? Where's the limit? How does
it end? What is the bailout? What are you bailing out?
Where does the bailout go? These are all questions that
would get answered. You can. You can sit there and say, oh,
(11:46):
the government's corrupt. Oh I feel this way and that way.
Here's the thing top, which is the thing that was
used to bail out banks and such and hedge ones,
and during that whole situation at least made more sense
because if a bank fails, people lose actual money. This
stuff fails, what do people lose? What is the service
that people lose access to? Chat GPT?
Speaker 3 (12:05):
That's it.
Speaker 2 (12:06):
I don't even think that Donald Trump could explain what
chat GPT is. And even then, he al really has
people to give him money, like these people don't have
anything to offer other than just this vague sense of
AI is important. I just don't buy it. And I
think that people if you're worried to get hopeful.
Speaker 3 (12:23):
I don't know.
Speaker 2 (12:24):
I'm hopeful that something's going to explode within this terrible
fucking industry. And I'm kind of hopeful because I think
it's time to ask a basic question, which is what
if generative AI just is not profitable? And I think
this is a question that we have to seriously consider
at this point because it's ramifications are significant. If I'm honest,
I think the future of large language models will be
(12:45):
largely client side on egregiously expensive personal setups or enthusiasts,
and in a handful of niche enterprise roles. Large language
models do not scale profitably, and their functionality is not
significant enough to justify the costs of running them. By
immediately applying old economics the idea that you would pay
a monthly fee to have relatively unlimited access, companies like
open ai and Anthropic immediately trained users to use their
(13:07):
products in a way that was antithetical to their costs.
Then again, had these models been served in that way,
had they been mindful of the cost and charged what
they actually were, there would likely have been no way
to get this far. If open ai is making billions
of dollars a month, or a billion dollars a month,
or whatever the fuck it is, it's possibly losing that
much or more after revenue, and that's the money it
can get selling the product in a form that can
(13:29):
never turn profitable itself. If open ai charged in line
with its actual costs, would even be able to justify
a free version of chat GPT outside of a few
requests a month. The revenue you see today is what
people are willing to pay for a product that loses money,
and I cannot imagine they would pay as much as
companies in question charge their costs. If I'm wrong, Curse
will be just fine. And that's assuming that Cursor's current
(13:51):
hobbled form is even profitable, which it has not said
it is. So you've got an entire industry of companies
that struggle to do anything other than lose a lot
of money. Great, and now we have a massive expansion
of data centers the likes of which we've never seen
or to capture demand for a product that nobody makes
much money selling. This naturally leads to an important question,
how do these people building data centers actually make money.
(14:24):
At the start of August, the Wall Street Journal published
one of the most worrying facts I've seen in the
last two years. The Journal quoted investor Paul Kodrowski is
saying that, and I quote as a percentage of gross
domestic products, spending on AI infrastructure has already exceeded spending
on telecom and internet infrastructure from the dot com boom,
and it's still growing. Kudrowski described this insane capex spending
(14:44):
as a sort of private sector stimulus program. The Journal
also quotes analyst Neil Dhuta as saying that CAPEX spending
for AI contributed more to growth in the US economy
in the past two quarters than all consumers spending. The
massive build out of data centers and the associate sheated
physical gear like chips, service some raw materials to building
them has become a massive dominant economic force, building capacity
(15:06):
for an industry that is yet.
Speaker 3 (15:07):
To prove it can make real money.
Speaker 2 (15:09):
And no Microsoft talking about it's a zero revenue it's
last quarterly earnings for the first time is not the
same thing as it stopped explicitly stating its AI revenue
in January when it was thirteen billion dollars annualized, so
about a billion a month. And by the way, that's revenue,
not profit. Just to remind you, I need to every
time remind anyone who ever mentions AI revenue that that's
(15:31):
not profit anyway. Aikapex allegedly, though I've heard some questions
about this figure, accounts to one point two percent of
the US GDP in the first half of the year,
and it accounted for more than half of the to
quote the Wall Street Journal, already sluggish one point two
percent growth of the US economy in the first half
of the year. Another Wall Street Journal piece published a
few days later discussed how data center development is souring
(15:53):
the free cash flow for big tech, turning them from
the kind of asset light businesses that the markets love
into entities burdened by physics, real estate and their associated
costs and I quote. For years, investors loved those companies
because they were asset lighting. They earned their profits or
on intangible assets such as intellectual property, software, and digital
platforms with network effects. This can be seen as a
(16:13):
metric called free cash flow, roughly defined as cash flow
from operations minus capital expenditures. This is arguably the purest
measure of business's underlying cash generating potential. From twenty sixteen
through twenty twenty three, free cash flow and net earnings
of Alphabet, Amazon, Meta, and Microsoft grew roughly in tandem,
but since twenty twenty three the two have diverged. The
(16:34):
four companies can by net income is up seventy three
percent to ninety one billion in the second quarter from
two years earlier, while free cash flow is down thirty
percent to forty billion. According to fact Sheet Data, Apple
a relative pika on capital spending has also seen free
cash flow lag behind. These numbers are all very scary,
and I mean that sincerely, but they also failed to
explain why how much was actually spent on AI capex
(16:58):
in the US. One would think two different articles on
the subject would include that number versus a single quarter's worth.
But from my estimates, I expect capital expenditures from the
Magnificent seven alone to crest two hundred billion dollars in
the first half of twenty twenty five, with Axios estimating
they'd spend around four hundred billion total this year. Most
articles are drafting off of a blog from Paul Kodrowski,
who estimates that total AI capex would be somewhere in
(17:20):
the region of five hundred and twenty billion dollars for
the total in the year, which felt conservative for me,
so I did the smart thing and asked him. Kudrowski
noted that these numbers focus entirely on the four big
spenders Microsoft, Google, Meta, and Amazon, and his own estimated
three hundred and twelve billion dollar capex and one point
two percent number came from the assumption that the US
GDP in twenty twenty five will be around twenty eight
(17:41):
trillion dollars, which is lower than other forecasts which put
it around thirty trillion. Even Kujoski, in his own words,
was trying to be conservative using public data and then
building his analysis from there. I personally believe his estimate
is too conservative because it does in factor in the
capital expenditures Maracle, which along with Crusoe is building the
vast abilenei Exis data center for Open AI or any
(18:02):
other private data center developers thinking cash into ar capex.
When I asked him to elaborate, he estimated that AI
spending all in was around half of three percent Q
two real GDP growth. He added that it could be
half of US GDP four year GDP growth if.
Speaker 3 (18:19):
Certain conditions are out. That's so cool.
Speaker 2 (18:22):
Half of the US economy's growth came from building data
centers for generative AI, which is the combined revenue of
a little more than the fucking smartwatch industry did in
twenty twenty four. Another troubling point is that big tech
doesn't just buy data centers and uses them, but in
many cases, pace for a construction company to build them,
fills them full of GPUs, then leases them from a
company that runs them, meaning that they themselves don't have
to personally staff up and maintain them. This creates an
(18:45):
economic boom for construction companies in the short term, as
well as lucrative contracts for ongoing support as long as
the company in question still wants them. While Microsoft or
Amazon might use a data center and indeed act as
if they own it, ultimately somebody else is holding the
bag and the ultimate responsibility for said data center. One
such company is QTS, the data center developer that leases
(19:06):
both to both Amazon and Meta according to The New
York Times, which was acquired by Blackstone in twenty twenty
one for ten billion dollars. Since then, Blackstone has used
commercially mortgage backed securities I know it's so great to
raise over eight point seven billion dollars since since then
to sink into qts's expansion, and as of mid July
said it'd be investing twenty five billion dollars in AI
(19:27):
data centers and energy. Blackstone, according to The New York Times,
see strong demand from tech companies who apparently are willing
to sign what they describe as airtight leases for fifteen
to twenty years to rent out data center space. As
a reminder that Microsoft has over one thousand lawyers in
the company anyway. Yet the Times also names another problem,
the unanswered question of how these private equity firms actually
(19:48):
exit these situations. Blackstone, KKR and other asset management firms
do not buy companies with the intention of siphoning off revenue,
but to pump them and sell them to another company
or dump them onto the public markets. Of course, much
like AI startups, it isn't obvious who would buy qts,
and what I imagine would be a twenty five or
thirty billion dollar valuation mean that black Stone would, as
I've mentioned, have to take them public. Similarly, KKR is
(20:11):
supposed fifteen billion dollar partnership with investment firm Energy Capital
Partners to build data centers and their associated utilities does
not appear to have much of an exit plan either.
And let's not forget Oracle Open AI and Cruso's abominable
mess in Abilene, Texas, where Oracle is paying for forty
billion dollars worth of GPUs and Crusoe is spending fifteen
dollars billion dollars raised from Blue Man, Owl Capital and
(20:32):
primary digital infrastructure. Though I think JP Morgan is involved
too to build data centers for open Ai, a company
that loses billion dollars.
Speaker 3 (20:39):
If yeah, why would they do that?
Speaker 2 (20:41):
Right? Why why would they do that? Well, it's simple,
you roube, you moron, you buffoon. It's so that open
Ai can allegedly, starting in twenty twenty eight, pay Oracle
thirty billion dollars a year for compute. And I am
being fully serious to be clear, open Ai, by my estimates,
has made only five point twenty six billion dollars this year,
and we'll have trouble meeting their twelve point seven billion
(21:03):
dollar revenue projection for twenty twenty five and will likely
lose more than ten billion dollars doing so. Oracle will,
according to the information, oh Crusoe one billion dollars in
payments across a fifteen year span of release. How does
Crusoe afford to pay back it's fifteen billion dollars in lungs?
Speaker 3 (21:18):
Beats me.
Speaker 2 (21:19):
The information says that Crusoe is raising a billion dollars
to take on the cloud giants by earning construction management
fees and rent and you can sell it's stake in
the project upon reaching certain completion milestones, while also building
its own Ai Compute, making the assumption that the demand
is there outside of hyperscalers. Then there's corp Eve, my
least favor company in the world. As I discussed a
few months ago, core Wave is burned by obscene debt
(21:41):
and a horrifying cash burn and has seen its stock spikee
to a high of one hundred and eighty three on
June twentieth, twenty twenty five, which has led to an
all stock attempt to acquire develop a core scientific for
nine billion dollars. And they're doing that with all stock,
and now that's falling apart because core Wave, as my
last check, and who knows where it will be by
the time this episode runs, is below one hundred bucks.
(22:02):
Because the IPO lockiles has gone.
Speaker 3 (22:05):
All the people that.
Speaker 2 (22:05):
Invested seem to be getting out. We'll see where they
go from here. Cor we've by the way, has since
gone public and had to borrow billions of dollars to
fund their obscene capital expenditures. And they are also by
the way, going to have to handle in October at
the beginning of their DDTL two point zero, their biggest loan,
and that's also when open Ai is allegedly going to
start paying them for their compute and by the way,
(22:28):
they lose hundreds of millions. They lose so much money
every quarter. They have no part to profitability. Ah, every
time I think about core Weave feel crazy and to
be honest, I'm gonna be honest when to put out
that blog, and people sent me some very unfair emails
about that. They said, there's no way that corl We've
can run out of money. If Corwe've runs out of money,
(22:49):
by the way, and they collapse, which I think is
very possible. I'm taking some fucking names on this one.
A lot of people very unfair to miss. The ztrum
corwave I had is also pretty much reliant on Microsoft
as one of its primary customers. While this relationship has
been fairly smooth so far as far as we know,
this dependence also presents the next eistential threat to core
we even is part of the reason why I'm so
(23:10):
pessimistic about their survival. Microsoft has its own infrastructure and
has every incentive to cut out middlemen when it's able
to meet supply with the demand itsself owns or lease
it's rather than subcontracts out simply because middlemen add costs
and shrink margins.
Speaker 3 (23:23):
If Microsoft walks, what's left?
Speaker 2 (23:25):
How does it service its its ongoing obligations and the debt.
In all of these cases, data center developers seem to
have very few options as to making actual money. We
have companies spending billions of dollars to vastly expend their
data center footprint, but very little evidence that doing so
results in revenue, let alone some sort of payoff event.
And similarly, the actual capital expenditures they're making are much
(23:46):
smaller than those of big tech. Digital Realty Trust, one
of the largest developers with over three hundred data centers
worldwide and five point five billion dollars in revenue in
twenty twenty four, only spend three point five billion dollars
in Capex last quarter, and Equinix eight point seven billion
dollars of revenue in twenty twenty four total, which has
two hundred and seventy data centers, but Capex at three
point five billion dollars for a quarter. Two NTT Global
(24:09):
Data Centers, which has over one hundred and sixty of
the fuckers, has dedicated ten billion dollars in Capex through
twenty twenty seven to build out more of them.
Speaker 3 (24:16):
Yeah, in many of.
Speaker 2 (24:17):
These cases, it's because these companies are, to quote a
source of mind, functionally obsolete for this cycle because legacy
data centers are not plug and play ready for GPUs
to slot into. And remember GPUs and the way they
work with GENERATIVEAI is just atypical. They burn them, they
run them hot constantly, They're running full fucking till the
whole fucking time. It's absolutely ridiculous, leading to a bunch
(24:38):
of replacements. But it just to be specific, requires a
bunch of specialized cooling, much more energy. And yeah, there's
just it's not a thing where you just take out
one computer and you put in another. And also any
investment in CAPEX by these companies would have to be
for both GPUs and said retrofitting. By the way, massive
amounts of money. And this means that the money flowing
(24:59):
into a ID data centers, it's predominantly going to neo
clouds like Corweven Crusoe, and all seems to flow back
to private equity firms that never thought about where the
cash out might be. Blackstone led corby seven point five
billion dollar a line with Magnetal Capital, who, by the way,
was involved with a bunch of subprime mortgages, and Crusoe
signed a deal a week ago with infrastructure firm and
they're also owned by Blackstone called tall Grass, and they're
(25:21):
building a data center in Wyoming, all of which seems
very good for Blackstone, unless you think, how does it
actually make money here? As private equity firms generally not
like to hold assets longer than five years, and the
timer is running. Even if it did, it's capital expenditures
that are dropping the bucket in the grand scheme of things.
Assuming Crusoe burns as the information suggests, it will as
much as four billion dollars in twenty twenty five. Corweave
(25:43):
spends as much as twenty billion dollars. Digital Realty Crust
spends as much as fourteen billion dollars. Global data centers
three point three three billion, that's ten billion over three years.
Equiniz spends fourteen billion. That's about fifty five point three
three billion dollars in AI CAPEK spent in twenty twenty
five from the largest developer of data centers in the world.
For some context, one hundred and two billion dollars were
(26:04):
spent by Meta Alphabet, Microsoft and Amazon in the last quarter.
Private equity may face the same problem as AI startups.
In the end, though, there is no real exit strategy
for any of these investments. Furthermore, the supposed AI capex
boom that's driving the US economy is not, as reported,
driven by the massive interest in driving AI infrastructure up
(26:24):
for a variety of customers.
Speaker 3 (26:25):
The reality is simple. The majority of AI capex.
Speaker 2 (26:28):
Is from big tech, which is a massive systemic weakness
for our economy. Now, while some might say that AI
(26:49):
capex has swallowed the US economy, I think it's more
appropriate to say that big tech capex has swallowed the
US economy. I also want to be clear that the economy,
which is the overall state of the country's production and
can asumption of stuff and the flow of money between
participants in said economy, and the markets as in the
stock market, well, those are two very different things. But
the calculations from Paul Kodrowski and others have now allowed
(27:10):
us to see where one might hit the other. You see,
the markets do not actually represent reality. While Microsoft, Amazon,
Google and Meta might want you to think there's a
ton of money in AI, their growth is mostly from
selling further iterations and contracts of their existing stuff, or
in the case of Meta further increasing its ad revenue.
The economy is where things are actually bought and sold,
representing the economic effects of both the things that happen
(27:31):
to build out the AI boom or whatever you call it,
and selling access to the services and the models themselves.
I recognize this is simplistic, but I'm laying it out
for a reason. As I discussed in the three part
Hater's Guide to the AI Bubble in video, is the
weak point in the US stock market, representing ninety percent
of the value of the Magnificent Seven, which in turn
makes up about thirty five percent of the value of
(27:53):
the U S stock market. The associated Magnificent Seven stocks
have seen a huge boom through their own growth, which
has been mistakenly in incorrectly attributed to revenue from AI,
which I've laid out previously is about forty billion dollars
of revenue in what in the last year or two,
maybe a little bit more. Nevertheless, the markets can continue
to be irrational because all they care about is number
going up. As the value of a stock is often
(28:14):
times disconnected from the value of the company itself, instead
associated with its kind of propensity for growth. It's fishy,
it sucks, and it's going to end badly. GDP and
other measurements of the economy aren't really something you can
fudge quite as easily, nor can you say a bunch
of fancy words to make people feel better in the
event that growth stores are the clients.
Speaker 3 (28:33):
This leads me to my principal worry that AI.
Speaker 2 (28:36):
Capex is actually a term for the expenditures of four companies,
namely Microsoft, Amazon, Google, and Meta, with Nvidia's GPU sales
being part of that capex too. What we include you
can if you want, include others like Oracle, XAI and
various neoclouds like Corweven, Crusoe, or who. According to DA
Davidson's gil Luria will account for about ten percent of
(28:56):
GPU sales from Nvideo in twenty twenty five. The reality
is that whatever our economic forces being driven by AI investment,
it's really just four companies building and leasing data centers
to burn on generatorvai a product that makes it relatively
small amount of money before losing a great deal more. Also,
want to be clear, Anthropic runs on Amazon and Google,
Open AI runs on Microsoft. These companies don't really own
(29:18):
their own infrastructure. That's what Open AI is allegedly building
with Oracle. Now forty two percent of Nvidia's revenue comes
from Magnificent seven per Laura Brand at Yahoo Finance, Big
up to Laura one of the best out there, which
naturally means that big tech is the lynchpin of investment
in data centers. I'll put it far more simply. If
AI capex represents such a large part of our GDP
and economic growth, our economy does on some level rest
(29:40):
on the back of Microsoft, Google, Meta, and Amazon and
their continued investment in AI. What should worry everybody is
that Microsoft, which makes up eighteen point nine percent of
Nvidia's revenue, assigned basically no leases in the last twelve months,
and it's committed data center construction and lamp purchases are
down year over year. Also, their data center capex is
not all to be AI, and all of these data
(30:02):
centers won't necessarily be used for that. In fact, nothing
they're doing suggests that they're super into it.
Speaker 3 (30:08):
It's not good.
Speaker 2 (30:10):
And while its CAPEX may not have dipped yet, in
part because the chip heavy nature of generative AI means
that CAPEX isn't exclusively dominated by property or construction. It's
now obvious that if it does, there will be direct
effects on both the US economy and the stock market,
as Microsoft is part of what amounts to a stimulus
packaged propping up America's economic growth. And not to repeat
(30:31):
the point too much, but Big Tech is yet to
actually turn anything resembling a profit on these data centers
and isn't making much revenue at all out of generative AI.
How exactly does this end?
Speaker 3 (30:41):
What is the plan?
Speaker 2 (30:42):
Here's Big Tech going to spend hundreds of billions of
dollars a year in CAPEX on generative AI in perpetuity?
Will they continue to buy more and more in video
chips as they do so, because it's always got to
be more. It can't be the same. They must get
more otherwise Nvidia doesn't grow At some point. Surely these
companies are built in enough data centers. Surely at some
point they will run out of space to put these
(31:03):
GPUs in. Is the plan to by then make so
much money from MAI that it won't matter? What does
Nvidia do at that point? How does the US economy
rebound from the loss of activity that follows?
Speaker 3 (31:12):
I don't fucking know. And it's worrying me.
Speaker 2 (31:16):
As I've said again and again, the genera IFAI bubble
is and always has been fundamentally irrational and inherently gothic,
playing in the ruins, patterns, and pathways of previous tech booms,
despite this one having little or no resemblance to them.
Though the tech industry loves to talk about building a
glorious future, its presence is one steeped in rituals of
death and decay, with the virtues of value creation and
(31:37):
productivity take a back seat to burning billions of dollars
and lying to the public again and again and again.
The way in which the media has participated in these
lives is disgusting. Venture capital, still drunk off the fumes
of twenty twenty one, keeps running the same old planebook,
shove as much money into a company as possible in
the hopes you can dump it onto an acquirer of
the public markets, only to get high on their own supply,
(31:59):
pushing valuations to the point that there's no possible liquidity
event for the majority of big private AI companies as
a result of their overstuffed valuations, burdens and business models,
and a lack of any real intellectual property, and like
the rest of the AI and bubble, Silicon Valley's only
liquidity path out of the bubble is big tech itself.
Without Google, character Dot AI, and WINDSORFS founders would likely
(32:20):
have been left for dead, and the same ghosts for Inflection,
And I'd even argue Scale AI, who's fourteen point three
billion dollar investment doing air quotes there for Meta effectively
decapitated the company, removing its CEO, Alexander Wang, leaving the
rest of the company to die playing are fourteen percent
of its staff on five hundred contractors mere weeks after
its CEO and investors cashed in. In fact, Generative AI
(32:42):
is turning out to be a fever dream entirely made up.
Speaker 3 (32:45):
Big tech.
Speaker 2 (32:46):
OpenAI would be dead if it wasn't for the massive
infrastructure provided by Microsoft at cost in returns for its
rights to its IP research and the ability to sell
its models, on top of tens of billions of dollars
of venture capital thrown into its billion dollar cache. Incinerator
Anthropic would be dead if both Google and Amazon, the
latter of which provides most of its infrastructure, hadn't invested
billions in keeping it alive. So that it can burn
(33:08):
three billion dollars or more in twenty twenty five while
fucking over its enterprise customers and rate limiting the rest.
The generative AI industry is, at its core unnatural. It
does not make significant revenue compared to its unbelievable costs,
nor does it have much revenue potential itself. It requires,
unlike just about every other software revolution, an unbelievable amount
of physical infrastructure to run because nobody but big tech
(33:30):
can afford to build the infrastructure necessary, creates very little
opportunity for competition or efficiency. As the markets are in
the throes of the growth of or costs rot economy,
they failed to keep big tech in line, conflating Big
text ability to grow with growth driven as a result
of the capital expenditures. Sensible reasonable markets would notice the
decay of free cash flow or ridiculousness of big text
(33:51):
capex bonanza, but instead they clap and squeal on oink
whenever Satching the Della jingles his keys. And I got
told the other day by someone that Satch in a
Della is not a is the citiot, that he's secretly
this big tech guy. And here's here's what I'd like
to say to them, it's fucking stupid. It's fucking dumb.
If this guy's actually technical, he's just a craven liar.
I actually think he's a bozo. Anyway, put in that side,
(34:14):
what's missing is any real value generation. Again, I tell you,
put aside any feelings you may have about generative AI
itself and focus on the actual economic results of this bubble.
How much revenue is there? Why is there no profit?
Why are there no exits? Why is big tech, which
has sunk hundreds of billions of dollars into generative AI,
not talk about the money they're making from it? Why
(34:35):
for three straight years have we been asked just wait
and see and for how long are we going to
have to wait? And when are we going to see it?
What's incredible is that the inherently compute intensive nature of
generative AI basically requires the construction of these facilities without
actually representing whether they are contributing to the revenues of
the companies that operate the models, like Anthropic or Open AI,
(34:55):
or any other business that builds on top of them.
As the models get more complet and hungary, more data
centers get built, which Hyperscaler's book as long term revenue,
even though it's subsidized by Hyperscalers or funded by VC money.
This in turn stimulates even more CAPEX spending, and without
having to answer any basic questions about longevity or market
fit longevity longevity.
Speaker 3 (35:16):
I don't know yet.
Speaker 2 (35:17):
The worst part of this financial farce is that we've
now got a built in economic breaking point in the
form of capex from AI. At some point, Kapex has
the slow if not because the lack of revenues are
massive costs associated, but because we live in a world
with finite space. And when said CAPEX slow happens, so
will the purchase of Nvidio GPUs, which in turn has
(35:38):
proven by Kadrowski and others slow America's economic growth. And
that growth is pretty much based on the whims of
four companies, which is an incredibly risky and scary proposition.
I haven't even dug into the wealth of private credit
deals that underpin build outs from private AI neo clouds
like core Wave, Cruise, So, Nebuis, and Lambda, in part
because their economic significance is so much smaller than the
big text ugly meanland the sprawl, and because I don't
(36:00):
like saying their names very much. To quote Paul Kodrowski,
we live in a historically anomalous moment. Regardless of what
one thinks about the merits of AI or explosive data
center expansion, the scale and pace of capital deployment into
a rapidly depreciating technology is remarkable. These are not railroads.
We aren't building century long infrastructure. AI data centers are
(36:22):
short lived, asset intensive facilities, writing declining cost technology curves,
requiring frequent hardware replacement to preserve margins, to which I say,
what margin's for? But nevertheless, you can't bail this out
because there is nothing to bail out. Microsoft, Meta, Amazon,
and Google have plenty of money and have proven they
can spend it in video is already doing everything it
(36:43):
can to justify people spending more on their GPUs. There's
little more it can do here other than soak up
the growth before the party ends. That CAPEX reduction will
bring with it a reduction in expenditures on those GPUs,
which will take a chunk out of the US stock market.
Although the stock market isn't the economy, the two are
inherently linked, and the popping of the AI bubble will
(37:03):
have down downstream ramifications, just that the dot com bubble did.
With the wider economy, expect to see an acceleration in
layoffs and offshoring in pop driven by a need for
tech companies to show for the first time in living memory,
fiscal restraint. For cities where tech is a major sector
of the economy thinks Seattle and San Francisco, there will
be knock on effects to those companies and the individuals
(37:23):
that support the tech sector, like restaurants, construction companies, building apartments,
uber drivers, and so on. We'll see a drying up
of VC funding, which is kind of already starting. Pension
funds will take a hit, which will affect how much
people have to spend in retirement.
Speaker 3 (37:37):
It'll be grim.
Speaker 2 (37:39):
Worse than that is the fact that these companies will
be by definition, non performing assets and ones that inflicted
an opportunity cost that will be impossible to calculate. While
a house, once built and sold, technically falls into that category,
doesn't add to any economic productivity. Obviously, people at least
need somewhere to live. Shelter is an essential component of life.
You can live without a data center the size of
(37:59):
Manhattan what would happen if companies like Microsoft and Meta
instead spent the money on things that actually drug productivity
or created a valuable competitive business that drove economic activity.
For example, Hell, even if they just gave everyone a
ten percent raise, it would likely be better for the
economy than this if we're factoring in things like consumer
spending instead. It's just waste, ugly, pointless waste. In summary,
(38:21):
we're already facing the prospect of a recession, and though
I'm not an economist, I can imagine being a particularly
nasty one. Given that the Magnificent seven account for forty
seven point eight seven percent of the Rustle one thousand
indexes returns in twenty twenty four. Even if big Tech
somehow makes this crap profitable, they won't. It's hard to
imagine that they'll be able to counterbalance any CAPEX reduction
(38:42):
with revenue, because there doesn't seem to be that much
revenue in generative AI to begin with. This is what
happens when you allow the rot economy to run wild,
building the stock market and tech industry on growth over
everything else. This is what happens when the tech media
repeatedly fails to hold the powerful to account catering to
their narratives and making excuses from are abominable billion dollar
losses and mediocre, questionably useful products. Well, for all you
(39:05):
want about the so called agentic era or anyonized revenues
that make you hot under the collar, I see no
reason for celebration about an industry with no exit plans
and needless capital expenditures that appeared to be one of
the few things keeping the American economy growing for effectively
no reason. I've been writing about the tech industry's obsession
with General IVAI for two years, and I've really never
(39:26):
felt more grim before. This was an economic uncertainty, a
way that our markets might contract, the big tech might
take a big haircunt, that a big a bunch of
money might have been wasted. But otherwise the world could
keep turning, right, it felt before Kutrowski said that thing,
and this was all sitting in public as well. I
don't know how I missed it. It felt until then
(39:47):
that it was like, Oh, this will really fuck up
big tech. It will hit their stocks. It'll be bad
for them now because the economy has slowed so much. Otherwise,
this data center, Capex and these four companies is holding
everything up and it won't lost. There's just no way
it can last forever, and there's not really anything to
bulk it up or and Microsoft's already pulling back.
Speaker 3 (40:08):
How does this end?
Speaker 2 (40:09):
I keep asking that the original title of this was
called how does this end? And I really don't know
how this ends. It feels as if everything is the
learning for disaster, and I really feel there's nothing that
can be done to avert in.
Speaker 3 (40:22):
I will be here to tell you what's happening. Is
it happens?
Speaker 2 (40:26):
And your third part for this week is going to
be about GPT five, which I'm going to be honest
doesn't really feel me full of hope or promise either.
Thank you for listening to Better Offline.
Speaker 1 (40:44):
The editor and composer of the Better Offline theme song
is Matasowski. You can check out more of his music
and audio projects at Mattasowski dot com, m A T
T O S O W s ki dot com. You
can email me an easy at Better Offline dot or
visit Better Offline dot com to find more podcast links
and of course, my newsletter. I also really recommend you
(41:06):
go to chat dot where'soead dot at to visit the
Discord and go to our slash.
Speaker 2 (41:10):
Better Offline to check out our reddit. Thank you so
much for listening.
Speaker 3 (41:15):
Better Offline is a production of cool Zone Media.
Speaker 2 (41:17):
For more from cool Zone Media, visit our website cool
Zonemedia dot com, or check us out on the iHeartRadio app,
Apple Podcasts, or wherever you get your podcasts.
Media.
Speaker 2 (00:05):
Hi, I'm Edzetron, and this is the second part of
this week's better offline. As ever, go to the episode
notes by some merch get the challenge coin, subscribe to
(00:25):
my newsletter if you like it, go premium. Either way,
you've got this episode for free if you can stand
the ads, which many of you can.
Speaker 3 (00:32):
Anyway.
Speaker 2 (00:33):
The thrust of this two part is that the AI
trade is going to, as you've probably guessed by now,
end badly, and as I'll also explain, I believe we're
now set up for potentially economic and market wrecking consequences
of this hysterical investment bubble. Today we're going to go
beyond just startups and into how the poison of the
bubble is crept into our economy and how dangerous things
(00:53):
are getting as a result. In the first installment, we
talked about Cursor, the AI coding startup that, despite having
no mote and zero sustainability, has somehow earned a valuation
of ten billion dollars, and then asked why none of
these generative AI companies are being acquired, at least not
in the traditional sense. And no, I'm not including the
acquisitions where a big company snags the talent and leaves
(01:14):
barely breathing bodies of startups in their wake. This matters
because if these AI companies can't get an exit, go public,
or get bought, it raises serious questions about how any
of this ends. Cursor is the best example, and the
fact that Anthropic is hopelessly dependent on Cursor for revenue
despite doing everything they can to kill it, makes the
entire situation feel a little bit weird. Cursor's paths to
(01:36):
viability seem frankly to be non existent, and the collapse
of Cursor will inevitably result in a vast chunk of
anthropics revenue is going up in smoke, which again will
also make its collapse seem inevitable, as how can it
fundraise that evaluation over its previous one when its biggest
customer just died? So much of what I'm describing as
structural and few companies have bigger structural obstacles to survive
(01:57):
all than open AI. As a reminder, open aia I
appears to have burned at least ten billion dollars in
the last two months. It has just raised another eight
point three billion dollars after raising ten billion dollars in
June according to The New York Times, and intends to
receive around twenty two point five billion dollars by the
end of the year from SoftBank, and that is assuming
it becomes a for profit entity by the end of
the year, and if that doesn't happen, the route gets
(02:18):
cut to twenty billion dollars total, meaning that SoftBank would
only be on the hook for a further one point
seven billion dollars. I am repeating myself, and I need
you to really get this. Open Ai just got ten
billion dollars in June, just in June, and had to
raise another eight point three billion dollars in August. And
that is an unbelievable cash burn, one dwarfing any startup
in history, raveled only by Xai, makers of Grok the
(02:40):
races dollar m which loses over a billion dollars a month.
I should also be clear that if open ai does
not convert to a for profit there is no path
forward to continue raising capital. Open Ai must have the
promise of an IPO. It must go public because a
valuation of three hundred billion dollars or more. Open Ai
can no longer be acquired because nobody has that much money.
And if Let's be real, nobody actually believes open ai
(03:03):
is worth that much. The only way to prove that
anybody does is to take open ai public, and that
will be impossible if it cannot convert. And ironically, SoftBank's
large and late stage participation makes exits actually harder, as
early investors will see their holdings diluted as a percentage
of total equity or whatever the hell we're calling it.
Because these aren't real equity shares, their profit participation units.
Speaker 3 (03:25):
They're a nonprofit. You can't do stock in them.
Speaker 2 (03:28):
While a normal company could just issue equity and deal
with the dilution that way, open AI's structure necess states
the negotiation, where companies can obstruct the entire process if
they see fit. Speaking of companies doing that, let's talk
about Microsoft. As I asked in my premium newsletter a
few weeks ago, what if Microsoft does not want open
ai to convert. They own all the IP, they own
(03:49):
all the access to open AI's research, and already run
most of their infrastructure, while assuming the best case scenario
that it would end up owning a massive junk of
the biggest tech stile up of all time. And I'm
talking about equity not open AI's current profit sharing units.
Microsoft might also believe that it stands to gain more
by letting open ai die and assuming its role in
the AI ecosystem. But let's assume that open ai converts
(04:12):
and now has to continue raising money at a rate
that will require it allegedly to only need to raise
seventeen billion dollars in twenty twenty seven. That number doesn't
make any sense considering the open ai already had to
bring forward its eight point three billion dollar fund raised
by at least three months. But let's stick with that idea.
Open ai believes it will be profitable somehow by twenty thirty,
(04:32):
and even if we assume that that means it intends
to burn over one hundred billion dollars total to get there,
is the plan to take open ai public and then
dumper toxic acx onto the public markets. Then it will
just flounder and convulse and die for everybody to seek.
Kind of sucks. Can you imagine open aiss one? How
old do you think this company would do when they
(04:54):
had a true financial audit from a major accounting firm,
and you can claim they already have one. I simply
don't belie leave you, this burn does not seem like
an accountancy firm would be good with them. And if
you want to know what all that looks like Google
we Work, which went from tech industry darling to joke
in a matter of days in part because it was
forced to disclose how bad things actually were when they
(05:14):
filed to go public. No really are linked to an
article I've linked in the spreadsheet called the Strangest and
most Alarming things in we Work's IPO filing. It's a
bad sign if your IPO filing is described as strange
and alarming. Those are not even terms of art, they're
just worrisome. With that in mind, I feel the same
way about Anthropic. Nobody is buying this company a one
hundred and seventy billion dollars, which is how much it's
(05:35):
raising it right now, and thus the only way to
access liquidity would be to take Anthropic public and show
the world how a company that made seventy two million
dollars in January and more than four hundred million dollars
in July twenty twenty five and also plans to lose
three billion dollars or more, and then you show them
that and let the market sign a fair price, especially
if the S one includes a bunch of strange and
(05:56):
alarming stuff that raises questions about whether anthropics value is
actually justified. I'm I'm stuttering here because everyone's acting like
these irrational things like yeah, our open ai, which burns
five billion dollars in twenty twenty four, probably ten or
more in twenty twenty five. If if not fifteen or twenty,
like that's fine, Like this is all good.
Speaker 3 (06:17):
They'll just go public one day, will they really?
Speaker 2 (06:20):
Really?
Speaker 3 (06:21):
Will they? No?
Speaker 2 (06:22):
Really, what's the plan? What's the plan here? Because the
arguments against my work always come down to costs will
go down and these products will become essential. Outside of
chat GPT, there's no real proof that these products are
anything remotely essential, And I'd argue there's very little about
chat GPT that Microsoft can't provide with rate limits via
co pilot. I'd also argue that essential is a very
(06:42):
subjective term, essentially in the sense that some people use
it as search doesn't mean it's useful for enterprises or
the majority of people, or could not be replaced by
anything else. And I guess chat GPT somehow makes one
billion dollars a month in revenue selling access to premium
versions of chat GPT, though I'm not one hundred percent
sure how Assuming it has twenty million customers paying twenty
(07:04):
bucks a month, that's four hundred million dollars a month.
Then five million business customers paying an average of one
hundred dollars a month each, that's nine hundred million dollars.
And is that is that really indicative? Is the average
really that good? Are there that many people paying thirty
five or fifty or two hundred bucks a month? How
many of those versus one hundred bucks? I mean, most
don't pay one hundred bucks. It's either thirty five, fifty
(07:27):
or two hundred. Open ai doesn't break out the actual
revenues behind these numbers for a reason. I believe that
reason is they don't look as good. I also, by
the way, when I was fucking around trying to play
with GPT five, I got offered a team subscription for
a dollar for one month. How how's how many people?
How many of those business customers got a month for
(07:47):
a dollar? How long does open ai counter customer? What
is open AI's churn like? And does it really as
I wrote in my news there, how much money do
open AI and Anthropic make in the year making more
than Spotify at one point five billion dollars a month.
We don't know an open AI, much like Anthropic, has
never shared actual revenues other than in June when they
said ten billion annualized revenue, choosing instead to leak to
(08:10):
the media and hope to wfiscape the actual amounts of
money being spent on its services. Anyway, long story short,
these companies are unprofitable with no end in sight, don't
even make that much money in most cases, and have
valued more than anybody would ever buy them for. And
they don't even have that much in the way of
valuable IP. And on top of that, the two biggest
players burn billions of dollars more than they make serving
(08:31):
these companies that also burn billions of dollars. I wish
someone would make this make sense because it doesn't. But
ed the government will give them more money forever, They'll
bail them out. I'm tired of this fucking point. I'm
so tired of hearing about bailouce. I'm so tired of
people saying they'll bail them out. I get that you're scared,
I get that things are grim, things are absolutely fucking grim,
(08:53):
But engage with reality here. You can't bail this out.
You can't bail it out even if this were going
to happen, will not. Who would they give the money
to and.
Speaker 3 (09:02):
For how long would they give it to? All AI startups?
Speaker 2 (09:06):
Is every startup going to get a paycheck protection program
for generative AI? How would that play out in rural,
red districts where big tech has never been popular, which
are being hit with both massive cuts to welfare as
well as the shock waves of betrayed war that has
made American agricultural exports like feedstocks, which previously went to
China by shiploads, less appealing worldwide. So they just, in
(09:26):
this scenario, by the way, bail out open AI then
stuff it full of government contracts to the tune of
what fifteen billion dollars a year? Right just in this example,
just to be clear, that's the low end of what
this would take to do. And they'll have to keep
doing it forever until Sam Wultman can build enough data
centers to what keep burning billions of dollars. There's no
plan to make this profitable. Say this happens, it won't
(09:49):
be Sam, Now what America has a bullshit generative AI
company attached to it, attached to the state that doesn't
really innovate. It doesn't really matter in any meaningful way
except that it owns a bunch of data centers, which
it doesn't yet. By the way, Microsoft owns all their infrastructure,
the core stuff is barely happening. Fucking hell, I just
don't think this happens. I think it's a silly idea,
(10:09):
and the most likely situation would be that Microsoft would
unhinge its daw and swallow open AI in its customer's whole. Hey, hey, hey,
did you know, by the way, the Microsoft's data center
construction is down year every year, and it's basically signed
no new data center leases.
Speaker 3 (10:22):
I wonder why it isn't building these new.
Speaker 2 (10:23):
Data centers or open AI could be anything. By the way,
Stargate isn't saving them either. As I've written recently, Stargate
doesn't actually exist beyond the media hype it generated, and
every single person who wrote big articles about Stargate being
real and Stargate being connected to Abilene Texas should be
fucking ashamed of themselves. It's deeply unprofessional anyway. That also
(10:45):
counts the hint of the involvement. There are several articles
that hint without directly saying that Abilene Texas was part
of the soft bank stargate thing. You can't wriggle out
of this one when the time comes or will be
ad by which I mean, I'm going to be pointing
at the articles aggressively. We have other stuff to do anyway.
(11:05):
By the way, does the government do this for everybody?
This very important question is everyone getting bailed out? Is
it just anthropical out and open AI? What about Google
and Microsoft? They're going to be burning and come on
a bunch of money. Why do they get the money
no one else? While their industries are going to turn
around and go wait wait, wait, wait, wait wait. Why
do I have to fucking make money when what's the
point is? Because AI is important, You're going to get
(11:25):
a bunch of fraud if this happens, it's not going
to Just to be clear, and what is the limit?
By the way, do they subsidize or compute for companies
like Cursor to what end? Where's the limit? How does
it end? What is the bailout? What are you bailing out?
Where does the bailout go? These are all questions that
would get answered. You can. You can sit there and say, oh,
(11:46):
the government's corrupt. Oh I feel this way and that way.
Here's the thing top, which is the thing that was
used to bail out banks and such and hedge ones,
and during that whole situation at least made more sense
because if a bank fails, people lose actual money. This
stuff fails, what do people lose? What is the service
that people lose access to? Chat GPT?
Speaker 3 (12:05):
That's it.
Speaker 2 (12:06):
I don't even think that Donald Trump could explain what
chat GPT is. And even then, he al really has
people to give him money, like these people don't have
anything to offer other than just this vague sense of
AI is important. I just don't buy it. And I
think that people if you're worried to get hopeful.
Speaker 3 (12:23):
I don't know.
Speaker 2 (12:24):
I'm hopeful that something's going to explode within this terrible
fucking industry. And I'm kind of hopeful because I think
it's time to ask a basic question, which is what
if generative AI just is not profitable? And I think
this is a question that we have to seriously consider
at this point because it's ramifications are significant. If I'm honest,
I think the future of large language models will be
(12:45):
largely client side on egregiously expensive personal setups or enthusiasts,
and in a handful of niche enterprise roles. Large language
models do not scale profitably, and their functionality is not
significant enough to justify the costs of running them. By
immediately applying old economics the idea that you would pay
a monthly fee to have relatively unlimited access, companies like
open ai and Anthropic immediately trained users to use their
(13:07):
products in a way that was antithetical to their costs.
Then again, had these models been served in that way,
had they been mindful of the cost and charged what
they actually were, there would likely have been no way
to get this far. If open ai is making billions
of dollars a month, or a billion dollars a month,
or whatever the fuck it is, it's possibly losing that
much or more after revenue, and that's the money it
can get selling the product in a form that can
(13:29):
never turn profitable itself. If open ai charged in line
with its actual costs, would even be able to justify
a free version of chat GPT outside of a few
requests a month. The revenue you see today is what
people are willing to pay for a product that loses money,
and I cannot imagine they would pay as much as
companies in question charge their costs. If I'm wrong, Curse
will be just fine. And that's assuming that Cursor's current
(13:51):
hobbled form is even profitable, which it has not said
it is. So you've got an entire industry of companies
that struggle to do anything other than lose a lot
of money. Great, and now we have a massive expansion
of data centers the likes of which we've never seen
or to capture demand for a product that nobody makes
much money selling. This naturally leads to an important question,
how do these people building data centers actually make money.
(14:24):
At the start of August, the Wall Street Journal published
one of the most worrying facts I've seen in the
last two years. The Journal quoted investor Paul Kodrowski is
saying that, and I quote as a percentage of gross
domestic products, spending on AI infrastructure has already exceeded spending
on telecom and internet infrastructure from the dot com boom,
and it's still growing. Kudrowski described this insane capex spending
(14:44):
as a sort of private sector stimulus program. The Journal
also quotes analyst Neil Dhuta as saying that CAPEX spending
for AI contributed more to growth in the US economy
in the past two quarters than all consumers spending. The
massive build out of data centers and the associate sheated
physical gear like chips, service some raw materials to building
them has become a massive dominant economic force, building capacity
(15:06):
for an industry that is yet.
Speaker 3 (15:07):
To prove it can make real money.
Speaker 2 (15:09):
And no Microsoft talking about it's a zero revenue it's
last quarterly earnings for the first time is not the
same thing as it stopped explicitly stating its AI revenue
in January when it was thirteen billion dollars annualized, so
about a billion a month. And by the way, that's revenue,
not profit. Just to remind you, I need to every
time remind anyone who ever mentions AI revenue that that's
(15:31):
not profit anyway. Aikapex allegedly, though I've heard some questions
about this figure, accounts to one point two percent of
the US GDP in the first half of the year,
and it accounted for more than half of the to
quote the Wall Street Journal, already sluggish one point two
percent growth of the US economy in the first half
of the year. Another Wall Street Journal piece published a
few days later discussed how data center development is souring
(15:53):
the free cash flow for big tech, turning them from
the kind of asset light businesses that the markets love
into entities burdened by physics, real estate and their associated
costs and I quote. For years, investors loved those companies
because they were asset lighting. They earned their profits or
on intangible assets such as intellectual property, software, and digital
platforms with network effects. This can be seen as a
(16:13):
metric called free cash flow, roughly defined as cash flow
from operations minus capital expenditures. This is arguably the purest
measure of business's underlying cash generating potential. From twenty sixteen
through twenty twenty three, free cash flow and net earnings
of Alphabet, Amazon, Meta, and Microsoft grew roughly in tandem,
but since twenty twenty three the two have diverged. The
(16:34):
four companies can by net income is up seventy three
percent to ninety one billion in the second quarter from
two years earlier, while free cash flow is down thirty
percent to forty billion. According to fact Sheet Data, Apple
a relative pika on capital spending has also seen free
cash flow lag behind. These numbers are all very scary,
and I mean that sincerely, but they also failed to
explain why how much was actually spent on AI capex
(16:58):
in the US. One would think two different articles on
the subject would include that number versus a single quarter's worth.
But from my estimates, I expect capital expenditures from the
Magnificent seven alone to crest two hundred billion dollars in
the first half of twenty twenty five, with Axios estimating
they'd spend around four hundred billion total this year. Most
articles are drafting off of a blog from Paul Kodrowski,
who estimates that total AI capex would be somewhere in
(17:20):
the region of five hundred and twenty billion dollars for
the total in the year, which felt conservative for me,
so I did the smart thing and asked him. Kudrowski
noted that these numbers focus entirely on the four big
spenders Microsoft, Google, Meta, and Amazon, and his own estimated
three hundred and twelve billion dollar capex and one point
two percent number came from the assumption that the US
GDP in twenty twenty five will be around twenty eight
(17:41):
trillion dollars, which is lower than other forecasts which put
it around thirty trillion. Even Kujoski, in his own words,
was trying to be conservative using public data and then
building his analysis from there. I personally believe his estimate
is too conservative because it does in factor in the
capital expenditures Maracle, which along with Crusoe is building the
vast abilenei Exis data center for Open AI or any
(18:02):
other private data center developers thinking cash into ar capex.
When I asked him to elaborate, he estimated that AI
spending all in was around half of three percent Q
two real GDP growth. He added that it could be
half of US GDP four year GDP growth if.
Speaker 3 (18:19):
Certain conditions are out. That's so cool.
Speaker 2 (18:22):
Half of the US economy's growth came from building data
centers for generative AI, which is the combined revenue of
a little more than the fucking smartwatch industry did in
twenty twenty four. Another troubling point is that big tech
doesn't just buy data centers and uses them, but in
many cases, pace for a construction company to build them,
fills them full of GPUs, then leases them from a
company that runs them, meaning that they themselves don't have
to personally staff up and maintain them. This creates an
(18:45):
economic boom for construction companies in the short term, as
well as lucrative contracts for ongoing support as long as
the company in question still wants them. While Microsoft or
Amazon might use a data center and indeed act as
if they own it, ultimately somebody else is holding the
bag and the ultimate responsibility for said data center. One
such company is QTS, the data center developer that leases
(19:06):
both to both Amazon and Meta according to The New
York Times, which was acquired by Blackstone in twenty twenty
one for ten billion dollars. Since then, Blackstone has used
commercially mortgage backed securities I know it's so great to
raise over eight point seven billion dollars since since then
to sink into qts's expansion, and as of mid July
said it'd be investing twenty five billion dollars in AI
(19:27):
data centers and energy. Blackstone, according to The New York Times,
see strong demand from tech companies who apparently are willing
to sign what they describe as airtight leases for fifteen
to twenty years to rent out data center space. As
a reminder that Microsoft has over one thousand lawyers in
the company anyway. Yet the Times also names another problem,
the unanswered question of how these private equity firms actually
(19:48):
exit these situations. Blackstone, KKR and other asset management firms
do not buy companies with the intention of siphoning off revenue,
but to pump them and sell them to another company
or dump them onto the public markets. Of course, much
like AI startups, it isn't obvious who would buy qts,
and what I imagine would be a twenty five or
thirty billion dollar valuation mean that black Stone would, as
I've mentioned, have to take them public. Similarly, KKR is
(20:11):
supposed fifteen billion dollar partnership with investment firm Energy Capital
Partners to build data centers and their associated utilities does
not appear to have much of an exit plan either.
And let's not forget Oracle Open AI and Cruso's abominable
mess in Abilene, Texas, where Oracle is paying for forty
billion dollars worth of GPUs and Crusoe is spending fifteen
dollars billion dollars raised from Blue Man, Owl Capital and
(20:32):
primary digital infrastructure. Though I think JP Morgan is involved
too to build data centers for open Ai, a company
that loses billion dollars.
Speaker 3 (20:39):
If yeah, why would they do that?
Speaker 2 (20:41):
Right? Why why would they do that? Well, it's simple,
you roube, you moron, you buffoon. It's so that open
Ai can allegedly, starting in twenty twenty eight, pay Oracle
thirty billion dollars a year for compute. And I am
being fully serious to be clear, open Ai, by my estimates,
has made only five point twenty six billion dollars this year,
and we'll have trouble meeting their twelve point seven billion
(21:03):
dollar revenue projection for twenty twenty five and will likely
lose more than ten billion dollars doing so. Oracle will,
according to the information, oh Crusoe one billion dollars in
payments across a fifteen year span of release. How does
Crusoe afford to pay back it's fifteen billion dollars in lungs?
Speaker 3 (21:18):
Beats me.
Speaker 2 (21:19):
The information says that Crusoe is raising a billion dollars
to take on the cloud giants by earning construction management
fees and rent and you can sell it's stake in
the project upon reaching certain completion milestones, while also building
its own Ai Compute, making the assumption that the demand
is there outside of hyperscalers. Then there's corp Eve, my
least favor company in the world. As I discussed a
few months ago, core Wave is burned by obscene debt
(21:41):
and a horrifying cash burn and has seen its stock spikee
to a high of one hundred and eighty three on
June twentieth, twenty twenty five, which has led to an
all stock attempt to acquire develop a core scientific for
nine billion dollars. And they're doing that with all stock,
and now that's falling apart because core Wave, as my
last check, and who knows where it will be by
the time this episode runs, is below one hundred bucks.
(22:02):
Because the IPO lockiles has gone.
Speaker 3 (22:05):
All the people that.
Speaker 2 (22:05):
Invested seem to be getting out. We'll see where they
go from here. Cor we've by the way, has since
gone public and had to borrow billions of dollars to
fund their obscene capital expenditures. And they are also by
the way, going to have to handle in October at
the beginning of their DDTL two point zero, their biggest loan,
and that's also when open Ai is allegedly going to
start paying them for their compute and by the way,
(22:28):
they lose hundreds of millions. They lose so much money
every quarter. They have no part to profitability. Ah, every
time I think about core Weave feel crazy and to
be honest, I'm gonna be honest when to put out
that blog, and people sent me some very unfair emails
about that. They said, there's no way that corl We've
can run out of money. If Corwe've runs out of money,
(22:49):
by the way, and they collapse, which I think is
very possible. I'm taking some fucking names on this one.
A lot of people very unfair to miss. The ztrum
corwave I had is also pretty much reliant on Microsoft
as one of its primary customers. While this relationship has
been fairly smooth so far as far as we know,
this dependence also presents the next eistential threat to core
we even is part of the reason why I'm so
(23:10):
pessimistic about their survival. Microsoft has its own infrastructure and
has every incentive to cut out middlemen when it's able
to meet supply with the demand itsself owns or lease
it's rather than subcontracts out simply because middlemen add costs
and shrink margins.
Speaker 3 (23:23):
If Microsoft walks, what's left?
Speaker 2 (23:25):
How does it service its its ongoing obligations and the debt.
In all of these cases, data center developers seem to
have very few options as to making actual money. We
have companies spending billions of dollars to vastly expend their
data center footprint, but very little evidence that doing so
results in revenue, let alone some sort of payoff event.
And similarly, the actual capital expenditures they're making are much
(23:46):
smaller than those of big tech. Digital Realty Trust, one
of the largest developers with over three hundred data centers
worldwide and five point five billion dollars in revenue in
twenty twenty four, only spend three point five billion dollars
in Capex last quarter, and Equinix eight point seven billion
dollars of revenue in twenty twenty four total, which has
two hundred and seventy data centers, but Capex at three
point five billion dollars for a quarter. Two NTT Global
(24:09):
Data Centers, which has over one hundred and sixty of
the fuckers, has dedicated ten billion dollars in Capex through
twenty twenty seven to build out more of them.
Speaker 3 (24:16):
Yeah, in many of.
Speaker 2 (24:17):
These cases, it's because these companies are, to quote a
source of mind, functionally obsolete for this cycle because legacy
data centers are not plug and play ready for GPUs
to slot into. And remember GPUs and the way they
work with GENERATIVEAI is just atypical. They burn them, they
run them hot constantly, They're running full fucking till the
whole fucking time. It's absolutely ridiculous, leading to a bunch
(24:38):
of replacements. But it just to be specific, requires a
bunch of specialized cooling, much more energy. And yeah, there's
just it's not a thing where you just take out
one computer and you put in another. And also any
investment in CAPEX by these companies would have to be
for both GPUs and said retrofitting. By the way, massive
amounts of money. And this means that the money flowing
(24:59):
into a ID data centers, it's predominantly going to neo
clouds like Corweven Crusoe, and all seems to flow back
to private equity firms that never thought about where the
cash out might be. Blackstone led corby seven point five
billion dollar a line with Magnetal Capital, who, by the way,
was involved with a bunch of subprime mortgages, and Crusoe
signed a deal a week ago with infrastructure firm and
they're also owned by Blackstone called tall Grass, and they're
(25:21):
building a data center in Wyoming, all of which seems
very good for Blackstone, unless you think, how does it
actually make money here? As private equity firms generally not
like to hold assets longer than five years, and the
timer is running. Even if it did, it's capital expenditures
that are dropping the bucket in the grand scheme of things.
Assuming Crusoe burns as the information suggests, it will as
much as four billion dollars in twenty twenty five. Corweave
(25:43):
spends as much as twenty billion dollars. Digital Realty Crust
spends as much as fourteen billion dollars. Global data centers
three point three three billion, that's ten billion over three years.
Equiniz spends fourteen billion. That's about fifty five point three
three billion dollars in AI CAPEK spent in twenty twenty
five from the largest developer of data centers in the world.
For some context, one hundred and two billion dollars were
(26:04):
spent by Meta Alphabet, Microsoft and Amazon in the last quarter.
Private equity may face the same problem as AI startups.
In the end, though, there is no real exit strategy
for any of these investments. Furthermore, the supposed AI capex
boom that's driving the US economy is not, as reported,
driven by the massive interest in driving AI infrastructure up
(26:24):
for a variety of customers.
Speaker 3 (26:25):
The reality is simple. The majority of AI capex.
Speaker 2 (26:28):
Is from big tech, which is a massive systemic weakness
for our economy. Now, while some might say that AI
(26:49):
capex has swallowed the US economy, I think it's more
appropriate to say that big tech capex has swallowed the
US economy. I also want to be clear that the economy,
which is the overall state of the country's production and
can asumption of stuff and the flow of money between
participants in said economy, and the markets as in the
stock market, well, those are two very different things. But
the calculations from Paul Kodrowski and others have now allowed
(27:10):
us to see where one might hit the other. You see,
the markets do not actually represent reality. While Microsoft, Amazon,
Google and Meta might want you to think there's a
ton of money in AI, their growth is mostly from
selling further iterations and contracts of their existing stuff, or
in the case of Meta further increasing its ad revenue.
The economy is where things are actually bought and sold,
representing the economic effects of both the things that happen
(27:31):
to build out the AI boom or whatever you call it,
and selling access to the services and the models themselves.
I recognize this is simplistic, but I'm laying it out
for a reason. As I discussed in the three part
Hater's Guide to the AI Bubble in video, is the
weak point in the US stock market, representing ninety percent
of the value of the Magnificent Seven, which in turn
makes up about thirty five percent of the value of
(27:53):
the U S stock market. The associated Magnificent Seven stocks
have seen a huge boom through their own growth, which
has been mistakenly in incorrectly attributed to revenue from AI,
which I've laid out previously is about forty billion dollars
of revenue in what in the last year or two,
maybe a little bit more. Nevertheless, the markets can continue
to be irrational because all they care about is number
going up. As the value of a stock is often
(28:14):
times disconnected from the value of the company itself, instead
associated with its kind of propensity for growth. It's fishy,
it sucks, and it's going to end badly. GDP and
other measurements of the economy aren't really something you can
fudge quite as easily, nor can you say a bunch
of fancy words to make people feel better in the
event that growth stores are the clients.
Speaker 3 (28:33):
This leads me to my principal worry that AI.
Speaker 2 (28:36):
Capex is actually a term for the expenditures of four companies,
namely Microsoft, Amazon, Google, and Meta, with Nvidia's GPU sales
being part of that capex too. What we include you
can if you want, include others like Oracle, XAI and
various neoclouds like Corweven, Crusoe, or who. According to DA
Davidson's gil Luria will account for about ten percent of
(28:56):
GPU sales from Nvideo in twenty twenty five. The reality
is that whatever our economic forces being driven by AI investment,
it's really just four companies building and leasing data centers
to burn on generatorvai a product that makes it relatively
small amount of money before losing a great deal more. Also,
want to be clear, Anthropic runs on Amazon and Google,
Open AI runs on Microsoft. These companies don't really own
(29:18):
their own infrastructure. That's what Open AI is allegedly building
with Oracle. Now forty two percent of Nvidia's revenue comes
from Magnificent seven per Laura Brand at Yahoo Finance, Big
up to Laura one of the best out there, which
naturally means that big tech is the lynchpin of investment
in data centers. I'll put it far more simply. If
AI capex represents such a large part of our GDP
and economic growth, our economy does on some level rest
(29:40):
on the back of Microsoft, Google, Meta, and Amazon and
their continued investment in AI. What should worry everybody is
that Microsoft, which makes up eighteen point nine percent of
Nvidia's revenue, assigned basically no leases in the last twelve months,
and it's committed data center construction and lamp purchases are
down year over year. Also, their data center capex is
not all to be AI, and all of these data
(30:02):
centers won't necessarily be used for that. In fact, nothing
they're doing suggests that they're super into it.
Speaker 3 (30:08):
It's not good.
Speaker 2 (30:10):
And while its CAPEX may not have dipped yet, in
part because the chip heavy nature of generative AI means
that CAPEX isn't exclusively dominated by property or construction. It's
now obvious that if it does, there will be direct
effects on both the US economy and the stock market,
as Microsoft is part of what amounts to a stimulus
packaged propping up America's economic growth. And not to repeat
(30:31):
the point too much, but Big Tech is yet to
actually turn anything resembling a profit on these data centers
and isn't making much revenue at all out of generative AI.
How exactly does this end?
Speaker 3 (30:41):
What is the plan?
Speaker 2 (30:42):
Here's Big Tech going to spend hundreds of billions of
dollars a year in CAPEX on generative AI in perpetuity?
Will they continue to buy more and more in video
chips as they do so, because it's always got to
be more. It can't be the same. They must get
more otherwise Nvidia doesn't grow At some point. Surely these
companies are built in enough data centers. Surely at some
point they will run out of space to put these
(31:03):
GPUs in. Is the plan to by then make so
much money from MAI that it won't matter? What does
Nvidia do at that point? How does the US economy
rebound from the loss of activity that follows?
Speaker 3 (31:12):
I don't fucking know. And it's worrying me.
Speaker 2 (31:16):
As I've said again and again, the genera IFAI bubble
is and always has been fundamentally irrational and inherently gothic,
playing in the ruins, patterns, and pathways of previous tech booms,
despite this one having little or no resemblance to them.
Though the tech industry loves to talk about building a
glorious future, its presence is one steeped in rituals of
death and decay, with the virtues of value creation and
(31:37):
productivity take a back seat to burning billions of dollars
and lying to the public again and again and again.
The way in which the media has participated in these
lives is disgusting. Venture capital, still drunk off the fumes
of twenty twenty one, keeps running the same old planebook,
shove as much money into a company as possible in
the hopes you can dump it onto an acquirer of
the public markets, only to get high on their own supply,
(31:59):
pushing valuations to the point that there's no possible liquidity
event for the majority of big private AI companies as
a result of their overstuffed valuations, burdens and business models,
and a lack of any real intellectual property, and like
the rest of the AI and bubble, Silicon Valley's only
liquidity path out of the bubble is big tech itself.
Without Google, character Dot AI, and WINDSORFS founders would likely
(32:20):
have been left for dead, and the same ghosts for Inflection,
And I'd even argue Scale AI, who's fourteen point three
billion dollar investment doing air quotes there for Meta effectively
decapitated the company, removing its CEO, Alexander Wang, leaving the
rest of the company to die playing are fourteen percent
of its staff on five hundred contractors mere weeks after
its CEO and investors cashed in. In fact, Generative AI
(32:42):
is turning out to be a fever dream entirely made up.
Speaker 3 (32:45):
Big tech.
Speaker 2 (32:46):
OpenAI would be dead if it wasn't for the massive
infrastructure provided by Microsoft at cost in returns for its
rights to its IP research and the ability to sell
its models, on top of tens of billions of dollars
of venture capital thrown into its billion dollar cache. Incinerator
Anthropic would be dead if both Google and Amazon, the
latter of which provides most of its infrastructure, hadn't invested
billions in keeping it alive. So that it can burn
(33:08):
three billion dollars or more in twenty twenty five while
fucking over its enterprise customers and rate limiting the rest.
The generative AI industry is, at its core unnatural. It
does not make significant revenue compared to its unbelievable costs,
nor does it have much revenue potential itself. It requires,
unlike just about every other software revolution, an unbelievable amount
of physical infrastructure to run because nobody but big tech
(33:30):
can afford to build the infrastructure necessary, creates very little
opportunity for competition or efficiency. As the markets are in
the throes of the growth of or costs rot economy,
they failed to keep big tech in line, conflating Big
text ability to grow with growth driven as a result
of the capital expenditures. Sensible reasonable markets would notice the
decay of free cash flow or ridiculousness of big text
(33:51):
capex bonanza, but instead they clap and squeal on oink
whenever Satching the Della jingles his keys. And I got
told the other day by someone that Satch in a
Della is not a is the citiot, that he's secretly
this big tech guy. And here's here's what I'd like
to say to them, it's fucking stupid. It's fucking dumb.
If this guy's actually technical, he's just a craven liar.
I actually think he's a bozo. Anyway, put in that side,
(34:14):
what's missing is any real value generation. Again, I tell you,
put aside any feelings you may have about generative AI
itself and focus on the actual economic results of this bubble.
How much revenue is there? Why is there no profit?
Why are there no exits? Why is big tech, which
has sunk hundreds of billions of dollars into generative AI,
not talk about the money they're making from it? Why
(34:35):
for three straight years have we been asked just wait
and see and for how long are we going to
have to wait? And when are we going to see it?
What's incredible is that the inherently compute intensive nature of
generative AI basically requires the construction of these facilities without
actually representing whether they are contributing to the revenues of
the companies that operate the models, like Anthropic or Open AI,
(34:55):
or any other business that builds on top of them.
As the models get more complet and hungary, more data
centers get built, which Hyperscaler's book as long term revenue,
even though it's subsidized by Hyperscalers or funded by VC money.
This in turn stimulates even more CAPEX spending, and without
having to answer any basic questions about longevity or market
fit longevity longevity.
Speaker 3 (35:16):
I don't know yet.
Speaker 2 (35:17):
The worst part of this financial farce is that we've
now got a built in economic breaking point in the
form of capex from AI. At some point, Kapex has
the slow if not because the lack of revenues are
massive costs associated, but because we live in a world
with finite space. And when said CAPEX slow happens, so
will the purchase of Nvidio GPUs, which in turn has
(35:38):
proven by Kadrowski and others slow America's economic growth. And
that growth is pretty much based on the whims of
four companies, which is an incredibly risky and scary proposition.
I haven't even dug into the wealth of private credit
deals that underpin build outs from private AI neo clouds
like core Wave, Cruise, So, Nebuis, and Lambda, in part
because their economic significance is so much smaller than the
big text ugly meanland the sprawl, and because I don't
(36:00):
like saying their names very much. To quote Paul Kodrowski,
we live in a historically anomalous moment. Regardless of what
one thinks about the merits of AI or explosive data
center expansion, the scale and pace of capital deployment into
a rapidly depreciating technology is remarkable. These are not railroads.
We aren't building century long infrastructure. AI data centers are
(36:22):
short lived, asset intensive facilities, writing declining cost technology curves,
requiring frequent hardware replacement to preserve margins, to which I say,
what margin's for? But nevertheless, you can't bail this out
because there is nothing to bail out. Microsoft, Meta, Amazon,
and Google have plenty of money and have proven they
can spend it in video is already doing everything it
(36:43):
can to justify people spending more on their GPUs. There's
little more it can do here other than soak up
the growth before the party ends. That CAPEX reduction will
bring with it a reduction in expenditures on those GPUs,
which will take a chunk out of the US stock market.
Although the stock market isn't the economy, the two are
inherently linked, and the popping of the AI bubble will
(37:03):
have down downstream ramifications, just that the dot com bubble did.
With the wider economy, expect to see an acceleration in
layoffs and offshoring in pop driven by a need for
tech companies to show for the first time in living memory,
fiscal restraint. For cities where tech is a major sector
of the economy thinks Seattle and San Francisco, there will
be knock on effects to those companies and the individuals
(37:23):
that support the tech sector, like restaurants, construction companies, building apartments,
uber drivers, and so on. We'll see a drying up
of VC funding, which is kind of already starting. Pension
funds will take a hit, which will affect how much
people have to spend in retirement.
Speaker 3 (37:37):
It'll be grim.
Speaker 2 (37:39):
Worse than that is the fact that these companies will
be by definition, non performing assets and ones that inflicted
an opportunity cost that will be impossible to calculate. While
a house, once built and sold, technically falls into that category,
doesn't add to any economic productivity. Obviously, people at least
need somewhere to live. Shelter is an essential component of life.
You can live without a data center the size of
(37:59):
Manhattan what would happen if companies like Microsoft and Meta
instead spent the money on things that actually drug productivity
or created a valuable competitive business that drove economic activity.
For example, Hell, even if they just gave everyone a
ten percent raise, it would likely be better for the
economy than this if we're factoring in things like consumer
spending instead. It's just waste, ugly, pointless waste. In summary,
(38:21):
we're already facing the prospect of a recession, and though
I'm not an economist, I can imagine being a particularly
nasty one. Given that the Magnificent seven account for forty
seven point eight seven percent of the Rustle one thousand
indexes returns in twenty twenty four. Even if big Tech
somehow makes this crap profitable, they won't. It's hard to
imagine that they'll be able to counterbalance any CAPEX reduction
(38:42):
with revenue, because there doesn't seem to be that much
revenue in generative AI to begin with. This is what
happens when you allow the rot economy to run wild,
building the stock market and tech industry on growth over
everything else. This is what happens when the tech media
repeatedly fails to hold the powerful to account catering to
their narratives and making excuses from are abominable billion dollar
losses and mediocre, questionably useful products. Well, for all you
(39:05):
want about the so called agentic era or anyonized revenues
that make you hot under the collar, I see no
reason for celebration about an industry with no exit plans
and needless capital expenditures that appeared to be one of
the few things keeping the American economy growing for effectively
no reason. I've been writing about the tech industry's obsession
with General IVAI for two years, and I've really never
(39:26):
felt more grim before. This was an economic uncertainty, a
way that our markets might contract, the big tech might
take a big haircunt, that a big a bunch of
money might have been wasted. But otherwise the world could
keep turning, right, it felt before Kutrowski said that thing,
and this was all sitting in public as well. I
don't know how I missed it. It felt until then
(39:47):
that it was like, Oh, this will really fuck up
big tech. It will hit their stocks. It'll be bad
for them now because the economy has slowed so much. Otherwise,
this data center, Capex and these four companies is holding
everything up and it won't lost. There's just no way
it can last forever, and there's not really anything to
bulk it up or and Microsoft's already pulling back.
Speaker 3 (40:08):
How does this end?
Speaker 2 (40:09):
I keep asking that the original title of this was
called how does this end? And I really don't know
how this ends. It feels as if everything is the
learning for disaster, and I really feel there's nothing that
can be done to avert in.
Speaker 3 (40:22):
I will be here to tell you what's happening. Is
it happens?
Speaker 2 (40:26):
And your third part for this week is going to
be about GPT five, which I'm going to be honest
doesn't really feel me full of hope or promise either.
Thank you for listening to Better Offline.
Speaker 1 (40:44):
The editor and composer of the Better Offline theme song
is Matasowski. You can check out more of his music
and audio projects at Mattasowski dot com, m A T
T O S O W s ki dot com. You
can email me an easy at Better Offline dot or
visit Better Offline dot com to find more podcast links
and of course, my newsletter. I also really recommend you
(41:06):
go to chat dot where'soead dot at to visit the
Discord and go to our slash.
Speaker 2 (41:10):
Better Offline to check out our reddit. Thank you so
much for listening.
Speaker 3 (41:15):
Better Offline is a production of cool Zone Media.
Speaker 2 (41:17):
For more from cool Zone Media, visit our website cool
Zonemedia dot com, or check us out on the iHeartRadio app,
Apple Podcasts, or wherever you get your podcasts.
Host
Ed Zitron
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
New Heights with Jason & Travis Kelce
Football’s funniest family duo — Jason Kelce of the Philadelphia Eagles and Travis Kelce of the Kansas City Chiefs — team up to provide next-level access to life in the league as it unfolds. The two brothers and Super Bowl champions drop weekly insights about the weekly slate of games and share their INSIDE perspectives on trending NFL news and sports headlines. They also endlessly rag on each other as brothers do, chat the latest in pop culture and welcome some very popular and well-known friends to chat with them. Check out new episodes every Wednesday. Follow New Heights on the Wondery App, YouTube or wherever you get your podcasts. You can listen to new episodes early and ad-free, and get exclusive content on Wondery+. Join Wondery+ in the Wondery App, Apple Podcasts or Spotify. And join our new membership for a unique fan experience by going to the New Heights YouTube channel now!