August 29, 2025 • 20 mins
In this week’s Better Offline monologue, Ed considers what would happen if he’s right about the AI bubble bursting - and the consequences of three years of myth-peddling journalism and markets addicted to growth at all costs.
FT: Microsoft talks set to push OpenAI’s restructure into next year
https://www.ft.com/content/b81d5fb6-26e9-417a-a0cc-6b6689b70c98
Sherwood: Nvidia, the asset manager, had a massive Q2 thanks to CoreWeave’s rally
https://sherwood.news/markets/nvidia-the-asset-manager-had-a-massive-q2-thanks-to-coreweaves-rally/
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 (00:01):
All Zone Media. Hail traveler, Welcome to this week's Better
offline monologue. I'm your host ed ZITRN, so I'm breeding
this the day after In Video put out there August
twenty seventh turnings, and I'm not sure anybody really knows
(00:21):
how to react. While in Video be estimates in general,
they for the second quarter running mist analyst estimates on
data center revenue, which is eighty eight percent of their revenue,
where all the GPUs are coming in at forty one
point one billion dollars on an estimated forty one point
three billion dollars. The market also has a very unhealthy
relationship with this company. In Video grew like fifty percent
or something year over year, but because they didn't grow
(00:43):
one hundred and twelve percent or whatever year over year,
people are the babies and the markets are soiling their diapers,
And honestly, it's because really it's unhealthy. The unrealistic valuations
that in Video has been given are finally causing problems.
And as a reminder, in Video is the company that
may the specialized GPUs used in effectively all generative AI
and has seen unprecedented stock growth as a result of
(01:05):
this hype. Cycle. The magnificent seven stocks Microsoft, Google, Meta, Amazon, Apple, Tesla,
and of course Nvidia make up thirty five percent of
the value of the US stock market, and as I
mentioned on the AI money Trap, they are sent a
capital expenditures, of which Nvidia makes up a large chunk.
By well, their GPUs do at least accounted for more
growth than all consumer spending combined. So on so many levels,
(01:29):
and Video really is the AI bubble. They're the only
company making a profit in AI, and forty two percent
of their revenue is from the magnificent seven companies buying
their GPUs, and their financial health and continued growth is
a barometer for the future growth of this nonsense, bullshit
bubble that I'm tired of talking about kind of, but
you know, there's always something to chat about yet. Luke
(01:49):
Carwo of Sherwood News discovered one particularly weird part of
Nvidia's earnings this week two point two billion dollars of
their revenue under net other income, which turned out to
be from the mark to market value of their investment
in unprofitable AI compute company cor Weve and by the way,
in Video invested in core Weave has a one point
(02:12):
three billion dollar GPU compute deal with them called Project Osprey.
They give them preferential rights their GPUs and core Weave.
When they raised debt, they raise by using the GPUs
they bought from a Videos as collateral in the loan,
and then they use that money from the from the
loan to buy more GPUs. Every time I think about
(02:33):
core Weave, I feel a little crazy. And in simple terms,
in Video book the increased value of their investment in
core Weave as revenue in a quarter where they barely
hit estimates. Hmmm. The reason I'm telling you this is
that I believe, as I've always said, that all of
this ends poorly in Video isn't doing anything dodgy per
(02:54):
se by including core Weave's stock. It's just that the
value of that stock and core Weave is covering up
in video is slow in growth. If Core weaves value
craters in the next few months, and Video will have
to book a massive loss in the same category. And
I know some of you are going to email me
and say, hey, where to hear mark to market before
And the answer is Enron. And I don't think this
(03:16):
is like Enron, because Enron was doing something somewhat different
they were valuing contracts in bizarre ways. This is not
exactly the same thing. Nevertheless, it's extremely weird to say, yeah,
the value of this stocker got over here went up.
Isn't that good? Yeah, that's revenue now. It's dangerous. It's
dangerous stuff. And it means that on some level, In
(03:39):
Video's financial health and future earnings are somewhat dependent on
core weave. It's all so good. I like it. It's very good.
It's normal. Everything's so good. It's great, everybody's doing well.
And what's even more better is that in Video is
effectively the only success story in AI, opening eye and
anthropic but billions of dollars each year and have no
(03:59):
path any kind of sustainability. And as I've argued again
and again, these products lack any kind of large scale
use cases other than the fact that every single media
outlet has been screaming chat GPT for the last three years,
and seemingly everyone just sees code generation goes, oh my god,
this is literally god. But I think after three years
of breathless hype, the worm is kind of turning on AI.
(04:21):
In the last week, I've been on national TV twice,
once on CNN and once on the BBC. And while
I'm still introduced as somebody that does not believed the hype,
the fact is that more people are beginning to wonder
what if Ed's right? So from here on now, I'm
going to extrapolate a little bit. I don't know the future.
I'm not saying, oh, this will definitely happen, but here
are some things that on my mind if I'm right. Well,
(04:44):
let's start with something simple. Open ai is toast. The
Financial Times reported that open AI's corporate restructuring is and
I quote likely to slip into next year as Microsoft
refuses to budge on the terms that of open ai
converting from a nonprofit to a for profit. Now. This
is because open ai went to Microsoft and said, okay, guys,
we need to convert to a for profit or we
(05:05):
will die. Microsoft said, okay, great, let's go ahead and
do that. We don't mind. Open Ai said, well, what
we're going to need is, we're going to need you
to end your exclusivity deal over our models. We're going
to need to cut your revenue share, We're going to
need to not share our IP and our research from
with you. What do you think? And Microsoft said, go
fuck yourself. Microsoft back in I think May said that
(05:27):
they were willing to walk away from the table. I
don't fucking blame them. Now people will say, well, Microsoft
wants open ai to go public, they'll get a bunch
of money. Kind of luck. Cool, we've different video, right,
Except Microsoft already owns all their IP and already runs
all their infrastructure. Remember open ai doesn't own any of
their info at all. They're trying to build that bullshit
out in Abilene, Texas, of course, but I mean that's
(05:50):
not going to be built this year or even next
at this point. So you know, Microsoft is not understandably
jazz to take this deal and could let open ai
die on the vie. And if open ai fails to
convert by the end of the year, soft Bank cups
they're funding round in half, meaning that by my calculations,
soft Bank's part of the remaining thirty billion dollars, which
would at this point become ten billion dollars, would drop
(06:12):
two seven hundred billion dollars thanks to open ai already
raising eight point three billion dollars from other investors. Some
people are saying it's ten billion dollars just from soft Bank.
I don't think people are good at math. I swear
to God, I've been looking at this ship for years,
and nevertheless things don't look good there. And again, soft
Bank has been very clear that if by the end
(06:33):
of the year open ai is not converted, they're cutting
that round in half. Even if open Ai and Microsoft
work this out today, I don't think it's possible that
they convert by the end of the year, if it's
even possible at all. Nonprofits are very specific things, and
no one has ever done what open ai is trying
to do. And if open ai cannot convert, they aren't
dead because they can never IPO and they're too big
(06:54):
to sell to anyone else. And thus investors would have
little reason to put further capital in other than believing
that open ai will so somehow stop burning billions of dollars,
something they have never proven is possible and nobody has
in fact, and fucking I hear the cost of inference
is going down again, and going to screen, it's going up.
I'm getting to it later and in a future episode.
And if the conversion fails, open Ai can probably cobble
(07:15):
together another fifteen billion dollars of funding before they shit
their pants and die. And as an aside, Open Ai
is currently doing an insider share sale at a five
hundred billion dollar valuation, with current and former employees selling
eight billion dollars worth of stock. This is not something
you do if you think you're going to go public
anytime soon. If ever, I need to be clear about this,
(07:37):
it looks bad and outside of Nvidia, AI is failing
to produce any kind of profit. And once venture capital
realizes this, they'll stop investing in these giant, loss ridden monstrosities.
This will naturally bring things to a headed Anthropic, another
company that burns billions of dollars, who will inevitably raise
maybe another ten to fifteen themselves, and perhaps another roundup after
(07:58):
that with nouffittability, though there is no reason, really no
reason to keep throwing cash in the furnace. And I
believe that Anthropic eventually gets absorbed into Amazon, like at
the end of that horrible game inside, and maybe it
will also flop around like a big blob in a field.
And if I spoiled inside of you, it's an old game,
you should have beaten it by now. Then when the
(08:19):
public markets, eventually the AI trade will fail. Despite what
the headlines may say, outside of Nvidia, the Magnificent seven
has not been making money on AI and has instead
been growing there already existing businesses, which is something that
reporters should know better. Yet they're claiming that any kind
of growth proves that AI bets are paying off. I'm
not singling you out yet, but you've gotta fucking stop.
(08:40):
Once other revenue sources stop papering over the flimsy or
unexistent revenue and I mean revenue, not profit from AI,
the street will realize and they will panic. But let's
get serious. If I'm right, we're all bearing witness to
one of the most catastrophic failures in the history of
the markets, journalism, and society itself. Generative AI was sold
as a myth. It's been blatantly obvious from the very
(09:01):
beginning that large language models do not have mass market
use cases and no coding llms do not goddamn count
they don't. They don't produce the kind of revenues. And
guess what. The coding lms are so expensive to run
that if you took away the subsidies, they would be
economically impossible. Stop arguing with me on this point. I
will crush you. Yet many many, many, many, many journalists
(09:24):
choose to blindly repeat the open AI and other companies
are building powerful AI that will replace human workers, choosing
to deliberately ignore the pisspoor outcomes from using these products
and I don't know, failing to define what powerful means
because powerful does not seem to mean goddamn anything. Journalists
(09:52):
had an opportunity to tell the truth, to look at
the products and say, hey, I don't really get why
this is so amazing or powerful, exactly, question the valuations
of companies that make little or no revenue, while burning
hundreds of millions or billions of dollars to interrogate the
outright lies of Sam Moultman and WARRYO Darryo ama Day. Instead,
they choose to fuel the fantastical fire, repeating the promises
(10:14):
of people like Sam Moultman in the hopes they'll be
the first to be able to parrot his next lie
or talk to his clammiass about nothing for three hours.
I worry that things are so much worse than we know.
I have an inkling that the price of serving GPU compute,
the underlying infrastructure of all generative AI, is horribly unprofitable.
And if you're a person who's going to say, oh,
(10:34):
it's only a few dollars an hour. Hey, what if
they're selling that few dollars an hour at a loss.
We know now that the cost of inference is increasing,
and that's coming up in future episodes, but it was
in the GPT five one, and thus the costs of
providing AI services, which were already unprofitable, is increasing too. Yet,
by constantly accepting a myth that AI would get more powerful,
(10:56):
that these companies would somehow work out how to stop
burning way more money than they may, the big tech
would not simply lie about what was possible. Journalism pedaled
outright lies and help sustain a hype cycle based on
little more than believing the last smart sounding guy you
ran into, or what that guy told someone you kind
of respect. You can dress this up by being an optimist.
(11:16):
You can dress this up by pretending that you thought
they would extrapolate from here, that these companies would just
work it out, that they were just like Uber, that
they were just like Amazon Web services, that they would
innovate and contribute you so much the economy based on
the things that they told you that they wanted you
to repeat, so their investors would give them more money.
(11:38):
You can pretend that all of this made sense at
any goddamn time, but you know you either half asked
it or you're lying to yourself at this point. And
I understand also that there are some people writing about
this who maybe don't want to write such fantastical things.
They feel pressured or they're scared, And I actually understand
the fear. Then, when everyone around you is pissed down
(12:00):
pants and they're saying, why haven't you pissed your pants?
Perhaps you wouldn't want to do it. But if everyone's
screaming at you, what are you meant to do? If
your editors are saying write a positive AI story, if
your editors are saying that thirteen percent of people in
the Stanford there was this. Sorry, I'm actually going to
reframe this point and not retake it, because this is important.
There was a study that went out that said that
(12:22):
thirteen percent I think there's a thirteen percent drop in
jobs that were affected by AI, and that this was
proof that AI was taking these jobs. When you look
at this goddamn study, you can see that it doesn't
actually prove that AI was involved and says that yeah,
thirteen percent of come of career paths and jobs that
(12:42):
were somehow being replaced by AI were dropping, and as
the result, AI was involved. It seems like a fumbled point,
but what it actually just said was that there was
a thirteen percent drop in certain employment and we'd like
AI to be involved. So it's on some level, I
realize i'm ranting. I realize I'm going a little bit offscript.
I'm just fucking tired of this. I'm tired because even now,
(13:07):
even today, even as you look at these companies burning
all this money, even as you look at the fact
that there are no real returns here, and when you
look at the products, you can see they don't really
do anything super futuristic. Even now, there are people who
still talk to me and repeatedly say shit like, well,
what if you're wrong, What if you're wrong about this? Well? Ed?
(13:28):
What if you turn around? And then the computer wakes
up and says, I love you, Ed, I love you.
Nothing happens because that won't happen, And we're living a lie.
And we built our economy on a lie. And it's scary.
It's scary because the economy is built on data center
growth and It's scary because so much of journalism just
missed this. They just missed it. And I myself have
(13:53):
done very well as a result of this, because I've
been elucidating hundreds of thousands words at a time, hours
a podcast a time. So I'm doing fine. I guess.
I don't have any stock positions in this crap. I'm
not investing. I'm not fiscally profiting off of this, but
my career is doing well, I guess. But on some level,
we've seen that at the highest levels of journalism there
(14:15):
are people who are willing to peddle bullshit. And as
this falls apart, as this collapses, there must be a
point at which people are held accountable for lying because
it is lying. And I believe the majority of journalists
are not trying to lie. I don't think that they
are deliberately doing this. I think they are weasel wording
around stuff so that they don't stand out from their
(14:37):
peers and get in trouble and face peer pressure for
not falling behind the AI bubble cycle. Except everyone's gonna
look fucking stupid if this collapses. If I'm right, everyone's
gonna look really silly, and you can feel however you
feel about me. This isn't about anyone specific other than
the very obvious hard forky folks. But if you're a
(15:00):
journous listening, this probably isn't abalue. What it is about, however,
is the truth, and it is about the fact that
everyone can say something that's wrong. And we as people
that report the truth, ostensibly must look at this when
it collapses, and I really do think it will, and
(15:20):
we must take an account of what has been said
and how it has been said, of how many affordances
were given to people like Sam Altman and Wario Ama Day,
of how many people said, wow, AI is going to
get more powerful. We must take an account of why
it was said it would get more powerful, and if
there was no reason to other than I think it
(15:42):
is number is going up that these people keep telling
me they have met smart people in the valley with
tears in their eyes, big strong men that told them
that AGI was coming. If that is the flimsy, fucking
reasoning we have used to send our economy into the toilet,
which is what I fear, we should all be goddamn
ashamed of ourselves. I'm not. I'm not ashamed because I've
(16:05):
admitted what I am from the very goddamn beginning, which
is just a guy with a microphone and a fucking blog,
and I'm doing my best out here to tell the truth.
If I am wrong, if I am truly wrong, if
I got all of this wrong, and AGI comes and
the computer eats me whole and shits me out like Kirby,
then I will eat crow and I will tell you
why I'm wrong in detail. Do you think other people
(16:28):
will do the same when the bubble bursts? Do you
think that these people will hold themselves accountable or do
you think they will sweep this under the fucking rug.
I think they will, and I don't believe listeners should
let them, and I will not be letting them either.
This isn't going to be an industry takedown, nor will
(16:48):
one be coming in the future, but I think journalism
needs to do a great deal better. And my frustration
is that in the end of all of this, the
people that will get hurt will be regular people. The
environmental destruction cannot be unwritten, the theft cannot be undone.
No class action suit can make up for the art directors,
(17:11):
the translators, the transcribers who are losing their jobs to
this right now, and who will even if these things
become too expensive to actually run at scale from here
on it, which is very possible. At some point, the
price pressure on these careers as they come back will
match the price of the AI, not the price of
the jobs that they had before. The labor destruction has
(17:35):
already happened on a much smaller level than they want
you to believe it, but it has happened. But fundamentally,
the AI bubble has damaged the concept of the truth
and has damaged the concept of what reporting is meant
to do, because an alarming amount of reporting has been
there to prove the AI bubble was real, rather than
(17:55):
prove what was actually real, to actually hold the accountable powerful,
to actually make sure that the truth was being said,
even if the truth was ikey, even if the truth
was wow. Tens of billions of dollars each quarter are
going into GPUs for products that only lose money, for
products that don't really resemble actual labor, and thus can
never do any of the things that people are promising.
(18:19):
And I like to believe that most people here mean well,
that they've just been going on with the market consensus
and I can understand that when everyone's honking in the
same way, do you want to oink? Do you want
to stand out? At a time when journalists have a
tough time keeping their jobs as it is, I blame
(18:40):
the people on the top. I blame the people with
the largest megaphones. I blame the people who are constantly
talking about this as the future and the people that
have derision for those who wouldare critique the powerful. And
if I'm right, really, if I'm right, there must be
(19:02):
some degree of accountability within journalism. There has to be
an apology to readers and listeners and viewers. There won't
be there, never will be there, never is, but something
has to change once the bubble bursts. And if i'm again,
if I'm wrong, I'll explain why I don't do this
to be right, I do it because I enjoy it,
(19:23):
and I'm very grateful you've give me your time. This
is meant to be a regular five to ten minute
monologue and it's going to go over twenty minutes. But
I'm frustrated, as you can tell, I'm furious. I'm furious
because the markets will take a nasty shit after this,
and once the markets realize that big tech doesn't have
any other ideas, and the one idea they had that
(19:43):
they've been wanking off about for years was nothing. Was
a myth that burned millions or billions of dollars. The
markets are not going to trust the tech industry anymore.
They're not going to trust them as the ultimate growth vehicles.
What do you think that does the startup funding? What
do you think that does two jobs in tech? Nothing good?
(20:06):
Nothing good at all. I want a better world, and
I think you do too. I'm very grateful for your time.
It's been a little different. I realized this week. Next
week we'll have Steve from Gamers Nextus, who will be
talking about the GPU black market in China and of
course the things that Bloomberg has done to pull down
that video. And then after that, I'm gonna have a
(20:29):
three part episode about how to argue with AI boosters
because it's about time we take the walter to these
fucking people. Thank you for listening.
All Zone Media. Hail traveler, Welcome to this week's Better
offline monologue. I'm your host ed ZITRN, so I'm breeding
this the day after In Video put out there August
twenty seventh turnings, and I'm not sure anybody really knows
(00:21):
how to react. While in Video be estimates in general,
they for the second quarter running mist analyst estimates on
data center revenue, which is eighty eight percent of their revenue,
where all the GPUs are coming in at forty one
point one billion dollars on an estimated forty one point
three billion dollars. The market also has a very unhealthy
relationship with this company. In Video grew like fifty percent
or something year over year, but because they didn't grow
(00:43):
one hundred and twelve percent or whatever year over year,
people are the babies and the markets are soiling their diapers,
And honestly, it's because really it's unhealthy. The unrealistic valuations
that in Video has been given are finally causing problems.
And as a reminder, in Video is the company that
may the specialized GPUs used in effectively all generative AI
and has seen unprecedented stock growth as a result of
(01:05):
this hype. Cycle. The magnificent seven stocks Microsoft, Google, Meta, Amazon, Apple, Tesla,
and of course Nvidia make up thirty five percent of
the value of the US stock market, and as I
mentioned on the AI money Trap, they are sent a
capital expenditures, of which Nvidia makes up a large chunk.
By well, their GPUs do at least accounted for more
growth than all consumer spending combined. So on so many levels,
(01:29):
and Video really is the AI bubble. They're the only
company making a profit in AI, and forty two percent
of their revenue is from the magnificent seven companies buying
their GPUs, and their financial health and continued growth is
a barometer for the future growth of this nonsense, bullshit
bubble that I'm tired of talking about kind of, but
you know, there's always something to chat about yet. Luke
(01:49):
Carwo of Sherwood News discovered one particularly weird part of
Nvidia's earnings this week two point two billion dollars of
their revenue under net other income, which turned out to
be from the mark to market value of their investment
in unprofitable AI compute company cor Weve and by the way,
in Video invested in core Weave has a one point
(02:12):
three billion dollar GPU compute deal with them called Project Osprey.
They give them preferential rights their GPUs and core Weave.
When they raised debt, they raise by using the GPUs
they bought from a Videos as collateral in the loan,
and then they use that money from the from the
loan to buy more GPUs. Every time I think about
(02:33):
core Weave, I feel a little crazy. And in simple terms,
in Video book the increased value of their investment in
core Weave as revenue in a quarter where they barely
hit estimates. Hmmm. The reason I'm telling you this is
that I believe, as I've always said, that all of
this ends poorly in Video isn't doing anything dodgy per
(02:54):
se by including core Weave's stock. It's just that the
value of that stock and core Weave is covering up
in video is slow in growth. If Core weaves value
craters in the next few months, and Video will have
to book a massive loss in the same category. And
I know some of you are going to email me
and say, hey, where to hear mark to market before
And the answer is Enron. And I don't think this
(03:16):
is like Enron, because Enron was doing something somewhat different
they were valuing contracts in bizarre ways. This is not
exactly the same thing. Nevertheless, it's extremely weird to say, yeah,
the value of this stocker got over here went up.
Isn't that good? Yeah, that's revenue now. It's dangerous. It's
dangerous stuff. And it means that on some level, In
(03:39):
Video's financial health and future earnings are somewhat dependent on
core weave. It's all so good. I like it. It's very good.
It's normal. Everything's so good. It's great, everybody's doing well.
And what's even more better is that in Video is
effectively the only success story in AI, opening eye and
anthropic but billions of dollars each year and have no
(03:59):
path any kind of sustainability. And as I've argued again
and again, these products lack any kind of large scale
use cases other than the fact that every single media
outlet has been screaming chat GPT for the last three years,
and seemingly everyone just sees code generation goes, oh my god,
this is literally god. But I think after three years
of breathless hype, the worm is kind of turning on AI.
(04:21):
In the last week, I've been on national TV twice,
once on CNN and once on the BBC. And while
I'm still introduced as somebody that does not believed the hype,
the fact is that more people are beginning to wonder
what if Ed's right? So from here on now, I'm
going to extrapolate a little bit. I don't know the future.
I'm not saying, oh, this will definitely happen, but here
are some things that on my mind if I'm right. Well,
(04:44):
let's start with something simple. Open ai is toast. The
Financial Times reported that open AI's corporate restructuring is and
I quote likely to slip into next year as Microsoft
refuses to budge on the terms that of open ai
converting from a nonprofit to a for profit. Now. This
is because open ai went to Microsoft and said, okay, guys,
we need to convert to a for profit or we
(05:05):
will die. Microsoft said, okay, great, let's go ahead and
do that. We don't mind. Open Ai said, well, what
we're going to need is, we're going to need you
to end your exclusivity deal over our models. We're going
to need to cut your revenue share, We're going to
need to not share our IP and our research from
with you. What do you think? And Microsoft said, go
fuck yourself. Microsoft back in I think May said that
(05:27):
they were willing to walk away from the table. I
don't fucking blame them. Now people will say, well, Microsoft
wants open ai to go public, they'll get a bunch
of money. Kind of luck. Cool, we've different video, right,
Except Microsoft already owns all their IP and already runs
all their infrastructure. Remember open ai doesn't own any of
their info at all. They're trying to build that bullshit
out in Abilene, Texas, of course, but I mean that's
(05:50):
not going to be built this year or even next
at this point. So you know, Microsoft is not understandably
jazz to take this deal and could let open ai
die on the vie. And if open ai fails to
convert by the end of the year, soft Bank cups
they're funding round in half, meaning that by my calculations,
soft Bank's part of the remaining thirty billion dollars, which
would at this point become ten billion dollars, would drop
(06:12):
two seven hundred billion dollars thanks to open ai already
raising eight point three billion dollars from other investors. Some
people are saying it's ten billion dollars just from soft Bank.
I don't think people are good at math. I swear
to God, I've been looking at this ship for years,
and nevertheless things don't look good there. And again, soft
Bank has been very clear that if by the end
(06:33):
of the year open ai is not converted, they're cutting
that round in half. Even if open Ai and Microsoft
work this out today, I don't think it's possible that
they convert by the end of the year, if it's
even possible at all. Nonprofits are very specific things, and
no one has ever done what open ai is trying
to do. And if open ai cannot convert, they aren't
dead because they can never IPO and they're too big
(06:54):
to sell to anyone else. And thus investors would have
little reason to put further capital in other than believing
that open ai will so somehow stop burning billions of dollars,
something they have never proven is possible and nobody has
in fact, and fucking I hear the cost of inference
is going down again, and going to screen, it's going up.
I'm getting to it later and in a future episode.
And if the conversion fails, open Ai can probably cobble
(07:15):
together another fifteen billion dollars of funding before they shit
their pants and die. And as an aside, Open Ai
is currently doing an insider share sale at a five
hundred billion dollar valuation, with current and former employees selling
eight billion dollars worth of stock. This is not something
you do if you think you're going to go public
anytime soon. If ever, I need to be clear about this,
(07:37):
it looks bad and outside of Nvidia, AI is failing
to produce any kind of profit. And once venture capital
realizes this, they'll stop investing in these giant, loss ridden monstrosities.
This will naturally bring things to a headed Anthropic, another
company that burns billions of dollars, who will inevitably raise
maybe another ten to fifteen themselves, and perhaps another roundup after
(07:58):
that with nouffittability, though there is no reason, really no
reason to keep throwing cash in the furnace. And I
believe that Anthropic eventually gets absorbed into Amazon, like at
the end of that horrible game inside, and maybe it
will also flop around like a big blob in a field.
And if I spoiled inside of you, it's an old game,
you should have beaten it by now. Then when the
(08:19):
public markets, eventually the AI trade will fail. Despite what
the headlines may say, outside of Nvidia, the Magnificent seven
has not been making money on AI and has instead
been growing there already existing businesses, which is something that
reporters should know better. Yet they're claiming that any kind
of growth proves that AI bets are paying off. I'm
not singling you out yet, but you've gotta fucking stop.
(08:40):
Once other revenue sources stop papering over the flimsy or
unexistent revenue and I mean revenue, not profit from AI,
the street will realize and they will panic. But let's
get serious. If I'm right, we're all bearing witness to
one of the most catastrophic failures in the history of
the markets, journalism, and society itself. Generative AI was sold
as a myth. It's been blatantly obvious from the very
(09:01):
beginning that large language models do not have mass market
use cases and no coding llms do not goddamn count
they don't. They don't produce the kind of revenues. And
guess what. The coding lms are so expensive to run
that if you took away the subsidies, they would be
economically impossible. Stop arguing with me on this point. I
will crush you. Yet many many, many, many, many journalists
(09:24):
choose to blindly repeat the open AI and other companies
are building powerful AI that will replace human workers, choosing
to deliberately ignore the pisspoor outcomes from using these products
and I don't know, failing to define what powerful means
because powerful does not seem to mean goddamn anything. Journalists
(09:52):
had an opportunity to tell the truth, to look at
the products and say, hey, I don't really get why
this is so amazing or powerful, exactly, question the valuations
of companies that make little or no revenue, while burning
hundreds of millions or billions of dollars to interrogate the
outright lies of Sam Moultman and WARRYO Darryo ama Day. Instead,
they choose to fuel the fantastical fire, repeating the promises
(10:14):
of people like Sam Moultman in the hopes they'll be
the first to be able to parrot his next lie
or talk to his clammiass about nothing for three hours.
I worry that things are so much worse than we know.
I have an inkling that the price of serving GPU compute,
the underlying infrastructure of all generative AI, is horribly unprofitable.
And if you're a person who's going to say, oh,
(10:34):
it's only a few dollars an hour. Hey, what if
they're selling that few dollars an hour at a loss.
We know now that the cost of inference is increasing,
and that's coming up in future episodes, but it was
in the GPT five one, and thus the costs of
providing AI services, which were already unprofitable, is increasing too. Yet,
by constantly accepting a myth that AI would get more powerful,
(10:56):
that these companies would somehow work out how to stop
burning way more money than they may, the big tech
would not simply lie about what was possible. Journalism pedaled
outright lies and help sustain a hype cycle based on
little more than believing the last smart sounding guy you
ran into, or what that guy told someone you kind
of respect. You can dress this up by being an optimist.
(11:16):
You can dress this up by pretending that you thought
they would extrapolate from here, that these companies would just
work it out, that they were just like Uber, that
they were just like Amazon Web services, that they would
innovate and contribute you so much the economy based on
the things that they told you that they wanted you
to repeat, so their investors would give them more money.
(11:38):
You can pretend that all of this made sense at
any goddamn time, but you know you either half asked
it or you're lying to yourself at this point. And
I understand also that there are some people writing about
this who maybe don't want to write such fantastical things.
They feel pressured or they're scared, And I actually understand
the fear. Then, when everyone around you is pissed down
(12:00):
pants and they're saying, why haven't you pissed your pants?
Perhaps you wouldn't want to do it. But if everyone's
screaming at you, what are you meant to do? If
your editors are saying write a positive AI story, if
your editors are saying that thirteen percent of people in
the Stanford there was this. Sorry, I'm actually going to
reframe this point and not retake it, because this is important.
There was a study that went out that said that
(12:22):
thirteen percent I think there's a thirteen percent drop in
jobs that were affected by AI, and that this was
proof that AI was taking these jobs. When you look
at this goddamn study, you can see that it doesn't
actually prove that AI was involved and says that yeah,
thirteen percent of come of career paths and jobs that
(12:42):
were somehow being replaced by AI were dropping, and as
the result, AI was involved. It seems like a fumbled point,
but what it actually just said was that there was
a thirteen percent drop in certain employment and we'd like
AI to be involved. So it's on some level, I
realize i'm ranting. I realize I'm going a little bit offscript.
I'm just fucking tired of this. I'm tired because even now,
(13:07):
even today, even as you look at these companies burning
all this money, even as you look at the fact
that there are no real returns here, and when you
look at the products, you can see they don't really
do anything super futuristic. Even now, there are people who
still talk to me and repeatedly say shit like, well,
what if you're wrong, What if you're wrong about this? Well? Ed?
(13:28):
What if you turn around? And then the computer wakes
up and says, I love you, Ed, I love you.
Nothing happens because that won't happen, And we're living a lie.
And we built our economy on a lie. And it's scary.
It's scary because the economy is built on data center
growth and It's scary because so much of journalism just
missed this. They just missed it. And I myself have
(13:53):
done very well as a result of this, because I've
been elucidating hundreds of thousands words at a time, hours
a podcast a time. So I'm doing fine. I guess.
I don't have any stock positions in this crap. I'm
not investing. I'm not fiscally profiting off of this, but
my career is doing well, I guess. But on some level,
we've seen that at the highest levels of journalism there
(14:15):
are people who are willing to peddle bullshit. And as
this falls apart, as this collapses, there must be a
point at which people are held accountable for lying because
it is lying. And I believe the majority of journalists
are not trying to lie. I don't think that they
are deliberately doing this. I think they are weasel wording
around stuff so that they don't stand out from their
(14:37):
peers and get in trouble and face peer pressure for
not falling behind the AI bubble cycle. Except everyone's gonna
look fucking stupid if this collapses. If I'm right, everyone's
gonna look really silly, and you can feel however you
feel about me. This isn't about anyone specific other than
the very obvious hard forky folks. But if you're a
(15:00):
journous listening, this probably isn't abalue. What it is about, however,
is the truth, and it is about the fact that
everyone can say something that's wrong. And we as people
that report the truth, ostensibly must look at this when
it collapses, and I really do think it will, and
(15:20):
we must take an account of what has been said
and how it has been said, of how many affordances
were given to people like Sam Altman and Wario Ama Day,
of how many people said, wow, AI is going to
get more powerful. We must take an account of why
it was said it would get more powerful, and if
there was no reason to other than I think it
(15:42):
is number is going up that these people keep telling
me they have met smart people in the valley with
tears in their eyes, big strong men that told them
that AGI was coming. If that is the flimsy, fucking
reasoning we have used to send our economy into the toilet,
which is what I fear, we should all be goddamn
ashamed of ourselves. I'm not. I'm not ashamed because I've
(16:05):
admitted what I am from the very goddamn beginning, which
is just a guy with a microphone and a fucking blog,
and I'm doing my best out here to tell the truth.
If I am wrong, if I am truly wrong, if
I got all of this wrong, and AGI comes and
the computer eats me whole and shits me out like Kirby,
then I will eat crow and I will tell you
why I'm wrong in detail. Do you think other people
(16:28):
will do the same when the bubble bursts? Do you
think that these people will hold themselves accountable or do
you think they will sweep this under the fucking rug.
I think they will, and I don't believe listeners should
let them, and I will not be letting them either.
This isn't going to be an industry takedown, nor will
(16:48):
one be coming in the future, but I think journalism
needs to do a great deal better. And my frustration
is that in the end of all of this, the
people that will get hurt will be regular people. The
environmental destruction cannot be unwritten, the theft cannot be undone.
No class action suit can make up for the art directors,
(17:11):
the translators, the transcribers who are losing their jobs to
this right now, and who will even if these things
become too expensive to actually run at scale from here
on it, which is very possible. At some point, the
price pressure on these careers as they come back will
match the price of the AI, not the price of
the jobs that they had before. The labor destruction has
(17:35):
already happened on a much smaller level than they want
you to believe it, but it has happened. But fundamentally,
the AI bubble has damaged the concept of the truth
and has damaged the concept of what reporting is meant
to do, because an alarming amount of reporting has been
there to prove the AI bubble was real, rather than
(17:55):
prove what was actually real, to actually hold the accountable powerful,
to actually make sure that the truth was being said,
even if the truth was ikey, even if the truth
was wow. Tens of billions of dollars each quarter are
going into GPUs for products that only lose money, for
products that don't really resemble actual labor, and thus can
never do any of the things that people are promising.
(18:19):
And I like to believe that most people here mean well,
that they've just been going on with the market consensus
and I can understand that when everyone's honking in the
same way, do you want to oink? Do you want
to stand out? At a time when journalists have a
tough time keeping their jobs as it is, I blame
(18:40):
the people on the top. I blame the people with
the largest megaphones. I blame the people who are constantly
talking about this as the future and the people that
have derision for those who wouldare critique the powerful. And
if I'm right, really, if I'm right, there must be
(19:02):
some degree of accountability within journalism. There has to be
an apology to readers and listeners and viewers. There won't
be there, never will be there, never is, but something
has to change once the bubble bursts. And if i'm again,
if I'm wrong, I'll explain why I don't do this
to be right, I do it because I enjoy it,
(19:23):
and I'm very grateful you've give me your time. This
is meant to be a regular five to ten minute
monologue and it's going to go over twenty minutes. But
I'm frustrated, as you can tell, I'm furious. I'm furious
because the markets will take a nasty shit after this,
and once the markets realize that big tech doesn't have
any other ideas, and the one idea they had that
(19:43):
they've been wanking off about for years was nothing. Was
a myth that burned millions or billions of dollars. The
markets are not going to trust the tech industry anymore.
They're not going to trust them as the ultimate growth vehicles.
What do you think that does the startup funding? What
do you think that does two jobs in tech? Nothing good?
(20:06):
Nothing good at all. I want a better world, and
I think you do too. I'm very grateful for your time.
It's been a little different. I realized this week. Next
week we'll have Steve from Gamers Nextus, who will be
talking about the GPU black market in China and of
course the things that Bloomberg has done to pull down
that video. And then after that, I'm gonna have a
(20:29):
three part episode about how to argue with AI boosters
because it's about time we take the walter to these
fucking people. Thank you for listening.
Host
Ed Zitron
Popular Podcasts
NFL Daily with Gregg Rosenthal
Gregg Rosenthal and a rotating crew of elite NFL Media co-hosts, including Patrick Claybon, Colleen Wolfe, Steve Wyche, Nick Shook and Jourdan Rodrigue of The Athletic get you caught up daily on all the NFL news and analysis you need to be smarter and funnier than your friends.
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
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