- 10 minutes ago
Category
🗞
NewsTranscript
00:00What are we getting, I don't want to say we, what are the bulls out there getting wrong about AI?
00:04Because you've made no secret that you don't see any there there.
00:09Yeah. So I think people are conflating a semiconductor rally with an underlying successful business, which doesn't really exist.
00:16Anthropics' current revenue growth, which is deeply questionable in the fact that they leaked profitability, manipulated by Elon Musk, of
00:22course,
00:23their growth is coming because people cannot measure how much an AI task actually costs.
00:28And a couple of months ago, Anthropics started charging their enterprise customers the actual token rates.
00:33What this has led to is suddenly businesses are going, oh, how much money are we spending?
00:38Uber's, I think it was Uber's COO, Andrew McDonald, said that they are having trouble justifying the AI spend based
00:46on the actual return.
00:47That one can actually measure it.
00:49So you've got a thing where you can't measure the costs and you can't measure the return on investment.
00:54What do you call that? You call it a thing without an ROI.
00:58Isn't it, though, safe to say, Ed, like, I think about this, that there is going to be, when there's
01:05any new technology, there's a lot of money thrown at something, right?
01:07A hot new idea, whatever it is.
01:10And that ultimately, like we've talked with some experts in the AI world, they're saying, oh, these large language models,
01:16not everything's going to exist.
01:17Because right now, it doesn't cost you anything to use it.
01:20But at some point, there will be a cost and people will start to select what they want to play
01:26with.
01:26And so is it a case of right now, I mean, at some point there will be some winners, some
01:32losers, and we'll see some things filter their way out.
01:35And that there will be some return on investment for some of the larger players when it comes to AI.
01:40Why do you not see some winners and losers in this, or some winners in this?
01:44I think the winner is Jensen Huang, and the winner is the construction firms who have got prepaid for all
01:49the data center construction.
01:51Maybe Sam Altman and Dario Amadei have become billionaires through this.
01:54But when it comes to the actual businesses, you can't find anyone who can measure the ROI, because you can't
01:59do it.
02:00So when it comes down to the model companies themselves, they're horrifyingly, horrifyingly unprofitable.
02:06I think OpenAI lost like 20-something billion dollars last year.
02:09Anthropik's probably not far behind this.
02:11And we don't have access to their books.
02:12Of course, of course.
02:13But once they file for an IPO...
02:15And that's the thing.
02:16When we see these S1s, I think it's going to be kind of a massacre.
02:21Because I think that people have this view that these companies are becoming more profitable, or even have a path
02:26to profitability.
02:27And they don't have one.
02:28But, you know, somebody could say, and Carol, we talk about this all the time, Amazon for years didn't really
02:34have profits.
02:35Well, that's the thing, though.
02:36Now it's a behemoth.
02:37And now it's a, you know, then AWS came along.
02:39And now it's this juggernaut that, you know, has made a lot of people very wealthy.
02:44But the thing with Amazon is, between 2003 and 2017, Amazon Web Services, actually all Amazon CapEx, was around $57
02:52billion normalized for inflation.
02:54Between that whole time.
02:55In that entire time.
02:56And they were cash flow positive in nine.
02:58The classic thing with the dot-com reconciliation thing as well.
03:02Things went poorly.
03:03But after that, there was useful infrastructure.
03:05No such story exists here.
03:07GPUs, AI GPUs specifically, have very few other use cases.
03:11And on top of that, if you're talking about incomplete data centers, it's going to cost just as much money
03:16to finish them today as it will in 10 years.
03:18And similarly, a Blackwell, a B200, GB200 rack, MVL72 rack, for example, will cost just as much, if not more,
03:25in electricity 10 years from now.
03:27There is no pick-up-and-fix this story.
03:30Perhaps there will be some on-device large-language model use.
03:33But we are talking an industry that has absorbed over a trillion dollars in the last three years.
03:38The media attention from everyone and these continual stories about the magic of AI.
03:44But when you say, hey, is it making any money?
03:46They go, oh, no.
03:47Oh, we couldn't possibly.
03:48We don't do that yet.
03:50But don't worry, Uber lost a lot of money.
03:52Uber burned about $32 billion, which is less than half of what Anthropic has raised in the last six months.
04:01So I think it's safe to say there's a lot of smart investors are in this.
04:05Smart companies are in it, right?
04:06Who are, I mean, pick your big hyperscaler.
04:10They obviously want to be around for a long time.
04:12So I'm just, like, how do you explain their bets on all of this?
04:18So I have this theory called the rock-com bubble, which is the hyperscalers are out of hyper-growth ideas.
04:24We haven't had a new iPhone.
04:25We haven't had a new cloud computing.
04:27We haven't really had a new platform that would turn into the next Google search or Amazon Web Services.
04:33So they thought that large language models were going to be the panacea, that they were going to be creators
04:38of API growth,
04:40that they were going to be the new infrastructure play, load up the GPUs, sell access to everyone.
04:44The problem is no one checked to make sure any of this was profitable.
04:48And we still don't know if GPU compute is actually margin positive, which is crazy considering the amount of money
04:55going into it.
04:56And when we look at the companies that actually are selling it, Corwee, for example, they burn billions of dollars,
05:0320-something billion dollars worth of CapEx this year, about the same amount in debt.
05:07And their customers are NVIDIA, OpenAI through Google, OpenAI through Microsoft, and Meta, who does not have an AI strategy.
05:16We're all, it's very strange, like that Twilight Zone episode with the scary child.
05:21Like, we're all trying to pretend that Meta actually has a plan here.
05:24They don't have one.
05:25So is Apple kind of smart?
05:26Because we've talked with our Mark Gurman about this.
05:29Have they kind of been relying on Google, right, and Gemini, and not necessarily doing that big spend?
05:34And they're not doing any CapEx on...
05:35No, exactly.
05:36Well, I think what happened with Apple is they released Apple Intelligence,
05:39which may be the most, the largest scale anti-tech radicalization in history.
05:44My non-tech friends, I've heard nothing about AI from them.
05:47Moment Apple Intelligence comes out, they're like, this is the worst thing I've ever used.
05:51And I think that made Apple go, maybe not.
05:55Maybe we should take a step back.
05:56But their CapEx is minimal.
05:58They're spending about as much in CapEx, ironically, as Amazon was spending on AWS way back when.
06:03So I think Apple is cautiously holding back and kind of absorbing the headline after headline,
06:08the ridiculous headline of Apple's behind in AI, behind in a losing proposition, more like.
06:13We're speaking with Ed Zitron.
06:15He's the founder and CEO of Easy Primary Research.
06:18He's got the Better Offline podcast.
06:19He writes the Where's Your Ed At newsletter.
06:24Ed, just so everybody's aware, do you have skin in the game?
06:28Are you shorting any of these companies?
06:29Do you have a financial interest in what you're talking about?
06:33I invest in words.
06:34I love doing the newsletter I do.
06:36I think I'm spreading a great deal of financial literacy and of how to pull apart the narratives
06:41of these companies.
06:41But no, I don't, because the market is irrational.
06:44And the market is inherently invested in something that I think is destructive, especially the
06:49SpaceX IPO.
06:50And OpenAI and Anthropic should not be allowed to go public.
06:53They are dangerous, lossy companies that are going to be added to indices that will sink
06:59401ks.
07:00You know, it's interesting that you say this, because I got a text on a group chat today
07:04from a friend of mine who's like not happy about the idea that he's, as an index fund
07:08investor, he's going to be forced to have to buy SpaceX, for example, which is sort of
07:14the subject of yesterday's big take.
07:15Everybody should check it out on the Bloomberg Terminal.
07:17There's a lot we want to get to with you.
07:19We only have a couple more minutes.
07:20You've got to come back and do some more with us.
07:22Before you go, what happens if markets, you said markets are irrational right now, if markets
07:29become efficient, we, you know, and I think a lot of people think markets are efficient.
07:33If markets view this the way that you're viewing this, what are the repercussions of that bubble
07:37bursting?
07:38I think the majority of gains based on AI success are going away.
07:43I think the neoclouds like Iron, CoreWeave, Cypher Mining, and TerraWolf as well, all of
07:51them, perhaps not going to zero, but they're going to be shells of the corporations they
07:54used to be.
07:55I actually don't know how CoreWeave survives this.
07:57I think NVIDIA is looking for the most aggressive haircut you've ever seen.
08:02For example, because they currently are not, the way I've, based on research I've done,
08:06it looks like it's taking about six to 12 months to install a single quarter's worth of GPUs
08:12from NVIDIA.
08:13They've yet to put out any kind of guidance about this.
08:15They clearly know.
08:17But if you go and look at Microsoft, for example, none of the data centers that they have announced
08:21have broken ground since 2023 have actually been finished.
08:26The Fairwater data centers, they are, they've said that that was fully complete.
08:31It's not even half complete.
08:33It's one of the strangest things I've seen in history.
08:36And I think what it is, is that these companies know that they can get away with it.
08:39They know that the SEC scrutiny isn't there.
08:42And they also know that the froth is big enough that everyone kind of loses together.
08:46It all comes out in the wash when it's proven that there is not a there there.
08:50Do you get calls from these companies?
08:52No, I don't hear back from Microsoft or OpenAI or any of them.
08:56But you reach out?
08:57I do.
08:57I ask for comment and they just, I guess my email gets lost.
09:00Sam Orman, Dario Amadei, there's PR people have mentioned coming on my show, but then they
09:05go quiet, probably because I'm quite critical of them.
09:07They know I'd ask the questions they don't want to answer.
09:09Why do you think, I want to poke at something that you mentioned earlier.
09:12Why do you think OpenAI and Anthropics shouldn't be allowed to go public?
09:15Because they are lossy companies.
09:18They do not have a path to profitability.
09:21Anthropics, they leaked or however, that they were profitable.
09:24They were profitable in two months when Elon Musk, they gave them a discount on their compute
09:29through SpaceX.
09:30Clearly planned leaks.
09:33Clearly planned financial engineering.
09:35No normal profitable company does things like this.
09:38OpenAI and Anthropics are both horribly unprofitable and they are not getting better.
09:44The information reported a few weeks ago, OpenAI had a negative 122% non-gap operating
09:51margin in the last quarter.
09:52But again, just because a company doesn't make money doesn't mean it shouldn't go public.
09:57Just because a company is not profitable doesn't mean it shouldn't go public.
10:00Let me reframe this then.
10:01I don't think these companies will survive.
10:03Okay.
10:04And I think that investors are doing exit liquidity for venture capital.
10:09And I think that these are dangerous, unstable companies that have faced basically no real
10:14scrutiny for most of the years of their investment.
10:17You have journalists and analysts who are actively cheering them on.
10:21This is a dangerous precedent and it's a dangerous thing for retail investors who are
10:26being conned by this.
10:27And I mean the whole ROI conversation, that is a result of most people experiencing AI
10:34via subscriptions that are subsidized, meaning you can burn anywhere from $3 to $13.50 for
10:39every dollar of your subscription.
10:41So people don't know how much this stuff actually costs and they are investing and believing in
10:44it and creating workflows around it based on services they likely cannot afford long-term.
10:50Just got 20 seconds here.
10:51So we have a story.
10:52Uber has set usage caps on some AI-powered tools used by its staff to manage costs after blowing
10:57through its AI budget earlier this year.
10:59We're going to increasingly see these stories in your view?
11:02Yes, we will.
11:02And we've already heard a similar story coming out of Walmart, I think Bloomberg reported yesterday.
Comments