Transcription
This video is brought to you by investing.com.
So, Michael Bur has been in the news a lot recently, and one of the things you've probably seen him talking about is fraud in the big AI companies. Now, I can already see the comments. Bur's predicted 100 out of the last two crashes. But, even though he's often early, a lot of the time, his arguments aren't wrong. So, when he starts posting a boatload of evidence that the AI companies are sneakily manipulating their financial data to make themselves look more valuable, yeah, I'm paying attention. And in fact, so is Nvidia and Palanteer, who have now both publicly responded to Michael Barry's claims.
So, what the heck is going on? Well, in this video, I'm going to break down what Barry has been saying, what Nvidia and Palanteer have come back with and analyze whether this puts the broader stock market in danger of falling off a cliff.
So, firstly, what has Barry been saying? Well, really, there are three main points he's been making over the past month. And the first is that there are a hell of a lot of circular deals happening between all of these AI companies which will really artificially inflate their revenues. For example, Barry recently shared this chart from Bloomberg that shows Nvidia, OpenAI, Microsoft, Oracle, AMD, XAI, Coreweee, and all the other major AI players. And they're just basically shifting money around between each other in a complex web of spending and investment. And on the surface, looks like a healthy ecosystem, right? Everyone is investing in each other. Everyone is spending billions on compute. Everyone's signing huge cloud deals. But according to Barry, this isn't organic demand at all. This is circular financing. The same dollars getting counted as revenue multiple times as they pass through the system.
You might have already seen this in my recent video, but there's a fantastic meme that explains this concept. Have a look at this.
>> And speaking of money, how about the 20 bucks you owe me?
>> Oh, yeah. Well, I only got 10, so here's 10. I owe you 10. Thanks.
>> Hey, Mo, you owe me 20. Well, here's 10. I'll owe you 10.
>> Uh-uh. You owe me 20. Here's 10. I owe you 10. Here's a 10 I owe you. Here's a 10 I owe you. Here's a 10 ow you.
>> Good, man. We're all even.
That is exactly how Barry sees the AI companies. But instead of settling debts by passing the money around in the case of the AI companies, they're using the $10 to invest in the next company who will then sign a deal and will spend $10 with the next company who will then do the same to the first company and so on and so forth.
About this charty poster, Barry noted, "The actual chart with all the give and take deals would be unreadable. The future will regard this a picture of fraud, not a flywheel. True end demand is ridiculously small. Almost all customers are funded by their dealers." And that line, almost all customers are funded by their dealers, is the key to understanding this because think about what's actually happening. Nvidia invests in Open AI. Open AAI uses that money to buy Nvidia's chips. Then Oracle jumps in. They spend tens of billions on Nvidia hardware. And then Open AI signs a $300 billion cloud deal with Oracle. Then startups like Coreweave, Figure, Harvey, Mistl, XAI. They're funded by the same AI companies that they're customers of. For example, Nvidia has invested in Corewave, Figure, and XAI. It's kind of like the picks and shovels analogy. The saying goes that it wasn't the gold miners who made money in the gold rush, right? It was in fact the companies supplying the picks and shovels. Well, in this case, it's kind of the same with Nvidia, but Nvidia's gone one step further and is investing in the gold miners to give them money to buy more Nvidia picks and shovels. So ultimately what looks like this amazing growth and this amazing revenue is really just the same pool of capital spinning through this web. And Barry's point is that almost none of this growth is coming from real end users. And this matters of course because the AI boom is being valued as if these revenues are very sticky. Like there is this enormous wave of new customer demand. But if the money is actually just flowing around in a loop between the same four or five companies, then the growth isn't real. It's manufactured.
Now, this leads us right into Bar's second point, which is that when you look at the real growth data from these AI companies, organic growth is actually slowing. Everyone is talking a big game with these circular deals. But if you strip all of that out and look at the core businesses that are supposed to be powering this AI revolution, like cloud computing, the picture looks completely different. Take the latest figures from Microsoft, Google, and Amazon. A few years ago, cloud growth was absolutely exploding. Amazon Web Services was growing nearly 40% year-over-year. Google Cloud was growing 45%. Microsoft Azure was still above 20% even at massive scale. But today, those numbers have started to fall. Michael Barry notes AWS growth has slowed into the teens. Google Cloud has decelerated significantly and even Microsoft, the biggest beneficiary of OpenAI, is not growing their cloud segment as quickly now. And this is exactly Bur's point. If AI workloads were genuinely driving a once- in a generation tech boom, you would expect to see cloud revenues rising exponentially. Instead, you're seeing the rate of growth slow. So going back to our prior point, Barry is just connecting the dots for us, showing us that this circular revenue is actually disguising something more concerning, the fact that real underlying growth seems to be slowing.
So those are Michael Barry's first two points. And before we get to the third, I just wanted to give a massive shout out to investing.com and their investing pro subscription, which is one of the main platforms I use for my stock research. I find investing pro to be a really handy tool, particularly to track my watch list. As you can see, you can track fair values very easily all in the one dashboard so that you don't miss a buying opportunity. Then in the ideas tab, you can also follow gurus like Buffett or Dalio or Clarman etc. And I also want to shout proix AI which compiles different portfolios based on investing themes and tracks them over time. For example, top value stocks, best of buffet or even financial fortresses. Then beyond that, they also have Warren AI which is very helpful when it comes to investing research. For example, it's even helped me do some research for this video, explaining how these circular deals can raise revenue without an actual uptick in user demand. Overall, if investing.com looks like something you'd be keen to try, use my referral code in the description or pinned comment or scan the QR code on screen and you will score 15% off. Definitely check it out if you're interested and thanks to investing.com for sponsoring the channel and supporting this content.
But for now, back to Barry. we have to talk about this third point he's been making and also Nvidia and Palanteer clapping back at him. So the third point he's been making in the fraud argument is actually quite interesting because this one is more directly manipulative and this is the one he's been talking about much more than anything else. Bur is calling out the AI companies for extending their depreciation schedules on old hardware aka artificially boosting their earnings. Firstly, shout out to Hamish, who recently did a banger video on this, breaking it down in great detail. Definitely worth a watch. But here's the simple version of what Barry was talking about. In this tweet, he shows that over time, the big tech companies are quietly increasing the useful life of their network and compute hardware. Meta's done it, Google's done it, Microsoft's done it, even Oracle's done it. Now, on paper, this sounds harmless. Companies are saying, "Our servers are lasting longer now." Okay, cool. But financially, this artificially boosts their earnings. Why? Well, when a company buys AI chips or servers, it doesn't take all that cost at once. It spreads that cost out over the number of years the hardware is supposed to be useful, that's depreciation. So, if you depreciate a chip over 3 years, the expense each year is quite big. But if you suddenly depreciate it over 6 years, the annual depreciation expense gets cut in half. And when your expenses fall, of course, your earnings magically rise. Nothing's changed in the business. No new customers, no extra profit, just an accounting decision. And Barry's argument is this. In a world where AI chips become obsolete faster than ever, why on earth are companies pretending they last longer? Especially when Nvidia itself is releasing a new generation of GPUs every 12 to 18 months, each one massively more powerful and efficient than the last. This is where Barry says the big fraud comes in. Because if new chips are two to three times more powerful and way more efficient, the economic useful life of old chips is surely shrinking, not growing. And we saw a perfect example of this with BYU recently, which Bur has drawn attention to. They extended the life of their service from 4 to 5 years and instantly added the equivalent of roughly $15 billion US to their profit. They then extended it again from 5 to 6 years and their earnings jumped again. Now this doesn't happen because the business improved but because the spreadsheet did. Meanwhile, in the same filing, BU admitted a lot of their old hardware is already obsolete and had to be written down. Now that contradiction, hardware becoming obsolete faster while useful life assumptions stretching longer is exactly what Barry is calling out. To him, this isn't a misunderstanding, it's fraud, one of the more common ones in the modern era in his words.
So, they are the three major points of Bur's argument. And it turns out he has gotten right up the nose of these big tech companies. Recently, both Nvidia and Palanteer have had very public clashes with Barry about his thesis. On CNBC, Alex K, Palanteer CEO, was asked about Barry's comments. And if I'm honest, he kind of just crashed out. Instead of addressing the core thesis, he basically just waved Barry off as a short seller who shouldn't be taken seriously.
>> You know, currently, as far as I can tell, the two companies he's shorting are the ones making all the money, which is super weird. Like the idea that chips and is what you want to short is batshit crazy.
>> You you were very defensive for instance around short sellers. Michael Bur putting the short on this weekend. Some people were wondering.
>> By the way, I I what I was defensive about, not defensive, I would say offensive about. I don't like short sellers. Pick some other company to screw with. But actually in the in the in the Bur case, I I actually think what was happening was market manipulation. I strongly suspect he was getting out of his position. And to get out of his position, he had to screw the whole economy by bismerching the best financials ever. Again, financials that are are helping the average person as investors on the battlefield, etc., etc.
>> Are you going to take legal action?
>> No, I'm just pointing it out. I mean,
>> yeah, if you ask me, that is what I would call a little AI tantrum.
And Michael Bur certainly didn't let that one go. He absolutely ripped him a new one following those comments. But I find this really interesting. And I don't know if Alex Karp is just like this, but when companies are genuinely confident about their numbers, they tend not to get riled up by naysayers. They're very open in walking through the data. They address the criticisms head on. But normally, it's when you've got something to hide. When someone hits a nerve like what we saw with the depreciation stuff, that's when you start to get this other set of behavior, the deflection, the indignation, and attacking the critic as opposed to the argument.
But interestingly, Alex Cup wasn't the only one because believe it or not, Nvidia also clapped back at Michael Bur through a seven-page memo to Wall Street analysts mentioning Barry by name. Now, this is extremely unusual. Nvidia is the largest company in the world right now. They don't usually punch down, but the fact that they felt the need to respond at all probably tells you that the criticism is getting a little too close to the bone. Now, unfortunately, I wasn't able to get access to the full document, which is pretty annoying. But from what I did see, it didn't sound like Nvidia actually addressed Bur's arguments in a meaningful way either. Take the depreciation issue. Their response was basically, "Customers depreciate GPUs over four to six years, and older chips like the A100 still run at high utilization and provide strong margins." I mean, okay, but they didn't provide any data to prove that these older chips are still economically productive. That doesn't refute the core argument that physical usage doesn't equal economic value and that extending the useful life just inflates earnings. And then on the circular financing topic, the memo didn't do much better. Nvidia said its investments are small relative to total global capital markets and that these companies earn revenue mostly from third parties. But honestly, that kind of misses the point entirely. The issue isn't whether these startups get money from Nvidia, it's that they spend money with Nvidia. It's that loop that artificially boosts Nvidia's reported demand. But from what I understand, they didn't really make any comment on that. And Barry's response on Substack was basically, "Yeah, they wrote a memo trying to discredit me, and I stand by every point." He even emphasized, "He's not calling Nvidia the next Enron, as Nvidia defended. He's more calling it Cisco, the perfectly legitimate, profitable company that still became the poster child of the overbuild because investors mistook a massive supply boom for genuine sustainable demand."
So, overall, that is Michael Barry's claim on fraud in the AI companies and what the AI companies had to say for themselves. I find this stuff really interesting. I would love to hear what you guys think as well. Please drop that stuff down in the comments below and please like the video as well if you did enjoy it. Thanks again to investing.com for supporting the channel. Remember, 15% off is yours to claim with the link in the description or pin comment. But with that said, thanks very much for watching, guys. Really appreciate it and I'll see you all in the next video.