Transcript
A (0:00)
This is an iHeart podcast.
B (0:04)
So you've always dreamed of that state of the art home theater system. Here's the thing, if you invest well, you could get things like that. With Empower, you can get money working for you so you can go out and live a little. Isn't that why we work so hard to splurge sometimes, like on a massive high res TV or surround sound that puts you right in the action. So use Empower to help get good at money so you can be a little bad. Join their 19 million customers today@empower.com not an empower client, paid or sponsored.
A (0:30)
Run a business and not thinking about podcasting. Think again. More Americans listen to podcasts than ad supported streaming music from Spotify and Pandora. And as the number one podcaster, iHeart's twice as large as the next two combined. Learn how podcasting can help your business. Call 844-844-IHeart. Every case that is a cold case that has DNA right now in a backlog will be identified in our lifetime on the new podcast America's Crime Lab. Every case has a story to tell and the DNA holds the truth.
B (0:59)
He never thought going to get caught and I just looked at my computer screen, I was just like, ah, gotcha.
A (1:05)
This technology is already solving so many cases. Listen to America's Crime Lab on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts. When your car is making a strange noise, no matter what it is, you can't just pretend it's not happening.
B (1:23)
That's an interesting sound.
A (1:25)
It's like your mental health. If you're struggling and feeling overwhelmed, it's important to do something about it. It can be as simple as talk someone or just taking a deep calming breath to ground yourself. Because once you start to address the problem, you can go so much further. The Huntsman Mental Health Institute and the Ad Council have resources available for you at loveyourmindtoday.org.
B (1:47)
Call Zone Media. Hail Traveler. Welcome to this week's Better Offline monologue. I'm your host, Ed Zitron. Better Offline. So I'm reading this the day after Nvidia put out their August 27th earnings. Then I'm not sure anybody really knows how to react. While Nvidia beat estimates in general, they for the second quarter running missed analyst estimates on data center revenue, which is 88% of their revenue, where all the GPUs are coming in at $41.1 billion on an estimated $41.3 billion. The market also has a very unhealthy relationship with this company. Nvidia grew like 50% or something year over year, but because they didn't grow 112% or whatever year over year, people are the babies in the markets are soiling their diapers. And honestly, it's because really it's unhealthy. The unrealistic valuations that Nvidia has been given are finally causing problems. And as a reminder, Nvidia is the company that makes the specialized GPUs used in effectively all generative AI and has seen unprecedented stock growth as a result of this hype cycle. The magnificent seven stocks, Microsoft, Google, Meta, Amazon, Apple, Tesla, and of course Nvidia make up 35% of the value of the US stock market. And as a mentioned on the AI money trap, they ascend 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, Nvidia really is the AI bubble. They're the only company making a profit on AI and 42% 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 Carwer of Sherwood News discovered one particularly weird part of Nvidia's earnings this week. $2.2 billion 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 Core Weave. And by the way, Nvidia invested in CoreWeave has a $1.3 billion GPU compute deal with them called Project Osprey. They give them preferential rights, their GPUs and core weave. When they raise debt, they raise by using the GPUs they bought from Nvidia 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 Core Weave, I feel a little crazy. And in simple terms, Nvidia booked the increased value of their investment in Core Weave as revenue in a quarter where they barely hit estimates. Hmm. The reason I'm telling you this is that I believe, as I have always said, that all of this ends poorly. Nvidia isn't doing anything dodgy per se by including Core Weave stock. It's just that the value of that stock in Core Weave is covering up Nvidia's slowing growth. If Core Weave's value craters in the next few months, Nvidia 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'd I hear Mark to market before? And the answer is Enron. And I don't think this 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 my, this stock I 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, Nvidia's financial health and future earnings are somewhat dependent on CoreWeave. 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 Nvidia is effectively the only Success story in AI. OpenAI and anthropic burn billions of dollars each year and have no path to any kind of sustainability. And as I've argued again and again, these products lack any kind of large scale cases other than the fact that every single media outlet has been screaming ChatGPT 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. 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 believe the hype, the fact is that more people are beginning to wonder what if? Ed's right. So from here on out, 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, let's start with something simple. OpenAI is toast. The Financial Times reported that OpenAI's corporate restructuring is, and I quote, likely to slip into next year as Microsoft refuses to budge on the terms of that of OpenAI converting from a nonprofit to a for profit. Now this is because OpenAI went to Microsoft and said, okay guys, we need to convert to a for profit or we will die. Microsoft said, okay, great, let go ahead and do that. We don't mind OpenAI 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 they were willing to walk away from the table. I don't fucking blame them. Now people will say, well, Microsoft wants OpenAI to go public. They'll get a bunch of money. Kind of like Core we've did for Nvidia. Right? Except Microsoft already owns all their IP and already runs all their infrastructure. Remember, OpenAI doesn't own any of their infra at all. They're trying to build that bullshit out in Abilene, Texas, of course, but I mean, that's not going to be built this year or even next at this point. So, you know, Microsoft is not understandably jazzed to take this deal and could let OpenAI die on the vine. And if OpenAI fails to convert by the end of the year, SoftBank cuts their funding round in half, meaning that by my calculations, SoftBank's part of the remaining $30 billion, which would at this point become $10 billion, would drop to $700 billion thanks to OpenAI already raising $8.3 billion from other investors. Some people are saying it's $10 billion just from SoftBank. I don't think people are good at math. I swear to God, I've been looking at this shit for years and nevertheless, things don't look good there. And again, SoftBank has been very clear that if by the end of the year OpenAI is not converted, they're cutting that round in half. Even if OpenAI 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 OpenAI is trying to do. And if OpenAI cannot convert, they are dead because they can never IPO and they're too big to sell to anyone else. And thus investors would have little reason to put further capital in other than believing that OpenAI will 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. I'm going to screen it's going up. I'm getting to it later and in a future episode. And if the conversion fails, OpenAI can probably cobble together another $15 billion of funding before they shit their pants and die. And as an aside, OpenAI is currently doing an insider share sale at a $500 billion valuation with current and former employees selling $8 billion 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, 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 head at Anthropic, another company that burns billions of dollars, who will inevitably raise maybe another 1015 themselves and perhaps another roundup after that, with no path to profitability, 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 for you. It's an old game. You should have beaten it by now though, in the 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 their 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 gotta fucking stop. Once other revenue sources stop papering over the flimsy or non existent 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 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 LLMs 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 choose to blindly repeat that OpenAI and other companies are building powerful AI that would replace human workers, choosing to deliberately ignore the piss poor 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.