At the end of last week Elon Musk’s much-hyped SpaceX shares took a tumble just over a month after the company went public.  But what does that mean for the big AI companies and for the so-called AI bubble.Today AI critic and host of the Better Offline podcast Ed Zitron on why he thinks it will all come crashing down. Featured: Ed Zitron, host of Better Offline podcast and EZ Primary Research CEO Subscribe to ABC News Daily on the ABC listen app.Program:More from ABC News DailyAustralia, Blockchain Technology, Science and Technology, Technology, Computer Science, AI, Chatbots, Business, Economics and Finance, Economy, Financial Markets, Stock MarketTranscriptSam Hawley: At the end of last week, Elon Musk's SpaceX shares took a tumble. Shine and hype had come off just over a month after it went public. What does that mean for the big AI companies and for the so-called AI bubble? Today's staunch AI critic and host of the Better Offline podcast, Ed Zitron, and why he thinks it will all come crashing down. I'm Sam Hawley on Gadigal land in Sydney. This is ABC News Daily. Ed, when SpaceX, Elon Musk's rocket and AI company debuted on the stock market, there was, to say the least, a lot of hype.News report: Three, two, one. SpaceX is about to go public and in the world of initial public offerings, there has never been a lift-off like the one investors will see on Friday.Sam Hawley: And well, initially it went really well for Musk, didn't it?Ed Zitron: I mean, kind of. He saw a 15-20% bump.Newsreader: SpaceX has raised $106 billion in the biggest ever initial public offering. The world's richest man, Elon Musk, has become the world's first trillionaire after investors got their first chance to buy shares in his company, SpaceX.Ed Zitron: And then fairly quickly, a reversal over the next few weeks. This stock was always a dog and I think anyone who didn't say that was trapped inside the Elon Musk industrial complex or just the hyperscaler reality created by the big tech firms.Sam Hawley: Yeah, but at the time, and you know, this is just over a month ago when it listed, it was thought that it was a pretty good sign for other AI companies like Anthropic and OpenAI. There were musings that if they listed, you know, these three companies could make over $200 billion US dollars this year alone. There was hype, but there was also hope.Ed Zitron: Right. I mean, but there's been hype and hope for this entire sordid era. The entire AI bubble has been about manufacturing consent for an AI industry that really has no business in existing. For LLM outcomes that don't exist, for statements about generative AI that have never actually touched reality, and for companies like OpenAI, Anthropic and SpaceX that will never turn a profit from their AI services.Sam Hawley: Just to note then, at the end of trading last week, as you allude to, SpaceX took a hit with its shares descending below its float price.News : After flying up to the moon, SpaceX shares seem to have encountered a rough landing. Prices down by a massive 33% from its record high.Sam Hawley: Well, Ed, let's consider then more deeply this AI boom and its future, particularly when it comes to OpenAI. Now, OpenAI and Anthropic, they both lodged IPOs last month, along with SpaceX. They said they intended to float this year, but that hasn't happened yet. Why do you think?Ed Zitron: Well, I think in OpenAI's case, they lost a little under $21 billion on $13.07 billion of revenue. These figures were confirmed by the Financial Times after I reported them. OpenAI's costs are massively outstripping its revenues. And yeah, I imagine that, plus the kind of dwindling hopes of the SpaceX IPO, have kind of made the idea of OpenAI listing kind of difficult.News report: OpenAI filed to go public last month, but the initial reactions from investors, it's true to say, giving the company cold feet and giving the whole market some shivers.Ed Zitron: OpenAI also, per the New York Times, were trying for a trillion dollar valuation. Apparently that was going to be more difficult, by which I mean impossible, than people realised. So, I think that's why that held back. I think Anthropic is in a weirder position, because they were doing gangbusters earlier in the year, and I'd be very curious to see where their growth is going. But fundamentally, the story about AI has moved from this kind of dream-like hysteria. Suddenly, we've gone from everybody saying that AI is the next big thing, to saying things along the lines of, well, you know, we can't spend too much on that. I think we're all coming to the realisation that the actual outcomes from AI, the actual ROI from AI, just isn't there. Maybe it never was. And the problem here is that nobody wants to really have the gruesome conversation here. The discussion of what if OpenAI collapses, what if Anthropic collapses, and what if all of this was a terrible waste.Sam Hawley: All right. Well, why don't we have this gruesome discussion then, Ed? Now, just to note, OpenAI, the reports are that perhaps it will delay its listing until 2027, until next year. And just to note that OpenAI, the figures that have been leaked, they have not confirmed. But let's just turn to your thesis then, that the AI bubble could very well burst. You say that the AI bubble is OpenAI. What do you mean by that?Ed Zitron: I mean that OpenAI is the vast majority of AI compute demand. I think that they represent anywhere between 70 and 90% of all compute demand. Their revenue makes up the majority of AI revenue with Anthropic. Nobody would have done any of this were it not for ChatGPT. And really, without OpenAI's continued subsidisation of its free users and even its paid users, AI would never have grown this big. There would have never been justification for all of these data centres, and there would have been no justification for investing in companies like Anthropic. These companies are only this way because they were enabled by hyperscalers, and they only continue to inflate because of OpenAI's influence over culture, which is directly a result of hyperscalers propping it up.Sam Hawley: Wow. Okay. So what you're saying is we only really care about this technology because of OpenAI, but surely it's moved on from that. I mean, it's quite beneficial, right? AI?Ed Zitron: Not really. I hate to say that. I hate to push back, but we actually do not have a tangible shred of proof that this is actually producing return on investment for anyone. We have engineers that swear by it, but we have studies that say that engineers are actually slower when they use generative AI. We have a lot of companies that have claimed they're laying people off because of AI, but that's AI-washing. Sam Altman himself said companies are AI-washing, which is they fire people claiming it's because of AI productivity, but what it really is is just they're trying to save costs, in some cases because of how much AI costs. And as far as what LLM-based services do, most people, ask AI boosters, they'll tell you code, search, and brainstorming. The same thing they said a year ago. That's because that's pretty much all these models can do, and they're all horribly unprofitable as well. I think fundamentally it's just an example of how much our reality and what good and bad and productive and ROI driving, how much of that is driven by what wealthy people think and want to believe, rather than what's actually happening. And at this point, the only way it ends is tears.Sam Hawley: And Ed, you are concerned, aren't you, that open AI, it could become the next Lehman Brothers, an investment bank, of course, which collapsed in 2008, which was a catalyst for the global financial crisis. What happens if open AI collapses? And there's no suggestion, of course, that it is going to, is there?Ed Zitron: Here's the thing. I have to push back. I don't even know why you needed to couch that by saying there was no... There are tons of reasons to believe this, because as I've said, OpenAI is the majority of AI revenue. It's the majority of AI use cases. It's the majority of AI writ large. Without open AI, if OpenAI can die, everyone who's currently been ignoring the sorry state of AI companies, saying that they'll just work it out, it's just like the dot-com bubble, which it is not, they will see these companies and say, oh right, none of these companies ever made sense. Perplexity, Cursor, Cognition, all of these companies that are perennially unprofitable. Suddenly, everyone will go from thinking, I'll believe anything these people say, to doubting everything they say. And everything falls apart from there, because the largest confidence game in the history of tech is OpenAI.Sam Hawley: Well, the head of OpenAI, Sam Altman, he told CNBC last month that while concerns are legitimate, he says companies will figure it out quickly.Sam Altman, OpenAI CEO : And I assume that the industry will figure that pretty quickly, but I think that is a fair issue. You do? Yeah. Quickly being? I would bet that by another year or two from now, there is a much better rationalisation of companies' spend relative to outcomes.Sam Hawley: I mean, is it possible that it is actually too early to say if this massive expenditure will or won't pay off?Ed Zitron: No. And that's the thing, that right there is exactly what the companies want you to say, which is, we're early, we're early, stop thinking about it. That is what the AI companies need you to do. We're not early. We're not early at all. More than a trillion dollars has been invested just by four companies into an industry that is yet to prove its worth. There is no magical fixing this. Maybe two years ago, they could have had some hope. But now we're at the point where they've taken up so much money, and so many people have signed their names on their bums and said, okay, well, mess me up. We're going to the moon together. Except there's not really a reconciliation point here.Sam Hawley: What about that counter argument, though, that the AI bubble, it's just too big to burst, that we're already in this sort of revolutionary moment that we can't really get off this train?Ed Zitron: I mean, we can, and we will. It's not going to destroy Microsoft, Google, Meta, and Amazon. But the AI bubble really ends when one of those or all of those companies agree to stop investing in data centers. When that stops happening, NVIDIA's revenue drops, and Micron's revenue drops, and SK Hynix, and all the RAM companies drops. All of the Taiwanese server manufacturers stop making AI servers as much, and so on and so forth. You can't bail that out. You can't bail out companies choosing not to spend money. Similarly, if open AI runs out of money, even if the government plugged that gap once or twice, it isn't going to save the fact that it can't afford to pay its bills long term. This means that you can't really bail it out. There's no bailoutable section. People misunderstand Too Big to Fail as a situation where they stopped companies collapsing. Yeah, they stopped AIG from going under so that AIG didn't crash the commercial paper loan market. Yeah, the whole layman situation was rough. Things were bailed out. They helped sort out things around layman. They still let many companies die. They still let the stock market go into 70%, 80% drawdown. They still let millions of people go homeless. They still led to massive unemployment. They didn't stop anything happening. They did that because if they didn't do that, banking would crash. Banking would stop. That was Too Big to Fail. There's not really troubled assets with AI to bail out. You can't bail out the fact that there really isn't real demand, and what demand exists is predominantly through subsidized subscriptions. This is not the same thing, and I think that when people kind of wish it to be, what they're trying to do is avoid thinking about hard things and thinking about dangerous and scary things. People aren't scared of a bailout. People are scared of something happening that they've never experienced before. This is a really unique situation, and uniquely dangerous because of the amount of retail money, regular people's money, that is now flowing into AI and AI-related things. There is no turning this around unless they manage to create a business that makes $300 billion of profit a year in the next two years, something undoable. All of this could have been avoided if everybody had bothered to scrutinize any of this before 2026.Sam Hawley: If you are right and there is this real potential that the AI bubble could burst, just tell me what would the AI sector look like after that, and what would it mean for all of us?Ed Zitron: I think that there is a real chance that AI as we know it goes away. I think that ChatGPT, the free version, I do not believe exists in the future. I don't know if Google will even have AI overviews. I think that large language models will either be boring, on-device things where you buy a massive $100,000 rig and you actually license it from a company, or just the kind of niche SKUs that hyperscalers only sell to specialists. They don't sell them to the general public. The idea of a general public LLM might actually be a thing of the past. And I think we will have an authority crisis in the media. I think we will have a legitimate problem with legitimacy within people who are covering these companies, because this is a catastrophic failure of the media's responsibility towards the general public. And the suffering will only be felt by regular people. It will only be felt by those who are most at risk, because the rich people will be fine. These people aren't going to suffer real consequences. And the only way that things change is if we, if anyone reporting on this subject, if anyone who has talked about AI, I think everyone involved in this needs to sit back and think, how did this inflate and what was my role in it?Sam Hawley: Ed Zitron is the host of the Better Offline podcast and the head of EZ primary research. This episode was produced by Ilaria Brophy. Audio production by Sam Dunn. Our supervising producer is Sydney Pead. I’m Sam Hawley.  Thanks for listening.