Transcription
I've heard you make a few predictions about the future of the economic markets. You know, you're famous for predicting 2008 and the financial crash that occurred then. I've heard you saying that you think because of AI there's going to be another financial crash around the corner within one or two years.
Yeah. What's happening with AI is a classic economic boom and bust cycle overlaid on the fact that AI can also eliminate a huge amount of employment which we've never seen that possibility in the past on that scale. But a common pattern in capitalism is that some new technology will be developed like railways for example. Some people see the potential profitability of railways. Everybody pours in creating railways. You get too many railways built. The 90% of the companies that create the railways go bust. But then we all have these rail systems that we benefit from afterwards. So that's the classic uh pattern of Joseph Schumpeter, who was the person who best described that, an Austrian economist from the early uh early 20th century. So he said you'll get the the banks will finance a new investment area. That investment produces a new technology which causes a boom while you're building the technology, but when the technology comes online, it undercuts existing businesses and causes a slump. So that's the boom and slump cycle, and AI is a a natural example of that. And what you get is massive overinvestment in the first instance because everybody who invests in AI has the ambition of being the only AI provider on the planet. Therefore, you get too many companies investing, there's too much money going into it. That's what causes a boom. But then when the technology comes online, because it undercuts existing technologies, you have a slump. And when you look at the investment taking place at the moment, the big tech companies, Meta, Amazon, Microsoft, Alphabet, and Google, Oracle is on track to spend 720 billion on AI infrastructure in 2026 alone, which is less than 20% of the revenue that they're making. We are seeing a 5:1 ratio of money being spent versus money coming in, which is historically unsustainable.
Yeah. And I think that's true. I there there has to be a slump coming out of this. And in in a sense, that's part of the natural cyclical behavior of capitalism, because if you want to make a profit, you've got to bring in technology that undercuts everybody you're currently rivals with. So that's the railways are a classic example there. You know, you have to get around by carriage instead. You undermine the carriage companies by bringing in the railways. But the ultimate benefit, society benefits because now you've got the railways for transportation. So that's the same sort of thing that AI is doing this time around. But companies, 90% of those companies are going to fail.
I mean, this is kind of what we're seeing already. So the failure rate of AI-specific startups has hit 90% in 2026.
Wow. That's luck.
Yeah. You you predicted that one correctly. Significantly higher than the 70% average for general technology. Roughly 95% of enterprise AI pilots fail to move into into production when they incur massive cost. The other thing I think a lot about is um a lot of startups now are raising a lot of money at crazy crazy valuations. I I can think of one particular startup I know, they're making like a couple of million dollars a year, they've raised at a billion dollar valuation, and because they've got the word AI on them. And the thing is, because everyone's so such in a frenzy at the moment about AI, they're probably going to raise a 2 billion dollar valuation 6 months from now.
When you think about what's going on there, someone somewhere is putting their money in, and they're going to lose it all. And they're going to lose it all. And when when everybody starts losing all their money very very quickly, you see this contraction, where everybody realizes that their paper gains, the gains they thought they had on paper because of valuation went up, have just evaporated. And when you see that, you have to quickly count your pennies, and get frugal, and pull back in again. Pull back in again. Lay people off and so on and so forth. So I I actually do personally believe that we're probably within 24 months of a pretty severe contraction. And that won't just impact these tech oligarchs, it'll impact all of us in different ways.
Yeah, it's a boom and bust cycle. I mean, the only thing which we've experienced in our own lives which is similar would be the telecommunications bubble and then the internet bubble between 1990 and 2001, 2002. Um, we don't get bubbles in the internet anymore because that's now a stable technology in that sense, uh, but it wasn't a big one. This is much bigger. What do we do as entrepreneurs, as team members and companies? What do we do at this moment if what you're saying is correct that there will be a boom and bust? A boom and bust? Which I, every smart person that I've spoken to agrees that there will be a bust soon.
Yeah.
Their timelines vary.
Yeah.
But what does one do right now in March, April 2026 to prepare for this?
Well, you put money aside if you can. You buy other assets you think are going to survive the the boom and bust cycle.
Like what?
That's the trouble. I mean, gold's been driven up. Gold's now been driven down. Uh, people are buying Bitcoin, but Bitcoin is collapsing as well. Uh, in some ways, you you really, you can't, it's like saying what do I do during an earthquake to not fall over in terms of insulating yourself? I really can't see a way of insulating yourself from the downturn. But I don't, I'm not as worried about that as the long-term consequences of AI, because this is the first technology which implies you can actually virtually eliminate labor as necessary for reducing output, because you can use AI rather than clerks. You can use, I know this is a long way from being feasible, but robots could replace process workers, and then suddenly something which employs 70% of the global population is no longer necessary. And then what do you do in that situation? What I've seen which I respect coming out of the tech bros in America is they're talking in terms of universal basic income.
You think a universal basic income is a good idea and we should probably explain what that is.
Yeah. Well, it's it's the state provides everybody with enough money to stay alive. That's the basic idea. Rather than having to work for a living, at the minimum, you get paid an amount of money that means you can buy the goods and services that are necessary to stay alive. You don't necessarily prosper, but you get enough to survive. And so that's the idea of UBI. Now, at the moment, to survive, you've got to have a job. And like that guy you mentioned is working at three three jobs right now. If he got a UBI, he wouldn't have to work at those three jobs. He might work at one, or he might actually consider his own business possibilities in that situation. So, I think universal basic income is a necessity given what robotics and AI can do to employment.
And you think up to 50% of working-class jobs could be wiped out because of AI and robotics.
Yeah.
I mean, that's um that's been a prediction from the leaders of some of the biggest companies in AI. I heard the the leader of Anthropic uh recently say the same thing that he thinks 50% of jobs could be wiped out. I think the shocking thing that we've talked a lot about on the show is just, you know, there's been other sort of economic or industrial revolutions in the past that have caused for job displacement.
Yeah.
But none, I would argue, at this speed.
No. And none that can replace virtually everything. I see one of there's a there's a classic story I read back when I was uh talking about the global financial crisis, came out of a New York Times article where they went to interview workers in an air conditioning factory. And there was one woman they found there whose job it was to place a thermocouple inside the air conditioning units as they went past. So there's 3,000 of these going past her a day. She's just placing one of these thermocouples where it needs to go inside the circuitry of the air conditioning unit. And she said, "You don't have to love your job as long as it pays you money." It was, you know, totally boring job. That's all she's doing. The thing is, the reason she got that job was you couldn't make a machine to replace her because the air conditioning units don't necessarily end up precisely at the same point. To make a machine that would do that is really difficult. Now, if you train a robot on it, the robot perception can ultimately get to the point where the robot can place that piece inside there. That particular unskilled job disappears. So people who work in jobs like that no longer have a possibility of getting a job.
I think even, you know, Anthropic released a report. Anthropic are the makers of Claude. They released a report saying that entry-level positions, they're seeing a 13% decline already in people getting those entry-level jobs. And actually, as an employer, someone that spends literally all last night I was looking through our inbox, our recruitment inboxes at candidates and talent.
I have noticed myself changing. I've noticed that um people that I would have given roles to maybe six months ago, I now have to think long and hard about whether there's going to be technology that can do those exact roles instead. And it's, it was a really shocking thing. I was saying to the team last night at like 1:00 a.m. in the office, I was like, this is a prime example of a candidate. I was looking at this particular candidate that 6 months ago I would have bitten their hand off, but now I have to pause because my innovation team in the corner of the office, they're they're able to do that now with these AI agents instead. And so I am, you know, you hear a lot about the theoretical impact of AI, but you're actually making the decision yourself. And then it's theory, it's theory, it's this thing on my Twitter feed like blah blah blah, whatever you hear on a podcast, you go blah blah blah, whatever. And then you find yourself actually behaving that way.
Your behavior is changing and you're going, "Oh, it's very hard to know the types of people to hire into our company." And I've kind of almost segmented them into these two groups where you've got people that have very deep expertise. Yeah. I'd say it's three groups. People that have very very deep expertise on a particular thing, you know, like my CFO group. Number two, I'd say, are people that are AI proficient, who can actually handle this stuff and be the people who manage the agents. Yes. And they can redesign our workflows across every department in the company to be agentic. Um, the word about AI agents, that's kind of like the word use, so agentic workflows. And then the third group of people are people who have skills that are highly beneficial human to human and in in real life, so like human to human sales people that deal with relationships, and are very good at it. And are very good at it, because there are still a certain type of sale where people want to meet the person, shake their hand, and say, "Okay, you're responsible for this deal." We're still in a situation where people don't want agents to do that. Those are like the three groups. What I didn't say is young people who have just come out of university, maybe don't know anything about agents. They don't have the deep expertise yet.
Yeah.
And when you look at the data, we'll throw some of the data up on the screen. It appears that these sort of entry-level white-collar jobs are the ones that are right now suffering. Yeah. Some of these investment companies would hire like three or 400 analysts to look at um companies and make decisions. That is one example of a of a role that's very at risk now. We've got an investment fund. We need one analyst, Molly. 6 months ago, we were interviewing more analysts. We now realize that we just need Molly and we need to give Molly AI.
Yeah.
And she can set up, I think Molly said to me yesterday when I left the office at when she left the office at midnight, she's now set up three agents, these AI agents, as her team. Those would have been three people.
Well, look, I I saw a demo of that. Like I've developed a software package called Ravvel, uh, which I've got one programmer for, and I I teach an online course as well, and I give Ravvel as part of that online course. And one of the members of the course said he's using an AI to build Ravvel models, and he's also using an AI to write code behind Ravvel. And he gave a demo this a couple of days ago, and you know, I watched it happen on screen as he built a model, a simulation system, and it was messy on one stage, but it produced the correct mathematics. So he's showing you can actually, he he's trying to tell me that we should uh get my main programmer to learn to drive agents to do the whole thing. Now, my main programmer has said, look, there's things that I can do that an AI cannot do. He's one of your highly gifted people, and he said, "It wouldn't be worth my while to have me telling an AI what to do because what I lose in terms of my own initiative, I can't, I just sort of balance out." It's really a, okay, but he hires a junior programmer, then the junior programmer would be one who's trying to drive the AI.
I I do think programmers are fine. Actually, there was some stats that I saw the other day that showed there's been this huge demand in people trying to hire programmers. It's interesting because you hear stats from Spotify. Spotify saying, "We haven't written a human line of code since December." And I'm very good friends with the guys at Spotify. I was actually with the CEO the other day, a couple of weeks ago, in in Austin. And you you hear that, and I did check that with them. That's true. So you assume that that means we don't need programmers anymore. But if you think about like Jevons paradox, when something becomes, you know, Jevons paradox is the old analogy, cheaper, you use more of it.
Yeah.
Yeah. So like when coal became cheaper, people were worried that maybe the coal industry was out of business or trains, whatever. But actually, what ended up happening is people just drove more trains and they used them for other things like transport. And the same applies, I think, for AI. When creating technology becomes easier, every company starts using more technology. So media companies, lawyers, you name the company, executive assistants, they all become coders. And actually, the demand for highly effort, really anyone who knows how to code or program,
Yeah.
We're seeing it. It's exploding.
Yeah.
But I just think the job disruption in the near for most people is going to be pretty.
Yeah. I mean, there's ways in which AI and robotics should be welcomed, because it means the possibility exists, and it's only a possibility, that we can no longer have to be exploited to get an income. Because if you look at the Marxist attitude towards capitalism is that cap, workers, capitalists exploit the workers. Okay. Um, the real world is we've been exploiting energy, mutually, but labor and capital exploits energy. We could have a future where we don't have to work for a living, and therefore you could do what you want to do for a living. That's the Star Trek future. That's the possibility that it promises. But at the same time, uh, it could actually eliminate the jobs that people currently rely upon. And what I fear is we have two possibilities. We have a future where Star Trek's high future, where you have replicators that make the goods and we consume and and we all live an energy abundant life, or the Hunger Games, where there's one little elite that gets has all the robots and lives extremely well, and we tolerate and oppress the vast majority, and they end up, you know, Hunger Games entertainment. Those are the two possibilities we face.
I do think the cost of goods and services will come down, which is great. Yeah. I think robotics, you know, if Elon is right, and I often say with Elon, like his timelines are not always accurate, but he does tend to deliver magic. He ultimately delivers, but it, you know, he always ever promises and delivers his lighter than he plans. And if he's right about when he says there's going to be more humanoid robots, his Optimus robots, than humans, and he says also in his predictions that there's going to be no need to study to be a surgeon because the robots are going to be so much more, uh, advanced and better than any living surgeon, that would imply that surgery and other sort of medical diagnoses and procedures are going to be incredibly cheap, incredibly quick.
Great.
How do you pay for them? That's the next question.
Yeah. So yeah, how do you pay for them? And do people want to, you know,
It's also, it's also the physical requirements. I mean, the amount of copper inside a robot, you're talking, you know, several kilos per robot. Um, do we have enough to produce 8 billion of them?
And maybe, you know, surgeons do much more than just operate.
Yeah.
There's a human element to the medical profession which I think is sometimes unappreciated. Like I would, I don't know if I'm quite ready to go talk to a robot about my health yet. Maybe I'll adjust. I wanted to come back to something you actually said earlier. You talked about Bitcoin briefly. I've heard you say that you think Bitcoin is going to zero.
Yeah.
Okay. This is worrying. I think I have some Bitcoin.
You're an economist. You're saying that Bitcoin is going to zero. Why?
Because ultimately because of its reliance upon energy. I mean, I, you know, you Max, Max Kaiser and Stacy Herbert, have you met them at all? No. They were sort of the original proselytizers for Bitcoin, and they're now living in, I think, El Salvador, um, which has adopted Bitcoin as a form of currency. When they told me about Bitcoin, I could have bought it for a pound a Bitcoin, which would have been, I would have been wealthier than you if I'd done that. The reason I didn't was they explained that the way that the public ledger is kept safe is that it takes too much energy to break it. So each transaction requires 10 minutes of computer processing time globally, by the looks of it, to actually create an extra Bitcoin, and that means it's too expensive for somebody to try to break the ledger. That means it's got a huge requirement for energy use, and I believe, knowing what I know from climate scientists, that at some point we're going to realize we're using far too much energy on the planet. We've got to cut the energy consumption, and the two easiest things to cut out to reduce energy consumption are cryptocurrencies and international travel.
But aren't you seeing that, you know, nuclear energy is becoming vogue again? And they're talking a lot, you know, about
It's the amount of time it takes to build that stuff. I mean, China is building nuclear power stations at a hell of a rate and much much cheaper, more cheaply than America is doing.
Solar has become a big topic of conversation.
Yeah. Again, there's a guy called Simon Michaux, whom I recommend you get in touch with as well. And Simon is an engineer who claims that we simply don't have the physical minerals necessary to support a completely solar and wind-based energy system. He's got people who criticize his analysis, definitely, but we still are using far more physical resources than we're aware of at the moment on the planet, and the availability of various critical elements that we need for the system we have right now, it's much less abundant than we would like it to be. So, a lot of these things about, you know, robotics taken over, do we have the minerals for it? Solar power, do we have the minerals? The answer is not is not yes. Okay. Sometimes the answer is no. Other times it's it's dubious. But I think that energy requirement alone is a problem.
You're saying that we're going to have we're going to have to cut back on our energy consumption.
But I mean, the direction of travel has been we've been able to produce more and more and more and more energy, and we're dumping it into the environment. The planet, the problem about the use of energy is it's happening on a planet. Okay. Can the biosphere cope with the waste that we dump into it as a result of using that energy? And that is something which economists are completely stupid on, beyond stupid. They've trivialized the dangers of the amount of resources we use and the amount of energy we use. So I don't think that energy future is possible on this biosphere at the moment. It's possible in the future if we get off the biosphere. So in that sense, I'm even more of a space cadet than Elon Musk is. I think we have to plan to take production off planet. But while we're constrained on the biosphere, the biosphere's constraints will stop us using as much energy as we used wish to use.
What are you, what are your closing statements on this whole situation with the war and Iran and everything that's going on from a geopolitical perspective?
Basic thing is our system is far more fragile than we've convinced ourselves that it is. And we can make observations about potential futures which presume a robustness we don't have. And if that robustness is destroyed, either by military conflict or by overextending what we put into the biosphere, then we can fall off what's called the Seneca cliff. We can go from an abundant future to a collapse.
And what would you say the people at home should be doing to course correct the path that you think we're on?
Stop electing fools. Um, electing Trump was an enormous mistake. We've got politicians who follow what's called neoliberal political philosophies. Therefore, put us in this problem. It hasn't worked. We need to reverse back to having a human-oriented and physically realistic view of how the economy managed should be managed and how the biosphere should be managed. We have to take care of our home, and in a central sense, we're destroying our home and thinking we can keep on doing that indefinitely. We can't. Our home is Planet Earth. Planet Earth has got physical restraints. We haven't respected them. Planet Earth will tell us what it thinks of that this century.
If you love the D CEO brand and you watch this channel, please do me a huge favor. Become part of the 15% of the viewers on this channel that have hit the subscribe button. It helps us tremendously, and the bigger the channel gets, the bigger the guests.