Is the AI boom actually a bubble?

The AI Gold Rush has been ramping up and the numbers are staggering. In July, Nvidia, the company whose chips are at the heart of the AI revolution, became the first tech giant in the world to reach a value of four trillion dollars. And in September, Nvidia announced plans to invest up to a hundred billion dollars into open AI. That's the company behind ChatGPT. Big Tech is also on a spending spray when it comes to hiring talent. This summer, Meta reportedly offered a 24-year-old researcher a jaw-dropping amount of money, a $250 million package over four years. But now, some economists are worried that things might be getting out of hand. Faith in the global AI boom has recently been tested by research which showed that 95% of organizations are getting zero return from their investments in generative AI. Meanwhile, some experts and banks are growing increasingly wary of what they are calling an AI bubble. Earlier this month, the UK's central bank did something unusual. It said that some AI companies might be overvalued and warned of a market bubble. Then last week, the International Monetary Fund warned that if the AI bubble burst, it could potentially rival the.com crash of the early 2000s. So what is really going on? From the BBC, I'm Tristan Redman. And I'm Asma Khaled. And today on the global story, we turn to Evan Davis. He's the BBC's "Go To Guy" for when economics gets complicated. And we ask him whether the AI boom is built on solid ground or headed for a deja vu of the.com crash. Inevitably, we asked AI, Evan what the most exciting way to introduce you would be. And it said, we should say that you have a dog called Whippy and you ride a motorbike and also that you are a man who somehow made economic sexy on national television. Is that right? Sexy, I'm not going to answer. I do have a dog, Mr. Whippy. Mr. Whippy, the Whip it. Mr. Whippy, the Whip it. It didn't get the Mr. Mr. Whippy is a brand of ice cream. That is the name we gave our dog. The key question, do you ride a motorbike? Right. The honest story on the motorbike is we got rid of the bikes when we got the dog. The dog really couldn't go on the bikes. And the dog looked so full on being left behind. If we kind of got the gear on and went out on the bikes, we just had to get rid of the bikes. I do have to ask, would it not have been possible to get a sidecar for the dog? They were on those kind of bikes, Tristan. They were right. They were kind of bikes. You steeped down on and ride fast on. So no, dog and dog. But I think we were kind of slightly growing out of the bikes anyway. Okay. All right. Well, I think that that introduction we got from AI Tristan tells me sort of the strengths. I will say as well as the weaknesses, perhaps the limitations of AI, which we're going to get into on today's show. And we actually haven't wanted to talk to you about whether there is currently an AI bubble in our economy. And I think it's worthwhile to backtrack and rewind to the most recent tech bubble, at least in my memory. And that was the.com boom and bust cycle of the late 90s and the early 2000s. You were covering the economy then, right? Yeah, I was working at the BBC all those years ago. I was covering the rise in those shares. And I don't know if you remember, it was late 1996. The then chairman of the Federal Reserve, Alan Greenspan, who has seen as something of a profit, really, in financial markets at the time. It is my pleasure to call upon Mr Greenspan, whose lecture is titled "The Challenge of Central Banking in a Democratic Society." He gave a speech and he used this phrase, "Markets may have irrational exuberance in them." How do we know when irrational exuberance has unduly escalated asset values, which then becomes subject? And that was taken as a warning that everybody's getting way ahead of themselves when it came to these new.com stocks. The internet was a thing. People were getting email addresses and getting very excited to have this kind of technology. He gave that speech at the end of '96 and the market kept on going up. Even though he thought it was irrational exuberance, potentially then the market went up all '97, all '98, all '99. At 11 a.m. this morning, the company's stock went public and Wall Street went bonkers. Initially offered at a price of $28 a share, NetScape shot up to 72 within minutes. The drugstore internet space Friday is going to own 25.3% of drugstore.com. Everybody's invested, everybody's happy, everybody's making money. Something's wrong here. It carried on up and the bubble eventually burst, you know, three years later in the year 2000. It's described as nothing short of breath-taking, a point-strop, never before seen on the U.S. markets. This closing bell might as well have been an alarm. Already more than 200.coms have gone bust, and another 20 die each month. And even when it burst, it kind of went back down to where it was where now in Greenspan said, "Oh, my goodness, maybe there's irrational exuberant." So it's very hard to call a bubble. That was one of the lessons then. It's very hard to know when the bubble is bubble. This is what I always say about bubbles. There's an element in which they're not a problem of capitalism. They're a feature of capitalism. Capitalism has found something. It could be the internet or it could be trains. Or if you want to go back far enough in Dutch history, it could be tulips. Because that's always the first sighted bubble of tulip mania when tulip prices got absolutely crazy in the Netherlands. So whatever it is, something has been discovered. There's excitement about it. There's a real investment boom in that thing. People really want to get it. It's fashionable, it's hot. And there's a reason for that, like the internet, like AI. There's a reason to think this might be very productive. We should be excited about it. And then it turns from that into a frothy over excitement. People chasing the last person. So instead of someone thinking, yeah, this is really exciting. They just think, oh, other people are going into it. I'm as dash into it. And then you just get into a kind of mad, mad scramble. That last bit, that's the frothy bit, which I think is the bubble. It's when you see too many people who don't really know what they're doing, spending large amounts of money. You kind of want to start getting very worried. It's the bit that you probably wouldn't want if you could take it away. Back in the 20s, what was the warning saying? It's when the bloke saying, when the shoe shine guy gives me a stock tip. I know that it's all gone crazy and that I have to get out of the market. But the first bit, the excitement over new technology. And the willingness to kind of throw money at it. And to say, we don't know. We really don't know which of these companies is going to work out. We have to take some risk. I think you would say that bit works for the world. So you're saying really it's kind of part of the natural selection almost of innovation. That is pretty well exactly what I'm saying. I think you've got a new technology. We need to invest in it in order to deliver the benefits of it. But the investments in it are also eliciting information about the world as to what the value of this technology is. And so that's the bit that you really want to keep. That's the bit you want to treasure. So you take some interesting examples in the dot com era when there's this new excitement. What can the internet deliver? And one of the kind of showpiece stocks, which everyone now looks back on in his embarrassed about, was a thing called pets dot com, which was basically pet food delivery. Remember that? Right. So that was a classic costume hotel. Yeah, and then collapses. But the funny thing is, the really funny thing is, you know, we get our Mr. Whippy gets his top food delivered via online purchases. So decades later, pets dot com might have been onto something. But capitalism, if it is anything, is about resolving uncertainty and unfolding which bits of things are going to work and which bits are not going to work. So people start making bets, good thing. And then I can't stress this enough. People follow on with just that I want to be in on the action. And you know, you might take a company. Let's take open AI, which is a company that makes about $12 billion a year in revenue at the moment. A billion dollars a month, it's growing. But let's call it $12 billion a year in revenue. Talking about making a trillion dollars of investment, which seems a lot of investment to make. Now, that's real investment. That's not buying shares. That's actually spending money on, you know, massive, massive amount on big data centers. In that process of spending, say, a trillion dollars on the back of a $12 billion of revenue, its value could go, it's what, it's about half a trillion now. Its value could go up to 10 trillion and back down to half a trillion. Which is just, that's the bubble bit. But what you fundamentally have to focus on is their spending money on actual data center. That is a bet. It's an important bet. And we don't know whether that bet is going to work till maybe a few people have done it. And we see whether the market will pay the price for the products that follow from it. So that's so twice. I just think it's really hard to know what is the case for this moment being a bubble. I mean, if this was a court, what would the prosecution say? The prosecution case would be when you look at the share prices of the tech companies in particular. And more widely. And you say, how much of these companies actually earning? How much are they actually producing in revenue? Which is what shareholders ultimately need to justify the share price. And you say, got the share prices are very high relative to those earnings. That is the thing, that is the real flashing red light on this being a bubble. That people are buying shares in these companies at high prices not because those companies are earning a lot of money, making a lot of money. They're buying their shares because they think someone else is going to buy those shares at this high price. And they think these shares are going to go up tomorrow like they went up yesterday rather than looking at the earnings. And it's when people are no longer looking at whether the companies are earning any money that you kind of say people are getting detached from reality. That's the bit you have to worry about because at some point there's no other sucker who's going to buy. And then the share price won't go up. And someone will say, I bought this very expensive share. I'm hardly got any earnings at the bottom of it. And the prices aren't going up anymore. So I'll sell and then the prices start unwinding and they go down. And then I sell because the guy next to me sold and then you get into the negative part of the bubble cycle. The worry is that we're not perhaps quite where the.com bubble was. But some of those measures of price of shares relative to earnings are touching those kind of.com era levels. Evan, you saw the.com boom and bust. When a bust like that happens, what does it actually look like? How do you recognize it when it's happening? At first slowly and then quickly, I guess is the. I mean, I think what happened was you started getting some announcements by companies and then you started seeing their share prices fall and then you realized that the index was going to fall a lot. And then, in fact, in the.com one, a lot of these were small companies. They were startups. It wasn't companies like Apple or Meta or Google who are already sizable, big, profitable companies. What you suddenly saw was companies going from elevated share prices to zero very quickly, close to zero. And that was like, oh my god, they're really down. Oh, they're down 80 percent. And that was then. Now, I feel it is kind of slightly different because these companies that are the ones who've been pushed up the highest are also quite big companies with existing businesses that seem to be very valuable. So you wouldn't expect Alphabet, Google, to kind of diminish into nothing. I mean, it's a pretty solid company. So it does feel a bit different this time. So Evan, we have also seen a couple of mega deals in the last few weeks involving hundreds of billions of dollars. And these deals have been flagged as a worrying thing called circular financing. And I wanted you to explain what that phrase means and why it is concerning to some. Right, so the deals you're talking about are Nvidia investing in open AI. So this has definitely been seen as a kind of flashing light, flashing amber, potentially flashing red light that you've had deals involving open AI very notably. Open AI produce the AI, but they need the chips on which the AI runs. And the company and video make a lot of these chips that are best for AI. So Nvidia's share price has been the most staggering story of kind of market expectation of the last couple of years. And Nvidia is worth more than the London, London FTSE 100 combined. It's a big question. I mean, these are staggering amount. So you've got the AI producer that needs chips being invested in by the chip maker. And then the circular that you've heard is open AI, take the money that they've been given by Nvidia and they buy Nvidia's product with it. So they give the money back to Nvidia and they say we buy some of your chips. So then the reason why this is sometimes seen as very worrying is that's like allows Nvidia to say, look how our sales are booming. And everyone says, oh, that's great. You know, Nvidia must be worth a lot. Pushes up the Nvidia share price. And it's like, but Nvidia paid for those products that they claim to be selling by putting the money into. Is it a bit some Ponzi scheme-ish potential? So that's the, that is the, that's the, that's the charge. You're trying to lift the value of companies up by their bootstraps. So yes, definitely that the kind of circular financing can be a red flag. However, it can also be a very wise investment decision. So it could also be Nvidia say, my goodness, look at OpenAI, you know, potentially our biggest customer, and they're brilliant, you know, and look at the products they're producing. They're good for them. They're good for us. Why wouldn't we invest in them? And, and we'll invest in them. They'll buy our product. They're going to build data centers that are going to be hugely profitable. And, and we're going to be in on it. And so there's a perfectly benign interpretation of these investments, which is Nvidia know more about this world than any of us. And they're putting their money into it, because they know what they're, they're getting, and by the way, they want to finance it, because it finances their expansion. So there, there are always in these areas, which is why it's so difficult to call the kind of proportion of bubble, as opposed to proportion of, of sensible investment. There's always a benign explanation of why a company might want to invest in its customers. Now, that's part one. Part two, of course, is, is you do have quite a small number of entangled players in this world. The magnificent seven they're often called in this stock market. Sort of a relatively narrow number of people are making the decisions on which a very large number of bets are being taken. The entanglement of Open AI with Nvidia, Open AI with AMD, who are a competitor in video. The kind of the way in which they're all bound up together, does mean you've got a very small ecosystem on which this AI investment boom, let's call it, is all built. And that potentially makes it a little bit more, just a tiny bit more scary. Yeah, because if one part falls, then the rest of it can go with it, essentially. Correct. Yep. I want to ask you, Evan, if I can about the idea of productivity and profitability, because the assumption, certainly for a layperson, such as myself, is that the reason why AI is assumed to be such a great economic boom is that it will improve productivity greatly and therefore profitability for companies. Has that actually been borne out, though, by the facts? MindStan is no, not yet borne out by the facts, that's why people are baking big bets on all of that. So I think it's important to say we should distinguish between two different things that are going on. There's the investment boom by the AI companies, which is Open AI spending a trillion on data centers, that kind of thing. That's the big part of the investment boom. And then you have the second investment boom, AI related, which is everyday companies, small and large, saying, what can we do with AI? How does that make us more productive? And will that make us more profitable, which is your question, Tristan? All I would say is that is not a two-year process. That's actually, you know, 10-year process, maybe even a 20-year process. You go back to the internet, internet really erupted mid-90s, and it's really the mid-2000s. You start seeing the productivity figures showing, oh yeah, it really did have an effect. It did actually mean travel agents became very different to the way they were, or some retail became very different to the way it was. So I would expect, if AI is going to have a productivity effect, that to dribble through over a decade. So Evan, are you saying that if companies are not yet seeing productivity gains because of AI, maybe that doesn't matter because it's just too soon? And the reason I asked that is in doing research for this episode, we looked at this report out of MIT that showed 95 percent of companies that had integrated AI, thus far it seems, have failed to see an increase in their productivity. And that makes me wonder, okay, well, is this actually yielding the results that are promised AI did, right? That is a worrying piece of evidence, Asma. I would still say, give it a little bit of time, you know. I mean, it can take time to re-engineer processes, because AI may not just be about adding 3% here or 4% there. It may much, much more be about, let's no longer do this whole chunk of stuff. Let's put our effort into doing something completely different. So I personally would be relaxed about looking for the benefits too quickly, but I would be surprised if there are not some industries where AI can have a transformational effect. And I would be surprised if there wasn't some false hope and some false dawns, because that's what you always get. Actually, today in the UK, we've got this announcement from Waymo, their driverless taxis that are whizzing around cities in the US, that they say are coming to London next year. That is basically an AI productivity enhancement. I mean, taxi drivers are not going to welcome it. It's my guess, but that is a car that uses less labor to be driven around. And we can get into whether, you know, the displacement of jobs is how we're going to cope with that. But, you know, I would expect things will, will happen and we may not even realise that there are AI when they happen, but I'd also expect that lots of things won't happen. If it were a bubble and it did burst, what would the consequences be for the economy? I mean, would it just affect people who have invested and got shares in these companies? Or is it hitting everyone at this stage? That's a very good and complex question. So firstly, when share prices go up, people tend to feel a bit richer, particularly people who've invested in shares and maybe spend more and they think, oh, yeah, life's good, I can have a holiday blah, blah, blah. And then when the share prices unwind, they feel poor and they feel miserable and they say, oh, my goodness, I thought I was rich. I'm not, I can't afford a holiday. So you do tend to get a bit of economic cycle drive through the, if you like, the wealth effect, people feeling wealthy or not wealthy with the ups and down to the market. So that is that is one effect. Most people's pension will have some exposure to AI. And most of them probably should have some exposure to AI because this is a very big technology. And if it, if the bets all pay off, you want to be a little bit exposed to that, right? You want to get some of the winnings and you have to face the possibility you'll take some of the losses. So yes, I think most funds will have, have, have some exposure. My guess is, if you're 15 of 20% exposed to AI and you lose half the value of your AI investments, these are sort of hypothetical figures. I'm making them up. These are not real figures, folks. I'm not giving any financial advice. If you've heard anything that sounds like financial advice, you misunderstood me. But if you, if you would, you would, you know, if you've got 20% exposed to AI and AI loses half its value, you've lost 10% of your pot. That kind, that kind of scale is very realistic for ordinary people and their, and their pension investments. And for young people, I don't think that's going to matter very much because you get good years and bad years and things pay off and they don't. Obviously, if you're very close to retirement and you're relying on that pot next Thursday and you lose 10% of it on Wednesday, that's, that's much more serious. But that would be the kind of scale of the picture, I think. But Evan, if you're some money who doesn't have money, invest in stock market, a lot of Americans don't. I guess I'm wondering, you know, you talked about the dot com bubble and this moment of things and ideas that came out of that era sticking around. We were joking, Tristan and I, that do you remember Netscape, that search engine that everyone used to use once upon a time? Netscape is not really around. No one, the sort of like under the age of 30, really nice Netscape. Including the producers on this show. Yeah. We were like, what is this thing you're talking about? But there were, there were winners and losers out of that dot com era. And I think ultimately, if you were not involved in the tech industry, if you're not working in the tech industry, perhaps you were younger, you think, okay, well, at the end of it, hey, I got better search engine. I got Google. That was one of the winners out of that long, long process. So if you're an average person, just using AI, consuming it, not working in it. And at the end of this, if this is a bubble, you just get better AI. What does it matter to you if this is a bubble in a purse? I mean, I think what, what, what your average consumer who's not an investor, what your average consumer gets out of this is other people are spending a lot of money on finding out whether this is going to be useful to you, the consumer. I mean, that's ultimately what, what's going on. Look, like all big economic shocks, dot com bubbles, the financial crash of 2000, 2007, 2008, 2009, of course, there are, there are always our aftershocks and they're always our consequences. And it ripples through, and it ripples through, yeah. But broadly speaking, I think, you know, this is ultimately, it is about the consumer and about what we're going to use it for and what our bosses and our companies are going to use it for. What is AI going to do for us is the big question capitalism is trying to answer. And it doesn't know the answer yet. It's working on that answer. And a lot of money is going to be made and a lot of money will be lost and a lot of money will be wasted on the way. And let's try and minimise the losses and let's not try to get sucked in ourselves personally to sort of mad scramble to and don't get fomo and, you know, fear of missing out and oh my god, I've got to be in there because my friend's in there and I missed out on Bitcoin so I better get in on this one. Just keep level headed about it, but the process is likely to be quite messy. Thank you, Evan. It's such a treat to have you on the show. Thank you so much. Thank you both very much. Really enjoyed that. See you soon. Bye then. That was the BBC's Evan Davis. He hosts a business podcast for the BBC called The Bottom Line, and you can find more of him over there. And by the way, if you enjoyed today's episode, please take a moment and rate our show wherever you listen to podcasts. It helps other people find the show. Today's episode was produced by Aron Keller and Sandra Ellen. It was edited by James Shield and engineered by Travis Evans. Our senior news editor is China Collins. And folks, that is it for today's show. We will talk to you again tomorrow.