DeepSeek, AI and the markets

0:05Pushkin. Big curveball hit global markets this week in the form of DeepSeek, a new alternative to ChatGPTGPT that has sprung out of China. It was, we're told, dead cheap to build, it requires fewer fancy microchips than the dominant US versions, and it seems to work pretty well. Its emergence whacked a massive$600 billion off the market value of chipmaker NVIDIA on Monday. Even for the biggest company on the planet, that is really quite a lot of money. Today on the show we're asking, why is this such a big deal? This is Unhedged, the markets and finance podcast from the Financial Times and Pushkin.

0:49I'm Katie Martin. I'm not an AI robot. I'm a human markets columnist here at the FT in London. And I'm joined down the line from New York City by fellow human Robert Armstrong from the Unhedged newsletter. Rob, how's it going? It's so great to have a day where markets go a little bonkers and it's not about American politics. It's about something else. There's still time. The week is young. Yeah, we're not going to run out of bonkers anytime in the next couple of months. Not soon. Yeah. Here's the thing. DeepSeek has been around for quite a while, right? The tech community has been abuzz with it for weeks.

1:29But the simple souls of like financial markets, traders and investors and stuff, and frankly, you and me only really got the memo this week. So we're not techies and we're not going to pretend to be techies, but why are market senators about this? Several days ago, let's call it a week ago, DeepSeek said, we have this new model that was cheap to build and works really well. And everybody was like, yeah, tell me another one, broadly speaking. Yeah. And then they released it, meaning you could download the model and run it on your own computers. and it slowly dawned on people in the AI industry, this thing really works.

2:11And that message took a day or two to get through to markets, but the message did get through yesterday, boy. And it's interesting listening to people talk about it. There is remarkably little noise from people whose net worth has radically decreased as a result of the appearance of this model. Very few of them are calling BS on this thing. Yeah. Instead, they're saying, wow, this actually is very impressive. We're working with this. We're trying to reverse engineer this. Yeah. This is very fascinating indeed. Yeah. I've seen some of the work that it spits out. So the economist Stephanie Kelton, she asked it, what would happen to Canadian GDP in the event that Trump brought in 25 % tariffs on Canadian exports?

2:59And it spat out a very credible answer, actually, and came up with a number that matches what a lot of actual humans have come up with. We're going to have so much time for our hobbies, Katie, when this thing takes our sad little jobs. I'm going to take up quilting, I think is what I'm going to do. I'm thinking paddle. So last week, Angela Zhang, Z-H-A-N-G. She's an expert in Chinese tech. She wrote in the FT about how DeepSeek is like the real deal. And here's a really interesting wrinkle that she was writing about. The reason I spell out her surname is that, listeners, you've got to check this thing out because it flew kind of under the radar when it came out last week.

3:42But she is dead right. And she was saying, first of all, DeepSeek is not a one-off. There's a whole bunch of Chinese AI developers out there doing really interesting stuff and coming up with high-quality products. The other is one of the reasons this has all happened is that the U.S. has starved China of good quality, powerful U.S. chips. And so Chinese developers, Chinese engineers, Chinese techies and nerds basically have had to work really hard and innovate really quickly to come up with new solutions of their own. Especially on the front of efficiency. Yeah. They've essentially been deprived of computer power.

4:23And so they've just been forced to think harder about efficiency. It is an important lesson in how technology evolves and how hard it is to keep it in a box, as it were. So OpenAI's ChatGPT cost whatever,$11 billion to build. And OpenAI is still constantly saying we need more bazillions of dollars, possibly even trillions of dollars, to keep building this out and create the energy platforms that we need to make this work, yada yada. a deep seeker saying yeah we did it for 5.6 million bucks and you're like you did yeah you mentioned before that we're not tech people this point bears emphasizing i'm a person who struggles mildly with the remote control on my four-year-old television but i will use excel so but i think something that you and i do know about that is worth thinking about is kind of how markets price things.

5:20And what's interesting, I think, is this moment has revealed the assumptions we had about what economic structure the AI industry would have. So broadly, the vision of the future that had most of the market's attention until yesterday was AI model, very tricky and extremely expensive to train up and pretty expensive to run to. And the person who does the training up now has the model operating on their own computers, right? This is why everybody is building a massive data center, whether you're meta or Google or Amazon or Microsoft or whoever, you're pouring money into having this gigantic computer.

6:11You can run this model, which you own and is proprietary to you. And then people will have to come to you and say, pretty please, can I ask your big machine questions? And you charge them money for that. Importantly, the really crucial pieces in this puzzle are all American. Correct. America has this absolute dominance over big tech, absolute dominance over AI. This is like a, as one investor put it to me recently, this is like a geostrategic imperative to be in US big tech because this is just impregnable and no one else can challenge this stuff. And that brings me to, so the thing that investors are talking about here and listeners, you should like listen out for this word, is that people are saying, okay, there goes the moat.

7:01So Rob, what are people talking about when they're talking about the moat in investment terms? A moat is the barrier to entry that keeps other people from copying your stuff. Classically, the difference between a good business and a bad business from the point of view of investors is the width of the moat. The classic example of this is the airline industry, which is amazing what they managed to do. They move millions of people around the world through the air every day with very rare accidents, incredible technological and logistical feats. And it benefits us all hugely. And it's a terrible industry and you should never invest in it because it does not have a moat.

7:45right anytime the airline industry starts making decent money some rich idiot is like you know what it sounds fun to have an airline and they can do that airlines this is what i'm thinking whereas if you're microsoft and you have the operating system that everybody has decided that is the one they're going to run their personal computers on and you can charge people for that IP, your moat is a mile wide and you just keep printing it forever. There's been this idea that like NVIDIA has a very deep, very wide moat around it that's full of scary animals that will eat you if you try and enter it. And it turns out that moat, actually, NVIDIA is just standing on like a bit of a puddle.

8:34Well, no, well, I don't know. I would take issue with you there. The stock fell 16 % yesterday. Because it's the biggest company in the world, 16 % equals a squillion dollars or whatever number you said earlier. 600 billion, man. 589, I think, to be precise. But frankly, at that point, who's counting? The NVIDIA is at the level it was back in September. And even then, it was maybe the most valuable company in the world. And we were talking about how incredibly richly valued it is. So what the market is telling you is NVIDIA, you know, is 16 % less valuable than it was. It hasn't been cut in half or anything.

9:10It's not a panic. And the stock has stabilized this morning. That's Tuesday morning. Yeah. As of Tuesday morning, this is still an almost unimaginably valuable franchise they have. It's just 16 % less valuable than it was on Friday. I mean, look, so there's two schools of thought here. One of them is, oh, it's just a flesh wound. The other is, so just before I came to record this podcast, I got a note on this from Man Group, which is a hedge fund group here in London. It says, the sell-off reflects fears of a deep air pocket in AI infrastructure spending. For years, the tech sector has built investment cases on the belief that developing competitive AI models demands massive hardware investments and seemingly unlimited budgets.

9:54So the launch of DeepSeek, for me, is the first thing that has pierced that hubris. It's the first thing that has challenged that narrative. And that's why this is important. We might decide further down the line that DeepSeek is too much of a security risk for people in the West to be using because who knows what kind of government interference there is in China. We might decide that it's just less good than chat GPT. Whatever doesn't matter to me. Point is, this is a challenge to U.S. tech majesty over the planet. It is a challenge to U.S. tech. And the point I would make is that it's a challenge to the view that AI was a winner-take-all industry.

10:32Yeah. That one company or another would have the best model, which was proprietary to them and ran on their computers. And everybody who wanted the best AI services would have to build those services on top of that person's proprietary model. And we all just give them our money forever. And we all just give them our money forever. And that picture has, to a certain extent, gone out the window. And it might be that the spoils are shared more broadly here. But what is interesting, let me just make one more point, is that the stocks that really fell yesterday were, as you intimated earlier, were like the American utilities.

11:10I was going to say, I was going to say, so Rob Armstrong, tell me about how this has affected other stocks. Well, I'm telling you, man, these stocks, we wrote about this a month or two ago. These utility stocks that are located in the areas where the data centers are being built, meaning Washington, D.C., Washington State, the Northeast, these stocks were up like 40%, 100%, 200 % last year because they saw the demand going up. But then if you have a more efficient, cheaper model, it means you drink less energy. And these stocks not only fell massively yesterday, more than NVIDIA did, but they're still falling today.

11:52So there's like this revolution going on and all the people who are building these data centers, building equipment for the non-chip equipment for the data centers, the utilities that we're going to build a gigantic wire into the data center, all of this stuff has fallen out of bed even more dramatically than NVIDIA. It is super, super interesting. Let me ask you two questions here that normies ask about this whole thing. The first one is, is this the dot-com crash all over again? It is interesting. What were the stocks that we were all worried about? The Magnificent Seven. People who thought we were in a bubble were very focused on Alphabet, Amazon, Meta, Microsoft, NVIDIA, and Tesla.

12:38Did I get them all? I think I got them all. You might. Yeah, something like that. Okay, something like that. Those stocks haven't moved that much. Well, 17 % on NVIDIA, but yeah. They were massively investing in AI. And in a way, there is an argument that they may be in an important sense better off now. Whereas they thought they had to get in this great money throwing contest over who would own the big AI model. Because if you're Google or Amazon or Meta and you don't have your own giant, hugely expensive proprietary AI thingy, then everybody else will eat your lunch. Well, now your lunch is safe and you can spend your money on something else because it turns out there's a possibility that you will be able to have AI applications that don't require you building this expensive fortress.

13:27The other thing to me that differentiates this from the dot-com crash from, God, we're old, Rob, like 25 years ago. Jeez Louise. But that was different because that was a bunch of like internet made up companies that didn't really do anything. And they were valued in enormous excess to the earnings they were making. Yeah, not necessarily true here. NVIDIA, yes, it's taken a big hit. but this is a thing that makes proper money like 30 odd billion a quarter or something no six of the seven of the magnificent seven are six of the best companies there has ever been from an economic point of view yeah in terms of return on investment profitability growth potential the width of the moat all of that stuff the seventh of course being tesla i thought you that one let's park that one park a tesla but i mean these are awesome businesses and you know maybe they're overvalued yeah but they're not going to disappear no right you know and that's very different now people have also missed i would say how expensive all u.s big cap stocks are you know what i mean costco is a great company right the bulk warehouse chain where i often go myself and very much enjoy.

14:50It's trading at something like 50 times earnings. It's more expensive than any of the Magnificent Seven, right? So stocks are really damn expensive. Large capitalization US stocks are very expensive and they could massively correct. Predicting win is a fool's game. But again, it doesn't feel quite like kidylitter.com or whatever was the nonsense businesses of our youth. Yes. So it doesn't feel like that. And that, Again, to recap, perhaps for our younger listeners, that was an enormous crash that took the entire market down with it and ended up with the Fed, you know, the US central bank interest rates super hard, bailing everyone out.

15:28Arguably stored up a lot of problems for another day by having interest rates too low for too long. This doesn't feel like that. But the other question that normies ask is, do I need to worry about this for my pension? And I guess the sort of related question that is being asked there is, is this going to turn into some sort of financial crisis? I'm going to say no, by the way, for what it's worth. I'm OK. I'll make the argument for elevated caution. And I'll say what I think that means for people's portfolio and certainly what it means for mine. We have, I would argue, an unusually uncertain moment in the path forward of interest rates, which means the price of money.

16:10And the price of money being uncertain makes everything unstable. At the same time, every risk asset in America is priced about as richly as it is ever priced. That is a volatile combination. I will let listeners reach their own conclusions about political risk, which is a fraught topic I don't understand very well. I mean, the point is panicking and having all your money in cash has almost never in the last 150 years been a good strategy. But is there an argument for maybe reducing your exposure to the United States a little bit, increasing your exposure to cash? Europe's been doing actually very nicely record highs in the DAX and the FTSE in the UK.

16:59Yeah, I think there is a case. I don't think it's probable. But I think it is, you know, if there was a pretty significant repricing in American risk assets in the next year or two, exactly zero people would be surprised by that because of the setup I just described. But it's very expensive. And I know this because I've made the mistake myself, trying to time exactly when that is going to be. So all you can do is look at your kind of portfolio allocation and say, am I going to feel like a bad human being when I look at this portfolio or an idiot when I wake up on the morning where, you know, the Apple card has been overturned?

17:41The one thing, so like I say, I don't think this is laying the groundwork for some sort of financial crisis, but I do think this could, So this whole shakeout in tech and in speculative tech and in AI stuff for all the kind of tourists who moved on from Web3, remember that? And decided that the future was all in AI. Was the Web3 the metaverse or was it crypto or was it? I can't remember which of the things we were excited about. Just idiocy. Anyway, there's a lot of venture capital money in AI and in the kind of energy complex behind AI. There's a lot of private equity money in AI and in that energy complex.

18:22I'm just going to say there must be some portfolios out there of like early stage incubator money that are getting pretty squeaky bums right now. Well, after yesterday's events, the prices we would almost like to see were invisible. OpenAI is a private company. If you had to mark the value of that company, how much did it fall yesterday? I have no idea. I'd sure like to know, though. Similarly, Anthropic. Anthropic? I don't know. I don't think I've ever said that word out loud before. But similarly, all these cluster of private companies in this space. How do those books look? I don't know. But I can tell you that Sam Altman, who runs OpenAI, which backs ChatGPT, he said that the emergence of DeepSeek is, and I quote, legit invigorating.

19:09Yes. That is tech pro speak for, ooh, lads, we've got a bit of a problem. I think this is what Mike Tyson's opponents used to say. Being punched in the face by Mike Tyson is legitimately invigorating. That is basically what is going on here. But we often like to wrap these things up with a little bit of a prediction. So let me ask you, DeepSeek, are we still going to be talking about it this time next year? Is this going to be like a part of the tech, just furniture? It is impossible to say. We will certainly this time next year be talking about the competition among AI models and which is best and which is cheapest and which one is good for what.

19:53But whether DeepSeek will be overtaken next week by something we don't even know the name of. Britseek. Some sort of, you know, maybe the Brits or the Europeans could really make a splash here. Yeah. So I just, I'm going to back away from that prediction because I don't know how to make it. All I know is it is possible to do this AI stuff on a, not necessarily a shoestring, but a lot more of a shoestring than we've been led to believe by Silicon Valley. So interesting times. If you are at a venture capital firm or a private equity firm and you don't like your job very much and you would like to tell us how much money you've lost on this stuff, we are all ears on Hedge.ft.com.

20:36And we're going to be right back with Long Short.

20:41Okie doke. Now it's time for Long Short, that part of the show where we go long, a thing we love, or short, a thing we hate. Rob, what you got? I am long Brazilian jiu-jitsu. Not because I will ever do that sport or any combat sport of any kind. Although I would like to see that. You know, and it's this huge trend. Is it? Like Zuckerberg does it. Everybody, every man in the world is doing Brazilian jiu-jitsu. Real man. And I was like skeptical of this whole thing. It's like, you know, macho nonsense. But my son, who's 15, is doing it and he loves it and it like burns his energy. It kind of has built his confidence.

21:20And now I have someone in the house who can kill me with his bare hands. And I just feel like this trend has legs. My 16-year-old boy is into Muay Thai. Maybe we should get them to fight each other. We should fight them, yes. I am short. You're a 16-year-old. My 16-year-old is likely to be somewhat shorter than your 15-year-old. But I think they'd be up for this. Let's make it happen. My short is really quick. I am short winter. It's too dark. Everyone is ill. The weather is terrible. I'm absolutely sick to the back teeth of this whole thing. It's time it's spring now. It's over for me. Only two and a half more months, Katie.

22:01Oh, for the love of God. Also dry January. It's really dragging on. Could really ease a drink at this point. Anyway, let's wrap it up there. Listeners, we will be back in your ears on Thursday. So listen up then. Unhedged is produced by Jake Harper and edited by Brian Erstad. Our executive producer is Jacob Goldstein. We had additional help from Topher Forges. Cheryl Brumley is the FT's Global Head of Audio. Special thanks to Laura Clark, Alastair Mackey, Greta Cohn and Natalie Sadler. FT Premium subscribers can get the Unhedged newsletter for free. A 30-day free trial is available to everyone else.

22:37Just go to ft.com slash unhedged offer. I'm Katie Martin. Thanks for listening.

22:54We'll be right back.