Speaker 1At long last, I welcome my friend John Hempton back to the podcast. So John is founder and chief investment officer at Bronte Capital, a Sydney-based global hedge fund. And today we're going to talk about AI from a finance guy's perspective. And selfishly for me, John, I think the three questions I personally want to get a better handle on are firstly, if there is a bubble in AI, where exactly is it located or where is it likely to emerge? And then two kind of ancillary questions, which we'll spend less time on than the first one. And they are, if there is a bubble, what does this mean for compute? And secondly, what, if any, are the implications for Australian public policy? But before we get into AI as a business, I want to talk about AI as a technology. What are some of the interesting ways you're using AI at the moment, either personally or at work?
Speaker 2It's. AI as a tech is freaking amazing. If you haven't decided that this is the most amazing technology of your lifetime, you have not seriously used it with cutting edge models. On a personal basis, I threw all of the plans for a house that we're building through AI. None of the engineering specs, just the plans. And I wanted to work out what the heating and cooling loads were. And so I put this in and it asked me questions about materials. It asked for the address. It then went and got the weather data from Sydney airport, which is hourly for many years. It worked out what the worst days were. And the worst days are going to be cold days in July and hot days in February. It worked out how many solar panels on my roof. It worked out the angle of the roof. It worked out the question of whether I could, how much of the house I could cool on the theoretically worst February night without touching. And it did all this using, and I was using Fable and it did a full engineering diagram, full heat loss stuff, everything from first principles. And it used about a hundred dollars worth of tokens. And the answer came out within 5% of the answers that were given to us by an engineering specification. So I put the engineering specification through the AI and the AI said, yeah, they had, Fujitsu was the supplier of the compressors and Fujitsu has some software that does it. And the software is open to anyone. So it had downloaded the software, run it for the house, worked out the calculation and got exactly the air conditioning one. And it did that as a check of its own numbers. And I mean, if you wanted to do that from first principles, it would take a good engineering grad a few days, you know, maybe even a week. And it did it for a hundred dollars. Now, AI can't get built. But I don't think it's very long before we connect AI to the robots. You think your jobs are not safe if you're an accountant or a financial analyst like me, your job's not even safe if you're a carpenter, right? Because we're going to connect this thing to the robot. Now in my day-to-day work, I use it a lot and I use it sometimes with great effect and sometimes with not. And it's been a puzzle as to why it's not. So I'll give you an example. One of the questions I want to do is find all of the bad banks in the world, all the banks that are lending money that they can't get back, all the banks that are the basis of the next financial crisis. I like shorting bad banks. I shorted Silicon Valley Bank. I shorted Signature Bank, right? The shorting banks is one of these beautiful things because you can do it in any size you like and there's almost always a bank somewhere in the world in trouble. So I wanted to work out how to teach an AI how to do it. And I'm going to show you how to do it. So I'm going to show you how to analyze banks. And I've done really, really well with the deposit side of the balance sheet, but I haven't managed to solve the problem with the asset side of the balance sheet, which is usually the more interesting side. So the deposit side, there are different ways that a bank can get deposits. ANZ Bank in Australia gets deposits by branch incumbency, right? It has branches on every street. Your father banked there, you banked there, you opened a bank account when you were 22, and now you're running a bank account when you're 22. And you're running a business and you're age 45. And guess what? It's still at ANZ Bank. And now your business has $7 million of deposits because it's a nice profitable business with a big turnover. That $7 million of deposits that ANZ Bank got through branch incumbency. That's probably the most common way that banks get deposits. But there are other ways. For instance, um, you can get deposits by being the highest rate payer. So if you, you look at what is the rate for a three-month term deposit in the US at $250,000, which is exactly where the FDIC cutoff is. And there's an order. And sitting at the top of the order is probably Banco ZK. And you look at six months and it's again, it's Banco ZK. And you look at a one-year deposit. And again, it's Banco ZK. And we know how Banco ZK gets its deposits. It pays for them. So a little bank that was originally Bank of the Ozarks has become a national bank in the US and it became a national bank by paying up for its deposits. That's a less good deposit base than ANZ. ANZ has large amounts of deposits. It doesn't pay for it. It doesn't, right? But it's a good way of analyzing the deposits. Then there are other ways you can get deposits. Um, to pick one we just talked about, Silicon Valley Bank. Silicon Valley Bank didn't have branches. What it had was a really good relationship with hundreds of venture capitalists in Silicon Valley. And the Silicon Valley venture capitalists will typically find two or three kids who come in with a good business idea and they give them $15 million. Now you don't give a bunch of 20 year olds who have never been in business $15 million if you don't want it to be snorted on cocaine or something. So what you do is you invite your friendly Silicon Valley bank branch, banker in. And the Silicon Valley banker acts as a sort of quasi-beginner CFO. So he sets them up with a payroll system. He sets them up with accounting systems. He sets them up with various compliance systems. The board member who is almost certainly a venture capitalist will get an automatic alert if transactions are unusual on the account. Effectively, he sets them up with the basic set of minimal controls that a Silicon Valley venture capitalist would want and exchange them with. And then the bank gets these big deposit bases. Now those deposit bases are actually quite a high quality asset. And they were the eventual reason why the bank collapsed. But it's an interesting story and it's a different way of getting your deposits. Now take a look at, say, Signature Bank. The way that Signature Bank got its deposits was it hired deposit-rich bankers. And these deposit-rich bankers would come with clients with large lists of deposits. Some of those large lists of deposits might be people who didn't want to comply with Know Your Client rules. Obviously, crypto firms were very high on the list. There are other ways you can get your deposit. In Indonesia, the way that you get your bank deposits is that your boss might bank, your company might bank with Mandiri. And if you want to get your payroll, you have to have an account with Mandiri. So all the employees have accounts at the same bank as their employer. And those employee accounts contain most of the savings of Indonesia. Now, what you do with this is you feed example after example where you know the answer into the AI. And you tweak it till the AI gives you the right answer. And you know it's the right answer. It's not hallucinating. These are things you have to know fairly well. And you do that with 15 or 20 examples. And then you say, write a mega prompt. And what you do is you write a mega prompt. What the mega prompt should do is it should delineate between all of those types of ways of getting your deposit. Then you try your mega prompt on three or four more banks that you know the answer for. And sometimes these are banks that have two methods or three methods of getting their deposit. And once you've got that to tweak out correctly, and it gives you a sensible answer, you then say, OK, I want to run that mega prompt over every bank in the world with a market cap more than $100 million. That's an agent. And anybody can go and make such an agent. Now, that agent, as I described, will cost you about $30,000 to run on a cloud on a high level. But it's a very impressive piece of work. And if you wanted to train an analyst to look at every bank in the world and work out the quality of its deposit base, this is not a task that is doable within a 12-person firm. It's a task that's doable within our firm. Now, we're very sure that we've got it to work for deposits. We absolutely emphatically and with bells and whistles have not got it to work on for loans. And the way I did it for loans was I took the credit card companies of the early 2000s, Providian, Capital One, Metris, CompuCredit, and ran those through using the data that was available six months, a year, two years, three years before they blew themselves up. And I never worked out a prompt that could delineate, say, Metris from Capital One. And Metris went to zero and Capital One went up and up and up, and then it went up some further. And if I had tried to use a prompt like that to delineate with them and therefore trade them, I would have got it wrong. Now, every regulator in the world would love a prompt that goes through all their banks and works out which ones are problematic. I mean, this is a regulator's dream. It's also a short seller's dream, and I haven't managed to work out how to do it. And I think the answer, and I'm, you know, I'm spitballing here, but I think the answer is that the deposit thing is measurable. How much you pay for your deposits is a number, and that number is in the accounts. How much the deposits are trading for on the open market, you know, what is the yield that they're offering to brokers for broker deposits in the US, or what are the yields that they're offering on websites that, you know, like Bankrate.com, which compared bankrupts. All of that is measurable. Hmm. Whereas when I'm looking at Metris, there's a lot of extend and pretend going on, right? Metris was every time a borrower was in trouble, it was extending, increasing their credit limits. And if they paid a loan, they'd get a bit more credit limit. And eventually they'd be in a lot more trouble with a bigger credit limit, but you couldn't see that in the numbers. The numbers in the accounts were deceptive, and the AI can't pick that deception up yet, it might have a little bit of a go at deception, but the AI's are very reluctant to tell you something is deceptive, right? Or to say nasty things about people. You can give the AI general instructions to be speculative about it, but their safety protections will hurt you. But I've yet to get to a situation where I can write a prompt and say, go analyze the asset base of this bank and tell me what the real delinquency is.
Speaker 1So what's a concrete example of deception with respect to loans? Well, that's hard. One of the
Speaker 2things that I've run, I ran NewBank, which is a big Latin American bank through to see if it was extending and pretending, and it couldn't do it, right? There's nothing in the accounts. And then I started trying to find checkboards where people from, poor people from Brazil talk about how they're ripping off the agent, the loans. I can't read Spanish, but you start finding places where the odd person is talking about, well, if you pay this, they'll extend your credit. And right, exactly what happened to Providian or Metris, where they'd give you a $2,000 credit limit, or they'd give you a $200 credit limit originally. And then if you paid those loans, they'd up it, $2,000. And then if you got into trouble paying the $2,000, they'd up it, and you then paid it, they'd up it to $4,000. And then you defaulted on the whole $4,000. And in America, there are predatory lenders, which the left talk about all the time, but there are also predatory borrowers. And the predatory borrowers look at a credit card offer in the mail and think, oh, goody, I can rip $4,000 off. Now, one of my questions was whether NewBank, which is a large bank, Metris on a grand scale for Brazil. And I can't see it in the accounts of Metris, but I know it was happening because I got involved in chatting with the people that were doing it. And I can't see it happening in NewBank's accounts, but these days, those people might be on chat boards in Spanish in Reddit. And so I found a whole lot of chat boards on Spanish in Reddit, and the answer was, yes, it's happening, but I can't tell whether it's happening this much or this much. And I can't tell whether it's happening this much or this much. And if it's happening this much, NewBank is not a very good short. And if it's happening this much, NewBank's a very good short. Now, maybe in the future, it will have much more access to information. It will find all those chat boards that will be able to compare them to other banks that blew up in the past. But at the moment, short of going to the favelas in the wrong parts of Latin America and talking to borrowers, I don't know how I would do it, right? There's no way that the AI can do it because the AI doesn't have the information. And the bank, if it's lying about it, is lying about it. And the AI takes the bank's statements as very high quality. Now, I have thrown at the AI things that I know to be grotesquely fraudulent. And the AI always wants to be nice about it. So there are ways that you can give general instructions. And the general instructions I gave them were trying to work out 70% or 80% right that it's abroad. And I'm right across 500 stocks. It doesn't matter, right? That's going to be a good list of shorts. So my standard of proof is not beyond reasonable doubt. It is a reasonably articulable suspicion. But the AI is exceedingly bad at doing that. Now, part of that is post-training. The AI is trained not to defame people. And I want to undo that training. And part of it is that the companies themselves are deceptive. So the AI is fed deceptive data and doesn't have a good sense of bullshit itself, right? There's a thing that people develop after a while, which you'd call a bullshit detector. And the AI is thin on that. Now, you can get it to areas where it knows it's bullshit, right? But it's not its default. Right. You still need a human here. In some sense, and I take a bit of solace in this, chess players first beat humans 25 years, chess computers first beat humans 25 years ago. But there was a 10-year period where a grandmaster plus a chess computer was better than a chess computer. The central model. Right. The grandmaster added value. The grandmaster doesn't add value. The place we're at at AI is quite clearly the very good stock analyst plus the AI is way better than the AI. And in fact, the very good stock analyst is still probably better than the AI. But the very good stock market analyst plus the AI is definitely a thing. But I still haven't managed to get the AI to do lots of things. And most of those things involve where the numbers have some level of uncertainty about them. Yeah. Yeah. It's all about the value.
Speaker 1It's also interesting in terms of how replaceable short sellers are. It's interesting how there's such a physical embodied component to your job. Like I'll often have a chat with you on the phone and you're in some country doing on the ground research, finding stuff out that you couldn't really glean just from using a computer.
Speaker 2Yes. But this is not AI related, but it's very funny how investor relations discussions differ between America and everywhere else. In America, investor relations is absolutely clearly a sales function and you get a very defined pattern that happens within the corporate office. And for the most part, there's no difference between getting an investor relations trip in America and doing it on Zoom. If you start doing it in Indonesia, the investor relations pattern is not really the same. It's not really a pattern. And when you go into the office, you'll discover that the culture is completely different to how you expect and you learn plenty on the ground. Now, Indonesia might be one extreme of that and America, the other extreme. But to the extent that you get a sanitized version, maybe the sanitized version leaves you in the same place as the AI. Right. Because the AI can get the sanitized version. It just, to the Goldman Sachs Buyer Side Conference with the CEO.
Speaker 1Yeah. Yeah. But to tie off this segment of the conversation, I'll put you down as extremely impressed with the technology. Unbelievably impressed with the technology.
Speaker 2I am- You expect the capabilities to continue to progress. I'm frightened about it because it's so good. Right. And I don't want to get too speculative because I have not a clue. I'm just a finance guy. I have not a clue where this tech is going. I mean, the rate of improvement of this technology is astonishing and it's going to upend large parts of the world in ways I don't understand. But I suppose you wouldn't entrust your money to a firm that had no humans in the loop. Not at the moment. Yeah. Certainly not at the moment. There are some very AI driven firms and I don't trust the judgment at all. And part of the issue is that your AI agent to analyze banks is only as good as you are. And I don't trust the judgment at all. And part of the issue is that you trained it to be. Yeah. Right. And that's actually going to be a more general problem with AI implementation though, which is- Should we get to that later?
Speaker 1Okay. Yeah. So I want to talk now about AI as a business. But first, just briefly, I wanted to ask you about situational awareness, this AI hedge fund that-
Speaker 2Sorry, I'm a finance guy and I'm laughing and I shouldn't laugh. But for a finance guy, situational awareness thing is just funny.
Speaker 1Do you want to? Do you want to? briefly recap the story. And the question I wanted to ask you is, if there was any avuncular advice you could have given the situational awareness team before the public markets book collapsed, what would you have told them? And is there anything you could have said to save the public markets book?
Speaker 2Leopold Aschenbrenner was a 22-year-old, maybe 21, and he wrote an astonishing essay in 2004 called Situational Awareness. And the astonishing essay has a single trick. It takes the scaling rules seriously. If you put this much compute on, you get an AI of this level of intelligence. If you put 10 times more compute on, you get more intelligence. If you put 100,000 times more compute on, you get this much more intelligence. And he predicted super intelligence on some scale. And he said that at some stage, somebody will build a trillion dollar data center. And he had a sketch of this trillion dollar data center, and it was this gargantuan pile of buildings with big smoking coal stations next to it. But ultimately, he was going to say this thing will take over the world. And he believed the scaling laws. He still does believe the scaling laws. When I first saw this and people talked about trillion dollar data centers, I thought, haha, I've heard many trillion dollar stories before. And here's one that's true. He was far more prescient about the technology than I was. And yeah, I know people who read the essay at the time and took him seriously at the time. But I read the essay at the time, took him as a joke, and he was right. And I was wrong. He has been more right about the tech than almost anybody. So he then starts a hedge fund with the name Situational Awareness. And the hedge fund just buys AI windows. And he's going to start a hedge fund with the name Situational Awareness. And short's AI losers. And his definition of an AI winner is whatever he thinks they're going to need. So for instance, he was long second tier power companies. And his iconic set of AI losers were SaaS software companies, because he thinks that the AI could just replace your SaaS system very cheaply. And I don't actually think he's wrong. None of this actually looked like he was wildly off on his prediction. He levered this thing four times. And there's a whole lot of variability in markets. And there was snickering amongst financial people who would look at his portfolio and just laugh and look at the fact that he's this 21-year-old kid who looks like he's 16 and laugh a little further and thought, this thing is going to come a cropper. And the time between when I started laughing at him and when it came a cropper was a long time ago. And I think that's why he's under two months. It was spectacular. He goes into financial markets and he makes a total and utter idiot of himself. And this happens both ways. You see finance people go into tech and make utter idiots of themselves. They buy techs that are going to be obsolete. They invest money in all sorts of ways. They're just wrong. I remember explaining to some very senior Hewlett Packard people that the company that they had just bought was a blatant fraud. That was when they bought Autonomy. And they didn't see it. And I explained why in very simple accounting terms. And they still couldn't see it. And two months later, they come back and tell me I was right. I wonder how the hell I saw it and they didn't. And it's just they have a tech view of the world and I have a very sophisticated accounting finance view of the world. So the tech people are spectacularly wrong when they go into finance. And the finance people are spectacularly wrong when they go into tech. And this has been a tech podcast series. And I'm a finance guy. And I'm coming out to talk to you. And I feel like I'm Leon Aschenbrenner in reverse. I'm the guy that's going to say things that are spectacularly wrong. And I'm going to be wrong in all sorts of ways that I don't know. And I'm hoping the tech guys tell me where I'm wrong. Because I know in advance that I'm right. But I do want to tell them that I'm wrong. And I'm going to tell them that I'm them where I think they're wrong. And that's telling a bunch of religious converts on the biggest and most important technology in the world that they're idiots is a dangerous thing to do for a finance guy. But let me try. And I'm going to try it by parable. My favorite explanation of this problem is actually the chemical industry. And the chemical industry covers a multitude of things. And I'm going to tell you what that is. It's an awful lot of money for your spoonful of pixie dust. And that's the extreme example of specialty chemical. And at the other end of the chemical business is a giant plant that costs billions of dollars to build that takes in natural gas at one end and sends out plastic pellets at the other end. And that business has very high capital cost, fairly low marginal cost, and is subject to competition on the output. And that could be very profitable or it could be very unprofitable. And the situation where you have excess supply, it becomes very unprofitable very fast, right? Because immediately your price drops to your marginal cost. You've almost invariably borrowed money for the capital cost and you go bust. And this is a lesson that I give people time and time again, which is if you have high capital cost, low marginal cost, and you're subject to competition the moment that you have excess supply, this just turns into the worst business in the world. The business, that parable is told several ways. Warren Buffett tells it about his textile company where he was selling the lining that goes on the inside of your jacket, which was a commodity. And the guys come in and say, hey, we're going to, we've got this technology that makes us a more efficient loom. We're going to build an extra loom and make everything 15% more efficient. And that's the moment he decides to close the textile company. And he's going to close the textile company. And he's got the second part of that story right. The second part of the story right is if you have high capital cost, low marginal cost, subject to competition, and you get two to five or 20% technical improvement every year, you're absolutely doomed. And the reason you're absolutely doomed is there's a 20% more efficient plant. Somebody's going to build it. And suddenly you're going to get excess supply. And then there'll be another 20% more efficient plant and somebody will build it. My favorite example is solar panels. Solar panels have completely changed the world. In my lifetime, solar panels have gone from about $5 a watt to about 15 cents. And that's only sort of 15 years, right? They're probably $50 a watt if I go back further. At the current price of solar panels, if you're anywhere where there's excess land and the sun shines, the cheapest form of electrons is solar panels, period. In Australia, almost all these data centers, will be largely powered by the sun and batteries. This is the cheapest form of electricity. You'd think that making solar panels, you know, this technology that completely changed the world and they're selling billions of dollars worth of them would be a good business, but you'd be wrong, right? I don't really care which solar panel I have. They're subject to competition. The building of the plants is high capital costs. The running of them is low marginal cost. I just buy them off price. And there's been a couple of percent technical improvement every year. So if you look at the Chinese solar panel manufacturers, every one of them has been bailed out by the Chinese state at least three times, right? This is like the worst business I've ever seen, even though it's changed the world, right? I can go back to the Buffett example. You have high capital costs, low marginal costs, subject to competition. And there was a technical improvement of this 15% better loom. And Warren Buffett was clever enough to say, well, I'm not investing in that, right? And he was right because that would have caused the next price drop. Now I look at data centers or I look at AI generally, and the capital cost of a new model is a billion dollars of training and then 10 billion of training. And in Ashton Brenner's vision, the capital cost of a model is a billion dollars of training. And in Ashton Brenner's vision, the capital cost of a new model will be a trillion dollars of training. These scaling rules work seemingly forever. and the marginal cost is inference, which is much lower than the capital cost of training. And this looks like it's high capital cost, low marginal cost subject to competition. And the moment I see this, I have a prediction that everybody's going bust. The absolute last place I want to invest in this world is in AI. Now, that's really sort of schizoid, right? At one end, I think this is the most impactful technology of my life, right? This is going to change the world beyond recognition. And at the other end, I think it's like the worst investment profile I've ever seen in my life. Now, I don't want to discourage you. I like this AI. I'm going to be a customer forever. I really like it. So please invest lots of money in it. I just will not be joining you. Now, the funny thing, though, is Silicon Valley has been here before. The last time that Silicon Valley invested large amounts of money in technology was in fiber optics. And I remember Global Crossing, but I also remember Enron did, you know, vast amounts of fiber optics, and Williams did, and QWest did. And, of course, WorldCom did. And more or less every penny that was ever invested in fiber optics went to zero. But fiber optics changed the world. I run a global hedge fund. I run it from Australia. I talked to somebody in America for an hour and a half this morning for the same price that I could talk to my wife at the suburban shopping center. Right. And the ability to globalize. The world completely and utterly depends on fiber optics. The whole modern Internet depends on fiber optics. It's just that everybody that invested in fiber optics lost their money. Yeah. Well, what did you have? You had a very extreme version of high capital cost. It cost billions of dollars to put these cables in. Very low marginal cost. Subject to competition. There were plenty of competing cables. And worse, there was technological improvements all the time. The particular technological improvement. That drove it last time was optonics optics and or tunable lasers was the product. And in the old days, you think of a fiber optic cable where you're sending information down as light and you'd have a single red laser. Now you can get if you up the frequency, you can get more information into your fiber optic cable. So a blue laser has more information than a red laser. This is multiplexing. No, just change just different frequencies. A blue laser. Is going to have more, more information than a red laser, which is going to have more information than an infrared laser. Now, we know a blue laser has got more information than the red laser. When we went from a CD player to a DVD player and high resolution, we eventually had this thing called a Blu ray player. And what was a Blu ray player? It was a higher frequency laser, a Blu ray. So you could change the frequency of the laser and you'd get more information down your fiber optic cable. Just like you get. More information out of the Blu ray player than a CD player. Right. But why stop there? Why not put a red laser and a blue laser down or a red laser, blue laser and 25 other colors of laser down? Now, there are some minuses. If you put 25 colors of laser down, you need a repeater more often. Your repeaters go from, say, once every 80 kilometers to once every 30 kilometers. But you're suddenly getting 50 times as much information down a fiber. And this technological improvement was so vast that many of the fibers that were laid at billions of dollars of cost were never lit. Right. The so-called dark fiber problem.
Speaker 1There was about $100 billion of dark fiber left behind.
Speaker 2I don't know the number. I just remember looking at the bankruptcies afterwards. Right. But the bankruptcies afterwards are sad. Right. I mean, capitalism produces. It produces scammers who will invent stories and sell you bullshit stocks. And I like shorting those. And I don't feel any sadness when one of those fails. Right. I mean, and I don't even particularly feel sadness for the victims because most of them are sort of willing dupes. Right. But I feel sadness about the fiber optic cable companies failing. Right. But the fiber optic cable companies built a technology. That made my life massively better. Right. They changed the world in all sorts of good ways and they still got paid with a donut. And that is the nature of capitalism. And it happens all the time when you see high capital costs, low marginal costs subject to competition. Right. If I can leave you with, you know, the whole purpose of me coming on this podcast was to say this industry. It looks horrible to me. Right. Right. It's high capital costs, low marginal costs subject to competition with technical improvement. And I've seen the movie many times before. And you, as an investor, should have the Warren Buffett reaction. Oh, everything is getting 15 percent better. I'm not going to invest. Now, I should also say that I get I see increasing alarm when I hear people saying that governments should be involved in investing for sovereign.
Speaker 1And can I pause you before we get to that? I've got a bunch of follow up questions on fiber, because I think I think the telecom boom is a really interesting historical analogy for the boom. So a few follow ups here. Firstly, were you short WorldCom at the time?
Speaker 2No, but I was about 30 and an idiot. I would have been shorted now. And in fact, I probably wouldn't have been short the stock. I would have been short the debt. I see. Now, the reason for shorting. This is a hedge fund thing rather than a tech thing. OK. If you're short a stock at 10 and it goes to 100 and then to zero, you're almost certainly going to lose money, even though you're right. And the reason you're going to lose money, even though you're right, is that at 50, you're now five times as short and you're basically forced to cover. Right. Your risk management rules tell you you have to cover. Yeah. You've got to have those, you know, otherwise you're Leaping Leo Ashbrenner. Right. But if you're short the debt, you just pay the coupon. OK. It only works when they're going to hit the wall spectacularly. Yeah. Right. But short the debt of WorldCom was a great trade. Right. Because it didn't matter what the stock did. And eventually, you know, this. The junior pieces of debt were all wiped out in that bankruptcy. But in those days, I was working for an equity fund that wouldn't have shorted the debt anyway. Right. Is that platinum? Yes. Right. It was a mutual fund. Right. I had different clients. Yeah. Right. So I had different freedoms. Yeah. Which, you know, at best is freedom to do good things and at worst is freedom to get yourself into trouble. Right. And so I get obsessive about risk management. Incidentally, going back to Leo Ashbrenner, Ashenbrenner, if I could have given him one piece of advice, I would have said, take all the corpus on risk management, the conventional portfolios and run his own portfolio through it on Claude Opus at max. Allow Claude Opus six hours to give him the range of explanations. They would have told him he was going to go bust with a 95% certainty. There was nothing unconventional in the way that Leo Ashenbrenner went bust. His mistakes were totally predictable. They were within the range of something that an AI could work out. And he's within the range of things that I could do in my head. Right. I just looked at it and said, he's going bust. Right. I laughed. You know, by contrast, the people at Citadel worked out how to make a billion dollars off his bankruptcy. So I'll give them, you know, the credit here. But Leo should have trusted the AI. Right. Now, again, knowing how to use the AI, in this case, he would have had to ask the AI the right questions. He would have had to ask it the finance questions. Just like the only way that I could analyze the bank was to know enough about banks to ask the AI the right questions. Right. But the AI could have. Solved his problem. One average finance person and one very good AI would have done him a lot of good.
Speaker 1Second follow-up question on the telecom bubble. Can you help me sharpen the analogy? Who are the fiber optic laying companies in the AI sector? Is it the AI labs? Is it the neoclouds, the chip makers? Is it the hyperscalers? Is it hyperscalers at a particular point in the deal cycle?
Speaker 2This is really. Really hard. The whole industry top down looks like high capital. cost, low marginal cost, subject to competition. If you look at the model, it costs you a billion or $2 billion or $20 billion or a trillion to develop one of these models. And then running it is low cost. Now, it may be that once you get to a model that costs a trillion to build, there's only one of them because nobody else is going to sink the trillion dollars. So the vast capital cost may actually wind up being your protection from competition.
Speaker 1Well, exactly. I was going to ask you about that.
Speaker 2Right. But certainly at a billion-dollar model, there's no protection from competition. There are plenty of people who will sink a billion dollars into building a model. Yep. Then if I go back down, the data centers are the most extreme thing, right? They're selling just the ability to do a mathematical calculation. And it doesn't really matter, provided they have the same reliability. They don't lose data, right? They have the same uptime, et cetera. That looks like the ultimate commodity. And it doesn't really matter if I buy from NextDC or Corweave or from Google. Now, there's somebody on the internet recently who is talking about the difference in quality of data center. And it's not the quality of data center you can buy now. It's the quality of data center you might buy in a Ford contract. Or Darted for compute to be delivered next year. And I can buy compute from NextDC, which is a somewhat problem or Firmus, which are both problematic Australian data centers, or I could buy compute from Google. Well, if I enter into a contract to buy compute from Google, I can be pretty sure that I'm going to get that compute delivered to me. But if I enter it into a contract to buy compute from one of the problematic Australian data centers, it's not obvious. So, it's not entirely commodity, at least in the Ford contract, right? And there is some advantage in being Google. But it looks awful like a commodity once built, right? I'm not sure that the compute that SpaceX is delivering Anthropic is any fundamentally different to the compute that Amazon is delivering Anthropic. The compute that SpaceX is delivering Anthropic is a much higher price than the compute that Amazon is delivering. Right. In fact, I
Speaker 1recall in Anthropics S1, that even developed some system for making compute across the portfolio somehow more fungible. Do you recall that? Yes. And this is the puzzle.
Speaker 2I've just spelled out the most ridiculously bearish AI hypothesis. And deep in my heart, I believe it, right? Deep in my heart, I believe that everybody, that this technology is going to be solar-powered, or fiber optics, it's going to change the world beyond recognition and beyond either of those technologies by a long way. And everybody that touches it will go bust. And then I read the Anthropic S1, and I have to say, I've not had a document that shocked me more in the last five
Speaker 1years. Yeah. So, the day it was published, you called me and you said Anthropic was a much better business than you'd realized. Do you- I still believe that. Still believe that?
Speaker 2I still believe that.
Speaker 1And what surprised you most about the S1?
Speaker 2Okay. The total amount, and the Anthropic S1 only has numbers going to the end of July, and we're currently in October. There'll be an updated one before the prospectus, the final prospectus, which should include October numbers, but not November numbers. Yeah. But the total amount that they had spent on compute, was about $19.5 billion. It was just under $20 billion. That compute could be broken into compute for inference, which is sold, and compute for training the models. The total amount that they'd spent on training was not delineated, but using some rough estimates and some AI, and this is all plausible, I get to about $12 billion. More than half of that compute was spent this year. And in fact, it was spent after they developed Fable.
Speaker 1I'm sorry, was the $12 billion, is that over the last 12 months?
Speaker 2No, cumulative in the whole history of the firm. Wow. Right? And that was after they developed Fable. So before they developed Fable, the total amount spent on training was only about $6 billion. And that number does not gel with the Ashen Brenner view of the world. OpenAI has, by all accounts, spent about $45 to maybe $60 billion on compute. Actual compute consumed at Google is probably in the $60 billion range. What you have at Anthropic is what is clearly, at the moment, the best model. And it was produced for lowish single billion dollars of training. And that shocked the hell out of me. Right? And it shocked the hell out of me for a couple of reasons. The first one is the total amount of compute. Well, I had Leo Ashen Brenner's paper in my head, and I just assumed, you know, in order to have the best model by this stage, they would have spent $100 billion. They just haven't. And if they've managed to build what is clearly the best model for lowish single-digit billions of dollars, then it's almost inconceivable that the $7 trillion of compute that people are putting in at the moment is going to have a market. I mean, my first reaction is, oh, my God, this, you know, just the CapEx at the NeoClouds and the compute suppliers just looks horrible. Just looks horrible beyond belief. The second thing was, Anthropic Claude is about five times more efficient than what I think the numbers are at Google and at OpenAI. Now, Anthropic's S1 doesn't say, we have made ourselves five times more efficient than the competitors, and we have some edge. They're just not that blatant. But that's what it looks like. And they do say, they have spent a lot of time and effort working out efficient ways of optimizing compute. And I've got to say, I believe them because they're five times more efficient. Now, I've been bashing this idea around with friends for ages, right? Which is, why are they five times more efficient? And we don't know. So let me give you some hypotheses. Hypothesis A is it's an illusion that they just have a really cute, cheap deal of compute, right? Because they signed a very, very big contract, for instance, with Amazon, which requires that they buy $150 billion of compute between now and the year 2035. And Amazon, at about the same time, bought a whole lot of equity in Anthropic at a price that now seems incredibly favorable. And so the compute might be a sweetheart deal. But I don't think so. Amazon didn't sign a deal out to 2035 at a price that doesn't give them a decent return. And the second thing is that Anthropic is currently buying spot compute from SpaceX, which is disclosed in the S1. And that deal is $3.5 billion per month, cancelable by either side on 90 days notice from March next year. Now, Elon Musk, and I don't disbelieve him. I usually disbelieve Elon Musk, but this time I think he's telling the truth. But Elon Musk says that the best spot.
Speaker 1Take him seriously, not literally.
Speaker 2Yeah, but this time I'm going to take him literally. Okay. Right. Elon Musk says that the best spot deals he's signed on compute have under a one-year payback. And the best spot deal he ever signed on compute was with Anthropic. Now, I'm pretty sure Anthropic will cancel that deal when they can, when the spot price of compute drops. The world is currently very, very short compute, right? But it won't be short compute in two years. Not the rate that this stuff's being built. And. I'm pretty sure Anthropic will cancel that deal. So, I don't think it's a sweetheart deal on compute that has made them so efficient. So, the next hypothesis was that they have an edge, but the edge isn't caused by anything that's really smart or clever. It's just caused by the competition being managed really badly. This is the Sam Altman runs a brothel thesis. Right. And. I've not heard this thesis. The thesis is that Sam Altman. I mean, I'll give you the kind version of it. The kind version of it is that Sam Altman's barely a businessman. He's more a promoter. And he's prepared to burn vast amounts of money on things that are promotion rather than business. And the exhibit is Sora, which was his video processing AI that he made very cheap or free to the public. And Sora used vast amounts of compute to make cat videos. And there's no commercial model here. It's just that Sora was really good advertising. And so Sam Altman burns $10 billion of compute on cat videos, right? Which is, you know, Sam Altman runs a brothel. Okay. Right? And that there really is, that Dario at Anthropic is just a really hard-ass businessman who would never burn $10 billion of compute on cat videos. And his competition is just awful. I think that thesis actually got some legs, right? You know, right? There really was a lot of billions of dollars spent on cat videos, right? The next possible thesis is that they have some kind of edge, but that edge will be caught. And that's the most likely outcome, which is that, you know, if there really is a ultra-efficient way of making these models and these models, you know, Anthropic can run Claude cheaper than OpenAI. Can run its chat GPT. Eventually, OpenAI will either use chat GPT or Claude to work out what the solution is, and they'll just catch up. And if they just catch up, then we're back to my horrible high capital cost, low marginal cost subject to competition. Yeah. And the fourth thesis, which I think is the least likely of them, is that OpenAI genuinely has an edge, that edge is technical and the edge is sustainable. And they ain't going to tell me what it is.
Speaker 1OpenAI or Anthropic?
Speaker 2Sorry, Anthropic really has an edge. Yeah. That edge is, right? And if Anthropic really has an edge, then what we're going to find is that Anthropic can build better models cheaper, sell them cheaper, and won't run out of cash, and the competitors all will. And that's an incredibly bullish story for Anthropic. And the reason it's incredibly bullish is not only do they win, but everybody else is building these data centers. which will soon be for sale from the liquidators for pennies on the dollar. And OpenAI, sorry, Anthropic will wind up not only winning, but it'll have somebody else bear a large slab of its capital cost. Now, I am like the most bearish person on AI economics as opposed to AI product that I know, I'm still going to buy some Anthropic in the IPO, right? And that's the sort of headline here, famous short seller buys Anthropic, right? I'm happy actually, if you run that, right? But it's like, I'm not that crass. But it's an interesting take because there is one way that you can solve the high capital cost, low marginal cost subject to competition, which is that you can make the capital costs so large that no one will ever go near you.
Speaker 1Exactly. And it's because the barrier to entry is so high, there's not much competition.
Speaker 2Right. And if you have an edge that lasts five years, but that edge that lasts five years is enough that you capture most of the business in the short term, then you buy your data centers for pennies on the dollar. And nobody else can do that after you've done it. I mean, Jack Teo became a billionaire in Australia when he started TPG. And the way he started TPG was literally he went to the liquidators' auctions for all the Sealex in America after the telecom bust, and he bought their capital equipment for $0.02 to $0.05 on the dollar. And then he set up his data centers next to, or his internet providers next to the highest demand users. So the first places he started were near the universities. And he would give people very high data contracts for not much money near the universities. And Telstra's view of it was that he must be loss-making and therefore they didn't bother trying to compete. And buying a capital equipment for two cents on the dollar is a very, very good way of getting ahead in business. That's the Jack Teo strategy, but I think the Jack Teo strategy could be run quite neatly here by Anthropic when everybody else goes bust.
Speaker 1So to turn the tables and inject some pessimism into the conversation, if we're sitting here in 2030 and Anthropic has failed, What's the most likely reason for that?
Speaker 2If it's either somebody else has got the anthropic position or the more likely scenario, and I think it's the most likely scenario, is that everybody's failed. Because frontier AI has become commoditized. Yeah, because you get to the point where it doesn't matter whose AI you use, right? They're all extremely good. And there's just price competition.
Speaker 1Scaling a business is hard enough already. So you don't want to be bottlenecked on things like compliance. And while achieving compliance can help you win bigger deals, enter new markets, and deepen trust with customers, it can also cost a lot of time and money. By automating up to 90% of the work involved in SOC 2, ISO 27001, and more, Vanta gets you compliant quickly. It also saves you money. A recent white paper by the International Data Corporation found that Vanta customers achieved $535,000 per year in benefits, and the platform pays for itself in just three months. A startup I worked at used Vanta, and I found it really reliable. Over 12,000 global companies like Atlassian, Cursor, and Replit also use Vanta to manage risk and prove security in real time. My audience can get a special offer of $1,000 off Vanta at vanta.com slash joe. That's V-A-N-T-A dot com slash J-O-E for $1,000 off. So I've got some questions about Frontier Labs as a business. Suppose GPUs get 10x more efficient per watt over the next three years. Is that bullish or bearish for the AI labs?
Speaker 2it's bullish for the labs and terribly bearish for everyone that does data centers now yeah that could be an overlap at the moment um anthropic owns no data center it just rents everything right it pays pays wholesale commute compute anthropic is building its own data centers in its ford plans now not all of them it's still entering you know it's still buying vast amounts of compute from amazon and things like that right but you if you get new optic chips for instance that are 10 times as electricity efficient then every every current chip is obsolete because the running costs of these things are so high right so then now whether the data center shell companies you know there are companies like digital realty who build the shells and the power supply and the shelving and the cooling, but they don't put the chips in. They don't put the computers in. Those might be fine, but the people who put the computers in are all stuffed. But by and large, if your inputs become cheaper, it's better for you.
Speaker 1But then it makes it easier for your competition unless we just keep extrapolating the scaling laws and then the
Speaker 2question is if you build the trillion dollar model which in ashen brenner's view right and anthropic you read the anthropic s1 and the first thing you have to notice is these guys are totally committed to the scaling laws they believe everything that leo said in that paper several years ago they believe that if you have a two trillion dollar model, right, it's just going to be smarter than Einstein. Yeah. There is a real question of how much of the time do you need to be smarter than Einstein, right? I mean, the task that I gave, that described at the beginning, where I feed the plans for a fairly elaborate house in and ask it to work out what my electricity draw, if I wish to air condition it is, that task is well within the range of a good first-year physics student at uni, right? You could just work out all the variables pretty fast. To actually do it for the first-year physics student would probably be a week or two work, and you'd probably make a mathematical error, but you certainly don't need to be Einstein. Now, if the task is take a look at this poison, like Scorpio bit. Well, there was a recent example, a discussion of scorpion venom, right? And see if you can design an antivenin, right, from first principles. Maybe you need to be Einstein. I don't know how to do that. I know how to do the air conditioning task. I've got enough university physics to at least know how to do that right i don't know how to do a biological task maybe it can be done maybe it can't be done maybe there's enough of a business model in having einstein in your pocket yeah but part of the problem is genuinely i don't know what you do with a model that's that smart yeah i don't know what you do with people that are that smart yeah um maybe that maybe the frontier is infinite. You know, the slogan is, first solve intelligence and then use that to solve all the world's
Speaker 1problems. Maybe this is a dumb question, but if the frontier is quite quote-unquote, infinite, and a frontier model is state-of-the-art for maybe six months, should we be thinking of the cost of training models as CapEx or OpEx?
Speaker 2The oldest question in accounting. Every time I see a CapEx, OpEx discussion, I'm reminded of the famous line of, what is pornography and what is art? And the answer is, you know it when you see it. If it's six months, it's OpEx. But that is a pornography question. You know it if you see it.
Speaker 1And when you look at it, what do you see? At the moment, I still see CapEx.
Speaker 2These models are getting better. There is no question in my mind that Fable is amazing. And I'm pretty sure the model that comes after Fable will be even better. More amazing, right? And they are products that will be used for very long periods of time. I'm building agents that I suspect I'll still be using in seven years, so that looks like CapEx to me. But at some point, it comes OpEx, right? And I guess at some point, your art becomes pornography, but I don't know where that point is.
Speaker 1To achieve an adequate rate of return, on that CapEx, what kind of, do you have any rough numbers around the sort of order of magnitude of revenue they need to generate over the next few years?
Speaker 2Yeah, that's the biggest joke of all. That is the funniest thing I've seen.
Speaker 1This is where you're back to being bearish.
Speaker 2Yeah, I get bearish here. Okay. The first question is, who's going to pay for this AI? Right. Now, I'm a very rich bloke by most people's standards, and I was perfectly happy. I was perfectly happy for a family account to sign up for a $30 a month Claude account. I've actually even got a $30 a month personal OpenAI account as well. So I've got two of them. Even that adds up to less than my mobile phone bill, right? The number of people I know who will sign up to $100 a month is relatively low. If you add up to $100 a month per household in America, you don't get that much. You don't get anywhere near trillions of dollars, right? There's probably at very best 100 million households that might do that in the US, right? It just doesn't get you there. In order to get there, you need to sell to business, right? Now, the current numbers, and I don't think these will be delivered, right? I think the crash will come well before then. But the current estimate is that the capex for AI, data centers, et cetera, between now and the end of 2030 is about $7 trillion.
Speaker 1Sorry, just for people's context, there was a good paper by the Brookings Institute, which I'll link to, which estimates that that spend between 2025 and 2032 is about 3.5% of US GDP on average in the US, which already dwarfs, I think it's probably three times the size of the telecom boom. Three times? Many, many times relative to GDP. I think the telecom boom was a bit over 1% of GDP, US GDP.
Speaker 2Really? I thought it was less than that.
Speaker 1Okay. And it also dwarfs the railway boom. The railway boom is the big one. Yeah. But it dwarfs even that in terms of the average annual percentage, although it's not quite a fair comparison because the railway boom went on for 20 years, so it lasted a lot longer than the AI boom has so far. But anyway, I interrupted you.
Speaker 2Anyway, you get to $7 trillion. $7 trillion. This is compute, right? You're going to have to assume that it has five-year lifespans, right? Every, you know, every.
Speaker 1Is that more because the chips are superseded or because they physically degrade?
Speaker 2But the chips will probably degrade. Google used to run its data centres full blast and it changed the estimated lifespan of its chips from four to six years eventually, right? And I believe Google when they say that, right? But that, and that change, had multi-billion dollar effects on Google's stated earnings. So it was something that if you followed Google's accounts, you noticed. But it never changed them to seven years. And they just burn out, right? There's, right? But the second thing is that Moore's law applies still, right? The new chips will be more efficient. They will be better, right? The value of the chip is heavier when it's cutting edge. So if you think about its depreciation, it's probably functional. Economically, a bit upfront, right? I mean, at the moment, compute's worth very, a great deal, right? But, you know, in six months time, the price will be lower. So how should you depreciate your chip is a real question, right? But call it five years, call it six years. I've got five in my head. It's not out, right? They use a lot of electricity to run, right? A lot of cooling and some other stuff as well. You know, there are people there, not many in the data centre, but there are people there. So in order to get, in order to get your $7 billion back, you need about $1.2 to $1.5 billion a year, depending on your lifespan in EBITDA, which means, trillion, sorry, which means you probably need $2 trillion of revenue. If you're going to get it from the business, the place you're going to get it from is Firing Star, right? I mean, this is a very, very good productivity tool. You know, don't kid yourself. The AI is after your job. And that applies whether you're a carpenter or a stock analyst.
Speaker 1But if you get it from Firing Star. It almost, it has to be after your job as a consequence of the scale of the build-up. Yeah, you've got it, right? The only way to get the revenue to get a sufficient return is to cut the wage bill.
Speaker 2Right. So let's put this in perspective. If you're going to get $2 trillion a year, you've got to save about $4 trillion of labour at the low end, right? Because people don't fire, somebody for a 10% saving, right? Because it's high risk. You've spent money training the person. The person fits in your business. You've got to restructure your business. You don't fire people, you know, 10%. So if you've got to find $4 trillion of labour savings, that's a big number. And I'll give you an idea how big a number. The total wage bill, every wage paid in the OECD is about $40 trillion a year. So you've got to save 10% of the wage bill across the OECD. Now, at the moment, it's probably going more after high-end workers than low-end workers. So it's probably less than 10% of the employees in the OECD. But of people who are my friends, it's more than 10% of them, right? I can't see how you get a return on that $7 trillion unless you cause a massive global recession and a massive, global restructuring. And if you do cause that massive global recession, you ain't getting the return anyway. There is the second problem. And the second problem is, I don't think that this can get into business at the rate that makes that return possible.
Speaker 1This is where the slow diffusion of the technology and the bubble question, it kind of flips sides of the same coin.
Speaker 2Yeah. Let's go back to my original example. My original example was, I wanted to train the AI to understand banks. And I took all of my understanding of banks and all the extreme examples I had in my head, Silicon Valley Bank or OZK or ANZ Bank as different ways of getting deposits. And I put all of those examples into the AI. And having spent many days putting just that bit of my knowledge into the AI, I had a working model that could replace me as an analyst of at least the deposit part of a bank. But that agent doesn't work unless I put my skills into the AI. And that's going to be a general problem across society. There are plenty of jobs that middle management in companies do that can and should be replaced by an AI. But the incentive structure is such that the middle management won't do it. Why would I do that if I thought it was going to result in me being fired? So this is, of course, going to wind up being an incumbent's curse. A new company won't have that problem, right? Because it doesn't have the staff with the incumbency problem. And so new companies will displace old companies. But I can't, in my head, make the AI spend get a reasonable return unless you sack 10% of the workforce of the Western world. And I can't see how you sack 10% of the workforce unless the workforce train the AI and I don't know how that's going to happen, right? This doesn't mean that AI won't eventually displace 10% of the jobs. I think it will displace more than that. I get bearish about the very structure of capitalism if the AI can do almost everything, right? I mean, how do you distribute that? I saw you start winding up thinking about fantasy worlds that look like Star Trek, where the replicator makes whatever you want. And because the replicator makes whatever you want, there's no factory. There's no work as such. And the only work is exploring. That's a nice world, but I don't understand how it's organized. Right? There's no, you know, I understand how a capitalist world is organized. I understand how a democratic world is organized. I don't understand how Star Trek is organized. R.I.P. Patrick Stewart. I binge-watched Star Trek The Next Generation when I heard he died. Oh, really? Yeah. Actually, that said, his management lesson is very simple. Make it so.
Speaker 1So I want to come back to, well, let's see. I mean, we have already touched on this, but do you have any just general takes on this question of how likely it is that frontier AI will be commoditized? I'm a finance guy.
Speaker 2Yeah. Ask. Ask Leo. Okay, okay. No, no, I mean that, right? No, no. If I start opining about where the tech is going, I'm probably wrong.
Speaker 1Yeah. Yeah. I think we all fall into that category. So just, I wanted to pop back. I had a couple more follow-up questions on the telecom bubble that I didn't get to ask. One was, so for me, the lesson to take from the telecom bubble is to never underrate the elasticity of supply. And you can see how this might apply to compute. For example, we might find ways of decentralizing training compute. We might find ways of building out orbital compute. We might find ways of doing photonic compute, which there's a good Substack post by Alex Turnbull on that I can link to. But I think that's the main lesson I take.
Speaker 2I think you overestimate just how much the supply is. The supply matters. In solar panels, you've bankrupted everybody three times with 2% or 3% efficiency gains each quarter, each year. In Warren Buffett gave up on the textile industry with a 7% or 10% efficiency gain staring him in the face. If you got 5% efficiency gains every quarter in a high capital, every year in a high capital cost business, everybody goes bust even if the product changes the world. If you've got 50%, it's just worse, right? The elasticity of supply just made it worse.
Speaker 1So the supply coefficient doesn't even need to be that big. Is that what you're saying?
Speaker 2No, just all you need to do is have an incentive that's big enough that someone builds the next plant every year, right? And then you'll guarantee the excess supply. Okay, yeah, interesting.
Speaker 1So John and I were just discussing how elastic the supply of compute could be. Well, as it happens, I worked through this same question, but from a public policy perspective with Greg and Ewan from E61, a nonpartisan economic research institute focused on Australian public policy. We wrote up what we found in a special edition of E61's Plus 61 newsletter. To receive a copy of our essay on data centers and the compute economy, go to e61.in slash joewalker. That's e61.in slash joe. W-A-L-K-E-R. Can you tell me what to do with this fact? One big difference between the AI boom and the telecom boom or the railway boom is the economic lifetime of the infrastructure. So optical fiber and railways are going to last for very, very long periods. The chips last for maybe five years.
Speaker 2I think that just compresses this thing. Okay. Right. No solar plant that has been built has lasted more than seven years. Right. And the reason is the technology improvement is fast enough that they don't last. Now, railway, part of the trick is that we managed to get more and more out of the lines. So we made the trains longer. We digitized their separation so that you can now run trains separated by only a few minutes. Right. Precision railroad was the jargon, which was their way of getting more and more and more out of their infrastructure. But the. The core infrastructure remained. In the solar panel case, every 10-year-old solar panel plant does not exist. Right. It's already obsoleted by the next generation. Chips used to be that way, but they're not anymore. And the only reason they're not anymore is that we found large amounts of use for obsolete fabs. Right. One of the, for instance, my favorite use for those is what's called MEMS chips, where I'll give you an example. You make a really bad capacitor. And the really bad capacitor has some hydrophilic plastic between the layers. And this can be a capacitor the size of a pinhead. And then you print that capacitor onto a silicon chip, which has a minimal amount of computing and a minimal amount of NAND memory. You put it in a room with 20% humidity at 20 degrees, 30% humidity, 40% humidity, etc. Then you calibrate it, sort. sort the calibrating in the NAND chip. And now you have a humidity detector that is the size of a pinhead. You might wonder what the hell the use of this is. Well, these are now so cheap that in a high-end car, they put them in the air conditioning. And in the old days, I remember, you know, trying to dial up and down the air conditioning to get the demist right. So that if you put the air conditioning up too hard, it was too cold in the car. If you didn't put it down, your front windscreen. and there was a magic point where it was just right. Well, now the car will work out that magic point because it has a humidity sensor in it. Now, that little device got used on an old chip, right? Because a 20-year-old or 15-year-old wafer is fine for that because it doesn't have very much. Those sensors are everywhere in our world. And the example du jour is that is in electric car batteries. This was the trick that Elon worked out or Tesla worked out, which is in the old days, you'd run all these batteries together and the cars blew up. And the reason the cars blew up was that you'd have one battery that ran away, got hot, misbehaved or would burst. And then you'd get a consecutive flame and the flame would wipe out the whole battery. And so the electric car company before Tesla was something called Fisker. And they only built 40 or something of those cars and more than half of them ended. And now electric cars don't end in fireballs. And the reason they don't end in fireballs is we put a MEMS sensor, which tests temperature and things like that on every single cell of the battery. And we know that this battery cell here is running away. So we turn it off or we moderate how much power is going into it so that we stop it bursting into flames. And the Tesla battery was a computer controlled battery. And so we put these. We put these little MEMS sensors, 7,000 of them into your Tesla. Now nobody, no Joe Sixpacker knows that 7,000 little bits of silicon are sitting on the battery in your Tesla, but I promise you they are. That sort of sensor technology is now everywhere in our society and we don't notice it. And so suddenly all these chips, these fabs that were not cutting edge, found a use. And so the return on old equipment. Equipment turned out to be much better than my model would expect. And I'm always puzzled by, you know, how often I'm wrong on things because I would have estimated that a 15 year old fab was worth zero and I would have been wrong. Yeah. And maybe one day I'll be wrong the same way, which is I will estimate that some 15 year old data center is useless and we'll find a use for it.
Speaker 1Well, but the interesting thing about that story about fabs is it, it makes me think, you know, if slash when there is overbuild in this boom, the nice productive infrastructure left behind might not, might not be the GPUs, but it might actually be the fabs and the energy generation.
Speaker 2I think it's more likely to be the energy generation, but I am waiting to be wrong.
Speaker 1A couple of questions about compute. So say an analyst was applying for a job at Bronte and you set them the problem question of forecasting compute demand for 20,000. 20, 30, what do you want to see them do in an answer that would get them hired? So how, how should they go about trying to break down that problem?
Speaker 2I don't know what I would ask if they're a techie person, I would actually ask them to explain all the tech and the scaling rules and explain where they can get to. And then I would ask them the questions of what do you think you can do with that model? And if they're a techie person, they're going to sound like, I would ask Leopold, Ash and Brenner, and I don't know how to assess such a person, right? Because they're just sort of wide eyed. And when I talk about tech, I'm an idiot, right? But, you know, at least one of the ways that I would assess such a person is I'd introduce them to one of my techie person friends and get their assessment, right? The finance person, I would ask them to cost it out, right? You know, what are the cost structures? What drives it, right? How do you learn enough? about the technology to produce a reasonable cost structure um what difference does it make if you put it in queensland versus the united states right would be a nice costing structure and in queensland your construction costs are higher but your electricity costs are almost certainly lower and they're lower for two reasons one is the queensland government wants to encourage coal but the more realistic reason is that solar panels have trump's stupid tariffs going into the u.s and you know if the chinese are going to waste large amounts of money building solar panel plants that make nice efficient solar panels i don't understand why you'd want to tariff them right thanks i'll have those solar panels do you think
Speaker 1it's easier to forecast uh inference compute demand or training compute
Speaker 2demand i have no idea um i mean i if i look at um anthropics s1 i've been wrong on both so right i really have no idea
Speaker 1um how problematic is the circular funding in ai and what's have you seen any examples that have struck you as particularly
Speaker 2egregious oh it's problematic like shit okay right um but i don't know how problematic it will wind up at the end right um i sometimes think the circular funding is there so that people can get control of the asset in bankruptcy right but that's not a standard view i i mean circular funding has almost always been bad news in finance but in this case people really do want to spend seven trillion dollars it's got to come from anywhere just
Speaker 1to um just to make sure i have this really clamped down um if there is a bubble in ai like where specifically is it located it's going to be
Speaker 2well the one thing that's unambiguous is the compute right high capital cost low marginal cost subject to competition and nothing's changed it may be in the frontier models too depending if i used to think it was in the frontier models and then the anthropic s1 where convinced me that anthropic was 5x as efficient as everybody else and now i'm not sure
Speaker 1and in australia i i think we when it comes to compute we mostly build the buildings and the power and then foreign hyperscalers would own the chips would that mean that we are more protected in a bust we're more
Speaker 2protected against change of chip technology we're not more protected but not as protected as you'd think if the new chips use a fifth as much power we won't be able to sell we'll build all this excess solar capacity and we won't be able to sell it right i guess we'll find some aluminium smelters in
Speaker 1light of everything we've spoken about today if you were advising the australian government on its data center push what piece of advice or pieces of advice
Speaker 2would you give it firstly i'm just going to make an observation apparently anthropic has made this to the australian government as well which is it doesn't matter if the data is sent to the australian government it doesn't matter if the data center is in suburbia or not right the idea of a data center put in central melbourne was crazy and once you started putting 70 foot high power transmission poles down suburban streets in victoria you could expect and deserve a not in my backyard response now when i use claude i'm often giving it a task that takes five minutes or i'm giving it an agent task that runs in the middle of the night anyway right i don't care when it runs and it doesn't matter to me whether the data center is in timbuktu or or suburban sydney because the difference is 0.15 milliseconds on a five minute task so put it in regional australia right now you should have no restrictions on putting it in regional australia though you probably do want to force them to generate most their own power and the reason you want to force them to generate most their own power is the politics which is that if they bought it off the grid they'd just put it in the middle of the night and they'd put it in the middle of the night put a push electric prices up and energy prices are dare i say it you know one of the hot button political issues for good reason and get an anti-data center reaction now there is no nimby argument against putting it in regional queensland there are country towns in australia that actually want to have a prison because it provides jobs and a data no matter how i counted a data center is a better neighbor than a prison right um also in large amounts of regional australia we have big advantages large amounts of sunlight relatively large amounts of land long transmission lines that travel across states that you can hook yourself to and no tariffs on the chinese batteries and solar solar is unambiguously and for the reasons we explain the cheapest form of electricity in the world batteries are going down the same path if you look at cattle which cattle is the world's biggest battery maker it's a chinese company they have finally cracked the issue of sodium batteries and they're promising batteries assembled in shipping containers ready to just plug into a grid for 75 a kilowatt hour next year right you connect those batteries to solar panels there's no question australia just has the beautiful combination very good fiber optic connections to the world very stable governments lots of land populations that would actually take this happily right most regional australia would take this very happily right um and no chinese tariffs right no donald trump you around all right so my my advice for the australian government would be to lean into that and to get the political buy-in in the places that want it and the places that wanted are going to be places with land that would otherwise want a prison all right clearly a better neighbor the second and also don't be ashamed of you know being restrictive in urban areas yeah the 50 foot or 75 foot high transmission lines going down suburban streets in melbourne really were ugly there is a genuine case for not in my backyard well
Speaker 1i think there was an afr article on friday which said that the government is uh preparing to announce a just a moratorium on data centers in urban areas fine it's
Speaker 2absolutely fine it's unnecessary for ai any way um that's so my first piece of advice is lean into it my second piece of advice is don't put your own money in it right the last thing i want to do is see is have the australian government do what the chinese government did with solar panels where it feeds money and feeds money and feeds money and for the benefit of the world that's my money i'm a taxpayer right so it's this combination of lean into it but don't fund it because of the bubble possibility because it's high capital cost low marginal cost subject to competition it's exactly the solar
Speaker 1panel and and even if there's no bubble picking winners is impossible no this is almost guaranteed
Speaker 2losers okay right but bubbles are such a loaded word yeah yeah yeah yeah a solar panel is a bubble no we're using them all they've changed the world they've made the world better yeah did everyone lose money yes right right did did all the manufacturers lose money yes is it a bubble well we're using them right there's a change it's a it's not it's not a
Speaker 1bubble it's a bad business like airlines it's a really bad business airlines aren't a bubble yeah but you don't want to you don't want to run in airlines were
Speaker 2exactly that they were high capital cost you had to build these really expensive planes once you were flying them an empty seat has a marginal cost of four fifths to three eighths of sweet buck all yeah in the old days they were subject to competition and there was three three to five percent technical improvement every year because the new planes were always better than the old ones yeah right airlines had this airlines had a pattern where every cycle they would make a fortune and then they would blow up losses larger than all the cumulative profits that they'd earned since opal and wilbur right and then that cycle they would wipe all that out again and make a fortune and then they'd blow it up again yeah right but in no sense has air travel been a bubble it's just been a bad business yeah right and this is roughly what i'm going to say about data centers it's a horrible business yeah
Speaker 1can i just summarize tell me if i'm under understanding this correctly there may or may not be a bubble or bubbles in the ai sector it's hard to know what will happen in large part because it's hard to forecast what's going to happen with the underlying technology but we can say that elements of the sector appear to be what we would call bad business which is to say high capital costs low marginal costs and competition
Speaker 2i mean it's a vastly bad business and the seven trillion dollars is clearly wrong right because nobody's going to pay for it i do i have no hesitation in saying that's too big the bubble word i like to apply to things that are complete nonsense yeah right tulips yeah right um it's not tulips right it's more like fiber optics yep right uh it doesn't mean everybody's not going to get bashed up right
Speaker 1and the policy implication is that even if it's more like fiber optics than tulips don't
Speaker 2subsidize don't subsidize it but happily facilitate it right you know lean into it accept the fact that people are going to invest largely if somebody's going to lay a fiber optic cable across Australia and make communication between Sydney and Perth cheap, let them. Don't necessarily put your own money in it, but let them. That's my advice to the Australian government. Let them. Why not? Yeah. Now I can say why not if it's a Melbourne data center, right? But I can't say why not if it's one next to the prison in Grafton. Yeah. John, it is always
Speaker 1a treat speaking with you. Thank you. Thanks for your time. Okay. Cheers. Thank you. I hope you enjoyed this episode. If you did, you can support the show by leaving a five-star rating on Apple Podcasts or Spotify, or by subscribing on YouTube. Thanks to this episode's sponsors. You can find them in the episode description. And thanks to Bill Manos and the Manos Foundation for their generous patronage of the show. If you'd like to become a sponsor or patron, you can go to billmanos.com. You can also go to jnwpod.com slash sponsor, or email me at joe at jnwpod.com. That's J-O-E at J-N-W-P-O-D.com. Thanks for listening. Until next time. Ciao.
PodcastTranscript AI![The Most Amazing Technology, the Worst Business in the World — John Hempton [Compute Series]](https://files.flightcast.com/workspaces/spjbrubahzl8ajwdo25jblkk/01M6J2TX3NMYETDZR1S15SQSZ3/162_-_andy_matuschak-_podcast_apps__16_.jpg)