Transcripts
Erik: Joining me now is freelancer.com founder and AI expert, Matt Barrie. Matt, it's been eight months since we spoke about AI last, it seems like there's been a disturbance in the force, some kind of a blow off. I don't know if it was a top or just an intermediate step along the way, but it seems like the enthusiasm is starting to abate a little bit on AI. Is this the beginning of the end? The end of the beginning, or something different? Let's recap what's happened since we spoke last.
Matt: Well, it's certainly interesting times. I think perhaps the best way to understand where we are in the space is probably to just do a bit of a recap first on the fundamental breakthrough that's happened in AI and how that translates to the underlying economics. Because, while the breakthroughs have been astonishing and unpredictable, even for the inventors of many other systems themselves, once one understands the economics, I think your listeners will get a good feeling for what's actually going on in the space and how it may play out. Now, fundamentally, what's going on, and the big breakthrough that's happened in the last few years, is the ability for effectively, machine learning or artificial intelligence, to consume very, very large data sets, to train and do so in a way where the more data you feed it, the better the AI gets. If you look at one of the very common forms of AI that are out there, which are these large language models like ChatGPT, which I think it is probably the AI that most of your listeners will be familiar with, essentially what these LLMs are, at the very core, a next word predictor. So, you take a lot of training data, you train the model, you then give it some new input, which is what you type into the ChatGPT interface. And all it's really doing, fundamentally, is predicting the next likely bit from your question, which was effectively the answer. So it's a bit like a next word predictor. So you give it a sentence, and ChatGPT will predict the next likely word, next word after that, and so on. Now, the fundamental breakthrough that happened in this space was really what's called a Transformer, which was invented by Google, which allows neural networks or machine learning to consume large amounts of training data without getting lost, and to do so in a very highly paralyzable way. So that reduces the time to train it and partly the cost. So you take a large amount of text, say, a large amount of English text that you scrape off the internet or get from books or other places, and you train it, and in the first instance, the AI will complete the sentence in a way that the output looks okay, but you could tell it was obviously written by a computer. But then what the Transformers allowed these models to do is step up the amount of data you feed in by an order of magnitude. Step up by an order of magnitude to compute and the parameters in the model. And then all of a sudden, the AI starts to get really good at completing that sentence. You know, the English is perfect, and so forth and it can carry a conversation. It becomes a little bit mesmerizing. But you step up again and again and again, and all of a sudden, that's where you get this sort of voodoo magic coming out the end of the model, where it can suddenly speak to you in Persian. It doesn't know math, it doesn't know math, it doesn't know math, suddenly, it can do University calculus. It can pass the bar exam. It can write the next Harry Potter book. It can do pattern recognition. It can bounce a pencil on a robot arm. It can control the HVAC system in the building. Because it's consumed so much data into this model that these abilities start emerging now. These abilities are emerging in a way where the inventors, they didn't predict them, they don't know how it's really happening. It's just kind of coming out with a certain amount of magnitude. And I think this is the important thing to remember, in terms of the fundamental economics, is that, the breakthroughs that we're seeing as we go from ChatGPT to GPT 4.0. And, we're waiting for GPT 5.0 to come out sometime, it's been at least, I think, since was it March 2023? We've been waiting for GPT 5.0 to come around. These models need to consume ever large amounts of data. They need to consume and do so with ever increasing orders of magnitude in terms of compute that has real constraints in terms of access to chips, access to data centers, access to energy, and fundamentally, you also need access to the actual raw data itself. I mean GPT 2.0, which was really only interesting to computer scientists, and came out many, many years ago that had about 1.4, 1.5 billion parameters in the model, and scraped out about 8 million web pages. And to give people probably a bit of an idea, there's probably about a billion websites on the internet, and about 200 million of them are basically active. GPT 3.0, when that came along, stepped it up by an order of magnitude. So, you went from 1.5 billion parameters, you can think of a parameters like a synapse in a brain, and it went from 1.5 billion to about 175 billion parameters. And the data it scraped down was about 400 billion tokens, which is about 3% of Wikipedia, about 8% of books, 22% of the web, and 60% of what we call a common crawl data set, which is what all these models are using the fundamentals to train on.
When GPT 4.0 came along, which is that big leap, where all of a sudden it could pass any exam that any human could take in the top decile, or sometimes in the top percent, it stepped up from 175 billion parameters, or synapses, to 1.8 trillion parameters, and it went from 400 billion tokens to 13 trillion tokens. And pretty much consumed very, very large percentages of the web. It added in Twitter, it's believed to add in Reddit, YouTube, etc., and so forth. But it consumed a huge, huge amount of data. Now we've been sitting around waiting for some time before GPT 5.0 comes along. And you think about, well, where does it go from here? Where does it get the data? Where does it get the compute? Where do the chips come from? And where does the money come from? I think the last estimate of the training model, the last training instance for GPT 4.0 was something like $80 million, just to do a training run. And I think Gemini is estimated to be something around $200 billion. So yeah, as we kind of step up going through the scale, you can see here, we're starting to reach some fundamental constraints in terms of economics and reality, in terms of where's the data coming from. If you scrape down the entire internet, I mean, there are some really good contemporary data sets, for example, in your phone, and there are many other data sets that we will consume. They're trying to get into video, which is obviously, there's a lot of data there we haven't really got into, in a very, very big way. But where's the money coming from? We've seen Nvidia's stock price go tilt, and I think they're punching about $300 billion a quarter of revenue, but 46% of that revenue that's coming in is coming from four customers, because you probably guessed the likely suspects of who those four are, because there's not many companies out there that can afford to buy the chips, set up the data centers and run these training runs. And so that's kind of why I think we're in this little bit of a lull right now, is because we're starting to see that we've kind of caught up to the available easy data, the available easy compute within the constraints of what companies have to spend on training runs. And now we're starting to jump up that next order of magnitude. And that's why, despite OpenAI promising all these things, I mean, what have they talked about since GPT 4.0 has come out? We've had talked about this advanced voice mode search GPT, which is their attempt to go after perplexity in providing a better version of Google. There's all this cryptic meme activity on Twitter, which I don't know if it's Sam Altman kind of just running a soft puppet account, or what have you, talking about strawberry, whatever that may end up being, Orion and obviously Sora, the video modality of the GPT model, where some pretty impressive videos were shown.
But then there's been nothing for months as OpenAI is just trying to figure out how to make it all work. And it's also very clear, I think, to anyone, the layman even out there, that there's a complete lack of sustainable competitive advantage, open source is catching up very, very quickly. Facebook is really leading the way there by open sourcing a lot of the Llama tooling that they're developing, in order to neuter the competition and make it really a war of attrition in terms of resources. And there's no business model right at the moment charging a few cents for an API call. The end reality is, there's no actual business model here with these foundational models. Now, there's going to be some incredible applications, which we'll talk about later with the AI and the incredibly lucrative opportunities. But in terms of these foundational models that are taking $100 billion to do a training run, and those numbers going up by laws of magnitude, you've got to think about it, if your cost of goods is $100 million to do a trading run, you've got to generate probably at least $500 million of revenue in order for that to be an economically viable activity as a business. So, we're kind of in this lull at the moment. There's certainly been some interesting things happening at Open Source. There's been some interesting bits and pieces coming out in the space around the edges, certainly in the image space, and the ability to generate high fidelity images of any particular type, and the reverse, in terms of being able to analyze your images and then kind of extract what's going on in the scene, some pretty amazing applications that are going to be quite possible. It's very clear that the text modality is pretty much solved, and now they're trying to chip away on video. So, some pretty amazing things have come out. But fundamentally, we're reaching these limits in terms of just the world. Compute, chips, access to data, and ultimately the underlying physics of all of that, which will get into energy and so forth, which I know we'll talk about later in the episode.
Erik: Joining me now is Bianco Research founder, Jim Bianco. Jim, it's great to get you back on the show. Last couple times that we had you on, you made the prediction that if the Fed did what they have now done, which is to wait until September before any rate cutting, they would risk the appearance of having become political. I'm going to go out on a limb here and say, I think that they've crossed that Rubicon. The Fed looks political. What are the implications of that? Does that really change anything?
Jim: Yeah, the Fed has definitely become political. Let me back up and remind everybody what I said in our last conversation in May, whatever the Fed policy was going into Memorial Day, were they tightening? Were they holding or were they cutting? Usually, will make it until Election Day. So to be clear, because people kind of misunderstand what I'm saying, if they were tightening going into Memorial Day, they could continue to tighten through the summer all the way to election day. If they were cutting, they continue to cut. They were holding because their last move was in July of ‘23 with their last rate hike. So from July 23 to Memorial Day this year, they were holding. That would suggest they would hold through Election Day. Well, as we talk a week before the Fed meeting with over 100% chance that the Fed will cut rates at their September 18 meeting, over 100% means we're debating whether or not it's going to be 25 or 50 basis point cut, not if there's going to be a cut. It looks like the Fed is going to break from its historical tradition, and it's going to cut rates in September of an election year. In other words, change policy in September of an election year. And that is unusual. It does look like it's going to be very political. The interesting thing about it is the Fed could cut rates, and both candidates could wind up, winding up, you know, criticizing the Fed for the same reason that the Fed could cut rates, and the Trump administration could put out something saying, see, the economy is falling apart. You know, the Biden-Harris administration has done a terrible job, even Jay Powell acknowledges it. And the Biden-Harris administration could say, why are you doing this before the election? To give them a talking point as well. So they're looking to be very political. So yes, they've decided to cross that Rubicon.
Now, I've talked to some Fed officials about this exact topic, and their comment to me was, no matter what we do in September, we're going to be criticized. If we don't cut, we're going to be criticized. If we cut, we're going to be criticized. We might as well ignore the politics and do what we think is going to happen. And that seems to be what they're doing. They're believing the data that the economy is cooling and that a cut is warranted at this point. Final thing about this topic, I also said back in May to the line that the Fed is political, but not partisan. So, what I meant by that back in May, and I think it applies here, is, no, they don't sit around the FOMC table saying, okay, let's cut to the chase. We really want Harris to be president, because they tend to be more Democrat than Republican at the Fed. So what can we do to get Harris elected? They do not, do not do that. So I don't think they're partisan, and they think of it in those terms, but they are political in that they are worried about their reputation, they are worried about the ramifications of their policies in Washington, and they care a great deal about that. But given what they've told me, and I think it's right, if they don't move, they're criticized. If they move, their criticized, whatever, if September 18 comes up, they're going to be criticized whether they do something or don't do something. So they've kind of said that all cancels each other out. We've just decided that the data is cooling enough that it warrants a rate cut, and that seems to be why they're doing it.
Erik: Joining me now is Mike Alkin, CIO and fund manager for Sachem Cove Partners. Mike has prepared a slide deck to accompany today's interview. Registered users will find the download link in your Research Roundup email. If you don't have a Research Roundup email, just go to our homepage. macrovoices.com, look for the red button above Mike's picture that says, looking for the downloads.
Mike, I can't believe it's been four years since I've talked to you. I want to just let our listeners know that because of my travel plans to the World Nuclear Association conference in London, we recorded this interview fully eight days before you'll hear it, way back on August 28th. So please excuse us if we don't have the latest market news. Mike, I want to start with the structure of this market, because I think, frankly, a lot of people other than yourself and a couple of other real pros don't really understand it. So let's start with the simplest question. You know, if I'm running a trucking company, I don't have to buy crude oil and hire somebody, contract with a refiner to refine it. For me, I just buy diesel fuel, if that's what I need to run my trucks. Why do utilities have to buy unrefined U308 yellow cake uranium and then contract with a conversion and enrichment? Why don't they just sell nuclear reactor fuel in a finished product market?
Mike: That's a great question. Well, first, thanks for having me again. It's a pleasure to catch up again. So, historically, the fuel buyer, and let's understand what a fuel buyer is, right there. Most fuel buyers are nuclear engineers. And if you think about this, the market is defined by one where about 80%, 85% of the pounds that are purchased in a year go through a long term contract market from three years forward from the date signed. And it could last 5 years,7 years, 10 years. That's what a term market is. That's the vast majority of the pounds sold. And on occasion, the spot market comes into play, which is 15%, 20% of the pounds, depending on the year, sometimes a little bit more. Post Fukushima, there happen to be some pounds backed up. That one went away at the spot market. But if you're running a nuclear reactor, you can't switch to natural gas or coal, it is what it is. So, security of supply is paramount.
So the folks running these are very smart people, but they're infrequently discovering price. They're not in there, they're not coal traders or natural gas traders or oil traders. They're in there very infrequently. And what they do is they want to control the process, right? If you think about the nuclear fuel cycle, if you're buying coal or gas, you're not talking about the coal cycle, the gas cycle. You buy it, and pretty soon it's at your plant gate. Here, you're talking, it could be upwards of two years, right? U308 is mined, it gets converted to UF6. UF6 then goes to the enrichment plant. That gets enriched, then it goes off to fabrication. And you're saying, like, why don't they just buy EUP? Sometimes they do just buy EUP. They can go into the market and buy enriched uranium product, but most of the time, they want to control the process. They want to be in charge of what shows up and where, and they want to know they have that capacity, so they contract at each step along the way. And to know the fuel cycle, there's a product in the fuel cycle that's U308, that's the uranium that comes out of the ground, and then from that point forward, their services. So, conversion is a surface converting U308 to, ultimately, U308 becomes UF6, and then UF6 becomes enriched, and then it goes to fabrication. Those are services that take the natural uranium and convert it in.
And so, the utilities want to control each step of that plan, sometimes they might, they'll do a little horse trading. They might take in some enriched uranium product, some EUP, and in that, there might be some conversion, there may be some enrichment, there's other things that they could strip out and they could sell to others. And you know, that's getting deeper into a route. I don't want to go down a rabbit hole, but they want to control all portions of that. And so, like I said, their number one job is security of supply, because there is no substitute for what they do. So they want to carry it. And prices of these will move at different times, because at each stage of the fuel cycle, Erik, there are different market shares, and there are different players. For instance, in enrichment portion of the fuel cycle, which is where uranium, U308, by itself is nothing. UF6 needs to be enriched up to 3% to 5% levels to be able to be fissionable and create a reaction. When you're looking about those, enrichment is controlled. Up 40% of the market is Russia, then the West has Urenco and Orano, and so you can get some, you'll have market share there. China has some enrichment capabilities, and so it's very controlled by state owned players in enrichment. So their capacity is tighter, and now, after Fukushima, it was looser. But buyers, in a market where prices are rising and capacity is tightening, they tend to flock to it. To put some context around that, about five or six years ago, the price of enrichment, which is priced in a unit of work called a Separative Work Unit, was about $35 per SWU would be the acronym. Today, a unit of SWU is $176, right? So, it's gone up multiples. If you look at conversion, the act of converting U308 into UF6, that was back in 2016, ‘17, $4. Today in the spot market, it's $68. Why? Because a lot of capacity had to be taken out of the market during the downturn, and the commitments haven't come back in full swing for the converters to not only just bring back the capacity, but add new capacity that is needed. And so, if you're a fuel buyer, it needs to be enriched, so you focus on enrichment. With the Ukraine war, people are self-sanctioning away from Russia. There's a Russian ban in place in the US, so they got to scramble to go get enrichment capacity. And remember, just like there was a massive investment, which we could talk about, in the mine, under investment in capital expenditures in the miners for a decade post the 2011 event in Fukushima, same thing happened in enrichment and conversion. So, scramble goes to get the enrichment services. It goes to get the conversion services. There's a few more uranium miners than there are enrichers or converters. So it's a pecking order. Where do they go? Where's the biggest pinch point first? Now, even with that, the price of uranium has gone from $16, $17 at the low, to $80 right? So it still had a big move. It's just that's how, that's the nature of the business. The fuel buyers control it, and I don't see that changing.
Erik: Joining me now is Victor Shvets, global head of desk strategy for Macquarie Capital. Viktor, it's great to get you back on the show. So much we've got to talk about. Let's start with the Federal Reserve and for that matter, we can expand the conversation to central banks generally. Have they become political? Heavens no, it could never happen in the United States, and have they committed a policy error?
Viktor: Well, thank you very much for having me. It's always an interesting question, are central banks political? It's like asking, is any human institution political or not? For the answer, they’re all political in some form. It's all about the degree of independence that you get, but nobody is truly non-political. So, I don't think Federal Reserve or any other central banks are purely technocratic institutions. However, in terms of whether they made a policy error or not, that's an interesting question. Because, to my mind, the inflation, or the problem of inflation was over and done with probably sometime in earlier 2023. That's when I published that team transitory has won, and team transitory was right all along, any inflation after that, pretty much globally, was really anomalies, either statistical anomalies or methodological anomalies, but the underlying inflationary pulse arising out of COVID was pretty much over in 2023. So, the question is, by waiting all the way into ’24, whether central banks have committed a policy error by keeping interest rates above neutral rates. To my mind, they have committed a policy error, but the consequences are not as significant as they used to be.
One of the things I keep emphasizing is that we live in the world of excess surplus capital. We've got abundance of capital, not a constraint shortage of capital. Now, that's quite unique. It only occurred in the last 15, 20 years, and it's been increasing ever since, primarily because money supply has been growing so much faster than nominal GDP. And the difference between the two is this excess capital we have, it's very hard to compute, but on a global basis, I think we have at least $500 to $800 trillion of capital, which means anywhere from 5 to 10 times GDP. So, when you have excess capital, prices don't work as well because there is excess of that commodity, rather than a shortage. Now that helps if you commit a policy error, if you keep rates too high, but there is too much capital moving around, that cushions the real economy.
The second thing we have that we never had before is really instantaneous risk repricing. You know, Federal Reserve makes one announcement, the answer is automatically on a screen within a split second, so you immediately know where there is a problem. And the third thing we have, once again, we didn't have it before, really, is ability of central banks to roll out new policies on a whim incredibly fast. Think of it this way. Federal Reserve, it took them almost a year to become comfortable with QE. It took them at least a month to understand that we need an emergency repo facility back in 2019, but in 2023 it just took 72 hours in order to create a brand new facility to rescue Silicon Valley Bank and essentially end the regional banking crisis. So, if you have too much capital, which we do, in other words, there is no shortage of capital if you have instantaneous repricing of risk, and if you have policies that are designed for specific problems and can't be rolled out incredibly fast, how can you commit a policy error? And more importantly, if you have committed a policy error, you can unwind it incredibly fast, pretty much in a split second, without really damaging the underlying economy so much. Now, I'm not saying that if you decided that inflation is a real problem, and you want to keep, say, policy rates at 5.5%, when neutral rates in the US are closer to 3%- 3.5%, if you keep basis points up for a long period of time, eventually you're going to crash the economy. But why would you do that? If inflation is not really a problem and is unlikely to be a problem as you go forward?
So to answer your question, are central banks political? I think every human institution in some form is political. Number two, have they kept the interest rates, whether Federal Reserve or ECB, too high? The answer is yes, they have. Have they committed a policy error? Yes. Is it going to lead to recession or extremely negative outcomes, either in terms of bankruptcies or bad debts or consumer spending? The answer, no, we're going to end up with a slower growth. We're going to end 3%-3.5%. In the case of eurozone, it's going to be a more like 1.5%-2% lower rates. And we're going to get more liquidity. And if you think of liquidity, if you look at federal reserve balance sheet, they're down to about $3.2- $3.3 trillion in Treasury securities. If you think of reverse repo, they're down to only about $300 billion. They need to continue to unwind the idea of QT, and, in fact, go beyond that and start injecting liquidity. So, my view in the next 18 months, slow growth, but no recession, lower rates and more liquidity. So from an asset classes point of view, and asset prices point of view, what is there not to like? It's almost like a goldilock.
Erik: Joining me now is Saxo Bank’s commodity chief Ole Hansen. Ole prepared a slide deck to accompany this week's interview. You'll find the download link in your Research Roundup email. If you don't have a Research Roundup email, just go to our homepage, macrovoices.com, click the red button above Ole’s picture that says, looking for the downloads. Ole, I love this picture of the 747 on the opening page here. Let's dive into the slide deck and talk about the next page, which, of course, is the table of commodities, lot of things in the green this year. But give us the rundown.
Ole: Well, hello Erik and thank you very much for inviting me back. We passed the halfway mark. And I think the first observation is that if we look at the commodity sector as a whole, we're back to square one. The Bloomberg commodity index, which tracks 24 major commodity futures, is especially unchanged on the year, following a relatively strong rally during the later part of the first quarter and into the second. And then since then, the rally has deflated, and questions are clearly being asked right now, whether that whether rally is done, or what's going on in the market. But as we can also see, it's as per usual, some big movements are unfolding. I think those that has really attracted a lot of attention this year are also the ones that we can see on this performance, where we have precious metals out in front. From a sector perspective, we continue to see gold making new record high. Silver is struggling a little bit recently because of the weakness in the industrial metals, but that seems to be returning as well, as we speak. And then this big divergence in the agricultural space between soft commodities like coffee, cocoa and orange juice and sugar as well. And then grains on the other side, which is really struggling, another bumper crop year across the northern hemisphere, leaving bread prices hopefully relatively subdued into the coming winter. And then in the middle, we got the energy sector, where, basically we are in the Big Three scheme of things. We are trading unchanged almost in crude oil for the past year and a half and the range there is getting increasingly narrow, so obviously something is going to happen eventually.
MACRO VOICES is presented for informational and entertainment purposes only. The information presented in MACRO VOICES should NOT be construed as investment advice. Always consult a licensed investment professional before making important investment decisions. The opinions expressed on MACRO VOICES are those of the participants. MACRO VOICES, its producers, and hosts Erik Townsend and Patrick Ceresna shall NOT be liable for losses resulting from investment decisions based on information or viewpoints presented on MACRO VOICES.
