Title:   Gavin McCracken (Pt. 2): Robotics, AI, and A Commodities Supercycle
Show:    Value Hive Podcast (host Brandon Beylo, Macro Ops)
Guest:   Gavin McCracken (AI/mathematics PhD; independent investor — oil & gas, precious metals,
         junior mining; runs a concentrated, margin-financed personal book)
Date:    2026-AUG-28
URL:     https://open.spotify.com/episode/5ISZIrK6WcnggEfKPyYPsO
Length:  1:13:42 (~1 hr 13 min)
Note:    Spotify auto-generated transcript (accuracy may vary). This is an AUDIO podcast with NO
         YouTube upload — the Spotify panel DOES carry (mm:ss) cues, so they are kept here and the
         analysis page's "At" cells deep-link into the Spotify episode (open.spotify.com/...?t=<sec>).
         There is no video id, so the consolidated ticker pages carry the thesis via the summary +
         "In plain English" rather than an auto-pulled excerpt.
         CAPTURE: the whole transcript was read out of the Spotify transcript panel's DOM in one pass
         (all 1,342 cue/paragraph nodes were present in the document), so there are NO seams and
         nothing was spliced or invented. Spotify chapter headings are kept as == section == markers.
         Speaker diarization was numeric only ("Speaker 1" / "Speaker 2") and has been relabelled from
         context: Speaker 1 = Brandon Beylo (host — "we bought wheat futures... at Macro Ops", two
         young kids), Speaker 2 = Gavin McCracken (the AI PhD back from the Boston topology/algebra/
         geometry conference, holder of the Brent calls). Each cue's sentences are merged onto one
         line; every (mm:ss) cue is kept exactly where it was.
         Fillers ("um", "you know", "I mean", "like" and "right?" where contentless) and stutters
         removed; wording otherwise verbatim, profanity included.
         ASR name/term corrections applied: "Rock Resources"->ROK Resources; "EQ or Al Monte"->EQR or
         Almonty; "Rentec and James St."->Rentec and Jim Simons; "Scott Besson"->Scott Bessent;
         "Jevin's paradox"->Jevons paradox; "unitary"->Unitree; "Nurips"->NeurIPS; "Velo 3D3D
         systems"->Velo3D, 3D Systems; "nutrient"->Nutrien; "CFSQM"->CF, SQM; "millennial
         potash"->Millennial Potash; "El Nino"->El Nino; "Robert Limbaugh"->Ronald Limbaugh (the
         tungsten book is "Tungsten in Peace and War"); "P verse MP"->P versus NP; "VLSCCS"/"VLCCCS"
         ->VLCCs; "SP Rs"->SPRs; "the great piece of our time"->the great peace of our time;
         "Independence, Oregon small businesses"->independents or small businesses; "price of
         everything double S"->doubles.
         GENUINELY AMBIGUOUS garbles left as spoken and flagged here: (a) at 28:49 Beylo says "solar's
         pretty strong, it crossed over 70 today" in the middle of a SILVER exchange — most likely
         "silver's pretty strong", but it is left as spoken because McCracken's reply ("we know solar
         demand is there") is genuinely about solar; (b) "There's like Cavi energy I think in Canada"
         (43:22) is an unidentifiable fertilizer/sulfur name and is NOT resolved to a ticker; (c) "I
         think they've hit like 50 years day" (48:43) is garbled tanker-strike arithmetic; (d) "Toilet
         King" (42:31) is a social-media handle, left as spoken; (e) "Centrini's guys" (1:01:26).

================================================================

== AI's Impact on Math and Academic Concerns ==

(0:03) [Beylo] All right, Gavin, this is our Part 2.
(0:10) [McCracken] Awaited Part 2.
(0:12) [Beylo] I feel like not much has happened besides literally everything in oil and gas, precious metals, and so much to talk about. And we're recording this late, late for me, 8:00 Eastern Time. I'm just wrapping up a bowl of ice cream with some cookies.
(0:29) [Beylo] So I'm in a pretty awesome mood and get to talk to you. And I want to start this conversation on AI because you said you had a lot of thoughts on it. You've done some work in this, I believe your PhD in AI or something like that.
(0:45) [Beylo] And I think we just start there. And I'm going to let you just take the reins. What are your thoughts? What are you seeing that's made you have some different opinions?
(0:59) [McCracken] Yeah. So I'm actually just getting back from Boston. I was at a conference there, the topology, algebra and geometry and data science. So it's an AI conference, it's very math oriented. Obviously a lot of mathematicians are pretty concerned at the moment that they're going to be out of a job because, I don't know if you've heard, but multiple major conjectures have been solved now by —
(1:23) [Beylo] By AI. I've not.
(1:26) [McCracken] Yeah, so there's been some tweets and there's even jokes going around because humans used to write a paper and they'd be like, here's this conjecture, here's the proof. And there's just been tweets dropped by OpenAI employees where they're just like, here's this counterexample to some conjecture that's 200 years old or 100 years old.
(1:54) [McCracken] And it was pretty interesting to be there and see that they're concerned. Personally, I'm not too concerned. I see why they are. I think definitely mathematicians are going to have to innovate here, but there's a lot to be said there.
(2:12) [McCracken] But so this is this —
(2:14) [Beylo] Is going to be a dumb question. How do you innovate in mathematics?
(2:18) [McCracken] So I have a very different approach to doing math because I taught myself everything. And I literally taught myself everything. I accidentally rediscovered a couple of major theorems in my masters and stuff.
(2:35) [McCracken] And I thought I'd done something big and then turns out it's a well known result. But the advantage is — there's a famous mathematician that I kind of live by his words without even knowing them until someone told me. His name is Alexander Grothendieck.
(2:53) [McCracken] And he said, we don't read books, we write them. And that's something that I found in my master's and PhD. Most mathematicians are mostly just memorizing all the results and doing combinatorial search when they try and prove something.
(3:13) [McCracken] So just adding things on top of some foundation you're building and seeing where it goes. And that's never how I've really done anything. When I jump into a problem, I'm trying to really understand it and break it apart and understand even maybe what the problem is, which is something no one really asks in math or computer science.
(3:35) [McCracken] And so for a lot of these problems, basically what I think happened is you take an enormous amount of compute and a large language model can search for the answer.
(3:51) [McCracken] And to some extent it's — I don't want to call it intelligent search, but it's not as bad as random search. It's a little better. And so as far as innovating goes, or at least where I think AI will come into play —
(4:13) [McCracken] So one of the solutions it found to one of the problems was really just a counterexample. That was very — I don't want to say high dimensional, but there was tons of stuff going on. And for a human to have come up with it you would have been working for a year straight and really had to keep track of a lot of stuff in your working memory.
(4:36) [McCracken] And so I think it's going to help with things like this. But also I would say it's a waste of a human's life if they're sitting there just constructing this thing. You're not really thinking. So some people are scared.
(4:52) [McCracken] I see it as kind of an improvement.
(4:56) [Beylo] So you go to this conference — anything there change your mind about any biases or best cases you had about AI?
(5:05) [McCracken] Don't think so.

== Is This the End of the Information Era? ==

(5:06) [McCracken] I've been pretty in the same line of thinking for a while now. And some of the questions, of course, are what do you invest in to play off this, because it's already so hot.
(5:23) [McCracken] And basically my current way of thinking about it is I think that we're at the end of the information era. So I don't know if you ever played Civilization or these games — we have a tech tree for each era.
(5:40) [McCracken] There's the ancient era, which would be the Bronze Age and the Iron Age and the Renaissance and classical era. And then you get into the industrial revolution, which really is the era of analog compute. And what I mean by that is you have the steam engine.
(5:57) [McCracken] That's the primary thing driving it. You get railway, you eventually get the car. Eventually ideas lead to the airplane turbine and it kind of ends with nuclear fission, which is the ultimate turbine. You can build this power plant that can be gigawatt scale and uranium is super cheap.
(6:20) [McCracken] It's awesome on a power versus cost level. And then we enter the information era. And so that's the digital computer. By analog, analog computers are things like sundials — it's the actual universe flows energy to compute something for you, so it tends to be really efficient, really fast, and you pay a cost when you go to digital compute. You gain this super general, easy to code program.
(6:52) [McCracken] But because of that, you have to flip bits. The Turing machine, it's all these bits. And that's why you end up needing all this memory and Micron stock's going up, because they're bottlenecked on — to make things go faster, you can parallelize it and flip bits, and it's a super general thing.
(7:14) [McCracken] So you can't really apply energy to it too efficiently. And so this is where we are with basically current artificial intelligence. So what people are doing is they're scaling it in parallel sideways.
(7:31) [McCracken] Just more GPUs stacking them, and that's coming with associated commodity demands as well as demands for silicon chips themselves. And so we're at a point now where it's kind of getting crazy.
(7:50) [McCracken] You're doing trillion dollar CapEx plans. And the question is, is it a bubble? Is it not a bubble? And so I'm going to say now, I still don't think it's anywhere near close to popping. I think this is the last part of the tech tree for this era, though. I want to call it more like virtual intelligence.
(8:12) [McCracken] I don't think it's actually going to be intelligence. It just can really do well with patterns and identify new patterns and stuff, which I've definitely heard people when I talk to them philosophically use that as a definition for intelligence, pattern recognition and stuff.
(8:32) [McCracken] But I think it's probably beyond that, at least human intelligence.
(8:35) [Beylo] So what separates the human intelligence beyond pattern recognition?
(8:39) [McCracken] I don't know. One thing is absolutely that it's always getting better. People are always talking, they want artificial super intelligence, which is it can self improve. It gets better. And I'm always like, well, by that definition humans are artificial super intelligence.
(8:57) [McCracken] Every day I am better than I was yesterday. Every day I'm seeking to improve. I'm trying to learn new things. I'm putting new information together. And sometimes I think we learn things that actually aren't as simple as patterns. A lot of trading or investing —
(9:14) [McCracken] It's incredibly difficult. You end up in situations where you can be like, this is kind of something historical or something I've seen before, but it's rare — you can sit there and go, no, I can go all in. This is absolutely gonna happen.
(9:28) [Beylo] Yeah, well, the complexity of it too is — I think about this too — so many things in our life are highly complex. It's just our brains have evolved in such a really interesting way to find patterns in things and to try to move things to automate mode.
(9:46) [Beylo] Like take driving your car, driving to and from work or driving to and from some place. That's a very complex task. Learning how to drive the car, learning how to interact with everyone else on the road, staying on top of what you're seeing. But yet how many times do I find myself just automating that process and driving and just getting in the car, going and coming back.
(10:06) [Beylo] And I'm like, I barely even thought about that. And it's just our brain's way of taking something highly complex and finding patterns and then being comfortable with those patterns.
(10:18) [McCracken] Well, to some extent, yes, but also if something out of distribution happens, you wake up. If something shocking happens, your brain can really kick on and do things. People even talk about near death experiences, driving and stuff.
(10:35) [McCracken] Time slows down.
(10:38) [Beylo] So what's after the information age?

== Robotics to Kickstart a Commodity Supercycle ==

(10:41) [Beylo] You're saying this is kind of the last wave of information age.
(10:43) [McCracken] So this is the thing I'm kind of excited about right now. Definitely there's going to be a huge rise in robotics, and the reason I mentioned the analog computers — I was hoping you'd ask me this — and I mentioned that counterexample in math, it was this really higher dimensional —
(10:59) [McCracken] It's put all these things together. So there was this very famous debate in computer science between John von Neumann, who people call the goat, and a bunch of other famous computer scientists and mathematicians about whether they should pursue digital computers back in basically the Manhattan Project time.
(11:23) [McCracken] And it was because they were like, oh, but analog is so much more efficient. And it is, it's incredibly more efficient. But it's that trade off I mentioned where it's not general. And so because it's not general, your sundial only computes the time of day or whatever, where you want obviously to be able to do general things.
(11:46) [McCracken] And so what I think is going to happen is we're going to see an absolutely roaring commodity super cycle. I'm making that call now.
(11:55) [Beylo] I'm super biased so that makes me feel really good.
(11:58) [McCracken] It's going to be because people are going to use AI to help them build analog computers.
(12:03) [Beylo] So do you think — I just had this thought. Do you think as technology becomes more and more advanced — because your sundial example made me think of this — it's like now we almost have the capability to go back and make the sundial for anything that we want.
(12:26) [Beylo] So what you just mentioned, analog is a lot better, it's a lot more precise, whatever the word you used to describe it — it's like we've now solved for building that, or something specific that you want, at scale.
(12:44) [Beylo] Exactly.
(12:45) [McCracken] So to precede that, you need an era of robotics. Because you need to have general workers that help you do things that would just be too overwhelming to do as humans. By general robotics, I don't mean robots like humanoid — could be tiny little things that help you position a cable or something, because really the issue is the complexity of building these things.
(13:14) [McCracken] The most complicated analog computer we ever built was the nuclear reactor. Uranium decays and we go through a process that requires a lot of safety in order to get energy out of it to spin turbines.
(13:32) [McCracken] And so there are things — one of the bottlenecks right now with compute is matrix multiplication. And you could definitely imagine analog circuits that can do this.
(13:43) [Beylo] Why is that a bottleneck, matrix multiplication?
(13:45) [McCracken] That's how everything works. So a deep neural network is just a sequence of matrices. And you do matrix multiplication and we parallelize that by putting more GPUs in parallel because matrix multiplication is perfectly parallelizable.
(14:02) [McCracken] So if I want to do it faster, just give me more GPUs, I can do it faster.
(14:06) [Beylo] Got it.
(14:07) [McCracken] To some extent it gets bottlenecked by the number of layers, but I can perfectly parallelize each layer of compute. So a deep neural network is a sequence of layers, it's a bunch of neurons and they feed into the next one and the next one. And so basically if I can compute each layer very fast, I can do the whole thing fast because there's not that many layers.
(14:27) [Beylo] Is there — and again, I'm just gonna be firing off some dumb questions because I don't have the PhD like you do. But is there like an upper limit to the number of layers where beyond a certain amount, the incremental improvement — I guess that's kind of the whole thesis here, right?
(14:43) [Beylo] I'm asking the obvious question, but —
(14:46) [McCracken] A huge question. There was a limit until a paper came out where they invented something called a skip connection. And it's the dumbest thing ever. It's a great idea, but it's so dumb that no one really thought of it.
(14:59) [Beylo] What is it?
(15:01) [McCracken] Basically the issue was if you kept adding layers — we use calculus, we take the derivative to update all the weights connecting the layers, and those are the matrices. The weights are in the matrices, and basically the gradient would go to 0, and that's the derivative.
(15:17) [McCracken] And it's because as you add all these layers, it gets hard to tell what's doing what. And so it would go to zero and then you can't update anything. And the gradient is the gradient of the error. You show it a cat and it's supposed to say it's a cat. And if it says it's a dog, you'd be like, no, it's a cat. And you use calculus to do that.
(15:34) [McCracken] And so if the gradient goes to 0, you're not actually changing anything. So to prevent the gradient from going to zero, they just added skip connections, which instead of it just being feed forward, you add skip 5 layers, skip 5 layers, and just let the information still flow.
(15:52) [McCracken] And it solved this. So it kind of solved it. It got us to the point that we're at now, which is you take the next result, which was a paper called Attention Is All You Need that really got Transformers going. And then you just scale this. You just add as many layers as you can, use a ridiculous amount of compute and train it well and you get GPT.
(16:14) [McCracken] Really it's not much. It's a very simple idea. We honestly don't know what we're doing. It's just keep on adding stuff.
(16:26) [Beylo] That's comforting, yeah.
(16:28) [McCracken] A lot of the research is of course how do you train it faster, how do you train it and get a better solution and stuff like this. But the core idea is incredibly simple.
(16:39) [Beylo] So then going back before I went off on that tangent — you made this call that we're about to have a commodity super cycle.

== AI's Role in Discovery and Innovation ==

(16:45) [Beylo] And so I'd love to have you flesh that out a little bit more.
(16:49) [McCracken] Yeah, so the first place we're going to see it is math, and it's going to be AI helping to find these things that would just take forever. It's not an intelligent thing to construct these counterexamples. And all of the big conjectures that I've seen, they've all been disproven by counterexamples.
(17:06) [McCracken] You have to come up with this counterexample that interfaces with the problem and makes it clear that the claim is not true. So it proves it false. And they've all been things that either humans weren't really asking what the problem was and really getting to the bottom and distilling, breaking that problem up.
(17:31) [McCracken] And other than that, it would be just a search thing — find the magic way of setting things where you'd see that the claim breaks. So you have a statement you want to prove is true and always true. That's a conjecture. And so AI has really just been proving things false.
(17:48) [McCracken] I haven't seen it do anything, is it true, that I was impressed by. And that's something in itself too, because we keep seeing all these headlines where Claude Mythos was really good at breaking cryptography — people would have some kind of security system and it would find a loophole and gain root level access and do whatever it wants to their server or compute, whatever it is.
(18:19) [McCracken] And then we're seeing that too in GPT 2. It's not just Claude Mythos that's good at it. It seems like LLMs are incredible at finding security gaps. And so in the case of math, it's exactly the same thing. You have this structure which has been built by the axioms of mathematics.
(18:35) [McCracken] And then if you're trying to find a counterexample, that's literally it. You're trying to say something's true and say, where would it break? And then it finds it. So they seem to be incredibly good at finding where things break, at least for now.
(18:49) [Beylo] And I think that went back to our earlier discussion where you're hesitant to say — LLMs, AI, they can't create intelligence. And that maybe goes back to the thing where they're good at finding these counters because they're just taking everything in the world and trying to pattern it on.
(19:09) [Beylo] And if the pattern doesn't fit, then they're like, oh, this could be wrong, this could be a counter. But they can't create something. They can't create their own conjecture yet, it sounds like.
(19:18) [McCracken] Absolutely not. No. And that's where they're going to interface with mathematicians. Mathematicians are still going to be needed to say what problems are worth asking and what objects should we try and construct to help us prove other problems too.
(19:38) [McCracken] What ends up being useful. So yeah, the first place I expect this to be is math. And that's of course not going to create commodity demand, but I think immediately after that, you're gonna start seeing things like what just happened with mRNA stock.
(19:55) [McCracken] People had calls go up by 400,000%. They used AI. The human body is a very complex system. And so to try and identify how can we make an mRNA vaccine for cancer?
(20:13) [McCracken] And it helps. Helps them identify things and pick things out. And so that's — again, not too much commodity demand there, but it's the next thing that I expect, which is when we start actually using it to make robotics better and we start using it to build new things, which we haven't actually done in like 50 years now.
(20:31) [McCracken] Nuclear power plants are still the most impressive thing humans have ever built. We've known how to do it since what, the 60s? So we haven't done anything impressive in a long time now.
(20:41) [Beylo] Why do you think that is?
(20:43) [McCracken] It's the same thing I was talking about before. It might be an arrogant opinion, but I think people were really handed a lot of amazing things by World War One and World War 2 mathematicians. And part of the reason for that was it was life or death, including the Cold War.
(21:00) [McCracken] So the most influential and important algorithm in my opinion is the Fourier transform. And so we knew about this thing called the Fourier transform since like 200 years or something. I forgot exactly when. But basically what it says is you can rewrite a polynomial function in a sinusoidal basis in the complex plane.
(21:24) [McCracken] And it's extremely important. An example of its application: so we didn't know how to do it fast. We knew that it could be done. But two students of John von Neumann, or people influenced by him, they worked with him during the Cold War — we didn't have GPUs yet.
(21:42) [McCracken] We didn't have crazy digital computers. It's like a calculator that takes forever. It was very slow, these computers. So it turns out that if Russia blows up a nuclear bomb, you need to multiply polynomials to find out. So they needed to be able to compute this faster and it takes n^2.
(22:03) [McCracken] So if the polynomials are of length N — X^2 plus BX plus Y, something like that, that'd be length three. It would take 9. You have to expand, do the multiplication, if you remember it's super simple, n^2. So the length of the polynomial being N terms and you have to do squared computation.
(22:21) [McCracken] Turns out with the Fourier transform, if you can do it fast, you can do this in N log N, so that's basically linear because log is very small. So as N gets large, log gets tiny. It's a logarithm. And so you're left with linear time, which is incredibly fast. That's like saying you can multiply 2 things as fast as you can read them.
(22:39) [McCracken] So this was obviously a scientific thing that needed to be done. And it was figured out because the Cold War inspired it. And war has inspired most of humanity's great inventions, like fertilizers.
(22:55) [McCracken] That's World War 2 and World War One. The gases they were throwing at each other to kill each other, to suffocate. And so why do I think it — it's because people stopped asking what problems are.
(23:12) [McCracken] That's my opinion. They stopped really breaking apart problems and trying to understand that problem in isolation from everything they know. Instead, people approach problems by saying, here's everything I know. How do I use that to solve this problem?
(23:30) [McCracken] And so that second way is something I've personally never done. I always want to have a problem. I pretend it's the only problem I know, and I start from scratch. And that's what LLMs can't do. LLMs literally take the fact that you've trained them with the entire Internet and they're going to just throw ideas at things.
(23:48) [Beylo] So —
(23:48) [McCracken] If you're just going to parallelize something with a ton of compute, it can throw ideas at it in a somewhat intelligent manner. It can find solutions fairly quickly. It's just good search.
(24:05) [McCracken] And of course, that's also why we need so much compute for it, because it's not intelligent. It's throwing ideas at things and looping and trying again and again and again, and you have something that takes a gigawatt running all by itself for a week to solve something.

== Distinguishing AI from True Human Thinking ==

(24:19) [Beylo] And I read this tweet from Dan Koe, I think it was, and he basically said when we invented cheap calories, people became super unhealthy and fat. And his argument is thinking with LLMs and with AI has now become cheap, thereby kind of equating this idea that OK, now most people's thinking is going to end up like most people's bodies, which is unhealthy, overweight, all that stuff.
(24:49) [Beylo] So how do you prevent that?
(24:55) [McCracken] I'll take the 100% counterpoint there and I'll just say I don't think people were thinking. I don't think thinking is just what have I read before, let me try and apply it. Thinking is this other thing. You deconstruct something, you have to truly understand it, reverse engineer it, know how to build it yourself. So I think LLMs are doing a really useful thing. And I really hope that they remain accessible to everyone because I think that they get rid of this look up table kind of thing where you have to read everything because you don't anymore.
(25:33) [McCracken] Now you can just be like, hey GPT, can you give me a quick crash course on, I don't know, the Fourier transform and polynomials? And boom, it'll spit out some stuff. And then if it's too complicated, be like, can you do that again, but for a high school student? So they make information cheap.
(25:49) [McCracken] I don't think they make thinking cheap. I think thinking is different to information.
(25:53) [Beylo] So how does someone — let's say someone that doesn't have kind of the background or maybe even the embedded skills that you have to be an AI PhD, or to even learn how to take things apart to then put them back together. I've got friends that will just rebuild an engine.
(26:09) [Beylo] And in my head, I'm like, that is just unbelievable. So how can an investor or someone that wants to learn how to think better, or think more deeply about the right kind of questions — what are some things that they can do in their everyday life to build that skill?
(26:28) [Beylo] Because to me it sounds like it's just like exercise. It's just something you got to work on.
(26:33) [McCracken] 100%. I think that you just have to do it. You got to just jump in. I think one of the reasons that I perform well at anything now is I've done so many different things in my life.
(26:49) [McCracken] And I think diversity is just really important for the human brain. You have to do a whole bunch of different tasks, especially when it comes to something like the market. Because the market itself is — who even knows what it is? It's an evolutionary system. And other than that, we're all just playing with it and seeing what happens.
(27:07) [Beylo] Saying something we made up to keep us busy.
(27:09) [McCracken] Yeah, I think that is actually the hardest system humanity's ever made. So at least from a complexity — something like soccer or football, they're closed systems. There's a limited amount of stuff that can happen, but the market will keep creating new things forever because humans will be like, oh, last time this happened, this happened.
(27:30) [McCracken] And then maybe they try and front run it, but then because everyone front runs it, somehow it doesn't happen.
(27:41) [Beylo] Well, it's weird because even take the bond trade for instance. Everybody's max short the bonds, but then because everybody knows that everybody's max short the bonds, everybody wants to go long bonds, but then everyone in theory wants to be long.
(27:57) [Beylo] So then should you be short? There's levels to it. You can dig yourself into some deep holes. But OK, so then let's think about the idea of thinking as it applies to this new commodity super cycle, which we left off at robotics.

== Investing in the Coming Commodity Supercycle ==

(28:19) [Beylo] So what are the questions that you're asking now? If you have the hypothesis of, OK, we're about to enter a new commodity super cycle, maybe it'll be kick started by robotics — what questions are you asking?
(28:34) [Beylo] What questions should I be asking, trying to decide how to put money to work?
(28:41) [McCracken] So we've seen some signs. Silver's inflection — I think actually it's front running robotics.
(28:49) [Beylo] Mean solar's pretty strong. It crossed over 70 today which I was kind of surprised.
(28:53) [McCracken] Yeah, this is definitely — we know solar demand is there, but robotics demand is the thing I've been saying for a while, I think the market's missing. And every Optimus Tesla robot needs an ounce of silver. So when Elon says something like, I want to put one of these in every household on the planet, you can very quickly be like, wait a second, how many ounces is that?
(29:15) [McCracken] And then things like EV batteries need silver. And there's a ton of demand coming for silver. And that's why I'm just sitting on miners that produce it.
(29:30) [McCracken] There's other things too. What are the robots going to be made out of? Definitely aluminum. You're going to want them to be light. So it depends really. We're going to have to keep our eyes open for other things.
(29:48) [McCracken] But I think there's just going to be a huge period of innovation because like I said, information is cheap now, so people can quickly find out how to do things and then set things in motion to actually do them.
(30:02) [Beylo] That's —
(30:02) [McCracken] Just going to create building demand. How do I build this? And boom, they can quickly find out and start to do the research faster.
(30:14) [Beylo] Well then it becomes the bottleneck is the actual hardware. If people can get up until that starting point, the only inhibitor is can I get the raw material to build what I need. So do things like 3D printers become an interesting idea?
(30:31) [Beylo] I know the 3D printer space has been obliterated since probably the SPAC mania. If you look at Velo3D, 3D Systems, you've even had some bankruptcies in there. But that to me seems fascinating. If the constraint is some sort of hardware, some sort of material, 3D printers could be an interesting stop gap for that.
(30:55) [McCracken] For sure. You probably have to ask yourself why that bubble collapsed the first time. And I don't know the answer for that. I have friends at 3D print though, and it definitely seems incredibly useful. And then you could probably ask what the raw material is they need.
(31:11) [McCracken] I don't actually know. I'm guessing it's plastic or some kind of plastic.
(31:14) [Beylo] It's some kind of plastic resin, maybe some sort of thing. But you've got 3D printers — I think I saw this — that can 3D print metal?
(31:22) [McCracken] Yeah.
(31:23) [Beylo] I mean, that's insane.
(31:25) [McCracken] Yeah. So this is what I mean by raw materials are going to be the thing.
(31:33) [Beylo] So then how do you think about this, throwing another wrench into this idea of you have this tremendous demand from let's call it human creativity or human ingenuity, that the barriers have collapsed to them being able to create things. However, the wrench that I'm seeing is this bifurcation in the market between US and US allies and then China and China's allies, and there's going to be two metals markets.
(31:58) [Beylo] You're seeing this in rare earths already. There's two prices. There's the China price and then there's the ex-China price. How much is that going to play into effect too? You may want to mine these to make these things. But we need it for the war.
(32:14) [Beylo] We need it to fight China. And China is not sending this stuff. So there's very limited resource here.
(32:20) [McCracken] Yeah, no, that's exactly what I want as a mining investor. Obviously a less efficient market kind of means that you can take advantage of it. The whole efficient market hypothesis is supposed to be it's efficient, so you can't take advantage of it.
(32:39) [McCracken] But two markets is absolutely going to give us so many opportunities. Especially if we start seeing — if we can model and predict that certain commodities are going to have huge inflections in demand, it makes a lot of sense that you'd want to suddenly invest in certain mines.
(32:59) [McCracken] I've been saying for like a full year and a half now, I think resource nationalization is where we're going. I think actually probably my portfolio is not reflecting that enough right now. I should probably have more miners of certain minerals.
(33:17) [McCracken] But it seems like that's absolutely where we're going to go. And that's because that's what is becoming important again. Raw materials. Commodities.
(33:29) [Beylo] Kind of anything that matters.

== Geopolitical Risks and Disconnected Oil Prices ==

(33:30) [Beylo] I was fortunate to get into this industry, the raw material space, in 2022. And it's one of those things where you look at the demand drivers, you look at what's happening geopolitically, it's kind of hard to not get excited from an investment standpoint.
(33:49) [Beylo] And then speaking of this resource nationalism, something that you've mentioned to me multiple times is this idea of Trump doing an export ban on WTI. And it's literally made me stop in my tracks. If I look at an E&P and I'm like, oh, this looks super cheap.
(34:05) [Beylo] I'm like, oh man, it's all WTI. And then there's this little Gavin on my shoulder that's like, Trump's going to do an export ban. Don't do that. Can you walk me through that rationale? How would it play out? What would be the ramifications? And then maybe some of the best ways to play ex-WTI — obviously there's Brent.
(34:23) [Beylo] You can buy Brent calls and things like that.
(34:25) [McCracken] Yeah. So obviously for me, because everyone knows I'm using margin, I have no choice but to have basically 5% of my net worth locked in Brent calls. And they're long dated. And they're out of the money. So if Trump did do this, I would not be margin called to death, because Brent would just skyrocket and I'd probably have an mRNA moment, where these calls go up by like 400,000 percent or something.
(34:51) [McCracken] Brent just jumps to 500. This recent trade war with Canada is actually boosting my worry about this because now Trump can be like, oh fuck Canada. Let's do this WTI ban as well as products ban.
(35:08) [McCracken] And Eastern Canada gets all their stuff from America because Canada sucks as a country and doesn't know how to do anything. So we export the oil from Alberta and Saskatchewan and it's refined mostly in America. So that kind of a ban would just absolutely crater Eastern Canada.
(35:26) [McCracken] The thing is, of course, Canada can always retaliate and be like, all right, fuck you, we're not going to export from Alberta and Saskatchewan. And this is the point where we're all just shooting ourselves in the foot, because now America is also going to run out of gas. There's no winning here.
(35:41) [McCracken] It's just shooting ourselves in the foot while people like China benefit. Of course, it sucks that we're in this situation because it means that we have to hedge and hedges cost us money. I'm paying a premium for options and I have no choice because like you said, all the good E&Ps are North American.
(35:59) [McCracken] So you want to be in Canadian oil. There's some shale stuff in America that's interesting. And there's all these risks with South American. There's like, fuck Africa. There's no way I'm touching it.
(36:16) [McCracken] It's just some military's gonna come seize it and there you go, 100% loss. Some militia. And also, you could be like, oh, what about the North Sea, off Britain? But 80% windfall taxes and communist governments — all pass.
(36:35) [McCracken] So really you're backed into a corner right now where the only good option is North America, and Canada is a really good — I don't know if you saw my recent Substack, but I wrote about how I just learned that Saskatchewan's giving a royalty holiday on the wells.
(36:54) [Beylo] So the first —
(36:55) [McCracken] The 1st 38,000 barrels of production from southeast Saskatchewan, which is where ROK Resources, which you were actually the one to tell me about first —
(37:04) [Beylo] Sometimes I got some new stuff.
(37:07) [McCracken] That's where they're drilling. And so their first 38,000 barrels from each well are only at a 2.5% royalty and that's a huge deal because it's usually like 25%. So it basically guarantees that these wells will pay off their own cost as long as they're not a dud.
(37:25) [McCracken] As long as it doesn't come online at like 1 barrel of production a day. So the wells pay themselves off. So obviously that's great. But then of course you're exposed to the fact that Trump could really fuck around and Canada could really fuck around and ban oil exports to America too.
(37:43) [McCracken] So you just have to be like, all right, well —
(37:48) [Beylo] If —
(37:48) [McCracken] This happens. There's no way Brent stays below 500. Honestly, I think Brent would go to 1000. The world would just be fucked. If Canada and America both started just not exporting — it's too much hydrocarbons to suddenly disappear, especially with Hormuz still fucked up.
(38:06) [McCracken] And of course Iran would seize that instantly. They would probably immediately be like, not a single ship is leaving, because that's just — you can squeeze America.
(38:17) [Beylo] So then with that logic and stuff, why do you think — in my head it's like, OK, why isn't Brent higher to reflect some of this potential reality? And is it just because people don't want to step in front of that risk? They're just still concerned?
(38:33) [McCracken] So I think the spread's pretty big. I just checked it, 6 bucks. So it's actually pretty small. I thought it was more like 10. Well, Brent definitely is at a higher than normal premium. Usually it's like $3.00. So it's double the premium, but I think it's because everyone's hedging the same way as me, which is out of the money options.
(38:49) [McCracken] So they don't matter until they matter. And then the gamma blows someone's head off. Someone's getting fired for selling those options.
(39:01) [Beylo] So you said maybe you don't own enough miners as well. How are you thinking about your portfolio now and how you want it to look over the next 18 to 24 months?
(39:12) [McCracken] Yeah, tungsten I think is my biggest mistake. Like EQR. I should have had more.
(39:18) [Beylo] Dude, you're telling me man, I'm so pissed I didn't buy EQR or Almonty. That chart dude, just like — oh I hate it.
(39:27) [McCracken] Yeah, no, in hindsight too, it's just dumb to have not bought it because Russia, Ukraine was constantly blowing shit up made of tungsten. And now we got, in the Middle East, they're blowing shit up made of tungsten. Just vaporizing it. So the demand is absolutely there.
(39:44) [McCracken] We're getting pent up demand. We're hearing about shortages with missiles and things. And Trump saying he wants to move to a military economy and really ramp up the production of all this stuff. I feel like tungsten is still great here.
(40:03) [Beylo] There was a book — oh shoot, what's it called? Tungsten. I just read it. It was tungsten during the war time. I think it was called Tungsten during World War 2.
(40:21) [Beylo] But it was a great book. Oh, it's called Tungsten in Peace and War. It's a really good book if you want to learn the tungsten market, by Ronald Limbaugh. Basically if Trump wants to do a military economy, they're just going to run this playbook back.
(40:38) [Beylo] And it's interesting because it weaves in Bernard Baruch during that time who was his own kind of wild mining entrepreneurial investor and things like that. But the missiles one is interesting because it's hard to actually get.
(40:57) [Beylo] I was talking to an investor friend who does a lot of defence investing. It's actually really hard to get direct pure play missile exposure. Everything has all these ancillary things. But the missile rearmament, it's actually kind of hard to do a pure play on that besides tungsten, I guess.
(41:18) [McCracken] This is why I like betting on raw materials. You can kind of guesstimate if demand is going to spike for them or if there's going to be a supply shock, and usually it matters. Somehow Hormuz still doesn't matter.
(41:33) [McCracken] Last time we talked, I was saying everything is just a Hormuz trade. And next time we talk, I'll let you know how it's gone. We're still here. The update is: it lasts 48 hours. Two more tankers, engine rooms destroyed, and we're still waiting to see if oil cares.
(41:48) [Beylo] Doesn't matter. Anytime I see your posts on my timeline, I always tell myself, dude, I just need to short oil when Gavin starts posting screenshots of his P&L and I just need to start buying oil when he starts tweeting in all caps how upset he is.
(42:05) [Beylo] It's like I should just find this spread trade here.
(42:10) [McCracken] Yeah, I made so —
(42:11) [Beylo] Much money.
(42:12) [McCracken] Maybe not the screenshots of P&L because there was a couple times where it would crash the next day, but there's just so much volatility. I was doing that because I was very confident I was about to break out and I did. I posted recent all time highs, just huge breakout, sudden like 50% on my net worth.
(42:30) [Beylo] You had those CLMT calls?
(42:31) [McCracken] Yeah, CLMT was a driver. I got Toilet King to thank for that one even though I didn't listen to him. From like $12.00 all the way to 27 or something.
(42:46) [Beylo] Are you looking at any of the chemicals companies at all, just kind of looking across —
(42:51) [McCracken] The chems are super hard because their input costs are distorted. It's so hard because things like naphtha are exploding because it all comes out of Hormuz. It's the same reason I avoided fertilizer. A bunch of people made mistakes where they bought fertilizer companies.
(43:07) [McCracken] It crashed because they're like, well, there's a sulfur shortage. But unless the fertilizer company had sulfur on hand, that's a problem. That's gonna hurt their inputs. So there's probably gonna be a fertilizer crisis and it's hard to make money off it.
(43:22) [McCracken] There's like Cavi energy I think in Canada, and there's some more energy, and that's like it.
(43:30) [Beylo] And then there's I guess the potash plays. There's Millennial Potash, which is the big kind of small cap one that's actually done super well. And then there's Nutrien which we've looked at, CF, SQM, but that's also more — actually, I think it's a great idea to jump into El Nino.
(43:48) [Beylo] Have you done any work on the El Nino?
(43:51) [McCracken] I've done a tiny bit. I think if tankers have to start going around Latin America, because Panama Canal has problems — obviously tanker rates will go up, but the guess is oil should be 200, but oil's not going to go to 200 for magic reasons.
(44:12) [McCracken] So it'd be insane to have tankers going around Africa and Latin America and Hormuz to be randomly disrupted and Ukrainians sinking Russian tankers every week, to the extent that Kazakhstan is once a month shutting in production and closing their CPC pipeline terminal.
(44:35) [McCracken] I think actually Ukraine blew up one of the loading docks.
(44:41) [Beylo] Dude, the stuff with Ukraine —
(44:42) [McCracken] So it's insane what's heating up.
(44:46) [Beylo] I was about to say Russia, Ukraine is heating up and I feel like nobody's paying attention. It reminds me of 2022 when it first started.
(44:55) [McCracken] So I'm visiting my parents at the moment and I asked them, I was like, so you guys are aware that Ukraine is hitting Russian tankers? No. It's not mainstream media. And I was like, what about Iran? You know that they're hitting like 1 tanker a day almost. They were like, no, haven't heard this. So it's very obvious there's a media blackout to manipulate the oil price.
(45:16) [McCracken] Who knows how this ends. So one thing I want to say, kind of my theory: China — really the Chinese like to just make money, to stabilize things. The first time Trump won the election, they had no plan for it.
(45:32) [McCracken] They thought Hillary would win. And I have a feeling they're gonna try and make him lose midterms. We'll see, but I don't —
(45:42) [Beylo] Even think China needs to do anything. I think they just need to stand out and watch Trump lose it himself.
(45:48) [McCracken] It could be a bit, but they can guarantee it. So because right now the only reason oil is not 200 is China. They cut their imports by 5 million barrels a day. And through rerouting and all the other shit, SPR releases, we were able to make up basically for the gap.
(46:03) [McCracken] And we're just drawing down at 1 or 2 million barrels a day right now. But of course SPRs are being just unloaded all out. And that includes China. And so China was the single largest SPR player, like 5 million barrels a day.
(46:19) [McCracken] And if they were to unwind that, they could just be waiting for October 1st and suddenly they pick up the phone and buy every tanker on Earth. Just bring the oil to China, and oil spikes to 200 and of course gas and diesel follow it immediately.
(46:36) [McCracken] And Trump loses midterms. Trump is actually in a position now where China controls the oil price 100%. And China's allied with Iran and Russia, and maybe this is in all of their interests.
(46:52) [McCracken] To get Trump impeached, because if he totally gets wrecked in midterms, he's gone. And there's nothing he can do about that because he's just made the wrong choices.
(47:02) [Beylo] Well, it's not even just him being gone. It's the fact that his whole lineage. The people that stake their reputations on this second term and everything that they're doing — your chances of getting into the White House just collapse.
(47:20) [Beylo] And then you've got —
(47:22) [McCracken] You have to start —
(47:22) [Beylo] Thinking like, what would the world look like under an AOC presidency or a new presidency?
(47:27) [McCracken] Like —
(47:28) [Beylo] These are the questions you have to ask because if it gets bad enough, the populace isn't going to want anything to do with anyone associated with Trump heading into the midterms, if that's how things end. Because the way I see it is, I just think we're sleepwalking into a 2022 with Russia, Ukraine, but now we have the Hormuz situation and now you have an El Nino.
(47:55) [Beylo] And you saw one of the craziest calls that we've had at Macro Ops: we bought wheat futures. We loaded up on wheat futures right before Russia started bombing. It was total luck. The timing was total luck. But I feel like that's about to happen again.
(48:13) [Beylo] If this escalates — I know that wheat and ags and stuff have started to move. I don't think they've priced in the potential for a dramatic re-escalation at all across the agriculture space.
(48:27) [McCracken] Yeah. We're seeing Ukraine hit Russian cargo and then we're seeing Russia hitting Ukrainian cargo. So basically all of the price action we saw in 2022. Brent went to 138.
(48:43) [McCracken] I think that was front running what's actually happening today. That was people assuming Ukraine would be hitting Russian tankers. Well, that didn't happen until starting a couple months ago and it's been insane. I think they've hit like 50 years day, and that's confirmed, we know it's 50.
(49:03) [McCracken] Some of them are VLCCs, but they have hit VLCCs. And they hit primarily product carrying tankers. So I don't know — all of this, this is the most disconnected from reality market definitely in my life, but I feel like it might be ever. It feels like the price of literally everything is wrong.
(49:30) [Beylo] That's what makes markets so interesting, because at the end of the day, there's an element of price is truth and you got to trade the market that's in front of you, not the one that you wish it was. But then it's very hard to see things literally blowing up and then to see no price response or even a negative price response.
(49:53) [Beylo] It's just been a wild time. I think Calvin likes to call it the great peace of our time.

== Market Dynamics, Bear Markets, and Gold ==

(50:03) [McCracken] It's unreal. Especially when you think about the fact it is kind of World War — Iran and Ukraine, just right there on the map.
(50:12) [Beylo] Yeah. It's like when you were in middle school and you liked someone and you guys wanted to be boyfriend and girlfriend, but you never wanted to say it. That's kind of how I feel about this global conflict. Everyone kind of feels like it's World War Three, but no one's going to come out and say it yet.
(50:27) [Beylo] No one wants to be the first one to say it.
(50:29) [McCracken] Yep. I'm hoping that doesn't happen, of course. But we got to wait and see.
(50:37) [Beylo] So how do you think — you went to this PhD conference. In my head, these are two totally different lives that you live: a PhD in AI, a total degenerate in junior mining and natural resource stocks.
(51:00) [Beylo] Where's the thread that connects them? And how does the stuff you're doing in AI make you a better investor?
(51:08) [McCracken] I don't think AI actually interfaces too much with investing. Maybe computer science. Actually, definitely computer science does. And I'd say it's really my thinking about evolutionary systems. I had two years where I disappeared and I worked on P versus NP and I realized that's the class of evolutionary problems.
(51:27) [McCracken] And that's what the stock market is. So I think that really helps me, especially because I've had this amazing timing with hopping out of commodities right as they're about to crash — 2023 December, I pieced out of oil in full, and 2024 bear market year.
(51:47) [McCracken] And that's this concept of the self avoiding random walk. So from computer science, I'd say this is helping me. And definitely I use this thinking to think about AI as well — continual learning, reinforcement learning, these things where we want to keep the agent going.
(52:02) [McCracken] But I don't know, like I was saying, we don't know much when it comes to AI and we definitely don't know how to integrate this kind of logic or thinking computationally.
(52:18) [McCracken] And so the people who actually might know more would be Rentec and Jim Simons, because the quants, they've been trying to do this. You have the system, which is the stock market, it's always evolving. How do you come up with algorithms where you're gonna juice some alpha out of the market on whatever the time spans are you care about?
(52:39) [McCracken] And even them, they end up competing with each other. Because they start eventually all having the same strategy and then they get a little mini bubble that pops, they all lose 5-10%. And that's because the market evolves. If everyone is doing the same thing, that means it's about to — like evolutionary systems.
(52:59) [McCracken] Don't let that happen. You get an extinction event afterwards. If everyone has the same, there's no diversity. And probably part of the reason this market keeps going up is monetary debasement. At this point everything is kind of going up.
(53:20) [Beylo] It's amazing how long it's been since a true bear market.
(53:25) [McCracken] Yeah.
(53:25) [Beylo] Like a sustained bear market. And I don't know if I count those V bottoms, because the technical definition of bear market is a 20% decline or something like that. But those are all V bottoms. When I think of a true bear market, I think of 5 to 10 years, no returns, and 2 to three years of negative.
(53:50) [Beylo] It's not just, oh, I'm down 20, but then Trump tweets and I'm back up to all time highs. It's like, no, you're down 20. You retrace back up, now you're only down 10. Just kidding. There's a dead cat bounce, now it's down 30. We have not had one of those in what feels like a generation or two.
(54:13) [McCracken] Yeah, no, I totally agree. It's like 2008's the last time. And that recovered pretty quick, honestly. Didn't take that long.
(54:27) [Beylo] But 2008 — you say recovered quick, but what I'm getting at is that recovery would not seem quick according to today's standards because we're so used to the V.
(54:41) [McCracken] Yes, yes.
(54:42) [Beylo] That 08 would seem like an eternity because it was 08 to like 2010 is when things started recovering and turning, not even getting back to break even. It's almost like the market has a TikTok brain where it can only do things in rapid succession.
(54:59) [Beylo] It can't take its time.
(55:03) [McCracken] Yeah, I totally agree. It's interesting times for sure. I think people need to be broken, because as I keep saying, evolutionary systems, not everyone's allowed to win, but everyone is winning right now. Literally everyone.
(55:22) [McCracken] You have so many people who are up more than 10X trailing five years. And so much of it was absolutely retarded, shit coins and crypto. It just doesn't make sense. You could just have bought basically anything and you're going to have made money.
(55:42) [McCracken] And that's definitely part of what I think is this debt bubble we're in. If everyone has cheap access to liquidity, then everything keeps going up. But people keep getting more and more indebted.
(55:57) [McCracken] And eventually maybe there's some kind of liquidity squeeze. That's why I think that $200 oil would be so dangerous right now. This might be why they're so panicked about it and why there's literally a media blackout on tankers being hit by missiles by anyone, whether it's the Russian allied Ukraine or the anti-Russian Iran. It doesn't matter who's doing it.
(56:19) [McCracken] It's like we don't talk about this. Because the thing that would cause a liquidity crunch is oil. If oil goes to 200, there's a massive inflation that hits absolutely everything, because oil is the modern economy. I say all the time oil is the blood of the modern economy.
(56:36) [McCracken] And that's exactly — this would be the liquidity crunch. And that's kind of what I'm waiting for. I cut gold by about 50%, like gold miners. And it's because I don't think gold would do well in a liquidity crunch. Historically, it hasn't.
(56:51) [Beylo] Well, you saw Turkey — when the war started, these countries are selling all their gold. They can afford the war.
(56:59) [McCracken] But not even that. Imagine the price of everything doubles. People are gonna pawn their jewelry. Especially with the current gold price, people are going to sell it so fast. So again, it's this question of can they just keep flooding the market with cheap debt and people just keep buying things. And there is absolutely an end to this.
(57:23) [McCracken] It's just when is it. No one knows. So for the time being, we got to be in the market.

== Navigating Oil Volatility and Portfolio Strategy ==

(57:27) [McCracken] Play the casino the best you can, even if you know it's going to blow up.
(57:30) [Beylo] Yeah. And that's like the gold thing. I think gold's kind of overextended here, but again, the whole debasement narrative — you've got the dollar that's flashing kind of some sell signals according to some trend stuff that we've got at Macro Ops.
(57:45) [Beylo] And I want to be de-risking gold, but then everything is telling me gold's actually pretty good. Fundamentals are pretty good here. It's just tough.
(57:59) [McCracken] Yeah. The fundamentals are there. And that's part of my argument for using margin at the moment: if everyone's devaluing their currency, then that is a nice thing to have. And that has worked very well for me in recent years.
(58:16) [McCracken] But I also hate it because I see everything and I'm like, oh man, what is going to be the straw that breaks the camel's back?
(58:24) [Beylo] Dude, I couldn't imagine being you in that portfolio every morning, just with the oil exposure. I'd be a nervous wreck.
(58:33) [McCracken] Yeah, let me think. I think my net worth pulled back from the top like 60% on that peace deal narrative. No, no. Memorandum of Understanding.
(58:43) [Beylo] I was about to say it was a memorandum because that's when crude dipped to like 65 or 64.
(58:48) [McCracken] Something hit 68, I think 68. Although that was also when I got balls of steel suddenly and just went fuck this. And I started selling my equity for calls. Of course, equity drops less than calls will go up on a rebound. So that's how I hit these new all time highs.
(59:06) [McCracken] And I kind of told people I was doing it. Wall oil — just a tweet by me where I was just so angry. Fuck it. I bought a bunch of calls today. And that was me selling my equity to buy calls. And then of course, that was when oil was like 75 and it just kept going.
(59:21) [McCracken] And at 68, I was like, fuck it. And I just liquidated like 5% of my portfolio, put all 5% in calls on oil, mostly USO at the time. And then that's what slingshotted me back and brought me back to life. That's probably the ballsiest thing I've ever done in my life.
(59:38) [McCracken] I was just convinced I was right, because that whole memorandum of understanding, there's no way this holds. Iran keeps upping their demands. They're getting more and more vocal about it, and although now, who knows.
(59:54) [McCracken] Now we're at a point where for the most part, Hormuz is open, apart from you have to be a madman to actually sail through it with a full tanker of hydrocarbons. I'd quit my job before I do that, because it's random. It seems like there's no pattern of who's getting hit.
(1:00:11) [McCracken] Iran hits a random tanker when they need to. I'd be terrified to go through there at certain times of the day. I'd be plotting the time correlations of when tankers are most likely to be hit.
(1:00:25) [Beylo] Is there a price that you could be paid to drive a ship through the Strait of Hormuz right now?
(1:00:31) [McCracken] I don't think so. Because if you die, it's over, then you can't.
(1:00:37) [Beylo] Yeah. And I'm happy that left tail risk is pretty serious.
(1:00:42) [McCracken] Yeah, exactly. So maybe part of Warren Buffett's compounding, it's just been how long he's lived.
(1:00:48) [Beylo] Exactly. How early he started. And then just the fact that he's not dead yet.
(1:00:53) [McCracken] Yeah, so I would definitely quit my job before doing it because it does seem random. Chinese tankers have been hit, everyone's tankers have been hit, lots of Greek ones, mostly Greek ones.
(1:01:11) [Beylo] Yeah, the poor Greeks, man, catching strays. Well, I think that they're —
(1:01:16) [McCracken] Also the crazy shipping guys that are probably just like, no, fuck you, run the tanker, and then they get hit.
(1:01:26) [Beylo] Oh, man. I mean, Centrini's guys didn't get hit, so that was good.
(1:01:31) [McCracken] True. I actually am tempted to go out there and set up a sonar because that would be the highest alpha there is. If I could count the tankers myself.
(1:01:43) [Beylo] But what's the risk reward?
(1:01:45) [McCracken] I feel like if I do it in a canoe or something, the odds anyone blows me up are low.
(1:01:53) [Beylo] They're like, who's this Ed Sheeran / Shaun White looking redhead Canadian in the Strait of Hormuz with a canoe right now?
(1:02:00) [McCracken] Exactly. Except there's a risk that when you drop the sonar, America thinks it's a mine and they blow you up. So I've had this thought and I'm like, fuck, I need to build a drone boat and send the drone out and drop the sonar off, and let it float under the surface and then count.
(1:02:20) [Beylo] I'm sure that'd work well.
(1:02:23) [McCracken] It would work better. At least if they blow it up, you weren't on it. But I'm amazed no one's done this. The amount of alpha there — if you know exactly how many tankers are transiting right now, you can make infinite money.
(1:02:40) [Beylo] Can you though, if they're manipulating the market, so to speak?
(1:02:43) [McCracken] Yeah. You can't do that forever. It's the George Soros thing, with the great irony being that his prodigy Scott Bessent is now the one manipulating the market. And Soros made all his money by betting against people who are manipulating the market. With the pound, made a pile of money.
(1:03:02) [McCracken] But yeah, so I don't know, I'm actually amazed how bad the tanker data we have is. I'm still convinced at least 4 million barrels a day are shut in globally. So I'm fine holding oil because we are drawing down tanks.
(1:03:22) [McCracken] But yeah, it's like what is gonna happen next? I don't know.
(1:03:27) [McCracken] Both sides think they can outlast each other.
(1:03:29) [Beylo] So that's a dangerous game theory to play.
(1:03:35) [McCracken] It does actually seem like Iran is doing worse this week. Seems like a lot of tankers are transiting.
(1:03:44) [Beylo] So yeah, crazy times, man, crazy times. I always enjoy our conversations. I like that we spent a lot of time with AI in the beginning, because when we ended our first conversation, I was like, man, we could have done a whole other podcast.
(1:04:00) [Beylo] And so that's what this one was.
(1:04:03) [McCracken] Yeah. I definitely will have more to say on that in the future.

== Memory, AI, and the Future of Commodities ==

(1:04:07) [McCracken] I have been looking — I actually bought some Micron on the dip, but very small position. And there's things I think — I think memory is still gonna send it. And part of the reason is actually personal: I want my own local LLM and to do that I got to drop like 4 million U.S. dollars and most of the cost is memory.
(1:04:32) [McCracken] So if I wanna do that, then I guarantee you there's other people that are gonna do this. So that's a huge amount of demand coming from independents or small businesses that will want their own LLM. I think memory can still send it.
(1:04:47) [McCracken] There's also some scientific advancements that would really resolve the compute bottleneck, but they wouldn't resolve the memory bottleneck. So one of those I want to work on, but I'm too busy with stocks at the moment.
(1:05:04) [McCracken] And so there's a lot of reasons to actually predict memory prices exploding. Have you heard of Jevons paradox?
(1:05:14) [Beylo] Yeah, that was kind of my whole thesis for going long Dell computer, HPE. All these people are just going to want to build their own stuff and that's going to be huge for demand.
(1:05:27) [McCracken] Yeah, so that's why I think this bubble won't pop. And that's why I was mentioning the whole, when does this end? How high can things go? Because the demand for data centers is gonna drive commodities and robotics is gonna drive commodities. And so at least I think the frontier, which is tech — super hard to bet on tech, because there's so many competitors, especially when they're all doing the same thing, Gemini, GPT, Claude.
(1:05:53) [McCracken] And that's why commodities makes more sense to bet on. And I think memory and compute are evolving into commodities now. Whether or not you have this thing, it's like a reusable commodity. And so I think memory has got another leg to its cycle that will probably break people who aren't long memory when they watch it go off.
(1:06:22) [McCracken] But we'll see what happens. Definitely in a month or two we should chat about that.
(1:06:29) [Beylo] It's funny because I feel like there was so much grave dancing during this pullback in the semiconductor space and memory and AI, which made me think there's no way this trade is done. So many people are so quick to dunk on the AI trade.
(1:06:43) [McCracken] And that's why I think gold's about to pull back too. Literally everyone on Twitter called the bottom. Everyone was like, oh, it hit 3900. It's going —
(1:06:50) [Beylo] Dude, I didn't even want to buy it and I was public about this. I posted, I was like, look, everyone's calling the bottom, I just don't think it's that easy. And then it literally broke the wedge and just couldn't stop, up from 4 to 700. I was like, OK, I guess I was wrong.
(1:07:05) [Beylo] Luckily we're able to get long some gold, but I literally was staring at the chart and I'm like, it can't be this easy. I can't just put a buy stop right on the breakout and then it goes up. But I don't know, man, it did.
(1:07:21) [Beylo] And then it's kind of holding in. I think the trend is up, but it's just a weird market. It's just a weird market.
(1:07:32) [McCracken] Yeah, I think gold pulls back a bit here. We'll see.
(1:07:36) [Beylo] So do I, and I was surprised by the strength today, honestly. In silver.
(1:07:42) [McCracken] Well, silver makes sense to me. Like I was talking about, there's a lot of tech demand coming that I think people don't have covered, like robotic developers.
(1:07:51) [Beylo] Where do you think silver could trade in the next two years?
(1:07:54) [McCracken] I think $100 is about to be the floor price. It may not go much higher. It might go there and sit there, and that's why I am not interested in owning physical silver here. I do own it. But that's my old stuff I bought when it was 40 to $50. And that's my rainy day fund. For now, I just want to own silver miners.
(1:08:21) [McCracken] I'm still unsure which ones are the best to play with. But we talked before about Kuya — I think if they execute, that's interesting.
(1:08:32) [Beylo] They're like the cheapest silver miner out there.
(1:08:34) [McCracken] Yeah, but I barely own any because there's still this "if they execute". I wish you could find an APM right now, where it was trading at 1.5 times enterprise value —
(1:08:46) [Beylo] Over cash flow.
(1:08:47) [McCracken] Yeah. Definitely the market is starting to behave a bit and price things more correctly, at least in miners. Still lots of cheap gold miners. And that's what I'm hoping I can buy if there's another dip. I really want oil to just send it, cause a little bit of a liquidity crunch, gold crashes and I get to rotate.
(1:09:07) [McCracken] It would be so beautiful. It'd be like a legendary play. But probably not going to happen because it doesn't seem like it wants to. It seems like both gold and oil want to go up at the same time, which is like what the hell is going on?
(1:09:20) [Beylo] Yeah. Well, the other thing too — I always kind of view this as, what's an upside risk to gold, upside case being what if I just wait for a pullback that never comes? The amount of M&A that could happen in the gold space is pretty incredible.
(1:09:38) [Beylo] So many juniors trade at like .3 or lower times NAV at the gold price. Look at the cash flow these majors are generating, and they can't find those ounces for a .3 times NAV. It's going to cost them so much more to find them organically.
(1:09:59) [Beylo] So I think — and you haven't even really seen a massive M&A wave throughout the gold space.
(1:10:04) [McCracken] Yeah, no, we really haven't.
(1:10:07) [Beylo] There's been a lot more capital discipline too, which is good to see, because I think everyone got scared of the last gold cycle. Everybody basically bought everything and then they got crushed. And I think that's left a lot of scars.
(1:10:22) [McCracken] Yeah, definitely makes it better to be a retail investor, because you can find these cheaper ounces.
(1:10:32) [Beylo] Awesome dude, great conversation. Let's do this again in like a month or two.
(1:10:36) [McCracken] Yeah, let's do it. We'll see if I have more opinions on what to buy for AI then. I'm still researching. It's just so hard because tech companies are so big. You need a team to actually dig into these and value them.
(1:10:54) [Beylo] Well, next podcast, let's do like a robotics theme. I think that'd be cool.
(1:10:59) [McCracken] We could try and do that.
(1:11:01) [Beylo] I've got some thoughts on robotics.
(1:11:04) [McCracken] Then let's do it. I think there's definitely going to be opportunity there where there's going to be a moment to rotate into that that'll be basically like buying NVIDIA in 2022. I think that's coming up. Assuming that they can get the commodities.
(1:11:20) [McCracken] And that's why I think the best book coming up is going to be some mix of robotics with miners. Try and get it on both ends where you own the people selling the robotics companies the materials. And then on top of that you have the robotics companies that everyone is flowing capital into because they're like, oh my God, it's the next NVIDIA. So you got to beat that.
(1:11:43) [McCracken] It's just a question of which ones, and are they already priced correctly?
(1:11:50) [Beylo] I was about to say, the hard thing is you have all these things, but then 9.9 times out of 10 you have to plug your nose and buy irrespective of value. Because all of those things don't also happen with a super cheap stock usually. NVIDIA was the exception, because NVIDIA kind of traded around 10 to 15 times next 12 months earnings for this entire ride, which is crazy.
(1:12:10) [Beylo] But robotics, nothing's screaming cheap. Which is the problem.
(1:12:16) [McCracken] Yeah, exactly. No I agree. That Chinese IPO that just happened — forgot the company name even though I met them at I think NeurIPS or ICLR.
(1:12:26) [Beylo] Was it Unitree?
(1:12:27) [McCracken] Yeah. That's it. I was looking at the price, it was IPO and I was like, yeah, no.
(1:12:35) [Beylo] I'm also just sketched about buying Chinese robotics companies because again, the whole bifurcation. I don't know if I'd be comfortable in this world having a Chinese robot in my house.
(1:12:50) [McCracken] True, true. It's a good point.
(1:12:53) [Beylo] I don't even think Trump would let a Chinese robot come into this house. Imagine if the trade war heats up, World War Three heats up, all of a sudden all your Chinese humanoid robots that are doing your laundry are now just slaughtering everybody.
(1:13:06) [McCracken] Yeah, or they just turn off and you lose a pile of money. And it'll be one or the other.
(1:13:12) [Beylo] You'd hope it just turns off.
(1:13:14) [McCracken] Yeah.
(1:13:16) [Beylo] Awesome. Well, let's do it, man. Let's do round 3, robotics. We'll get down the calendar. Thanks so much for the pod. Thanks for doing it late at night. Set my schedule with two young kids, so I appreciate it.
(1:13:29) [McCracken] Yeah, no problem. My schedule hasn't been the easiest either with all the travel and AI conferences.
(1:13:35) [Beylo] For real man. But we made it work.
(1:13:37) [McCracken] Yep.
(1:13:39) [Beylo] Thank you. I'll talk to you soon.
(1:13:42) [McCracken] Have a good one.