Transcript

Dylan Patel - The Infinite Demand for Tokens, Claude Mythos, and Supply Constraints - [Invest Like the Best, EP.469]

0:02 And welcome everyone. I'm Patrick O'Shaughnessy and this is Invest Like the Best. This show is an open ended exploration of markets, ideas, stories, and strategies that will help you better invest both your time and your money. If you enjoy these conversations and want to go deeper. Check out Colossus, our quarterly publication with in-depth profiles of the people shaping business and investing. You can find Colossus along with all of our podcasts at Colossus.com. Patrick O'Shaughnessy is the CEO of Passitive Sum.

0:29 All opinions expressed by Patrick and podcast guests are solely their own opinions and do not reflect the opinion of Positive Sum. This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions. Clients of Positive Sum may maintain positions in the securities discussed in this podcast. To learn more, Visit PSUM dot VC Um This is my second conversation with Dylan Patel, known as the founder and CEO of Semi Analysis, where he tracks the semiconductor supply chain. and AI infrastructure build out.

1:01 This conversation is about the supply and demand of tokens. On demand, Dylan described something completely explosive. He explains why the frontier model is the only model anyone wants and willingness to pay for it is nearly unbounded. His own firm has gone from tens of thousands of dollars in AI spend last year to seven million run rate this year. On supply, we walk through the bottlenecks across memory, logic, and fab equipment that will determine how fast any of this can scale.

1:25 We also cover methos and what the leading labs need to do to fix their growing perception problem. Please enjoy my conversation with Dylan Patel. You told me this incredible story about how your own team's use Of tokens has changed dramatically this year. Yeah. Can you retell that story and what it is teaching you about what's going on in the world? Last year we thought we were heavy users of an AI. Everyone's using Jet GPT, everyone's using Claude. Providing whatever subscriptions anyone wants.

1:53 On the order of spend of tens of thousands of dollars for our firm. This year the spend is just skyrocketed. And it really started in late December. With Opus. That included Doug O'Loughlin, who's president. He's very much like leading the charge in the sense of non-technical people using AI.

2:10 Four. Coding Basically pilled the whole firm slowly over time. I think he's been the leader in doing that. Obviously the engineers were using it anyways, but Spend in January just started to inflect and rock it and rock it and rock it and rock it. We signed an enterprise contract with Anthropic and it's gone to the point where now I think when I last talked to you it was five million spend rate. It's actually seven million spend rate now. So we're spending seven million last week, by the way. But a lot of that is just the usage.

2:38 People who have never coded before. Uh using clawed code and spending thousands of dollars sometimes a day. And it oscillates some people spend thousands of dollars one day or spend a couple hundred dollars a couple of days and then they go back to thousand dollars. It's very variable across each individual user, but across a firm, we're spending seven million dollars a year now on clod code. at the current rate versus our salary expense being in the neighborhood of twenty five million dollars. So or north of twenty five percent of spend

3:07 On Cloud Code. As a percentage of salary. If this trajectory continues, then you know we'll spend more than hundred percent by the end of the year. Which Is a bit terrifying. Thankfully I don't have to decide between people and AI because our companies growing so fast, it's more so like, okay, well, I don't have to hire nearly as fast and I could spend a lot more on AI and it works and we just grow faster. But I think

3:26 other folks will start to reckon with the fact that Huh. If this person can do the work of five to ten to fifteen people using claud code. Then all of a sudden

3:37 I should probably cut people. And the use cases are so broad. Here are a couple examples. Okay. So for example, one thing is we have a reverse engineering lab in Oregon that we've been building for a year and a half. We have a bunch of fancy microscopes, scanning electron microscopes. The whole purpose of this is you reverse engineer chips, you get the architecture out of it, you get the materials that they're using to manufacture, and this is some of the data we sell. This is a very slow process of analyzing that data. Instead one person on the team

4:02 They've been able to spend with a couple thousand dollars on Cloud tokens, they've been able to create this application that is GPU accelerated, runs on a server that we have at Core Weave, and any time we send it an image, it will take the picture of the chip and overlay where Every single material is, oh, this part is copper. Oh, this part of the gate is tantalum. This part of the gate is germanium, this part of the gate is cobalt. And so you can do a finite element analysis of the entire stack up of the chip. Very, very quickly, visual with a dashboard, GUI, it's everything, a few thousand dollars of to Claude. The person previously worked at Intel and he said that was an entire team's job to build that and maintain that.

4:37 Oh rack that up across the entire firm. It's insane. Another example that I think is super fun is Malcolm is an economist at a major bank. Before their economist department was like a hundred or two hundred people. What he built was the most incredible thing ever. He piped all of this different data, threat data and all these other data, sorry, employment reports and all these other things from various APIs. We signed a couple of contracts with folks to get API access to data, pulled it all in, started running regression started looking at the impact of various economic revolutions

5:09 on the economy. From a deflationary, inflationary perspective. The Bureau of Labor Statistics has this entire set of 2000 tasks. And so he did that with AI, which ones can be done by AI, which ones can not, and grading them across a rubric. About Three percent are doable now with AI.

5:24 And so he's created this metric so that you can measure Things that can be done by AI, what the cost of being able to do those with AI and therefore the deflationary aspect of it. Phantom GDP is what he's called it. Output can go up, but because cost falls so much, actually GDP theoretically shrink. So he created this whole analysis and a brand new benchmark of language models, a set of evals across two thousand different evals. A year

5:50 He's like completely cracked out on Claude. He's like everything has changed. How do you think about it as a business owner? Going from close to zero to twenty five percent, accelerating towards whatever percent of total spend. At what point are you like, Whoa I need to put the brakes on this? And be careful how much we're spending. Maybe we don't need to spend on the most cutting it on Opus four point seven, which came out today. Maybe I can throttle it back to something that's a little bit cheaper. I think I'm in the information business. We sell analysis. We do consulting. We create data sets. I don't see why this wouldn't be completely commoditized.

6:22 On a pretty rapid basis, if I'm not constantly improving. My first product that I was selling as a data set. There's more people trying to do it. Now we've made it constantly better and better and better and more detailed. And so therefore it sells a market. But the way we were doing it. In twenty twenty three. It's not terribly different. It's basically what everyone else is doing now. If I don't move up the bar. Then I will be commoditized. If I don't move fast enough, I will also lose my edge. So the question is.

6:48 Yes, AI commoditizes things just like it commoditizes software. Those who can move fast and keep control of their customers and keep providing them an awesome service and keep improving the service won't shrink, they'll grow faster. those who are incumbent and not doing anything, they're gonna lose. And so It's a bit of an existential. If I don't adopt AI, someone else will and they will beat me. Another easy example is the energy space.

7:11 So we've had a few energy analysts for like a year now. We've been trying to build out this energy model. It's very complex. Energy's data services market is something like$10 million. So obviously a huge market for me to try and break into. And we've been like slowly grinding at it and it's been helpful for our data services business. We really hadn't broken into the energy data services business despite a year of having multiple people on the team. Then Cloud Code, Psychosis hits. One of the people who leads the data center energy and industrial business at semi-analysis, Jeremy, it hits him. А не оливсад і три вікс. He spent a lot. He's spending like six thousand dollars a day.

7:44 It was an insane amount, but he scraped every single power plant in the US, every single transmission line above a certain voltage, and created this entire mapping of the entire US grid, as well as a lot of demand sources, all from various public sources of data. And it's got like this dashboard where you can view and check. You can see all the micro regions of the US where there's power deficits and surpluses, all of these details. Built in a handful of weeks. We started showing some of our customers who buy our data center data set, but are energy traders. We showed some of them and they're like

8:14 Wow, how long did this take you? This is really good. This is better than XYZ company. And then we like g dig deeper. XYZ company has a hundred people and have been working on this for a decade. Obviously our thing is not fully as robust, but in some ways it is better. I'm gonna commoditize these energy services companies. Data services company.

8:30 Who's gonna come commoditize me if I don't move faster? And so the question from a business owner's perspective is. Yeah, I'm spending a lot, but What does that spend getting me? Is it getting more revenue? Are you worried that In the limit.

8:42 the people that control capital and invest in capital who are often hiring you for what you do will just say, Well, we have analysts too who are really smart about this. Like we'll just build this ourselves. If it's getting that easy. At what point does it just all pool into the investment firms that stand to gain the most because they have the most leverage on top of the data or the insights that they glean? First of all, any information services business. Obviously.

9:06 I don't generate as much value as my customer do from said information. Because if I sell you information for a dollar, you're only buying it for a dollar because you know that information helps you make a decision that lets you make More than one dollar. And so therefore you have made more money off of me than I did from the information myself. These investment funds all have their own information services, you know, especially like the super the Jane Streets of the world and the citadels. They're really detailed on their data. And yet these sort of folks also purchase data from us and continue to do so and continue to grow with us because I think there's just some it factor, right? We move faster, we're more nimble. A smaller team that's

9:41 Focused on just one specific thing. AI infrastructure and the huge revolution that causes in AI and tokenomics and all these things. And we see where it's headed and so we're moving faster and building faster. I think investment professionals, yes, they'll try and build some of the stuff we do, and more likely they'll just buy the data from us and it's cheaper for them to buy the data from us and then build on top of it than it is to build it themselves.

10:05 I feel like every conversation I have with you, what I'm always getting at is just supply and demand of tokens. That's the thing that's interesting to me in the world right now. What has this experience taught you about the demand Has it changed your view on the demand side of that equation, just feeling it viscerally yourself? If we take a stick back and look at the macro lens, right? Anthropic has gone from nine billion revenue to what, they're at thirty five, forty billion now. Probably by the time this air is forty, forty five billion, who does air ours.

10:32 Compute has not grown to the same degree. And if you do the calculations and you assume they didn't decrease their research and development compute, they clearly didn't they're released they have methos. They have Opus 4.7. So they clearly didn't decrease their research compute spend. So ultimately. What they've done, even if you assume all incremental compute they've gotten has gone towards inference, their margins are at a floor of 72%. In reality, some of that incremental compute they've got probably went to research and development at maybe higher than 72% gross margins. To be clear, at the start of the year, there was a leak from their funding round document someone leaked it. Thirty something percent gross margins.

11:06 Where on earth does a business like this grow margins like that? It's in principle, right? Their demand is so high, they're able to cut back on usage limits, rate limits, all these things. What really matters is having an anthropic rep and having an enterprise contract with them and getting the rate limit increases that you need because otherwise Tokens are ultimately super, super in demand. Whoever can pay for them, Anthropic has the same problem.

11:29 I mean not problem, it's just the reality of how capitalism works. Yes, people are sending in forty billion dollar ARR. And tokens. But those tokens are generating way more than forty billion dollars in value. Various businesses will have different value generation per token, but as we get more and more intelligent, what really matters is access to these most intelligent tokens and leveraging them at things. You as a person deciding what is the best way to leverage these tokens to grow business and generate value. Because a lot of folks will want tokens and generate tokens, but the shitty SaaS startup and SF who is using Claude.

12:04 to generate their software product is not necessarily actually creating a ton of value. And therefore they're gonna get priced out of tokens soon enough. I had this experience just today where on the flight here I got really limited out on something I saw four point seven came out. And what I immediately wanted was to be on four point seven.

12:21 That second. I couldn't think about using four point six anymore now that this four seven is out. I was perfectly happy with four point six for the last many weeks. It's amazing. Are you surprised that people are so insistent on going to the most expensive leading edge thing to the degree they are? Without a doubt. I think One funniest memories in the past month and a half is my self.

12:42 And uh buddy of mine Leopold being on our knees in front of an anthropic co-founder begging him for access to meet us and then pretending it doesn't exist. Cause we knew it existed and we're like, please give us access. And he's like, I don't know what you're talking about. What was your reaction to that rate card or that eval card coming out? It was rumored in the Bay Area. We knew it was supposed to be really good, but

13:07 If you just look at the benchmarks, and obviously benchmarks change over time. Mythos is potentially the biggest step up. in model capabilities in two years. I think that's really, really an important detail that It's so good that they're like don't want to release it, even though they already announced the price to their

13:24 people that they did a selective release for cyber for and it's five or ten X the token cost. They just don't want to release it because they're worried about the impact on the world. And they're releasing a worse version Opus four seven to us. And they explicitly said in the model card, Hey, we actually Preferentially made it worse at cyber. I don't know if you read that. Whoever you are, if you have enough capital You should get a freaking enterprise anthropic subscription where you pay per token, not with these subscriptions, because then you won't get rate limited much. And then you need to figure out how to leverage those tokens to the highest value task and make money off of it. Because ultimately what you're doing, maybe like a year from now or two years from now, the business is actually just arbitraging tokens. Right. The tokens are amazing, but let's

14:03 figure out what direction to point them in and then three or four years from now the model will know what to do with the tokens and how to make the most value. You need to go look at this retroactively, pick any benchmark. The cost to hit a certain capability tier. used to cost X, and now it cost one one hundredth or one one thousandth of that. Deep Seek, for example, on GPT four. was one six hundred the cost. And since then the cost have fallen further for GPD four class models.

14:28 Of course, no one gives a crap about GP four class models. They want the frontier, because the frontier lets them create the economically valuable things, but Jupyter four class models can still be used in stuff. And so people are using them in some like tiny use cases. It's just the cost have fallen so fast. But it's not really what's driving the demand. What's driving the demand is all these new use cases. Yeah. current four point six Opus or four point seven Opus tier models. A year from now.

14:54 My spend for the same exact quality the model. would probably be like seventy K. I I bet you it'll be a hundred times cheaper. Irrelevant because I'm going to be using a way way way better model, which can do way, way better things. Enthropic Methos is more expensive as a model, but it spends a lot less tokens to do the thing. And therefore it is actually cheaper in most tasks than four six Opus because it's just way more efficient.

15:16 even though each individual token is smarter. So Yeah, there's crazy geniuses. creating huge cost efficiency improvements every day. They work at the labs and they're making the models way more efficient. You see it every generation. What was it, five nano or ever was better than GPD four. Five mini was better than GPT four. It was like one one hundredth the cost.

15:36 This just happens. And We accept it at its face value. Ultimately you keep making things cheaper and then you keep scaling them up and you keep getting humongous improvements. When I last saw you meet us had just come out, maybe the day before or something. For the car to just come out.

15:50 And you said something like It actually made you feel like a little scared it was so good. What did you mean by that? Anthropics whole goal. in twenty twenty five and even a lot of twenty twenty four, they're like, hey, by the end of twenty twenty five, we need an L4 software engineer. in their model. And they by and large achieved that with four six opus.

16:10 What they didn't say is that And if you look at mithos, if you compare benchmarks, it's like an L six engineer. So L four is like pretty new. Well, six is like quite well experienced. I think Anthropic said that the model internally was available in February. So in two months they've gone from L four engineer to L six engineer. What's next?

16:29 When you think about the model progress, it's only accelerated. Anthropics release cadence has compressed. Open air's release cadence has compressed. Why? Because generally to make a better model, you need a few things, right? You need Amazing compute. Compute is very expensive and it has a timescale that we track and it's like it's growing, but it's set in stone for the next short term. It's like set in stone what you've already signed. And there will be delays and shifts and somehow you can find a little more, but it's generally pretty set in stone. There's amazing researchers that people are paying tens of millions of dollars for. And then lastly

16:57 There's implementation. And implementation historically have been very difficult. If I have an idea and I have to implement it, implementing is hard. Now Ideas Are there. Implementation is very easy. It's expensive.

17:09 But it's very easy. So how does one decide what ideas to implement? And it turns out If your implementation is just so much easier, now you can just implement more ideas and move on the treadmill faster and faster and faster. Whether that is AI model research, and so now your model release cadence is shrunk down to two months from where it was six months before. Or I wanna take every power plant in the US and every transmission line and model it and run regressions and see the micro supply and demand.

17:34 I can also do that. The idea is cheap. Which idea makes sense, which idea is worth the capital that you have to spend on the tokens, because the implementation is there. That's the key learning. If implementation costs continue to tank, which they are. We don't even have Methos yet. It's only been a handful of hours since Opus four seven launched, but my team is pretty excited about it internally.

17:57 What now comes to the world it's a complete reordering of how economies work. What used to matter a lot was execution was very, very fucking difficult and ideas were cheap. Надея зер. And plentiful, but execution's very easy. So really only the good ideas are the ones that can justify the spend on super cheap alimentation. So are you actually scared or does it just introduce an uncertainty that's hard

18:21 to grapple with. Uncertainty is there but I do Think that causes Some fear in terms of how does society reform itself. How does one

18:35 exist in a world where actually your ability to implement something is not actually that important. Your ability to just choose the correct idea for AI to implement, and then your ability to sell that idea. or sell what the AI has implemented is what matters. Your ability to garner capital towards that is what matters. And going back to the point of it's very important to have the newest model always. Who's gonna have access to the newest model? Anthropics project. I know it's not called Earwig, but I troll anthropic people by calling it Earwig.

19:02 Glasswig. Anthropic earwig. Where they only release meetos to certain companies for cyber. That's just gonna be something that continues. Models will have Less broad and less broad.

19:14 Deployment. I know Open AI and Enthropic and all these people are like, We want to have great AI for everyone. Yeah, it's very fucking expensive. Who's gonna pay for the trillion dollars of infrastructure? People who have money and we can build useful things with AI. And then you don't want people to distill your model, so you don't release them broadly. You release them to fewer and fewer set of customers. Those customers.

19:34 Are also now wrestling over the tokens. Unless Anthropic jacks them. They could double their pricing on Opus and I would continue to pay and I bet most users would continue to pay. I bet that wouldn't solve their humongous capacity problem that they have. So then the question becomes Where does this cycle end where token usage and therefore the benefits of those tokens, the additional value generated on top of those tokens, aggregates among fewer and fewer and fewer companies?

19:58 I don't have Mitos. You know who has Mythos? Cop. Fricking banks. Now they're only using it for cybersecurity, but at some point I can envision a world where, hey, maybe I, because I have an enterprise anthropic contract and because Entropic people kinda like me. They're willing to give us

20:12 slightly earlier access or slightly higher rate limits or something for a model. I hope that's what happens. And then my competitor, whoever that is. doesn't have that and I'm able to fucking crush them. There are people like Ken Griffin of Citadels super well connected and super rich. He goes and signs a deal with open air anthropic that's like, Yeah, I'm gonna get access to your models and I'll buy the first Ten million dollars worth of tokens each year. So whenever you release the model.

20:37 I'll spend the first 10 billion tokens and then everyone else can get the model after that. Yeah. And it's like, okay, well, now what does that do? Now he's gonna crush everyone to the markets. That's just an example. It could be any number of things. It could be cyber, like anthropicas worry about, oh now I can hack people. Can be information services business like myself, or I crush someone else. I think it's such a broad base. We don't know what these models can do. Anthropic doesn't know what these models can do. No one knows what these models can do. It's up to the end user to figure out where they can leverage the tokens to see what they can build and imagine, which is tremendously productive and uplifting for humanity, but then what happens to the concentration of resources and usage of it. Presumably right now robotics or robots consume relatively zero tokens versus everything else.

21:14 What's your view of that? If that's like a second demand curve that could start to ratchet. There's a new startup every single day within a mile of here. Trying to build something interesting in robotics.

21:25 So there's this concept of software only singularity, which is that The world has AI singularity, but only in software. And now what about the rest of the world? The vast majority of the world is Physical. You can see the world.

21:38 orient around hardware, not software. That's actually why I think software only singularity is like just a blip. And not like uh we do get everything else because once software's super easy, what makes robots really hard? It's programming microcontrollers and actuators and controlling all this stuff is very difficult. And Right now, the interesting thing about models, AI models, is they're actually really inefficient in learning. It's just we're able to give them so much data that they're able to learn and pass us in certain ways. currently the robot models, VLAs, vision language action models, which is very popular right now, is probably not going to be the thing that ultimately scales beyond. They're inefficient in data.

22:17 And we can't scale the data for them fast enough. There is going to be some way to large scale pre-trained robot models where just like humans see all this data throughout their lives. And what's interesting is humans The reason why we're so good is we're sample efficient. One example, two example, we're good. So applying that to robotics. So once you have this software only singularity implementation is super cheap, anyone can start to build these models that now. robots are actually useful. So I think

22:43 In the next six to eighteen months, we'll start seeing real breakthroughs in robotics. that enable few shot learning. i.e. there's a pre-trained robot model and now there's a robot that you have hired or bought or whatever. You showed a few examples and it's able to do it. Right now, you know, there's a lot of companies doing robots for advertisement or robots for simple stuff like that, but it'll be like, oh, folding clothes, sure, sure, sure. No, but it's gonna get really niche. Robots just for cleaning chalkboards and it's a rental service, or it'll be a model package that you download onto your standard robot that then does that. And anyways it'll be a huge explosion in

23:16 physical good acceleration and deflationary effects there, but That's ultimately going to keep Token demand going crazy. I don't think Token Miniman slows down personally. Did you learn anything else about the world based on Methos' results and how it was built?

23:32 This is my way of asking if you break down the components of the scaling laws things. Than prior models. And so yes, it is a much larger model. What chip it's trained on is not really relevant. It's the scale. And obviously to a hundred thousand black wells is equivalent to hundreds of thousands of prior generation chips. TPUs and training have their different release cadence, so it's not exactly like mirrored one to one.

23:54 But ultimately yes, Mitos is a significantly larger model. It's proof that the scaling laws still work. Everything about it shows the trend line continues of models more compute into model makes model better. And along the whole way, it's not just more compute into model makes model better. Along the whole way, we're also getting these compute efficiency wins, which are as all this research compute that the labs are spending is actually turning into If I want X capability to your model. Every six months that cost or every two months that m cost is dramatically decreasing. But then if I scale it up massively, I get a humongous capability jump as well. And so yes, it's proof that this is still happening.

24:29 Google and Anthropic are not heavy, heavy users of GPUs on the training side, but Open AI they'll start having their new class of models. I think they're taking a more sensible principle approach to scaling in small steps. Anthropic really went for a huge jump. We'll see better and better models throughout the year, and the release cadence is only gonna get faster. We've gone a long way in the conversation with saying almost nothing about open AI, which would have been so strange 12 months ago. Interesting thing. Everyone's like, Okay, so Anthropics just won, right? They had Methos in February, they never even released it because they didn't feel the need to. They're already sold out. The revenue's already adding 10 billion dollars a month.

25:02 And then you've got Opus four seven today, all before Open Eyes alleged SPU release, which media such as The Information and others have posted about. So clearly Anthropic is in the lead and open AI is cooked. Well what's interesting is because Anthropic has such Bounds on compute. And they can only grow it so fast and to the point of

25:23 Dario used to gloat about how OpenAI was being too aggressive on compute and Anthropic was more sensible in their scaling. And now Anthropic is like, fuck, I wish we had a lot more compute. OpenAI is able to pay the bills perfectly fine. In fact, they raised a ton of money to get incremental compute in addition to the irresponsible levels of compute that they were buying from Oracle and Core We and SoftBank and all these people and Microsoft, such as training. Now they're getting Tranium as well from Amazon. They've done this. Insane thing on compute. They also know they need more, but what's interesting is if you were to say Opus four six. Let's ignore models getting better over time.

25:56 Let's just take diffusion of this technology. You and I may jump on the model immediately day one, but other businesses take time and it takes time for people to learn and the spark of Oh shit. Claude psychosis moment doesn't hit everyone at the same time. And so by the end of the year, let's say a four six Opus chair model the economy would spend a hundred billion dollars on. I don't think that's unreasonable. It's spending forty billion right now. That's like a linear extrapolation. It's a linear extrapolation, not an exponential.

26:24 to get the exponential, you need the better models. Anthropic won't have enough compute to do that. And presumably OpenAI and Google will hit that tier. Soon enough. Whoever hits that tier next, sure Anthropic may get to charge seventy plus percent gross margins. But if OpenAI hits it next.

26:39 They charge 50% of gross margins. They still get all of this incremental demand, and probably they also won't have enough compute to serve all the users. Sure, maybe Mythos is a model where if the world had enough compute it'd be five hundred billion dollars of revenue or something crazy. There is such demand for these tokens and such limitations on compute. We see this with H one hundred prices skyrocketing. And all these other things.

27:02 the useful life of these GPs continue to extend extend. It's pretty clear even the tier two lab is going to be sold out of tokens. Let alone the Tier One lab. The Tier One Lab will have Better margins. But the Tier Two Lab will be sold out and probably the Tier Three Lab will also be close to sold out.

27:16 The economic value that the best model can deliver is growing faster. than our ability to actually serve those tokens to people via the infrastructure. And so this gap will continue to grow, and the mobile labs will continue to have expanding mergers. Until people in the hardware supply chain, infrastructure supply chain are like, Wait, no, why don't I just jack up my margins? So suffice to say, I think the assessment today or your assessment of the demand side is completely explosive in your own particular example here at semi analysis, but just more broadly, that You call it AI psychosis. As people fall into this experience of what they can do, the implementation difficulty.

27:47 going completely away. I've certainly felt that. My own token spend is just through the absolute roof, just in the matter of weeks. So that feels like a pretty good assessment. Anything we're missing on the demand side? If you don't use more tokens, you'll never escape the permanent underclass. Either you use more tokens and you generate economic value outside economic value for the use of those tokens. A lot of people are doing it the boring, lazy way. Oh, I guess I'll just work one hour a day instead of eight hours a day and I'll have AI do most of my job. That's the boring way. The cool way is I'll still work eight hours a day and I'll do eight X the work and maybe I'll make five X the money. You can't do this with a job, obviously. There's people who have multiple jobs. There's people who start companies and start selling stuff. There's people who are hustling.

28:26 Which is what I view like you and I as doing is we're mostly hustling. Get that economic value on this AI before everyone is using it and it's table stakes. Because it's still not table stakes. So If you don't use more tokens and generate the value from them. And capture that value. There's three different problems here. Using more tokens, generating value from those tokens, and capturing from the value that you created from the tokens. If you don't do these three things, you'll never escape the permanent underclass, i.e., as

28:51 models continue to skyrocket capability and the concentration of resources potentially happens. All right, let's talk about supply. What is changing at the frontier of supplying the entire stack that's required to serve all these tokens as the demand curve explodes? As demand skyrockets, prices are going up for everything on the supply side, whether it be the NGPUs. Their prices are going up.

29:13 In addition, their useful life is extending. H100 prices look like this. Yeah, exactly. There's people who have argued GPUs full lives are less than five years, complete nonsense. There are clusters now re-signing, three or four year old Hopper clusters resigning for three or four more years. There's A one hundred clusters that are resigning for another couple of years. So the useful life is clearly not five years. It's maybe even seven or eight years, arguably. We don't know yet. We'll see when Hopper gets there, but it's clearly not five years. So the useful life is extending and the prices are going up on that renewal. In effect, the gross margin was not thirty five percent on a cluster, it's beyond that. So margins are expanding in the cloud layer. Margins are

29:51 Extremely healthy on the hardware layer. with NVIDIA still charging seventy five or whatever percent gross margin. As we move down the stack memory, obviously margins have skyrocketed there. Places like optics. And logic, there are large prepayments and margins are growing slowly, more so the companies that are making chips like NVIDIA are paying huge prepayments, so it affect the cost of capital or timing of cash flow as a return on invested capital is going up, even if the gross margin isn't.

30:18 And you see this across the whole supply chain. You see ASML. is completely sold out. And they need Carl's ice to expand faster. Everyone's either sold out And margins are going up or they're getting prepayments, which increases the return on invested capital because the invested capital is lower. And so this is a consistent trend across any part. It's even like to make a P C B requires copper foil.

30:37 And that copper foil is sold out and people are making prepayments for it. Anything and everything that has a pulse. And it sold out, people are all jumping to get more incremental supply and fighting over the supply for the years after. What do you think are the most important bottlenecks? Typically in economic history, when there's this kind of demand, supply reorients and rises very, very quickly.

30:58 to meet the demand. It seems like it's almost impossible for supply Right now in this moment to keep up. Famous last words, every shortage is followed by a glut historically. But what are the most interesting bottlenecks to you across the supply side.

31:12 Supply chains are usually very fast to react. One unique thing is that our supply chains now are more complex than ever, and the things we're building are more complex than ever, and therefore the lead times are longer. And it's not like we haven't seen eighteen month long lead times in other industries. It's just Building your commental supply didn't take years. And this is the case with memory.

31:31 Memory can only grow capacity, low double digit percentages a year, right? Twenty, thirty percent a year. Even less for NAND, a little bit higher for DRM, but whatever. Even though the demand signal was very strong at the end of 2025, the memory companies immediately started reacting. None of that incremental capacity really gets here until the second that they've decided to do, in addition to the typical twenty to thirty percent. And they can stretch a little bit, but really the true incremental supply doesn't come until twenty eight, which is a very unique thing. Even if they wanted to build as fast as possible, it doesn't come till twenty eight, late twenty seven at best. So The result is memory prices of

32:06 Gone through the roof. And guess what? They're gonna double and triple again. At least on D RAM, especially. People like oh the memory story is overplayed, everyone gets it. It's like no no no, you don't get it. Dram

32:17 Well double or triple from here still. Because That's how much capacity is required and they have to steal capacity from somewhere else. And the only way to steal capacity from somewhere else in a capitalist economy is demand destruction via higher pricing. We're not rationing stuff here. And so ultimately that's what's gonna happen. And so margins continue to go up. I think Logic also has humongous capacity problems. T SMC just have their earnings. They keep up in CapEx. Ultimately takes them quite some time to build fabs and they're trying to do everything they can to squeeze every little output out of every fab that they have. But ultimately

32:49 They're not raising prices fast because they're good people, that seems like. single digit price increases instead of triple digit price increases like the memory guys have had. So you ultimately have this market where, yeah, T SMC is a great company, but are they actually gonna extract all the value? I mentioned things like copper foil, glass fibers for PCBs, lasers. These are things that are well understood in niche supply chains, but they're very, very tight. And ultimately upstream the semi conductor waiver fabrication equipment supply chain is one that's gone up a lot, but it's still very underappreciated. T SMC

33:20 CapEx this year, they say fifty six. We've had fifty seven point four billion since January, and we may up it slightly more just because we see some ways that they can get incremental capex, but What people aren't focusing on is what does that mean next year? And what does that mean the year after? And it turns out Three years from now, TSMC is gonna spend$100 million on CapEx.

33:38 Maybe two years from now. It might be twenty eight. Sincerely, they may spend a hundred billion dollars on CapEx in twenty twenty eight. And people just can't fathom that. But what does that mean for their downstream supply chains? Companies like Lamb Research or Applied Materials or ASML, or f their further downstream supply chains like MKSI and all these other companies, the tail whip it just gets whip harder and harder and harder. And that's a shortage if T SMT wants to spend a hundred billion dollars in twenty twenty eight. Which is a real possibility.

34:05 I think people would think that's insane, but that's a real, real possibility. What about other parts of the chip ecosystem where GPUs have been completely dominant? What about like CPUs or ASICs or things that start to pop out as both opportunities and bottlenecks beyond just Nvidia's GPU dominance? Yeah, I mean A6 are obviously taking off, but I'll pivot away from AI chips to talk about these other things. There's a project we did on FPGAs and it turns out there's a hundred and twenty FPGAs per

34:30 Next generation rack. A Iraq and then what about all the FPGA names? CPU eyes, all these reinforcement learning environments, plus all the slot code you and I are generating that is now running on some versal instance or whatever it is. Or some AWS instant or some bucket that we've spun up. All of that requires CPU. And so CPUs are completely sold out and demand is skyrocketing there.

34:51 Yeah, how people understand the role that CPU plays and everything? There's two main reasons why you need type of CP. One is when you're doing reinforcement learning. The CPU is very critical to that. So before you would throw all the internet's data into the model, train it, and it spits some stuff out. Now you train all the world's internet, you put all the w internet data into the model, then you put it in this environment. This environment is like hey model. Try this out.

35:13 And it tries stuff out, tries a bunch of different things. And in the end there is an environment which whether or not what it tried out is successful and it grades it. And these environments can be anything. It can be hey, check if the text was output in the right way, structured output. It can be very simple stuff or can be very complex stuff. And people are starting to get into very complex things. Like hey

35:35 I want you to open this file, change it, edit it, update it, submit it to this website. I want you to open up this physics simulation from Siemens and edit this CAD model. So the environments can get more and more complex. And those environments run on CPUs. They don't run on GPUs, they don't run on ASICs. The ASICs run the model. That takes the input data from the environment. Runs it through the model, the model creates outputs of various different trajectories, ways that it think it could solve it in different instances. those trajectories are graded slash scored, and the ones that are successful you train on and you update and you iterate, iterate, iterate. And so CPUs are very useful for that one. And then once you have these great models,

36:14 And you're deploying them, those models are generating code. They're generating useful output. That useful output. It doesn't go from a GPU straight to the human brain. It goes from a GPU or an ASIC through to a deployed app that you're deploying somewhere that actually just runs on CPUs. That's another area where there's a lot of demand and things are sold out. In a large, large way.

36:35 As you continue to assess and try to be the world's best informed person on both the trajectory of supply and demand. What are things that you wish you knew? to make that understanding that you don't know. I think the hardest Area for us.

36:50 And for everyone is understanding Economics of tokens. I think we have a really tremendously good insight into how much it costs to run infrastructure, what the cost of tokens are, what the cost of models are, what the margins of these labs are. But the usage and adoption is what's really difficult to model. Continuously, right? January, we had crazy estimates for February. Anthropic smashed them. How do we calibrate this model? What are the data sources for this? February, we had crazy assumptions for March. I know people were like, You're crazy, Dylan. And then they smashed it.

37:19 Everyone sees the number of ten billion and they're like, What the f how do they add ten billion of revenue? Who is using all these tokens? Why are they using them? What are they building with them? And then more importantly. with what they're building with these tokens, how is that actually diffusing into the economy and what value is that generating? 'Cause it's not really something that you can capture in any

37:36 GDP statistic. all of the value of the tokens that I use get transformed into better information, which I then sell at a discount to what people used to sell information for, relatively. And therefore That information is now making its way throughout the economy and people are making better investment decisions or better competitive decisions if they're a semi-gar company or a data center company or hyperscaler.

37:57 What is the value of this? What has that done to the economy? It's clearly by every subjective metric amazing. But where is the phantom GDP? What is the phantom GDP? How do we track the real economic value because the GDP metrics are not. accurate if you were to say, what is the GDP that Dylan Patel is making? It's tiny.

38:17 compared to the value that I think is being created. I think you would say the same for Patrick. What is the value being created by these tokens? Not on a basis of simple What is the knock on effect of all the things that these things are doing? I think that's the real question and challenge. It's hard to measure. I think we've got a tremendous reading on the supply side of things. I think we've got a tremendous reading on even a lot of the demand side signals, but it's what is the value these tokens are generating? That's hard to quantify and measure.

38:43 I hope we get a chance to do this every three months because this changes so quickly. What do you think's gonna happen next? When I come back three months from now and we're in San Francisco together again. What do you expect? Large scale protests? Really? Yeah, I think there will be a large scale protest against Anthropic and up at AI. People hate AI.

39:00 AI is Less popular than ice. Less popular than politicians. With Enthropic are adding so much revenue, that's gonna start causing business changes downstream. People are gonna get more and more scared of AI. They'll start blaming more and more of their own problems and things that are global have been deep seated problems for a long time. Those will bubble up and be blamed on AI. probably some politician or some influencer will be able to start taking and weaponizing AI against people. You look at the comments of news articles where Sam Altman had a Molotov cocktail thrown at his house twice in two weeks.

39:35 People are cheering it on. And this is just the beginning. So I think we'll see large scale protests against AI in three months. What is the counterweight to that? How should the AI industry Head that off.

39:46 First of all, Sam Altman and Dario have to stop getting on interviews. They're so uncharismatic. I don't know what they're doing. Every interview they do is like Well, normal people are gonna hate you even more. Sam being on Tucker Carlson probably made all Republicans hate open AI. I'm just guessing.

40:01 Same with Dariet. I think that's first two they need to start showing uplifting things that can be done with AI. Three, they need to stop talking about how the capabilities are gonna change the whole world constantly.

40:12 Because then people are gonna get fear. of that capability'cause they have no connection. There's no connection to it either. The average person doesn't know an anthropic employee. The average person doesn't know an open AI employee. Average person doesn't know who these people are, what their goals are.

40:25 And they just view them as a sneaky cab all of five thousand people at this company, they're gonna change the world and automate all the jobs and destroy society. That's what they view it as. And as people who are funding the building of all these data centers and power plants that are gonna pollute the world, they don't quite understand what's happening. So They have to stop talking about the future thing that's gonna happen and only talk about present, how uplifting AI is. I think it's a huge reorgan rebranding that needs to be done. This is so much fun. I love doing this with you. Thanks for your time.

40:55 If you enjoyed this episode, visit Colossus.com. You'll find every episode of this podcast complete with hand edited transcripts. You can also subscribe to Colossus, our quarterly print, digital, and private audio publication featuring in-depth profiles of the founders, investors, and companies that we admire most. Lear more at Colossus.com slash subscribe.