20VC: Why Google Will Win the AI Arms Race & OpenAI Will Not | NVIDIA vs AMD: Who Wins and Why | The Future of Inference vs Training | The Economics of Compute & Why To Win You Must Have Product, Data & Compute with Steeve Morin @ ZML

28:35And there's two companies I see going in that direction. One is called etched and the other one is called the V -sore. That's the two I see. Because if you can deliver this at the, I would say, the price that is comparable to GPUs, you've won. is minimizing SRAM the only way to reduce unit cost on these chips. Really. It's hard to say. I mean, you need some SRAM, but if you can, you know, have a smaller process node, but if you can hook yourself with external memory, then yes, you can do that a little better. But the thing is, if you go like full blown SRAM, then, you know, there's no magic. You will have to pay the price.

29:14I'm so enjoying this. I'm also loving my notes here are just expanding by the day. If that's today, how do you think the inference market evolves over the next three to five years? Pushed by reasoning. So reasoning, not in the sense that you see on deep seek and whatever, right? Reasoning and what's called latent space reasoning. Latent space reasoning and agents will push the market towards a different types of compute. What's latent space reasoning? So the way model's reason today is the reason it was in tokens. So it's as if if you think to yourself you would say out loud what you're thinking.

29:53So yes, it works, but it's a bit inefficient, right? And you lose information doing this. Latent space reasoning is this without going, I would say to English or whatever, right? So staying, you know, and what's called the latent space, which is where all the information of an LLM, Let's say, an LLM. An LLM lives. So this is very much how we work as humans. And we move toward what Jan LeCancel's energy -based model, in which we have different types of longer or shorter, I would say, thinking times, if you will. So that fundamentally, GPUs cannot deliver this. Plain and simple. At scale. Why can't GPUs deliver it?

30:37Because the access to external memory prevents it. So HBM is all the rage, right? But HBM compared to S -RAM is absolutely, you know, dark slow. So this is the problem you get. So HBM is like the best we can do, but it's still slow versus versus S -RAM. So when I had Jonathan on, he was like, Ashi, in video, I have such a stronghold because they're one of the only buyers of HBM, and that gives them this unique position. Ashi is being a sole buyer of HBM irrelevant. if the world needs that RAM instead. No, you want HBM to be clear. No, SRAM, this will not deliver. It's a dead end in terms of scaling SRAM, mean scaling the surface, mean you get depreciating problems.

31:24It explodes everywhere, right? So you need some SRAM, right? So we'll have bigger amounts of SRAM into chips. And of course, bigger was called external memory into chips. The issue with HBM is that it's still slow and yes, maybe Nvidia has a strong hold and they can prevent you from getting some. So that would be like, I call it the Nutella situation in which Nutella owns 80 % of the azoneats market, right? So yes, you can do a competitor, but who with your buy the nuts from, right? So there will be a need for HBM, there will be a need for SRAM. I would say better or more dedicated architecture will be able to deliver these things.

32:04And then there's like the next frontier after that, which is called compute in memory. There's two companies that are on that market. One is called Rain, Rain .ai. Samaltman is one of the investors. There's no surprise. The other one is called Fractile. So this is an X -Rontier. The idea is that instead of transferring the data between external memory and the CPU and do the compute there, you actually bring the CPU to the memory and you do everything. it's crazy stuff, but it's coming. How does it? Maybe not this year, but... How does that change the situation? It makes it much more efficient, but what does that actually mean in reality?

32:44It means you get maybe not a SRAM level performance, but you get very a lot faster performance in terms of compute. So, and if you translate that to LLM's, let's say, you get much, much higher tokens per second. In ACE, InGhostream, which is exactly what you want, when you go into reasoning. You want your model to maybe think, let's say, for like half a second and then boom. You don't want to wait 50 seconds and you know, context switch to some other thing, which is the problem everybody has today, mind you. So yeah, I think inference will be pushed, the compute landscape will be pushed to change because of these two constraints.

33:20I know I'm working on it. If you would describe value between training and inference out of a pool of 100, is it 80 inference 20 training? What does that look like? In five years I would say 95 % inference, 5 % training. Do you think Nvidia owns both of those markets in five years time? Depends on the supply. I think that there is a shot that they don't. Because here's the thing, even if we take same amount of, let's imagine we have a new chip from Amazon, right? That is the same amount. Oh, wait, we do. It's called Tranium. You know, why would I pay 90 % margin of Nvidia if I can freely change to Tranium?

34:04My old production is runs on AWS anyways. And like if you run on the cloud and you're running on Nvidia, you're getting, you know, squeezed out of your money, right? So if you're on production on dedicated chips, of course, you know, so maybe through commoditization, but you know, Hey, I'm on AWS, I can just click and boom, it runs on AWS's chips. Who cares, right? I just run my model like I did two minutes ago. With that realization, do you think we'll see Nvidia move up, stash, and also move into the cloud model? They are. They have a protocol name sort of does that. The thing with Nvidia is that they spend a lot of energy making you care about stuff you shouldn't care about.

34:49And they were very successful. Like who gives a shit about CUDA? I'm sorry, but I don't want to care about that, right? I want to do my stuff and Nvidia got me into saying, hey, you should care about this because there's nothing else on the market Well, that's not true, but ultimately this is the GPU I have in my machine. So you know off I go if tomorrow that changes Why would I pay 90 % margin on my compute? That's insane. This is why I believe it ultimately goes through the software Because the software like if my entry this is my entry point to the ecosystem So if the software abstracts away those in your synchroces as they do on CPUs, right, then the providers will compete on specs and not on fake modes or circumstantial modes, right?

35:35So this is where I think the market is going and of course there's the availability problem. There is, if you piss off Jensen, you might need to kiss the ring to get back in line, right? But I mean ultimately this is I don't see this as being sustainable from we chatted before you said about AMD And I said hey, I bought Nvidia and I bought AMD and Nvidia thanks to Jensen I made a ton of money and AMD I'm up 1 % versus the 20 % gain You said that Nvidia AMD basically sold everything to Microsoft to matter and had a GTM problem Can you just unpack that for me? So all I would say chipmakers have a GTM problem all of them whether you know It's Google whether it's AMD whether it's 10 store the problem is is that there's I would say Probably two fundamental problems the the number one is if you maintaining multiple stacks today is very very very hard So you don't so let's say I buy you know AMD I want to buy AMD, right?

36:41That means I'm going to abandon Nvidia. Oh crap, you know, I have a six year amountization plan on that. Oh man, what do I do? So do I need to support both stacks? And clear maybe and until AMD tells me, hey, you know, you have, I don't know, let's say a thousand and video GPUs You're about to buy a hundred thousand of AMD. I mean, come on, right? And I'm like, okay, that is you know makes it worse my while, right? But that is ultimately the fundamental problem is that the steps are very high, right? I need to have a lot of incentives to buy into the ecosystem. So I need to buy a lot of them. So if you're AMD, that is already a problem.

37:19But then Microsoft comes along and buys it all, makes by the way, OpenAI, or at least on the inference side, puts OpenAI in the green because of the efficiency gains. I'm just showing on the sunset. You're saying the switching costs are really high from one provider to another. Oh, yeah, absolutely. Which is why you don't or are you saying that to get into one of these bi -processes, you have to buy so much that it prohibits you. It's actually both. The buy -in is very high. So to make it worth it, you have to buy a lot. And if you buy a lot, this is what we talk to all of them. They always have the same questions and it's completely understandable.

37:57They say, this is great. But who's the customer? Because on the other side, let's take Amazon, for instance, with Ranium. Apple just came and said, hey, we're gonna buy 100 ,000 of them. So you want to buy, you know, 10 ,000, you feel like the big shot, right? Yeah, but go back to the queue because there's Apple before you, right? So they have to have very high commitments. You cannot be incrementally better. It's very hard, right? And also very hard. I can give you a, I can give you one metric if you want. I know for a fact that being seven times better and And whatever, take whatever metric you want, whether it's spend, whether it's whatever, is not enough to get people to switch.

38:37People will choose nothing over something. So this is a very hard market to enter into because you cannot also compete of incremental gains. It's very hard, right? You have to convince a lot of people. Maybe you can go the Middle East route in which they sprinkle everything and they evaluate everything. That's not, you know, very sustainable, I would say, strategy in the long term, but at least in the long term. What is the right sustainable strategy then? You don't want to go so heavy that you can't ever get out and you have that switching course. Right. But you also don't want to sprinkle it around and do, as you said, absolutely.

39:12The right approach to me is making the buying zero. If the buying is zero, you don't worry about this. You just buy whatever is best today. How do you do that by renting? Oh, because this is what we do. This is our promise. Artisus is that if the buy -in is zero, you completely unlock that value. Because you're free. When you say the buy -in is zero, what does that actually mean? It means that you can freely switch to compute freely. You just say, hey, now it's AMD, boom it runs. You just say, oh, it's 10 store and boom it runs. How do you need it then? Do you have agreements with all the different providers?

39:50Oh, yeah, yeah, yeah. Not agreements, but we work with them. to support dirt, dirt chips. But the thing is, my at least as, you know, I would say a user myself of, you know, our Varteck is that if it's free for me to switch or to choose whichever provider I want in terms of compute, right? AMD and VD, whatever. Then I can take whatever is best today and I can take whatever is best tomorrow and I can run both. I can run three different platforms at the same time. I don't care. I only run, you know, what is good at the moment. And that inlocks to me a very cool thing, which is incremental, you know, improvement.

40:26If you are 30 % better, I'll switch to you. So are you taking the risk on those on that hardware then? If you're the one providing them to turn off and on, on demand, provisioning, you name it, you take the risk. This is actually a great question. I think that if you are doing it bottom up and for two applications, you will lose because nobody will care. As they don't today, right? If you look at TPUs, they're available, they're great. Nobody cares. Why does nobody care about TPUs? Because the cost of buying, it's always the same, right? You have to spend six months of engineering to switch to TPUs.

41:02And mind you, TPUs do training. They're the only ones with training now. But AMD can do training, but it's also, but in terms of maturity, the by far the most mature software and compute is TPUs and then it's Nvidia. Right? So the buy -in is so high that people are like, we'll see, right? I'm not on Google Cloud, I have to sign up. Oh my God, right? So these are tremendous chips, these are tremendous assets. Now in terms of the risk, I think if you want to do it, you have to do it top to bottom. You have to start with whatever it is you're going to build and then permeate downwards into the infrastructure.

41:41take for example, Microsoft with OpenAI, they just bought all of AMD supply and they run you know, ChadGPT on it. That's it and that puts them in the green. That's actually what make them, you know, preferable on inference. So, or at least let's say not lose money, right? I'm sorry, how does Microsoft buying all of AMD supply make them not lose money on inference? Just help me understand. Because I can give you like actual numbers. If you run 8, 800. You can put two 7 TB models on them because of the RAM. That's number one. Number two is, if you go from one GPU to two, you don't get twice the performance.

42:19Maybe you get 10 % better performance. Yeah, that's the dirty secret nobody talks about. I'm talking inference, right? So you go from let's say 100 to 110 by doubling the amount of GPUs. That is insane. So you'd rather have two by one, then one by two. So with one machine of 8, 800, you can run 270b's model. If you do four GPUs and four GPUs, that's number one. If you run on AMD, well, there's enough memory inside the GPU to run one model per cal. So you get AGPUs eight times the throughput. When on the other end you get AGPUs to maybe two and a half times the throughput. So that is, you know, a Forex right there.

43:04Just, you know, by virtue of this, of this. So that is, you know, the the compute part. But if you look at all of these things, there are tremendous amount of, you know, we talk to companies who have chips of coming with almost 300 gigabytes of memory on it, right? So that is more, you know, a model, like one chip per model. This is the best thing you want. If you want 70b's, right? So which is what, But I would say not the state of the art, but this is the regular stuff people will use for serving. So if you look top to bottom and you know what you're going to build with them, then it's a lot better to do the efficiency gains because four times is a big deal.

43:42And bind you, these chips are 30 % cheaper than Nvidia's. It's like a no -brainer. But if you go bottom up and say I'm going to rent them out, people will not rent them. Simple. So that's why, you know, I think it's a good way to attack it from the software because ultimately, do you really care about that your MacBook, let's say, is an M2 or an M3? It's like, oh, it's the better one. And that's it, right? And imagine if you had to care about these things, that would be insane. When I listen to you now, I'm like, shit, I should sell my Nvidia and buy more AM. Okay. If you were forced to buy one, I'm not saying sell the other, I'm not saying like this on the other, buy one, which would you buy and why?

44:27Stock? Yeah. I used to think the market was efficient, so probably I would go today at least, I would go within video still, because the supply. But you know, if we play our cards right, we ship our stuff, hopefully I will come back and tell you to buy AMD as much as you can, or a 10 store, you know, if they go public or whoever else. These chips are amazing, by the way. What does everyone think they know about inference that they actually don't? Or what does everyone get wrong about inference? Probably not a lot of people are accustomed to what it entails to run production. So that inference is production and production is hard.

45:09Somebody has to wake up at night. And I used to be that guy, right? I don't want to do it again. So production is hard. Thankfully, we have a lot of software nowadays to do that a lot better But there's not a lot of reuse because the AI field at least is not really a custom to that yet It's changing, but you know the discussions I had you know a year ago and the discussions I had today are not the same They're going into the right direction, but they're not there exactly yet So it's probably that would be the number one thing that is only you know training code running only for what best right?

45:45This is not what it is. Can I say how do you evaluate the data center investment we're seeing being made? You know when you look at Facebook doing 60 to 65 Microsoft doing 80 and some of the intense CapEx expenditure that you're seeing how do you think about that on the data center side? I mean, there's still going after training. So there's still this frontier probably. It's why also and VDIS is the better buy right now. Because on the Nvidia side, if you do training, it's incremental. If you have bought a thousand Nvidia GPUs and you buy a thousand new Nvidia GPUs that gives you 2000 GPUs, right?

46:19But if you buy a thousand and a thousand AMD that gives you twice a thousand, right? It's a bit different. So they're still going after training, definitely, and they're very pragmatic in doing so. But I mean, they have the KAPX to spend. They're not making their money out of it probably. The only one, by the way, that owns their compute are Google. There's like this triangle of, I would say, of wind that I, this is my mental model, mind you. You have the products, the data, and the compute. Who has all three? And you get everything flows from there. Product is data compute. He has all three. Google, Amazon.

46:55Amazon, they don't have products. They have Amazon, right? They have AWS, but they don't have actual products. Google, that's like, you know, Android, Google Docs, whatever they have, everything, they can sprinkle everywhere. This is the sleeping giant in my mind. If they're not busy doing a rig, they might... It's fascinating, because everyone, if you are a shallow thinker, you think the OpenAI challenges that Golden Goose, which is such a Googleist threat all the never -know. Well, I mean, OpenAI is amazing, but it's not their compute. It is Microsoft's compute. And if you own your compute, you own your margin is essentially what you're saying.

47:32Yeah. Microsoft, even Microsoft, they bought, when they weren't running Nvidia, they bought Nvidia, you know, some outrageous margins. I talk to a lot of people that build the dissenters and I tell them, you know, they're mind you, these people like by tens of thousands of GPUs. And I ask them, hey, do you get at least a discount or something? And they're like, no, the only thing we get is the supply. So, I mean, ultimately, if you don't own your compute, you're starting with, you know, something at your ankle, definitely. And so, this is why I like to think in this, like, this triangle product data compute.

48:07And you can see where everybody sits and their weaknesses and their strengths. Can I ask you if we move a little bit? You said, it's totally rational that everyone's focusing on training still. When we think about that, it's rational if you think that efficiency and scaling will continue to continue play such emphasis on it. How do you think about model scaling and scaling more coming into place? How do you think about it? There's like a brute force approach to this. It is a very American approach. More and more but the thing is you look at for instance the XAI cluster. It's not a hundred thousand GPUs.

48:43It is four times twenty five thousand. You're starting, you know, see some because in Finni band and Nikis Rocky, which is anyways, the technology they used to bridge their GPUs together, you have a profound, right? At some point you're fighting physics. So you can push, it's like, you know, trying to get to the speed of light. As you approach it, the amount of energy you need is a lot higher and a lot higher and it grows and grows. So there's two, I would say, counter to that would be that number one is we still scale, but there's a lot of waste and and access, you know, spending on the engineering side, which is the deep seek approach, right?

49:21Very successful at that, mind you. They said, yeah, if we do this and this differently, then we get, you know, multiple sometimes, right? So virtually you increase your compute capacity because you're more efficient. And the other approach is Jan's, Jan LeCanel's approach, which is, this is not scaling, and at some point, you need, we need to look at the problem in the face and do something better, right? So of course we push and push, push, because there's capital still, but I'm more of these two approaches. I think you can do more with less. At what point do we stop and say, hey, there is a lot of wastage and we could do more better.

49:58I think until somebody does it, DeepSeek was a good wake -up call, right? Suddenly, efficiency is in. That's number one and number two is until there's a new architecture that comes out and changes the game. So in the case of a LLM, for instance, you have these what's called non -transformer models that changes fundamentally the compute requirements. So that might be a frontier that completely absolutes the transformers. And if the transformers are the, I would say, the building block by which current model work, right? So the way they work is that for each token or a syllable, if you will, the model will look at everything behind it.

50:36So you can see that as you add more text, you have more work to do. So there are these new architectures that do not require this that might change, you know, these things and probably shift the amount of compute needed to do training or to do inference. And then there's the new thing which is Yann's thesis, which is the word model. As in LLM's are at the end, what we need is something that understands the world fundamentally, and this is it's JEPA thesis, it's called. I'm very bullish on this, but it's very frontier. Why you bullish on it and why is it so frontier? Because it's young look on.

51:14It's hard to. He's no bullshit, right? So he explained to me how it worked and I was blown away. But it makes a lot of sense. We are creeped out because the machine talks back to us. But it's not a new thing, right? It used to, you know, this is not new technology when it came out, like when it exploded, it was a new technology. But suddenly it was talking back and that freaked us out and we got crazy on it, right? But language is one form of communication, but it is ultimately a very narrow window into the world. We use it to describe the world, arguably with some loss, right? And so the JIPA approach is, long story short, is that you have essentially two things you want to do and you try and minimize the energy to do them.

51:59And from this understanding emerges, physics emerges, etc. Because you're trying to minimize the amount of energy to go from one state to the other. And that actually makes sense. Like if you try and pick this air pod case, I'm not going to go around trip around the block to get it, right? I just get it. And in my brain it's wired to just do the thing. If I go and talk to myself out loud, put the hand down, move to the left and whatever, that feels very inefficient. So probably this will be something that changes. And in the case of the LLMS, there's good work also on what's called diffusion -based LLMS, which means like instead of thinking what's called autoregressively, that means you get a new token, you re -inject, and you redo, et cetera.

52:46They think more like what we do, which is in patches, right? Imagine a paragraph of text and words appear until it's done. Is distillation wrong? And if we're all progressively moving towards a better future for humanity, more efficient models, it's dilation, not effectively open -source in another round -per. I think it's fair game to be honest. I will not share the tier. It's fair game. If you, there were some people who tried to ask, I don't remember if it was an OpenAI model, so a diffusion model image, right? They asked it to generate an image from a Star Wars movie at whatever time stamp and it came out with the Star Wars movie, you know, screenshot.

53:30Obviously it was strained with it. I think it's fair game because there's no free lunch, right? It was strained with data. You had a good ride, somebody was sneaky and took it, but you took it from the beginning too. So let's just accept it's fair game. And you can also learn from their advancements. Absolutely, absolutely. I take my cup and enjoy it very much that movie every single day. You mentioned the training there. Obviously data and data quality dictates a lot of training ability. When you think about the future of data that feeds into training, how do you think about how that will be between synthetic data versus real data?

54:15I'm a bit split on this. There's a part of me that said that if you re -inject data into the system, the system deteriorates. That feels a bit, I would say, intuitive, but if you look at AlphaGo for instance, the moment it's ramped up in its skills is when they started generating games, synthetic games, right? So I'm a bit, you know, split, but there are some verticals that very much benefit from this code, LLM, for instance. We can run code, right? So this is the poolside thesis. Just so I understand why it is it work for coding and not for other things. Because you don't use the mod, the AI model to generate output.

54:53You use the machine, you just run the code. And you see what it makes and you run all these code and you create data out of it. Whereas if you run the LLM and you say to an LLM, or I generate me two trillion tokens of text, it will do it with it's, you know, you may inject and stuff. So there's a lot of tricks, but ultimately my guts tell me that it feels wrong, right? Because you re -inject Data that I was there and so it will detail you right there's you know, there's loss So yeah, I'm a bit bullish. I'm not sure exactly on what vertical code is one We'll see distillation is in some sense a bit like that you create synthetic data from a bigger model into a smaller one Probably the most, I would say mind blowing thing about distillation is that sometimes the smaller models become better than the bigger model through distillation.

55:45So all the models become better than bigger models purely because of the quality of the data that's inputted through them. The one theory is that the smaller model is better at generating output that you would want it to generate essentially. It's not better in the general sense. It's better at the task at which you were measuring it. This is what it learned to to imitate. How do you think about the future in terms of large monolithic models versus more dynamic architectures smaller models? Sometimes it's wasteful to run big models a lot of times. It's actually wasteful to run big models I think there's going to be a lot of smaller models for efficiency reasons, but there's a there's a but which is you talk to people at DeepMine and they don't even find tune anymore.

56:35Because they have such what's called big context window, which is what the data at the model you inject at runtime, that nowadays they just dump data into it and just say, do whatever that data tells you to do instead of fine -tuning as we used to do. So if the efficiency gains were not there yet, but if the efficiency gains, I would say pass that threshold, we'll just do it at runtime. We'll just have a great model that we'll just specialize at each request. But that's not for tomorrow, I think. What is retrieval augmented generation fast? It's a very clever trick. What you do is you represent knowledge into what's called the vector space or latent space.

57:23And what you do is through what's called vector search. So imagine you have, Let's say a 3D space that represents all knowledge, all of everything. Let's say a cat sit here, a dog sit close because it's an animal, but it's far from some other property and so on. What you do is you run the user's request through this same system. It's called an embedding. That will give you a vector and you will take whatever is closer to you. What's called symmetrically close. And then it's actually very clever. You actually insert those pieces of text before the request. So it's as if you would say knowing the following and you give the data, let's say it's law or whatever, please answer my request.

58:12And that's it. So that's a bit of a clever trick. It's a bit dirty because of course, you know, you are limited by the amount of data you can input. right? So there's this problem in which how do you chunk, you know, the data that you input? A lot of things we do not retrieve relevant generation, and when we say here's a link, summarize it to the key points. Is that not right? Because we're inputting the data. It is. It is. It depends on how it works, but yes, sometimes it is. But think of it as in, it's like a preamble to your question. Knowing the following, and the following is a tiny window So into the content, please answer my question.

58:52And of course, as you talk more and more, it will forget because that window is fixed. And how does that shift the movement from large generalized model to smaller, more advanced models? What pushes smaller models are efficiency, roughly speed. You know, less is better. So if we can do it less, then less it is. Simple as this, right? In terms of RAG, the key frontier is what we call attention level search. But this is something we're working on. You have the exclusivity now, I'm putting it out there. It doesn't push, I would say, model sizes. What really pushes model sizes are the efficiency rather than specializing.

59:29Meaning that if you can do the same performance with a smaller model that is fine -tuned, with a rag or whatever, then you'll do it with a smaller because, again, less is better. Kashi, before we move into a great far -run, I just want to ask you, when we had deep seek as we mentioned, And to what extent were you surprised that such innovation, I would argue, and I think many would agree with me, came from a Chinese competitor, not from a Western competitor. Oh, I love it. Constraint is the mother of innovation. Yes, we can, we can, you know, troll a bit about, you know, the Singapore, you know, grey market and all of these things.

1:00:07But ultimately, like, they had no choice. Here's the thing. If you can buy more, why would you give a damn? You can just buy more. So if you are pushed to efficiency, then you will deliver efficiency. These are very, very skilled people. This is the coolest thing to me about AI, honestly, is the geography doesn't matter anymore. You can just do things. You appear out of nowhere, boom. You're on the map. So I'm very, very glad that they did. I found the reaction very entertaining, to be honest. So yeah, I mean, constraint is a very good driver of efficiency. Do you think it is a meaningful threat to open AI and chat GPT?

1:00:45Bluntly, they still have the consumer loyalty, the consumer brand. To what extent is it actually a long -term threat? I'm not sure who is a threat to open AI at the moment. Here's why. You look at the numbers. I mean, we live in a bubble. We follow every new episode, the whatever new model, whatever who said who it and so on. But I go to my mother and I ask her, you know, do you know chat GPT? and she says, yes, and you know, I don't know. I don't want to done anybody, but do you want to know some other model? And she says, what it is? What is it? Right? Even Gemini, right? Like a Google, right?

1:01:18So they have a strong brand. They have a strong product, but there's a balance between the product and the models, honestly. So this is Gary from fruit stack, actually, who told me that is mental model in terms of model providers. It'll be like car makers. There's no winner tickle. Everybody will have their own, because ultimately, also human knowledge is Everybody has everything. So we're converging. But I liked the analogy. Yes, deep seek made a very good, you know, made waves, but it was it was, you know, waves that were amplified by the media and the narrative and the drama. Do you think export regulations inhibit China's ability to compete in any way?

1:01:58Today, maybe, tomorrow, I'm not sure. They're a bit late in terms of, you know, AZIC. They are like A100 level, but they have probably, I would say, one of their unfair advantage is that it's like, you know, when you do exercise in the water, right? It's like this, so this is their state. They are constrained, so they are bound to do better. They can just not buy their way into better compute. So I think it hinders their success, Yes, but I think it's short term, do you think that way? Are you fearful that Europe are going to regulate ourselves into constraints in a world of AI? No, I don't care.

1:02:40This is something that makes me wonder sometimes, I understand the narrative and so on, but I am absolutely not fearful. Let's be successful first and then we'll talk about the politics. I have so far, but again, I'm not mistral. I'm not, you know, I'm not building gigawatt that is the tensenters and so on. So if you build gigawatt that the centers you're running through these problems, but maybe you're running through these problems. But the thing is, if you're successful, everything flows from there. From there. Steve, I'm being direct here, but I'm asking you for the pros. Everyone says, Mr.

1:03:12Al, just doesn't have enough money to compete. That is kind of word on the street. To what extent is that fair? There are very competent. I think it's easy to spread further. There's a lot of fun going around, especially about regulation and everything. But here's the thing, I look around me and I don't see what I read. So I am hardly convinced about everybody was saying that they were dead and boom, they came out with their release and it was insane. So what I know is that I hope they don't have too much money. That's for sure. You want to be clever, right? Final one if we do a quick five, so enjoyed this Steve.

1:03:49Final one if we do it. where Stalgate was a $500 billion dollar announcement. How did you evaluate that? My first impression was that I don't buy it. I would say, you know, American start, right? You start with the claim and we'll figure it out later. I don't buy it. And ultimately, ultimately, I'm not sure I cared that much about it. Let's imagine it's true, right? Congratulations, amazing. But it is more of the same. It is a vertical scaling. And as you know, my days are spent on efficiency. So I look at these things as being like, all right, this is a bigger, you know, this is an American car of AI It's big it consumes a lot of gas, but ultimately, you know, it's not a good car I think there has to be you know sufficient capital, but at some point I'm not sure it is really a differentiator That was piled to deep seek then deep seek came there was always you know my thesis, but you know you need money You need infrastructure, you need, but what is ultimately the probably the two limiting factor today is talent and energy?

1:04:54That's it. The rest, you know, yes, of course you can buy 500 billion of GPUs. I by the way, 90 % margin. So if we work on that margin, we can you know, shrink that number probably. So I'm not easily entertained by these numbers. I've seen how the sausage is made way too many times. Dude, I want to do a quick fire with you. So I say a short statement, you give me your immediate. Sure. If you had to bet on one major shift in AI infrastructure over the next five years, what would it be? Oh, yeah, latency, reasoning, definitely. This year, what does that mean? So the shift from throughput, so how speed my answers to how long it takes for my answer complete to appear.

1:05:38That is probably one of the fundamental, like this year, right? Longer term, I'm very rooting for non -transformer models that will change the compute also landscape. And of course, you know, world models, right? Yes. And or energy -based models. What's one piece of advice you give to AI startups navigating the changing landscape of training inference in hardware? Probably the number one thing I would say is do not resell compute if you can. A lot of AI startups that are building on top of AI are trying to make a margin on top of a very big cake and Ultimately what they sell is compute if you look at the dollar of spend, you know for one dollar of spend Maybe 98 % of it goes to somebody else's margin So if you do AI as much as you can try to verticalize on on the product But not on the compute if you are you know if your business model implies buying a lot of tokens It's a very hard circle to square to put that into $20 a month.

1:06:42I always say, please look at it from that angle. If you can, try and avoid it. What's the biggest challenge that Jensen Huang faces today? The highs are very high, but they don't last forever. So probably it's how to navigate the downslope. Blackwell is probably something that keeps him awake at night. Why would that keep him awake at night? with that not re -energize him, more orders, new enthusiasm, new product, baby. Because orders are getting canceled. Why are they getting canceled? They have a lot of problems with this chips. So a lot of people are canceling their orders. These chips are like on the frontier of scaling.

1:07:23And so they were supposed to come out last summer. But that heat dissipation and matter bending problem, it used to be called the people who are very privy to Silicon told me, this is what we call a pretty big fricking problem. And quote, probably how to navigate the downslope. Maybe you don't know, but the supply of H100 was actually a smooth out over the year so that they decided that they didn't have like a big spike in deliveries and then a quarter or less, right? Which pissed a lot of people, mind you, who bought a lot of them. Some of them even haven't received their order from last year.

1:08:00And they already see like the new chip, the B200, and then the one after, you know, and they're super pissed. There will be a downslope at some point. The question is, you know, when, how, like if there's like the H100 bubble, of course, it will impact Nvidia. But Blackwell is, I'm probably going to get a lot of flag for this, but, you know, I've seen some very worrying numbers about it and varying testimonies about people who operate these things, right? So that ride will stop. or at least, you know, slow down. Steve, I'm not sure I've ever learned quite as much in one episode. Seriously, we said before.

1:08:38Oh, wow. No, I love what I do because I'm able to ask anything to the smartest people in their business. And I so appreciate you unpacking so much with me today, man. I'm thrilled to say that I actually finally get what you do after years. That's the thing. That's the thing. But you've been a star. So thank you, man. Thank you, appreciated. Thank you. I mean, I said it there. I think I learned more in that episode than I have done in the last 1000 when it comes to technical specs and the future of AI. Steve was incredible. If you want to watch the episode you can find it on YouTube by searching for 20VC, that's 20VC.

1:09:17But before we leave you today, turning your back of a napkin idea into a billion -dollar start -up requires countless hours of collaboration and teamwork. It can be really difficult to build a team that's aligned on everything from values to workflow. But that's exactly what Coda was made to do. Coda is an all -in -one collaborative workspace that started as a napkin sketch. Now, just five years since launching in beta, Coda has helped 50 ,000 teams all over the world get on the same page. Now at 20VC, we've used Coda to bring structure to our content planning and episode prep, and it's made a huge difference.

1:09:53Instead of bouncing between different tools, we can keep everything from guest research to scheduling and notes all in one place, which saves us so much time. With Cody you get the flexibility of docs, the structure of spreadsheets and the power of applications, all built for enterprise, and it's got the intelligence of AI which makes it even more awesome. If you restart up team looking to increase alignment and agility, Cody can help you move from planning to execution in record time. To try it for yourself, go to coder .io slash 2 -0 VC today and get 6 free months of the team plan for startups.

1:10:27That's coder .io slash 2 -0 VC to get started for free and get 6 free months of the team plan. Now that your team is aligned in collaborating, let's tackle those messy expense reports. You know, those receipts that seem to multiply like rabbits in your wallet, the endless email chains asking, can you approve this? Don't even get me started on a month and planet when you realise you have to reconcile it all. Or Plio offers smart company cards, physical, virtual and vendor specific, so teams can buy what they need while finance stays in control. Automate your expense reports, process invoices seamlessly and manage reimbursements effortlessly, all in one platform.

1:11:06With integrations to tools like Zero, QuickBooks and NetSuite, Plio fits right into your workflow, saving time and giving you full visibility over every entity, payment and subscription. Join over 37 ,000 companies already using PLEO to streamline their finances. Try PLEO today. It's like magic, but with fewer rabbits, find out more at pleo .io -4 -slash20vc. And don't forget to revolutionize how your team works together. Rome, a company of tomorrow runs at hyper speed. With quick drop -in meetings, a company of tomorrow is globally distributed and fully digitized. A company of tomorrow instantly connects human and AI workers.

1:11:46A company of tomorrow is in a Rome virtual office. See a visualization of your whole company, the live presence, the drop -in meetings, the AI summaries, the chats. It's an incredible view to see. Rome is a breakthrough workplace experience loved by over 500 companies of tomorrow for a fraction of the cost of Zoom and Slack. Visit Rome, that's O -R .am for an instant demo of Rome today. Nobody knows what the future holds, but I do know this. It's going to be built in a Rome virtual office, hopefully by you. That's Rome -R -O .am for an instant demo. As always, I so appreciate all your support and stay tuned for an incredible episode coming on Wednesday with Oscar, found at Glovo, on turning Glovo into a $2 billion business.