All right, everybody. Welcome back to the number one podcast in the world. It's your favorite podcast. It's your podcast's favorite podcast. It's the All-In podcast. Episode 279. With me, Fredberg, Sax, Chimath, you know the squad. You know the squad. We're here. It's the summer and we're ready to rock and roll. We got a power docket. We got a rocket docket. Palanteer and Nvidia have announced a sovereign AI partnership. Where have we heard that term before? Palanteer is going to use Nvidia's Neotron. Neotron like the Pixar film open models to build a custom frontier quality model to serve the US government. Palanteer is calling this new platform sovereign AI operating system. The US government agencies will own the hardware, the data, and the model waits. Palanteer also shared a viral tweet manifesto laying out the concept, "Data retention is your treasure. Transfer it at your own peril." Transferring that data hands over access to your pre-existing winning plays and yields the means of production for new ones. CEO Alex Karp went on CNBC to announce the partnership in a classic Karp Robin Williams style monologue. Here's a clip from his 20inut interview where he where he basically went after the frontier models like Anthropic. Play the clip. Our clients are just to say they're unhappy with the Frontier Labs is to say I'm welcome at the Berkeley faculty. It's like there's just a level of discomfort and loss of trust. Sam and and and Daario, there's nothing more fun than debating Daario in private. So this is I'm not throwing shade at them, but something has gone completely wrong. And the basic view among enterprises in this country is I'm going to chill lax uh and waste my time with tokens. I'm going to get no value and they're going to get my IP. When the do department of war goes to you and says I need this application, do they get to control the weights to do it or do you get to control the weights? Are we really going to outsource the battlefield of this country to the consensus view in Silicon Valley? That is effing insane.

All right, everybody. Welcome back to the number one podcast in the world. It's your favorite podcast. It's your podcast's favorite podcast. It's the All-In podcast. Episode 279. With me, Fredberg, Sax, Chimath, you know the squad. You know the squad. We're here. It's the summer and we're ready to rock and roll. We got a power docket. We got a rocket docket. Palanteer and Nvidia have announced a sovereign AI partnership. Where have we heard that term before? Palanteer is going to use Nvidia's Neotron. Neotron like the Pixar film open models to build a custom frontier quality model to serve the US government. Palanteer is calling this new platform sovereign AI operating system. The US government agencies will own the hardware, the data, and the model waits. Palanteer also shared a viral tweet manifesto laying out the concept, "Data retention is your treasure. Transfer it at your own peril." Transferring that data hands over access to your pre-existing winning plays and yields the means of production for new ones. CEO Alex Karp went on CNBC to announce the partnership in a classic Karp Robin Williams style monologue. Here's a clip from his 20inut interview where he where he basically went after the frontier models like Anthropic. Play the clip. Our clients are just to say they're unhappy with the Frontier Labs is to say I'm welcome at the Berkeley faculty. It's like there's just a level of discomfort and loss of trust. Sam and and and Daario, there's nothing more fun than debating Daario in private. So this is I'm not throwing shade at them, but something has gone completely wrong. And the basic view among enterprises in this country is I'm going to chill lax uh and waste my time with tokens. I'm going to get no value and they're going to get my IP. When the do department of war goes to you and says I need this application, do they get to control the weights to do it or do you get to control the weights? Are we really going to outsource the battlefield of this country to the consensus view in Silicon Valley? That is effing insane.

And so some folks uh refer to this as a televised nervous breakdown. We here at Allink call that Alex Garp on a Tuesday. And we talked about this Chimath uh Sax Freedber. We talked about this a whole bunch back in February. I coined the term intelligence sovereignty here. Here's your victory fact. Do I want to give all of the secrets in our organization, every piece of intellectual property to Sam Alman who's got to make a billion dollars a year to keep up with his spend, right? He's going to build every application. I've been talking about AI sovereignty here for a bit just in terms of how much more cost effective it is and how you're not training other people's AIs with your knowledge and your insights. This is why it's super important that open-source open-source agents and local hardware be able to run these models and that consumers and companies learn how to roll their own language models. The intelligent sovereignty is different than privacy. Privacy is oh, you can't see my photos. You can't peek into my notes app and what I wrote there in my journal. Intelligent sovereignty is you can't tell me what to think. You can't use your AI to analyze my photos, to analyze my emails, to analyze my messages, and tell me how to interpret the world. That's actually going to be the next key piece here. All right. And uh Sax, you are in your long post era. Another long post this week from you on this very topic. And obviously as uh AISAR there for the first half of the Trump administration, you uh been very involved and very close to this. I would love to hear inside I'd love to hear your take on this and your long posts over 300 words. You can follow x.com/davidsax, but also like the palace intrigue here and what this means in terms of the relationship with the government, which we're going to get into in our second story. Well, look, Jay Cal, I got to give you some credit there. The first part of your take was spot on. After that, it was kind of diminishing returns. I'm not sure why we had to listen to the next 30 seconds of it, but anyway, uh, it started off really strong. But let's go back to the supposed crash out by Karp on CNBC. It was nothing of the sort. It was all these legacy media types making that claim. And that's the first clue that he's actually saying something insightful and maybe kind of brilliant. And I think the thing that he said that I hadn't really thought about in quite those terms is he started talking about AI safety in the enterprise and what that really looks like. And what he said is that what technical customers want is control over their compute, their models, their data stack and their alpha, meaning their proprietary knowledge. They want to know they own the means of production, he said, and it's not being transferred to someone else. And what he's referring to there is that these enterprises are at risk of transferring their knowledge, their knowhow, their trade secrets, their customer data to these model providers who might eventually decide to compete with them. Like you said, JCAL and you can see that enterprises are waking up to this threat and they're not happy about it. And I think KARP is exactly right about that. Now, I think this is a really interesting take on AI safety because what safety means for an enterprise is again that they get to control their own data, their model weights, their compute. So, a frontier lab can't hoover up their proprietary knowledge, their alpha and turn it into their next product. And if you don't think that can happen, just look at what happened to Figma. So according to the information, Anthropic quote unquote blindsided its then business partner with the launch of Claude Design. So this was a new vertical app that Anthropic launched to compete in the design category. And Figma's founder said that Anthropic had not been completely honest with them. Anthropic chief product officer had actually even served on Figma's board and didn't resign until 3 days before the launch of Claude Design. So obviously, Figma again felt blindsided by this and you can see the resulting impact on their stock price. Figma's stock has fallen something like 50% this year while Anthropic's valuation has surged. This is not an isolated example. Anthropic has also launched Cloud Science, Claude Security, Claude Legal, Claude Financial, and of course Claude Code. And every single one of these vertical apps expanded into categories that was previously served by companies building on top of Anthropic's own models. And really, if you want to go back to when Anthropic's revenue explosion began, it was with the launch of Claude Code. And how did they know to launch that product? Because they saw that Cursor was doing extremely well. Cursor was one of their biggest customers. They created the coding assistant first. they created that category and anthropic said oh like why don't we vertically integrate so in other words they're watching where the value is being created on top of their models then they're moving in directly and this is a formula that I think is very Microsoft like you could say it's very Google like they want to dominate the model layer you could call that the operating system and then use that position that monopolistic position to capture the most lucrative verticals so if you want to think about like the Microsoft example they had the Windows monopoly and then systematically they went and dominated every lucrative category of business software. It started with spreadsheets and word processing and then eventually it went to the browser so forth. If you want to look at Google, they basically had a monopoly or dominant position in search. And if you go back to the early days of Google, the search results picked you off site. And in fact, they really pride themselves on how quickly they could send you off site. But gradually over time, they use that traffic to tell themselves where to build properties. And today, fewer than half of searches kick you off site. You stay on Google properties. And I think something similar is happening with Anthropic here。 the pattern is clear，他们将利用他们在模型中的主导地位，然后获取任何有趣且有利可图垂直领域中更多的领土。

Speaker A: again, back to Alex Karp's point, if you're an enterprise customer or a developer, why in the world would you ever want to share any proprietary data with them? You are mortgaging your future. You are sealing your fate. You are going to lead to disaster for your company. Just one last point then I'll turn over JCL is that Daario at the same time that they pursue this business strategy has been arguing that open-source models are dangerous and need to be restricted. Well, dangerous to whom? Not to enterprises that want to retain control over their data.

Speaker B: It's dangerous to his business model because his business model requires that customers don't have a lot of choice at the model layer. And what Karp is pointing out here is that if you want to have true AI safety as an enterprise, you have to retain the ability to choose at the model layer who gets to see and use your alpha.

Speaker A: Yeah, this is well said. I think you picked that carcass to the bone a bit, but Chimath, you're actually doing specific examples of this uh at 8090. You've been testing uh some of the open source models. I saw you share that on Twitter X. So maybe you could give us a little feedback on what you've learned as uh the CEO of 8090 and you know your firsthand experience now with using open source for the first time in the last couple of weeks for this specific use case in enterprises. When you start a company I mean you guys all know this you're not starting it for the moment that exists today. you're almost sort of trying to forecast if such and such a set of things happen then here is the scene that that gets created because it takes time to build something and it takes time to get enough reps to know what you're doing in your go to market.

Speaker B: This for me was the moment that I thought would arrive which is the point where everybody wakes up and realizes wait hold on a second two things are true. The first is that my business is complicated. I want AI to be able to accelerate it, but I want to be able to protect myself in doing so. And then the second is I want the flexibility where there's an independent third party control plane that I use to get all these benefits so that I don't leak and seed my advantages away. And I think Alex is an incredible, smart, brilliant guy and he completely nailed it. And I think he called out on its face the huge risk of this. So let me just give you this narrative in three tweets. The first one is I read this really interesting study from BCG and what they looked at was the return on capital employed or ROCE of various businesses. And this is what's incredible. The cost of capital has now with long-term rates moved back to what its long run average is which is around 8 to 11%. What that means is like that is the actual cost that you would borrow money at effectively. The problem is that half of large US companies now cannot deliver returns that exceed that. That is a really big problem. And then second there's a further problem which is that persistently low returns. So in the you know 12345% is about one in seven companies all around the world. Okay.

Speaker A: So why is this important to note? It means that being in business is complicated. It's hard. Not everything works all the time. There's a bunch of underperforming businesses. There's a bunch of underperforming segments. So in that lens, when you, you know, think about what Sax said, which is you have this company that comes to you and says, I have a magic box and all you have to do is tell me everything you're doing, and this magic box will make everything better. But then all of a sudden, from the shadows, the magic box says, you know what? I've decided to compete with you. That is a huge risk. And now that you've seen enough examples of it, I think you have to figure out a different way to do it. So then you go to the next tweet that I saw, which I thought was interesting.

Speaker B: And this is a woman who's an ex meta PM. And what she essentially says is like, hey, hold on a second. There is this assumption that you can't use an open-source model because it's all Chinese. And what she says was, well, even if it's 100x cheaper, and their response is, no, because we care about safety and security. And her perspective is, don't you understand that you can actually host open-source models with your own GPUs in US data centers that doesn't share any data back to anybody? And instead, what you're accidentally or purposefully doing is giving away all your data to a couple of frontier labs rather than owning it privately yourself. And 100x, it turns out, is a really big number to pay to do all of that by accident. And that's what Alex Karp is saying. He's like, why would anybody do this when there are alternatives?

Speaker A: And so the third post that I'll talk about is something that we did. So we took our software factory which is an agnostic third party control plane and we just wanted to see and we ran it on a very typical enterprise task which is you have an old piece of code you want to migrate it and you want to maintain it in a new framework so that it's easier and more flexible. pretty straightforward task. And so we ran it and we ran an experiment where we did Claude by itself, then we did us plus Claude, and then we ran it on the best frontier open-source model and then us plus that model. And the data is crazy. So when you use our harness with Claude, it was simultaneously 1.4x 4x cheaper and 1.5x faster than just using anthropic opus 48 alone. But if you wrap the open source model with our software factory, it was 16.4x cheaper. Now it was three times slower. But you know, you're talking about a couple of extra hours to save 16.4x on that one.

Speaker B: But if you wrap the open source model with our software factory, it was 16.4x cheaper. Now it was three times slower. But you know, you're talking about a couple of extra hours to save 16.4x on that one. The slowness Chimath, is that slowness because of the hardware being served up by Claude or is it

Speaker A: But if you take Sax's points and then if you take Alex Karp's point and just this actual data there is a very legitimate question which is if you are a reasonable company why are you not finding an independent way to access this intelligence in a way that doesn't leak your edge away to do so at this point now is kind of becoming derelict and irresponsible. 以前你可以进行实验，因为你不知道更好的。但现在当你了解所有这些数据点时，继续做出同样的决定，我认为是极其愚蠢的。

Speaker B: All right Dave Friedberg you are also a CEO of the surging oh and you have a lot of proprietary data. Let me ask you point blank. Do you trust your data to the frontier model companies or are you doing what there seems to be consensus here protecting your data protecting the crown jewels so to speak and using open source are you experimenting with it and just yes or no do you trust the frontier models with Ohio's data so I'll tell you there's been an effort by anthropic to go around and sign up life sciences companies to contributing to a new life sciences focused model that effort has been they're approaching these large companies with large proprietary data sets and saying hey if you share your data we will give you early access some sort of proprietary value sign this NDA and you can participate with us and I think nearly everyone I've spoken with has woken up to the fact that they are basically trying to commoditize everyone's business because fundamentally if all of the tens of billions of dollars you as a life sciences company have invested in experiments and product development and you've generated all of this proprietary data along the way. That data is a true asset of your organization. It's an asset that you've spent billions of dollars developing. And by handing it over to a model company to then combine with other people's data, you are effectively commoditizing the asset that you have, the one kind of core differentiation that you have. And so everyone is largely saying no. The way I see this evolving is very much in line with what Alex Karp suggested on CNBC. If you go back a couple of years, I think we all assumed there was going to be this

large hub, large spoke model for AI model development and deployment. Meaning there would be these very large clusters. These large clusters would be ultimately capital advantaged. So those who had the most capital, which is why everyone's raised tens and hundreds of billions of dollars, would be able to train models. And then there would be these large spokes, these large clusters for deploying those models with inference. So everyone's using the NeoClouds and the hyperscalers and whatnot to run models. And then maybe they've got their own proprietary data layer that sits in front of that. But I think what everyone's realizing is they're better off developing their own weights and their own models using either an open-source basis or there might be some intermediary business model that evolves. Meaning there will end up being several large hubs that do all of the core foundational model development. Then smaller hubs meaning like clusters for training that enterprises will use to train and develop their own proprietary advantaged models using their own data and then there will be these much more distributed spokes because I think everyone's also realizing the value of on-prem by putting a set of servers and building a cluster in your own data center or even in your own enterprise you know IT closet you can run a lot of the workflows that you're using AI for for your enterprise locally. And so I think the model is shifting where we're going from large hubs, large spokes to large hubs, medium hubs, and then a distributed spoke model where there will still be, you know, NeoClouds and hyperscalers that are being used for inference, but people will also have their own inference instances that they're going to run for their own enterprise setting. And everyone I think is walking this path and they're going to walk this path over the next couple of months because they're realizing quite quickly that in order to compete in a world of commoditizing knowledge and commoditizing capabilities, you have to leverage the core differentiating assets that you have, which means you have to build your own models and you will likely end up having to run your own inference with your own proprietary models.

Yeah. And to just give a a little bit of texture to how that's going to look, if you follow the Microsoft example, which we've talked about here before, Lotus 123, Word Perfect were their partners. They were replaced with Excel. They were replaced obviously with Microsoft Word. You don't even know those other two. Word Perfect, Wordar, Lotus 123, Vizzyalc. That's exactly what they have to do. And they have no choice but to do that now because they have a trillion dollar market cap. They must win the application layer. And Sam Alman went to Y Combinator and he said, "We'll give you $2 million worth of free tokens." And I came out and I said, "Listen, there's nothing personal against Sam. You know, Sam's a very aggressive dealmaker." And he wants to get access to those startups cuz he knows having run Y Combinator that if he can get their innovations, those founders latest thoughts about what's around the corner, he can incorporate them into the platform. There is no free pizza. There's no free beer. When somebody like Sam Walman comes to you and says, "Here's some free tokens. Your alarm should go up." Zuckerberg did the same. He said, "Hey, I'm going to give people a bunch of access. Going to give them money. Come to the Facebook platform." Nobody who went to bed with Microsoft in the 80s, Facebook in the 2000s, or Sam Alman now in in the 2020s did not wake up with their throats slit. This is a message to founders. If you partner with any of these people, they will slit your throat and take your business wholesale. There is nothing to discuss here. Don't trust them. Use your own models.

I think I think it's less Sam and Open AI to be honest. I think that the diversity of Open AI in terms of its revenue streams and specifically its consumer business may actually be its savior in a relative value basis. Right now, OpenAI equity I think is more reasonably priced than entropic equity. And the reason is not because of the quality of the models or the teams because they're both excellent teams, but the reason is that OpenAI can fall back on a really healthy consumer business.

Sure, the difficulty that Anthropic is going to face is that I think what Sax said is true that they have lost this fundamental trust about being able to stay within their sandbox. And if you consistently demonstrate this tendency to learn and then to try to disrupt your host organism, eventually you get sort of pigeonholed and you get cornered and people find ways to work around it. And so, look, I sent that text that I had about the or that post that I had about uh our testing of of our harness on these Chinese models to somebody well known in the industry. And he says, "Look, if you also add some post training with all of the telemetry that you're going to get from the harness itself, he's like, I suspect you'll find that it gets as good as Mythos." And and I thought, well, if that's true, then why don't I just take GLM, control it entirely, soup to nuts, on my own hardware inside of the United States with only US citizens that can touch it. It just seems like the brain dead obvious thing to do, and it's much much cheaper.

Yeah, and 100% correct. I will say if you're going to do this, you'll eventually wind up rolling your own LLM. I mentioned a company, Abacus, goabacus.co, that we seated in our accelerator. What they're doing now for HIPPA client people is they're actually giving you this go one box. They're literally saying we're going to make your own model for you. So once you start this you you start with you know Claude and use their wrapper and everything or open AIS. Then you move on to the next step. The next step is I'm going to use an open source model use my own try to find a harness etc. Where you will eventually wind up is you're going to fork these models you're going to build your own. That is the end state on-prem on your own hardware. Don't trust anybody because there's too much at stake. You cannot risk your entire business. You might as well be part of the vanguard and start investing here. You're going to slow down to speed up is what's going to happen. Whether you're using 8090, Abacus, or any of these other solutions or rolling your own, you must have AI sovereignty. You must have intelligent sovereignty or you're just giving your business over to your competitors. The only company, the only company at scale that's ever respected the developer community in this way is Apple. Apple took a very strategic approach to wanting to build an app store business to wanting to support developers. And they explicitly told people, if you make something super obvious that you can build in a week or two, it eventually might wind up in our basic collection of apps, the stock app in your iPhone, the notepad app in your iPhone, those apps are incredibly basic. But if you looked at, you know, Robin Hood's uh or Google Finance and then you looked at something like say Evernote back in the day, which was an advanced notetaker, it took 10 years for Notepad to add the features that Evernote had 7 8 9 10 years ago. They specifically slow their apps down to make them simple for users and not f with their ecosystem because they want to take the 30% tax. That's your choice here. There is no 30% tax here when it comes to Anthropic. There is no 30% tax equivalent with open AI.

The thing with Apple is that Apple was renting distribution. This is not renting distribution. This is where you're renting intelligence and judgment. And so the problem that a company has is you can't rent the same in this is why your point is actually right. I think it's for a different reason. You can't rent intelligence from the same place that rents it to your competitor.

Correct.

You just can't. You can't. It's just stupid.

It will overflow. it will go over the wall in your competitor's lap.

It's it becomes the lowest common denominator problem where you and your competitors now look exactly the same. Why would you do that? And again, if you go back to that that thing that BCG identified, so many companies are already

Speaker A: teetering on a very difficult position where they cannot generate returns on their invested equity. And so why would you then go and pay all this money so that you end up with the same answer as your competitor?

Speaker B: You can't do it. It's just not a choice. and Sachs, part of what's going on here is the deflationary nature of technology. Every single one of these tools is getting cheaper. Tokens are getting cheaper, Sachs. And if you look at Nvidia's role in all of this, they have something called Neotron. You can try it if you uh use Perplexity. You can just pick the drop down menu. It has deep thinking. You will not be able to tell the difference between Jensen Wong's open-source LLM sacks and Claude for 95% of your searches. I guarantee you now why why why has Nvidia and Jensen downplayed their open- source model until this moment? Why would he do that? Why would he never bring it up in the all interview? Never bring it up because his top customers were uh very concerned from what I understand about the fact that they had made so much project progress on their open source model. But suddenly after open AI announced their jalapeno chips, after Anthropic started uh making chips, after AMD did successful uh projects with both of these companies, after Elon said he's going to do his own fab, Nvidia's taken the gloves off, David, they are going to own the whole stack. They are going to be talking about open source a whole bunch. So maybe you could talk a little bit about Nvidia and their role in this as the open-source atscale full stack provider. You get the hardware from them and you're gonna get a model that's competitive with open AIS for free and all you have to do is use one of their hosting companies, Cory, whoever is buying Nvidia's, Colossus, etc. What are your thoughts on Nvidia suddenly being willing to talk about their open source projects today?

Speaker A: Well, look, I think part of it is they needed time to make the offering compelling and you know, they're up against some pretty great AI labs and I think some of it is just, hey, it takes time to train up these models. Now, one question is why is it that Nvidia and Palunteer are partnering? Like what makes them natural partners? And I want to just explain that. If you want to think about the AI stack for a minute at the most basic level, there's three layers to the stack. It's basically the chips, it's the models, and then it's the applications. Well, right now we have in the middle layer at the model layer of the stack, you've got two dominant companies. You've got Anthropic and Open AAI. We know that Enthropic is around 60some billion of ARR. Open AAI is at 40 something billion of ARR. As far as we know, no one else is really generating meaningful revenue at the model layer. So, we already have, let's say, an emerging duopoly situation. We have Anthropic pushing for a regulatory capture agenda that would probably enshrine that duopoly situation at a regulatory level because they're pushing for a safety agenda where Daria explicitly says that these other models are not safe. You shouldn't have access to them. So you've kind of got that situation. You got the market producing duopoly. You got the government now potentially leaning not to bust up the duopoly but maybe to enforce it. So that's sort of the emerging situation at the model layer. And so if you're an application at the top of the stack like Palunteer or you're a chip company at the bottom of the stack, that's the last thing you want. You want a competitive model layer. Why? Because if you're an application, you don't want to be beholden to one model provider, right? You want to have a choice. And if you're an enterprise, you want to have a choice because you don't want to have to give up all of your proprietary knowledge. And if you're a chip company, you don't want a monopsiny buyer situation where there's only one or two companies who can buy your chips and by the way, they're producing their own. You want to have as diverse and healthy an ecosystem as possible where there's lots of potential buyers for your chips. And if enterprises are rolling their own using open models, that's kind of an ideal situation because now there's like a long tale of buyers. So I think really the whole ecosystem in a way the chip companies, developers, applications, enterprises, everybody has an incentive for a competitive layer of the stack at the model layer. Really the only companies who don't are anthropic and open AI because obviously they want to dominate. They want to be a duopoly. And my view is, look, if you earn a monopoly or duopoly in our system, we don't ban monopolies in the United States. We ban anti-competitive tactics, but if you lawfully achieve a monopoly through amazing performance, we don't, you know, nationalize you or make you illegal, and I think that's fine. However, the government, in my view, should do nothing to make monopoly or duopoly more likely. They should make it harder for these companies to engage in monopoly tactics and they should do everything they can to keep the model layer competitive because competition is what brings out the best and it's good for the ecosystem and it ensures our civil liberties and consumer choice.

Speaker B: and it's going to be amazing for pricing sacks if you think about what this competition is about to do and you've been very vocal about this in your time in Washington DC. We want to have a level playing field. We want to see massive competition. I'll give you your flowers and your lay. You did a great job of now setting the table for this in 2026 and 2027 going forward by letting people compete. The cost of tokens, Freeberg, is going to go down 90% a year for the next 3 years. You're going to be able to buy a thousand times as many tokens that are more intelligent because you're going to have free options. The price will be free or close to free for many of these and that could be incredibly disruptive. Yes, Dave.

Speaker A: Yeah. And I think people are going to again deploy their own hardware against it. I think there's going to be a buying frenzy in the enterprise, not just with the Neoclouds and the hyperscalers. I think the the enterprise is going to be a buyer. And when that happens, you do the simple math and you don't want to have a dependency on server availability and cloud downtime and you don't want to put these models on some third party cloud and you want to do stuff that's very cheap. Like a lot of people are running I mean you guys do this. You're running day-to-day workflows for your enterprise and you realize, hey, I could run these workflows on an open source model on a machine in my office and we don't need to be sending this stuff back to some fancy scalable cloud service provider.

Speaker B: Right? The industry spent so many years convincing everybody to flip to the cloud and the realization may be that all this this idea of shared infrastructure may not be the best idea in a world of intelligence.

Speaker A: and this is what I mean by like a distributed spoke model because I do think it's not going to be all or none. And I think you're going to end up being like 702011 in how you're going to allocate your resources for model inference and running models. You're going to probably be 70% in some big cloud. Maybe you'll do 20% local 10% you'll try other clouds. You know, you'll kind of mix stuff up, but I don't think you're going to end up doing things the way you've been doing them historically. You you'll very quickly realize that it's okay to waste tokens. It's okay to let your employees make stupid apps that last for a couple weeks but burn through billions of tokens. If it's running on your own hardware, then all you're paying for is the electricity in your office or in your

Speaker B: 你刚才在这里一直在谈论的，大家都在意识到。是的。

Speaker A: 就像我完全没关系，你知道，呃，为了让我的孩子吃糖，如果我不用处理他们的话，你知道，就这么说吧。

Speaker B: 你们注意到吗？在过去的3天里，我们现在看到其他人试图锁定自己的系统了吗？微软刚刚宣布了25亿美元的投资来建立一个全职员工岗位（FTE），或者亚马逊正在花费10亿美元……

AI 的应用与工作岗位流失的辩论

Speaker A: 有一个开放的AI。

Speaker B: 是的。当他们敲门时，就像“嘿，我可以派我的工程师来研究你的业务并把它放入我的模型吗？” 我是说，人们会把这些门砰地关上。这就像在你的门口遇到一个耶和华见证人一样。我不想成为你教派的一部分。我想拥有它。

I think everyone is realizing that, you know, I was wondering where you were going with that buttering it up. So is Truman. But I think, you know, the idea that everyone had in their head two years ago, the narrative that was formed and everyone clutched on to, and by the way, you will not see the media reverse on this narrative. I've realized the importance when when if you're a reporter or you're a journalist and you come up with some story that says something that's happening or is going to happen or has happened or is about the future, you destroy your own credibility if the narrative shifts. You cannot ever let go of the narrative. And I think this is one of the things that that we can kind of acknowledge is going on with this job loss narrative problem is that even as all the data comes out as all of the reconfiguring of how enterprises are using AI as it's revealing to all of them that they're actually not just going to cut cost but they're going to grow revenue and it's going to be a kind of clunky way of getting there. It's not going to happen overnight. It's not this slippery slope to job loss to nealism. Everyone's gonna kind of wake up and be like, "Wait, this reality that we all thought we were living in is not really the reality that we are living in. We're living in a reality where AI is clunky. It takes some putting together. It's a little more complicated than we thought. It's definitely going to deliver value, but it's not about just turning off all the jobs and letting the genius AI solve all my enterprise problems and scale me into infinity without humans." And I think that that's a big kind of narrative shift. And you will not see the media accept that their narrative is wrong. Because as soon as they have to acknowledge that they were wrong in what they were saying about job loss and all the other senators and people that are proclaiming job loss, job loss, job loss, by the way, the reason they're making that proclamation is so that they can step in and control AI and they can drive their systems of socialism, which is what they're all looking to deploy. But if they had to come in and say, "Look, the data doesn't map to the narrative, their credibility is destroyed." So they'll double down on it. and they'll double down on it. And I'm telling everyone that's listening, look at the data. There is no job loss with AI. It is an absolute scam to tell the world that AI is taking away jobs and destroying jobs and the world is shifting. It is clunky. It is valuable. It is going to take some time and it is going to create far more jobs than it is destroying.

We're a bunch of monkeys. We've been given this new tool. We're going to solve more problems with the new tool. There will be job displacement. And the problem is people do not understand the nuance between these two terms. Certain jobs will be retired and they're going to be retired at a faster rate than you can imagine, but other jobs will happen. And you pointed this out, Shimoth. Hey, it's the it's the most I don't know if you did it here or on your personal channel on YouTube, but you know, it's the most empowering tool ever. If you can What jobs are getting replaced?

I challenge very simple, but you're saying some of them are going to go away faster than you can see. What? What are those jobs that are going to go away faster than you can see? Cuz I'm just I'm still not seeing the categories.

Totally. Customer support jobs. Those are going to go away very quickly. Where? Where? You're just making [__] up cuz it's not I see it in enterprises all the time. Where do you see it? Show me the customer support enterprise shutdown that's happening. Just show it to me. Is it in the room with us right now? Like where is it? I will literally, you know, I'll I'll do that research and and I'll give it to you.

I'll give you some research. Hold on. Can I can I provide some research? He asked me a question. Just let me finish my answer and then you can take it. You asked me the question. Customer service jobs are going to go away because consumers prefer to talk to the AI and it's perfect at that. Entrylevel jobs around data entry around business process outsourcing those are going to go away. So will driving cabs and those kind of jobs. Those ones are obvious lowhanging fruit. Just like we had the typing pool and messengers and other products go away when we had word processors on every desk. Those jobs will be displaced 100%. And those jobs

year we're on year 20 of cab drivers going away, dude. Like when Uber came out, everyone said all the driving jobs were going to go away. It's all over. Well, no. Whimo is stuck at 3,000. Tesla's stuck at like 30 cars. But that's all going to change in the very near future.

Everything you're saying is perspective. This is my point. You are still being perspective in the future with everything you're saying. And every point that we turn, every time we turn a page, it's like, wait, do you happen to think self-driving is real? Is that your premise? Is that self-driving is not going to get rid of cab drivers? Is that your premise?

I think it's real, but like like I think it's real for sure. Do you think it will get rid of cab drivers? You're arguing as if like I'm wrong here, but I think that every day I get in the car every day and my car drives me and I get on my phone and I do work.

Okay. But do you think cabs and cities are not going to be self-driving? Do you think zipline's not going to deliver? Bailey thing again. Jal, it's not Mont belly. I see this from my investments. I have a company auto lane that is doing this right now.

But just go to customer support again cuz you've been saying Yeah. I literally we've had this debate verbatim and displacement not job loss is is my only point.

Yeah. Yeah. Yeah. Yeah. What Freeberg is pointing out is that whenever anyone pushes back on the fact that you don't have any data to support this in the present, you say you're talking about the future. Okay, fine. It's a prediction. You know, there's no data to support it in the present. That's fine. You're saying it's gonna happen in the future?

No, no. I'll give you the data point. You can dismiss I want. I will. Yes. If you talk to Uber and Whimo, in the cities where Whimo is present and has gotten past a couple hundred cars, they've stopped recruiting drivers. Drivers are either static or going down in those markets. So that is absolute evidence that this is happening and the roll out is going to be fast and furious. So it is not a future prediction. You can talk to the CEO of Whimo. You can talk to the CEO of Lyft and Uber and they will tell you this explicitly. Market interview Darren. He said jobs were increasing at Uber.

Speaker A: Yeah, cuz you're delivering more like those because productivity is going up with with delivery. Productivity goes up means people drive more% lying. You're being dis you're being last time we no the last time we disingenuous in markets where Whimo is in markets where Whimo is you're you're there might be other markets where things are growing because they don't allow self-driving or they haven't gotten there yet. In markets where Whimo has hit critical mass, the number of human drivers is going down. This is just a fact. Stop recruiting in some narrow arbitrarily defined subset of their jobs. You're pointing to flatness.

Speaker B: Meanwhile, Dar told us that hiring a cost of the company has gone up in the face of these drivers.

Speaker A: No, no, you're wrong. Okay. Anyway, we we'll agree to disagree here.

Speaker B: Let me give you some actual data. Can I provide us this study? So, RAMP and Rellio Labs just released a new study in the past week.

Speaker A: Yeah, that was great.

Speaker B: And it was an actual study of over 21,000 firms in the US and they looked at their payroll data combined with their spending on AI. And what they saw is that firms that spent the most on AI actually grew the fastest and they tended to grow their headcount roughly 10% in the two years following the adoption of AI. And entry level headcount rose even faster, it grew at 12%. So all this stuff about how entry-level headcount's going to get wiped out, not true. The more firms adopted AI, the more hiring they did. At least that was a correlation. Obviously, they can't prove causation, but that was the correlation. And then companies that were not high intensity adopters of AI, they were either didn't adopt or they were low intensity adopters, they just saw flatness in their headcount. So, no one is really showing a trend here of job loss. What they're showing is that the more you adopt AI, those companies also tend to grow headcount. So I you know there's just no data to support this idea that in the present AI is causing job loss.

Speaker A: Job displacement will happen. Jobs will increase because of AI. Sovereignty will increase and people will make more money. Corporations will be more will have more earnings. But there will be jobs being retired. And I think if you look at translators, if you look at telephone operators, if you look at bill collectors, if you look at Whimo drivers, these very simple tasks are being automated. How do I know that? Because I invest in startups that are raising hundreds of millions of dollars to replace these jobs with AI. We had the pilot uh we we had people building the perfect pilot, the perfect driver, the perfect customer support rep. That is the point of AI. Every single sales pitch of these AI companies is to retire these jobs that are not great jobs. There is no way in God's green earth that humans will be at Amazon sorting packages. There's not. It's not going to happen. All those factory jobs are being replaced. Elon is explicitly building Optimus for this purpose. And when I say it's future, I have always contended it's future. Sachs, you cannot use your silly debate techniques to say that I'm using some crazy thing. It is happening as we speak. We're soaking in it. This is the transition period. What I'll tell you is that in every environment in which 8090 operates, every customer, I don't think we've seen layoffs. Not a single one.

Speaker B: Yeah. In this in this RAM card study, they said, quote, "Gains emerge gradually and are broad across roles, including engineering, sales, administration, and customer service. So, they're not seeing like one category of jobs getting wiped out." By the way, it could happen in the future. I mean, none of us can say for sure what's going to happen in the future, but if you're looking for data in the present, we're not seeing it yet.

Speaker A: I think there's going to be a strong human alpha. You know, like when vending machines came along, it's not like bartenders went away. Like I like going to a bar talking to the bartender and they make a great drink. And those bartenders make better money today than they did 20 years ago. I don't and and maybe the vending machine replacement model isn't the right one in that case, but I I think that there's a lot of value in human interaction that's going to get a premium in the future because as more stuff gets automated, I want to have an actual driver when I go to Las Vegas. I don't want to go in a Whimo. I want to have a great driver in a nice car that takes me to the dinners, drives me around New York City, same thing. I don't want to go in a Whimo, per se. I want to have, you know, people that can do things that could otherwise be automated. I'll pay a huge premium. I could sit in a freaking massage chair. Why do I go pay someone $200 for an hour massage? You know, I think there's there's an aspect to the alpha of we could cut that one. But uh but I do think there's a premium to humans and I don't think that um and I actually to humans

Speaker B: no the count JL this is really important. I think that the counter-narrative to automation generally is that we're going to realize the importance of human interaction and humans in the loop and we're going to pay a premium for it.

Speaker A: Well, do you remember like a year ago when CLA said they're going to replace their whole customer service department with AI and they did they did a lot of hype. It was peak RF farming and well lo and behold they flipped the decision back after a year. They said that from a brand perspective, a company perspective, it's just so critical that you are clear to your customer that there will always be a human if if you want it.

Speaker B: Human in the loop will be the premium and we will all realize in the next year and a half the importance of human in the loop even in the context of AI and automation working.

Speaker A: Human in the loop is going to be more valuable and then you're going to pay a premium for human in the loop. There's so many CEOs who want to tap into the media's hype cycle. This is the thing is that whenever the media gets a hold of one of these narratives, there's so many CEOs who are not like Karp, right, who just are not original and they don't really have anything important to say. So, the only way for them to get publicity is to glom onto that press cycle and try to use it. And then they'll chime in with, you know, how they're going to do crazy things to eliminate their whole customer support department and then lo and behold, it turns out just to be total totally fake.

Speaker B: I think the ramp data is directionally accurate. In our experience, I would say the same thing. I think people that use AI tend to grow faster. They tend to hire more. They tend to make more%.

Speaker A: Yeah, 100%.

Speaker B: By the way, I actually agree with 100% agreement on that.

Speaker A: From a career standpoint, I would not want to be a level one customer support rep because I do think that like if all you're doing is like answering forgotten password requests, that's not a great place to be. However, those jobs were basically outsourced to the Philippines and places like that a long time ago. I mean, most enterprises are not doing kind of that level zero, level one customer support in the US. They do have customer support centers, but they're doing like level two, level three. They're doing the escalations, the higher value stuff that's going to be a lot harder to replace with AI.

Speaker B: Totally. I do kind of worry about well what happens in some of these other countries that are doing the super entry level stuff because I think that is job loss but I don't know if it'll be massive but that's where the risk is. I mean frankly it's not in the US.

Speaker A: This is um so maybe we're getting to some consensus here. Job displacement versus job loss. Here is Brett Atcock sharing what they're doing at Figure. They had an 8-hour challenge to have this uh robot they're building sort packages and they did it for 200 hours. That's today. This is the least good it will ever be. This is the worst it's going to ever be. These things are getting really good. I spent time with Elon. I've looked at Optimus. I can tell you Optimus is better than

Speaker A: what Figure is doing. What Optimus is going to do when Bezos and Andy Jasse unleash Optimus inside of the Amazon factories is it's going to get rid of every, let me state this very clearly, every single package sorting, every single package delivery will not be done by humans in 10 years. And it's starting today. Just like the idea that you would put together some of this consumer electronic stuff in factories with humans has gone away over time and other countries. I think you're making a very good point here, Saxs, are going to experience it first. In America, knowledge workers, the entrepreneurial spirit. I think we're going to see a Cambrian explosion in startups, which means we're going to be the ones who benefit from it. But if you're looking at those other countries, they'll see it acutely first. We're just going to see it here with drivers, and that's 5 10 million people, like we saw it with cashiers. The idea of going to a cashier at a fast food restaurant, which I know you guys don't go to all that often, where Starbucks, the idea that you would interact with a cashier has been going away slowly because it's just consumers don't even want it. We'll agree to disagree here to a certain extent. Maybe we may

Speaker B: Look, I went to McDonald I went to McDonald's recently and I ordered from one of those monitors. Zero.

Speaker A: What did you order? You filet a fish guy? I'm a flay of fish guy. What'd you get? What's your

Speaker B: I'm a Big Mac. I like a Big Mac.

Speaker A: You like a Big Mac? What about you, Sax? What? What's your order? What's your Did you get a McFlurry? They had a thing called a homegrown burger which was some new I think it was kind of like more like an In-N-Out type burger situation they were testing.

Speaker B: Smash Patty.

Speaker A: Yeah, it was good. Anyway, my point is just the automation obviously they didn't have a cashier there already, but there's zero robots involved. Similarly, there's already a lot of automation going on in those Amazon warehouses. I don't know if you've ever been to like a FedEx packaging or routing depot. It's all conveyor belts and you know

Speaker B: Yeah, robots pushing it and directing. But it's not it's not humanoid robots. There's already tons and tons of automation at these places.

Speaker A: The last but this last mile part is the one I think that is going to shock people. When you have an optimist walking the Tesla self-driving car and then they walk it to your porch, ring your doorbell and knock on your door and put it on your front step porch. That's where this is going to get super interesting.