Sosin Shuns AI Infrastructure to Back Physical Moats, Carvana Stake

索辛避开AI基础设施，转投实体护城河与Carvana 充沛的智能将使前沿模型商品化并引发基础设施泡沫破裂，使持久利润转向实体分销网络、专有数据与最低成本运营方。 克里夫·索辛（Cliff Sosin）认为，智能仅仅是预测动态系统的工具，而非无限的标量，这意味着复合误差和物理限制将始终制约纯粹的计算推理。随着原始智能迅速降价并使软件商品化，价值将果断转向稀缺的实体互补物——例如注意力、专有数据集以及复杂的现实世界物流网络。索辛详细阐述了他为何避开资本密集型的AI基础设施竞赛，自主智能体将如何打破传统的消费者惯性与聚合理论，以及他为何凭借深厚的领域护城河来证明其投资组合向Carvana重仓80%是合理的。 由于混沌理论和复合截断误差，纯粹的人工智能面临着严格的数学和物理限制，需要进行现实世界中的实验而非仅靠纯算力来解决动态问题。 (0:00) 随着基础智能的成本趋近于零，经济价值将从软件转移到稀缺的实体互补物上，例如用户注意力、专有数据集、实体分销网络以及复杂制造。 (4:59) 自主消费者智能体将系统性地瓦解依赖客户认知惰性的企业，直接将购买决策导向成本最低、效率最高的生产商。 (32:01) 由于严重的资本周期风险、可预见的过度建设以及长交付周期资产中的赢家诅咒，索辛避免对AI基础设施建设进行直接投资。 (24:22) CAS Investment Partners在Carvana上保持着80%的投资组合集中度，优先看重该公司具有防御能力的实体物流基础设施，而非传统的学术性分散投资。 (50:56)

Cliff Sosin argues that artificial intelligence is fundamentally a predictive tool constrained by chaos and physical reality, which will rapidly commoditize baseline cognitive tasks. Consequently, enduring economic returns will accrue not to frontier model developers or speculative compute infrastructure, but to scarce complements such as human attention, proprietary data, and physical distribution scale. Physical Boundaries and the Limits of Intelligence (0:00) Sosin refutes the narrative that machine intelligence scales infinitely into omnipotent problem-solving. Instead, he characterizes intelligence as a predictive gadget operating in complex dynamic systems where initial conditions inevitably suffer from truncation errors, causing forecasts to diverge rapidly over time through the butterfly effect. Beyond theoretical deduction, solving real-world challenges requires direct physical experimentation governed by the laws of nature—such as the time needed to breed mice in biological testing—which cannot be accelerated simply by stacking neural parameters. Sosin also challenges assumptions of infinite demand for frontier models. Drawing parallels to optical telecommunications, he notes that demand follows an inverted U-curve where extreme deflation can ultimately contract total market revenue. Because few real-world tasks require a "300 IQ," the bulk of demand shifts toward "good enough" intelligence. As benchmark evaluations saturate, pricing power for frontier capabilities collapses, driving rapid commoditization across lower-tier inference tasks. Capital Cycles and AI Infrastructure Risks (20:39) Sosin avoids direct investment in artificial intelligence infrastructure, such as data centers and compute hosting, citing the impossibility of forecasting supply and inference demand within orders of magnitude. He warns that reallocating roughly 3% of U.S. GDP toward speculative AI infrastructure—pulling specialized trades like electricians and plumbers from the broader economy—mirrors historic capital cycles where long project lead times lead to severe overbuilding and the "winner's curse." Higher interest rates represent the necessary premium to convince society to defer immediate consumption for long-dated capacity. Technical volatility heightens this fragility. Dominant hardware designers like Nvidia produce cognitive flat files sent to foundries, an intellectual advantage that automated AI tools could theoretically compress. In contrast, Taiwan Semiconductor Manufacturing Company (TSMC) maintains a resilient physical moat rooted in tacit, illegible process engineering. However, TSMC's supply constraints have simultaneously created a vulnerability by incentivizing global competitors and governments to pour billions into replicating its manufacturing capabilities. Value Shifts to Attention, Data, and Physical Density (4:59) As software intelligence becomes a cheap commodity, Sosin argues that value concentrates in complementary assets: proprietary data, finite human attention, physical proximity, scale, and regulatory capture. For instance, he analyzes Meta as "personalized TV" whose multi-billion-dollar operating and capital expenditures are dedicated to buying human and machine intelligence. Because Meta commands billions of hours of consumer attention and vast behavioral data, cheap compute simply reduces operating costs and sharpens engagement algorithms without threatening its core distribution. In the physical economy, autonomous personal agents will eliminate consumer friction and inertia. Boundedly rational consumers currently tolerate uncompetitive bank deposit yields or unexamined service contracts out of cognitive lethargy. Proactive AI agents running continuous micro-optimizations will dynamically bounce contracts to the lowest-cost provider. In route-dense businesses like local garbage collection or delivery logistics, this dynamic funnels marginal volume directly to the lowest-cost operator, expanding route density, lowering unit costs, and reinforcing an unassailable physical scale advantage. Organizational Transformation and the Investing Landscape (46:42) Ubiquitous AI agents also threaten digital distribution models like Ben Thompson's Aggregation Theory: if personal agents survey an entire market rather than clicking top search results, digital placement moats erode in favor of pricing, quality, and service. Concurrently, cheap intelligence lowers internal coordination overhead, allowing highly integrated vertical stacks—such as Carvana's reconditioning and logistics network—to outperform fragmented horizontal supply chains. For enterprises deploying AI, Sosin advises a four-stage playbook: company-wide "vibe coding" hackathons to reset expectations, universal read-only internal system access, role-based write permissions, and governance for autonomous system code modification. Regarding asset management, Sosin views modern investing as a transient "freestyle chess" era where human real-world intuition complements machine processing. While probabilistic models will eventually outperform human discretionary underwriting on longer horizons, fundamental investing lacks the short-cycle verification loops that enable rapid algorithmic takeover. Defending his fund's roughly 80% concentration in Carvana, Sosin argues that conventional diversification prioritizes short-term price volatility over operational asset quality and extreme asymmetric upside. Sosin's thesis frames artificial intelligence as an engine of cost deflation rather than unconstrained digital power. The critical unresolved challenge for market participants is navigating the capital cycle's inevitable infrastructure overbuild while securing the scarce physical assets, proprietary attention feeds, and dense distribution networks that machine intelligence empowers but cannot replicate.

投资人 Cliff Sosin 提出，人工智能并非无所不能的神性存在，而是受制于混沌系统与物理现实的预测工具；随着智能本身迅速沦为廉价泛滥的大宗商品，资本的真正护城河将转移至专有数据、人类注意力、物理密度与工程规模等“智能互补品”。 复杂动态系统的物理硬边界与智能局限 (0:00) Sosin 质疑了将 AI 视为无限进化“全知实体”的叙事，主张智能本质上更像球体的圆度或一种比率，是利用稀疏事实推测宇宙的工具。在纯数学等具备闭式解的领域，智能可推导至完美；但在天气演变或流体混合等非线性复杂动态系统中，混沌理论与蝴蝶效应决定了极微小的初始舍入误差会在计算中级数放大，使得纯理论推导迅速失效。 面对这种不确定性，即便拥有超级智力也无法仅凭推演解决问题，必须依托耗费时间的物理实验直接与现实交互（例如生物制药中测试小鼠生命周期无法凭空加速）。Sosin 据此推论，超高智商（如 300 IQ）在现实中的高价值应用场景可能极度狭窄，大部分商业需求将停留在“足够好”的水平，促使智能本身迅速演变为近乎免费的日常工具。 智能互补品：注意力、物理资产与专有数据 (4:59) Sosin 认为，许多传统软件与白领工作本质上只是“结晶智能”（即固化的人类分析力），未来随时可被通用模型动态生成并被大宗商品化。因此，具有持久壁垒的资产必然是智能的互补品，包括专有数据、人类有限的注意力、情感归属、物理邻近性、工业级规模以及监管壁垒。 以 Meta 为例，Sosin 将其核心商业模式定性为“个性化电视”，其每年约 2500 亿美元的运营与资本支出几乎全用于采购智能（芯片、算法与研发人员）。当智能成本下降 100 倍，Meta 不仅不会被颠覆，反而能以极低成本生成更具黏性的内容与精准广告，进一步抢占消费者的有限注意力；类似地，如 Grindr 等受限于工程预算的高黏性细分平台，也能借助泛滥且廉价的开发能力释放以往被压抑的商业变现空间。 资本周期陷阱与算力基建的下行风险 (20:39) 对于 OpenAI、Anthropic 等前沿模型实验室，Sosin 抱持怀疑态度。他指出，一旦特定任务的基准评估饱和，下游客户便绝不会为简单任务支付前沿溢价，技术扩散将迅速抹平超额利润。针对“杰文斯悖论”（效率提升刺激需求无限扩张）的乐观论调，Sosin 反驳称技术需求遵循倒 U 型曲线（类似早期的光纤通信泡沫），当智能价格趋近于零时，过度降价反而可能使整体行业总产值萎缩。 在基础设施端，Sosin 明确避开数据中心、算力芯片和电力能源的直接投资。他认为，预测未来推理需求的弹性以及长远供给成本极为困难，而漫长的工程交付周期叠加资本追逐，几乎注定会触发“胜者诅咒”与严重的产能过剩；即便如英伟达这类行业龙头，其护城河依然建立在人类设计智慧之上，理论上同样面临芯片设计流程被 AI 自动化压缩复杂度的长线冲击。 Agent 抹平交易摩擦与重塑垂直整合 (30:00) 个人 AI Agent 的普及将从根本上瓦解依赖“消费者认知惰性”获利的传统商业模式。以往用户因嫌麻烦而容忍较低的银行存款利率或固定费率的日常账单；全天候运行的 Agent 则会为了微小的利差毫秒级切换供应商，令传统价格粘性彻底瓦解，本·汤普森（Ben Thompson）的“聚合理论”亦可能因此失效——当 Agent 穿透所有搜索结果时，占据第一页分发入口的数字垄断租金将荡然无存。 这种无摩擦环境反而将强化具备极致物理规模和最低成本的实体运营者（如拥有最高路线密度的区域废品回收商或即时配送网络）。此外，廉价智能极大降低了企业跨部门的协同成本，使得过去因人类管理带宽受限而难以做大的端到端垂直整合模式（例如涵盖从食品生产到医疗保险的全链条宠物企业）重新展现出高利润率与高效率优势。 投资范式转移、极端集中与双峰教育 (55:00) 在资产管理领域，Sosin 认为当前行业正处于人机协同的“自由人象棋时代”，人类基于现实生活体验的直觉仍有微弱互补价值，但中长周期下纯定价风险的人类投资者终将被机器概率模型超越。针对其基金将约 80% 资产集中于二手车零售商 Carvana 的激进持仓，Sosin 辩护称，传统分散理论侧重于规避短期波动，而极少数建立在不可替代的物理物流网络之上、归零风险极低的优质标的，理应享有极高集中度带来的非对称回报。 对于组织进化与教育，Sosin 主张企业推行包含“全员黑客松（Vibe Coding）”、只读系统贯通、授权写入与系统自治的四阶段落地路线；在人才培养上，他提出“双峰教育模型”：一方面必须脱离屏幕死磕基础运算、核心概念与思维模型，因为缺乏深厚领域素养的人根本无法精准指挥前沿模型；另一方面则要求学生全面借助 AI，以百倍于传统模式的雄心去推进复杂的现实应用。 Cliff Sosin 为当下狂热的 AI 投资热潮提供了冷静的反共识视角，指明在智能自身迅速大宗商品化的未来，超额利润将向物理现实、人类注意力与专有数据等互补资产集中；不过，前沿模型需求是否会如其预测那样迅速遭遇倒 U 型曲线天花板，以及垂直整合企业能否在高度自动化的时代全面战胜水平分工，仍有待更长周期的产业实践检验。

[35:00] Cliff Sosin: Price is a miracle because it's a signal wrapped in an incentive. [25:00] Cliff Sosin: Instead of betting on which railroad tycoon is going to make money and how many railroads we need, I'm trying to think about, OK, once you have railroads, I'd love to have the foresight to figure out CPG brands come next. [4:59] Cliff Sosin: And as I've been thinking about companies and investments, I found it helpful to assume everyone has really great intelligence and ask what's left. [45:00] Cliff Sosin: The really great outcomes are the ones that make no sense. You shouldn't do it at all. But then it turns out there is a path or the world turns.

[0:00] Ramp is the only platform built to make your finance team leaner, faster, and better, saving businesses 5% annually on average so you can stay focused on growth. Ramp just opened for business in the United Kingdom, so if you're running a business in the UK, you can now use Ramp's AI-powered finance platform to manage cards, expenses, bills, approvals, and accounting all in one place. I run my business on Ramp, and so should you. Learn more at ramp.com slash invest. Thank you. Open AI, Cursor, Anthropic, Perplexity, and Vercel all have something in common. They all use WorkOS. And here's why. To achieve enterprise adoption at scale, you have to deliver on core capabilities like SSO, Skim, RBAC, and Audit Logs. That's where WorkOS comes in. Instead of spending months building these mission-critical capabilities yourself, you can just use WorkOS APIs to gain all of them on day zero. run on WorkOS. WorkOS is the fastest way to become enterprise ready and stay focused on what matters most, your product. Visit WorkOS.com to get started. Rogo is the AI platform purpose-built for financial institutions, serving tens of thousands of bankers and investors at hundreds of leading firms worldwide. Rogo's AI agents autonomously execute large chunk of your firm's workflows. They can screen deals, draft sims, run buyer outreach, and do diligence on data rooms, all with full security and compliance handled. stronger than the last. To learn more, visit rogo.com slash invest. you called me saying it's just intelligence was sort of the theme of our conversation describe that simple idea there's this idea that if intelligence starts an amoeba and it gets better and better and then it's like a monkey and then it's like a person and then it's like einstein and then goes zoom and it becomes god I think that that's something of an intellectual mistake. What people have wrong there is they're missing what are the real limits of what intelligence can do. And maybe another way to put it, if you're thinking about intelligence as like an unbounded scaler at the height of a tree, isn't correct. What you should really think about intelligence as is more like a ratio. like the roundness of a ball. And intelligence is a gadget. And we use it to make predictions about the universe. And so we start with this thin scattering of facts about the universe where we have relatively few of them, and the universe is complex. And we use intelligence to interpret it between them. So as your intelligence goes up, your ability to color in the spaces gets better. And some domains are really amenable to that. In mathematics, you can solve it, those sorts of things. And in those areas, you get to perfect. But for most of the interesting stuff that is most of the questions in the world, the mathematical term is they're a complex dynamic system. So what that basically means is they're kind of like weather in the sense that you can't just guess your way. You can't make a closed form answer. You can try to simulate it, but if you simulate it, invariably your simulation will have some degree of error in your initial conditions or in each step going forward, because at a minimum your steps, for example, are going to be truncated. And so what happens is your error will build and invariably after the simulation runs for some period of time, you'll be off by some amount. And so for mixing cream into your coffee or predicting the weather or thinking about how groups of people who are all very complicated are going to interact with each other, intelligence just isn't that powerful. And so my sense is that what you get is this increasing... Intelligence. to the point where you are making all the predictions as something that SuperSmart could make, which it turns out is a relatively small set of all the interesting things you'd want to know about the world. To get past that, you then need to go out and get more information about the world, and that just takes time. You have to do experiments, you have to touch reality. One interesting corollary to that is that the Zoom actually becomes one of the things you can turn your intelligence to is reducing the cost of intelligence, which we're seeing. So what you actually end up getting is everyone gets unlimited, really great, very round ball, so to speak, intelligence for essentially free. And obviously that's a limit statement. It won't actually be free. But the point is that what we're seeing is that really good enough intelligence is rapidly falling in price. And it turns out really good enough intelligence will solve the vast majority of questions where people need to answer. Yeah. Can you say more about why this notion of a complex adaptive or dynamic system exists? means that actually even perfect intelligence or whatever intelligence we have in a bottle. 10 years from now or Pick Your Horizon, will still be limited in its applications. Yes, nonlinear dynamic systems. The famous story here and how these were discovered is came into the branch of mathematics became chaos theory is back in the day, there was a guy who was building a weather forecasting model back on big computers that took up rooms and that sort of thing. So he runs the model and it makes predictions of pressure and temperature over some span of some forecasting period. And he's like, cool. And so then, just to make sure it's still working, they reset it and they run it again. And it starts out basically the same results. But then the results start to diverge and by the [5:00] third, fourth, fifth day of the forecast. the results were entirely different. And the guy thought some version of, darn, I must have a bug in my system, and begins the process of trying to figure out why the computer isn't producing the same results correctly. in repeated runs. And it turned out that the technician who had loaded the data into the computer... in the second run had rounded off at some decimal, like one decimal less, all of the initial conditions for pressure and temperature and such. And you would have thought that, OK, maybe that would have produced a small but consistent error over the whole forecast period. But what had actually happened was that small error through all of the repeated steps. You take your current conditions, you calculate your next set of conditions, they compound, and you had an entirely different forecast by some small distance into the forecasting period in the unrecognizable weather. One model said it was 70 degrees and sunny, and the other model said it was 20 degrees and pouring. People called it the butterfly effect because, at least in principle, a butterfly could flap its wings in Africa, and that could lead to a hurricane in North America or whatnot. And it just turns out that many systems that we're interested in have this characteristic. You can know everything about pressure and temperature, but you can't necessarily write down an equation that you can solve easily that would tell you that there's going to be a hurricane. You have to model it. And the model entrant is going to be greened at some level, unless you actually go to the step of actually having... the whole universe. And so ultimately, in the limit, if you want to simulate the whole universe, what you actually need is the whole universe. If you don't actually do that, there are many, many circumstances where simulation is super useful for approximations. And in those circumstances, you can use intelligence to solve them. But there are many more where it doesn't work. And so intelligence is limited in that fundamental unknowableness, I guess, about the universe. So back to this notion of it's just intelligence. Yeah. What does that mean about where you think we're going from a more practical standpoint? We're all going to have a team of 200 IQ people. people working full-time in our pocket, and that team will have a team that it can deploy many more. And they're going to be trying to figure out all the little optimizations that we're too lazy to do. And every company will make similar teams available to their employees in order to do their jobs. There's just a lot of businesses where that's not good for them. Consumers are boundedly rational at present. They only have so much cognitive space. And so you go to a consumer and you say, look, we'll take your money. We won't steal it. We'll pay you a decent interest rate. You don't have to think about it. And you can go about your life. And the consumer doesn't say, look, if I think about this every day, I could save $400 a year. So therefore, I'm going to think about this every day. The consumer says something like, sounds good enough. I'm going to move on. But with an agent, you would think the agent will continuously optimize for them. Then there's also companies where what they produce fundamentally is some sort of artifact that is just some sort of crystallized intelligence. I'm by far not the first person to realize that AI could be bad for certain software companies. But just this idea that what an AI model is, is it's kind of like this general purpose intelligence tool. And what many software tools are, are crystallized and hardened pieces like projections of intelligence. And so as I've been thinking about companies and investments, I found it helpful to assume everyone has really great intelligence and ask what's left. The pushback is often some version of, but you don't understand how hard this is, you don't understand how complicated this is, you don't understand how many smart people have had to think for so long about this. And my hand wavy and approximately right response is, OK, sure, but that's all just intelligence. And to follow this logic all the way through to the end is if intelligence becomes cheap and ubiquitous. What you want to own are the complements to intelligence. And so the complements to intelligence are – their data, attention, love, affection, whatever sort of emotional attachment you want. It's physical proximity, it's scale. It's a little cynical, but it's regulatory capture. It's these sorts of things where for some businesses, what happens is their finite ability to deploy intelligence, be it economics, just the cost of it. They're only a certain size company and they don't have access to 10,000 of Google's best engineers for 10 years. or be it just the organization is behind or whatever, the ability to deploy way more intelligence to your business and your competitors to deploy more intelligence to their business and your customers to use lots of intelligence. It can work out to be better for you, or if intelligence is a barrier to you serving your customers, the lack of it. Or it can be bad for you if your customer is lethargy is the fundamental element of your economic value. Maybe pick Meta or something as the perfect example of a company that I think you believe owns perfect complements to ubiquitous cheap intelligence, just to bring the example to life. I'm actually going to set aside for a moment Muse and agents and that whole thing and focus on business. And the core business at Meta is personalized TV. It used to be social media, but now it's personalized TV. And it turns out that people really like watching Instagram or whatnot. Discovered that I love dark British humor and it's great. People talk about the 10,000 foot river. Let's talk about meta from space. the highest view. They have $250 billion in revenue. And they spend $250 billion a year. Some of it gets accounted for its cost, some of it gets accounted for its capex, it doesn't really matter. Just, they spend it all. [10:00] The question you might ask is, what do they spend it on? And a really reductionist view is that they spend it all on intelligence. What do I mean by that? Well, they're buying data centers. What is a data center? Well, it's kind of an intelligent producing tool. What else are they doing? Well, they're actually literally buying inference. That's just intelligence. What else are they doing? Well, they're paying really smart people to write algorithms. That's intelligence. They're writing code, also intelligence. They're surveying videos. That's intelligence adjacent. Sure, it's intelligence. And so in some highly reductionist sense, and you shouldn't take these things literally, but I'm just trying to highlight the really big directions here. All they buy is intelligence. And if You think forward with the fact that the price of intelligence is falling by a factor of 100 or something over the next five years. First principles... their costs kind of go away. And That's just thinking about the cost side of things. On the revenue side of things, there are all of the creators. You are a creator. How much more... quality creative can every creator that makes content for meta make? If they have access to intelligence, that's one one hundredth the cost and 10,000 times more effective. The point is probably better. And then meta in turn, maybe they don't take their costs to zero. Maybe they spend more or more than zero, and they do a far better job of picking the videos to show people next. which should make the content better targeted and more enjoyable. And then, of course, on the ad side, Advertisers could make far better designed, far better targeted, personalized ads that then meta in turn can target much better and advertisers can also build much more slippery funnels or whatever to convert. The whole point of all this is that maybe instead of taking their cost to zero, With much better intelligence at much lower prices, maybe they can sell a few more apps. All of this is downstream of what they possess – which is not destroyed by ubiquitous intelligence, which is data and attention. They just have this enormous amounts of data about all these people and how they view videos and whatnot. And they have an enormous amount of attention and these things compound each other. And so I don't think that the availability of cheap and ubiquitous intelligence is going to cause people not to be interested in watching funny 20 second videos. And so if anything, they can win some time on the margin from other things people do with their time. I don't actually own it, but a company that benefits... in theory from having a lot more intelligence is Grindr. Basically, Grindr is the gay dating app, but it has been constrained over its life by the fact that it's just not that big of a company. And it turns out they've got lots of attention. And they've got lots of data. But their ability to build all the things that you could imagine building on what ultimately amounts to... A finite size team and budget and ability to create things has meant in some ways that Grindr can't be all the things you could imagine it could be. And again, I don't actually own any Grindr. I study lots of companies and that just sort of occurred to me. This is another example of something where here, this mental model is what if you had 10,000 of Google's best engineers to go work on your problem for five years? What could you do? And Grindr is one where maybe they could do more. I want to talk about the labs. So back to the original point, I think lots of great investors would say intelligence, specifically frontier intelligence, is not a commodity. Historically, the conversation has been about the gap in intelligence between closed source and open source. Increasingly, the conversation is about the gap between what's publicly available from the labs and what's privately used by them inside their own businesses. They seem to have figured out how to stop or slow distillation. Lots of people would say that there's this escape velocity thing happening and that the frontier token is going to be way more valuable and useful and that the labs themselves can then distill their own models and provide the best version of the chief intelligence to. I'm curious for you to riff on the labs as companies, through this mind of an investor. If they have models that are better and the ability to produce adequate intelligence at way lower cost than everybody else and nobody can copy it, then it's going to be a great business. I guess I have a couple of thoughts. One is, as I've watched models get deployed, It feels like one way to think about it is you start out with these very inadequate models that are basically able to do a very small subset of jobs. And what happens is the models get better and they work their way up the stack. And with each successive step up, The frontier model has unlocked a larger, call it, addressable market. But behind it, what's happening is that once evals or whatever get saturated, you would never use a frontier model for a chat bot in a customer service application. You'd be an idiot. And so what's happening is those areas, they clearly commoditize. And so there's this wave of hard tasks where you really need a frontier model. And then behind that, you don't. So the friends I have who are big frontier labs companies, bulls or whatever, and obviously these companies' financials aren't public and they might have an information advantage over me. But- would say something like, don't worry, we can also, along the whole efficient frontier, produce cheaper tokens. And maybe they can right now, maybe they can't right now, maybe they will in a few years, maybe they won't. I find these very hard to think about. Certainly the trend has been rapid commoditization of these things. If you were to look at the rise of Chinese models and everything, [15:00] Three steps. And at a minimum, the cost has been collapsing. even if they do have a position where the tokens necessary to do tasks that are well behind the frontier. It might not be a very big market. because it just doesn't cost that much in the scheme of things to deliver and It is competitive. even if they are a cost leader in the shrinking commodity market. And then the other thing is that I think that programming was this huge use of models. As you push forward, the question becomes, for the next, next, next layers of intelligence. Do you still have bigger and bigger places to go after? And the bull story would be some version of, yes, people will pay anything for a model that can figure out how to cure cancer. Maybe that's true, but to go to our prior conversation, I think figuring out how to cure cancer requires touching reality, and I'm not sure that more IQ gets you that. If you don't believe that there's some super incredible use cases where you can just charge an enormous amount for more intelligence. Then you start getting to this point where the next step up in terms of you go coding and then maybe something else, but you get into smaller niches. It's like this diamond-shaped demand frontier where initially you get wider and wider uses, and then actually it starts to get more and more rarefied. How often do you really need a 300 IQ? One might start to ask. And look, we're going to find out. And I don't want to be too dogmatic. But it strikes me as totally reasonable that what we end up with is some version of there are very few use cases for super elite intelligence. There are lots of use cases for very good intelligence. That very good intelligence is highly competitive and commoditized. And if these frontier labs can somehow become cost leaders in doing that, there's a margin to make in being a cost leader. But then there's also just rampant deflation. And then this question becomes a lot like, How big is the market if good enough intelligence is very, very cheap? then how much good enough intelligence are people going to buy? I'll just keep playing the devil's advocate. So the two things people might say in this case are Javon's paradox that actually gets cheaper. We're just going to want to use more of it. I'm curious what you think about that. The second would be Historically, the new uses of intelligence have typically been discovered by highly intelligent people. My good friend says intelligence creates its own demand. What about that? What about the fact that maybe it's diamond-shaped for humans if humans are driving the new questions that the intelligence is set after to answer? But what if the thing itself is creating the new questions? And we train it to do that really effectively and just explore in an open-ended way in the way that all of humanity would be doing, led by the people with 150-plus IQ or something historically. - Shevins Paradox is just a narrow example of price elasticity and demand. In general, all markets are U-shape. So initially, when the product is terrible, There's no demand. And then the product gets better. and people will pay a lot. but they won't buy very much. And then as the price goes down, the size of the market will rise. because it becomes Jevin's paradox, essentially. But then you get to a point... where further price decreases don't actually drive enough demand and the market shrinks. And eventually, of course, if you take the price to zero, then the size of the market goes to zero. One example that I've thought about is the cost of fiber optic communication or just communications. The cost of moving information used to be super high and people moved very little of it. And then through a whole series of innovations, the cost of moving information went down a lot. And the amount of information moved went up a ton. And the market grew a lot for a long time. And then the market stopped growing. And the cost of moving information kept falling and the market kept shrinking. And nowadays, Cogent Communications, I think, routes 30 or 40 percent of all IP traffic across the Internet. And it's through one of its divisions, which is not even its biggest division, and Cogent Communications is... a few billion dollar market cap company, which is not a very big industry. Now, I'm not saying that we're going to get quite that small. in terms of intelligence. But Jevons Paradox is a statement about a particular spot on a curve, which doesn't necessarily have to be true. And as your other question about can more intelligence create more desire for intelligence, I mean, I have a couple of thoughts on that. Certainly, maybe. I started out by saying you need to touch reality to answer questions. And so if you had a model and it was going to think more about physics, it could certainly think lots of interesting thoughts about physics, but eventually it's going to need to do experiments. And that's not to say we couldn't also do experiments, but now we're bound by our ability to physically do experiments. I think there's going to be this incredible time where there's going to be so many cool discoveries, which we're already starting to read about, where this paper was written and this paper was written and this paper was written. And if somebody had read all three papers and thought about it, they would have realized this, but nobody had. But it turns out the models read all the papers and they'll figure that out. There's going to be all kinds of great stuff that falls out of that. But at some point, we're going to finish that. And then it's going to be about like, hey, there's some way to think about things that are outside of what we know. And we have a hypothesis, but a model, even a super smart model, won't be better at guessing that necessarily. We'll be better at guessing that than the smartest person. But the smartest person's odds are super low anyway. So it'll be able to come up with great ideas for experiments to do, and they'll be worth doing. But that will limit our ability to expand. The other thing I just think about is my own use of AI on a lot of levels. One, just to go back to the commoditization thing, I've totally commoditized my use of models for it. [20:00] It's because I'm lazy about how I do a lot of my building for stuff. I'll tell a GrokBot, to build things because I don't want to use up all my GrokBot wheat tokens. I then have it run all of them. I subscribe to all the things. and I'll have it build using those subscriptions. on a server. And so I actually don't know what model I'm using anymore. I used to know this model is better, this one's worse. I no longer know. All of this is being optimized on the back end for me. And all I know is that stuff is getting built. and it's not using up my subscriptions and so we're good. Vanta automates compliance so your team can spend less time on security reviews and more time getting customers. Vanta cuts audit prep by 82% and gives you instant up-to-date proof of trust without all the manual work, so you can close deals faster and with less friction. Customers report a 526% return on investment, and more than 16,000 companies use Vanta, including Ramp, Harvey, and Snowflake. Ridgeline is the first end-to-end system of record with embedded AI for investment management firms running portfolio accounting, reconciliation, reporting, trading, and compliance on one unified platform. Firms are moving off legacy technology and on to Ridgeline because of how far ahead Ridgeline's AI features are compared to anything else in investment management software. Thank you. There's this batch of companies that I'm curious what you think about, which is effectively real world experimentation companies that create... a loop that's automatic. There's this happening in robotics, there's this happening in several places in bio labs and automated wet labs, automated testing labs. There's, Any commentary on that land of company? Yeah, super great idea. Closing the loop in the physical world is going to be great and it's going to accelerate the process of learning. And this is a great idea. But again, there are speed limits. The thing I keep coming back to is some version of This is amazing. And I'm super, in a sense, AI-pilled in the sense of what I think these are things are going to do and they're useful for. But I'm also cognizant of the idea that things bump up against the way I like to tell this to people. As I said, look, you're really smart. So you must understand being smart has limits. Being smart has limits. And so the ability to just think your way to solutions is limited. And so if you're going to build an automatic biology testing laboratory, and you're going to connect it to a computer, and the computer is going to come up with an idea... and then it's gonna run an experiment, and it's gonna iterate. That's awesome. But if that experiment requires breeding mice, and mice need a year to grow up and die, Bad experiment, gonna take time. And it might be a great experiment. You might learn a lot. And that's great. Still going to take time. And the lab's only so big. And so you might have 5,000 great ideas. You're going to pick 20. And the model may be better. Your odds of that experiment marking might even be 1%. Turns out that might be 10 times better than the average researcher, and that's amazing. It's still 1%. None of this is to be pessimistic. This is super amazing. Discovering all the things that we should know because it's already in our corpus that we forgot is awesome. Discovering all the things and better directing our research is awesome. Automating that research to move faster is awesome. We will learn, it will push the frontier. But this is not like suddenly the model looks around and says, you guys missed it. Here's some new world that you guys can't comprehend or whatnot. I don't think that's where we're headed. It seems like the market has been, especially in private markets and the leading edge of investors thinking about AI, very focused on investing in supply related stuff, like how are we going to build and serve all the intelligence? versus what I'll call demand-related stuff. someone watching their personalized TV on their meta app or whatever. What do you think about that first batch of companies as investments? Super hard. I've joked I've been willfully ignorant about them and probably I'm sort of worse off for it. But there's this big buildup. And then you're trying to forecast inference demand. I can't forecast supply or demand within several orders of magnitude. This idea of how much more inference will people use as intelligence gets better and cheaper is a really, really hard thing. And equally hard is how much better and cheaper inference is going to get. And then even if you could somehow guess at those things, then there's the whole value chain and you need to be at the bottleneck in order to capture the rents. And then, of course, finally, you get to your actual competition with your competitors in that particular piece of the value chain. It's just a super hard thing. As I think about my use of AI as a consumer, increasingly I'm doing stuff like instead of having the AI chatbot talk to my daughter or text with my daughter about how to do metastatic, [25:00] Now I have it make a three blue, one brown style video explainer so that she can see a nice five minute or 10 minute video on powers or area circumference or something. It's super cool. I don't think I would do that if it was going to be $5. And so there's this thing where there's real pricey elasticity. The point I'm trying to make is I'll film myself working out with my Muse app and it'll show me tip pointers on my form. And I'll do that for very little money. But I won't do that if the price point is real. And I'm probably much less sensitive to the price than most people. AI demand is going to be really sensitive to the price. which intrinsically I think does limit in some sense the amount in total consumers are going to be willing to spend on this. Maybe this is a time to talk about capital cycles or something. It sure does seem like the smartest people in the world and many or most of the best entrepreneurs in the world are heavily personally invested in this supply build out thing. If you watch SpaceX's Investor Day yesterday, this is basically the theme of it. Arguably, maybe the greatest entrepreneur of all time is dedicated to effectively doing this, building out the data center supply, et cetera. This is just a historical question. If that's all true, and there's so much uncertainty about supply and demand, And there are other ways to go build businesses, products, make money. Why do you think so many of the best entrepreneurs, the smartest people, are so completely dedicating their lives to the build out of the supply. This is a huge infrastructure build out of a huge and important technology. And it would be impossible for there to be a huge infrastructure build out. of a huge and important technology. if there weren't a strong price signal. and a ton of talent going at it. So you may have your causality backwards. Obviously, these things work both ways. Could you imagine a telling of the story? where nobody interesting who'd made no money was building out the AI, right? That's probably the best answer. And to be clear, this is a huge and important project. It's only happening because super smart and talented people are focusing their lives on it. It's great that they are. They will have and deserve to make lots of money. It's just very, very hard. from where I sit. And there are probably people who are smarter and figured it out. From where I sit, it is very, very hard to figure out where one would invest along this and whether or not you'll give me money doing it. It's not to say that there isn't plenty of opportunity there. I've just sort of decided to focus my attention more on the longer term equilibrium results that come from what the world goes to. Instead of betting on which railroad tycoon is going to make money and how many railroads we need, I'm trying to think about, OK, once you have railroads, I'd love to have the foresight to figure out CPG brands come next. That's the way I've been trying to think about it. Can you riff a little bit about what you've learned about capital cycles and innovation cycles like this? I would love your perspective on having studied so much market history. if we're in the middle of one of these, maybe the biggest one ever, What does history teach us about these cycles and how they work that's interesting to you? You could think about American society right now as a household that's saving money for a house. We're building a house and we're all kind of feeling pinched because all these resources are dedicated to building out this big AI infrastructure. All these electricians and plumbers and everything else are going to do these tasks. And as consequence, there's higher costs elsewhere in the economy. And undoubtedly, we're going to get a lot of value for it. and we're going to keep pushing it. Because no one really knows. We're going to keep pushing it until eventually we overdo it. And... Along the way, people have to make tons of money. because otherwise people won't keep pushing it. And because things have lead times and because it's a winner's curse, it's a winner's curse. invariably the last person to stop. will regret it. And it's just so hard. It's just so hard. And then you combine that with the fact that there's just so much dynamism in the technology here, which is... I think was less true for other historic infrastructure buildouts. There's just so much change in terms of how you do these things. I mean, other people have made the point, I don't think it's necessarily wrong, that we talked about the models, but even at the chips level or whatever, that the potential for AI to commoditize even things like NVIDIA or... Recursive self-improvement everywhere at the... Everywhere. And a lot of it ends up being just intelligence. Like NVIDIA quite literally makes flat files that they send to Taiwan. Their flat files are very hard to make, but what they have is a lot of people who think really hard about it. And so in theory, you could make them. And also, then the question becomes, what if you step back a little from the frontier? Then the cost of making something comes way down. In that world, would your model be that physical and scale are the things that you'd want in a business that you were to back on the supply side of this equation? DSMC. I don't want to carry myself around as some sort of TSMC expert. I mean, certainly what they do has better attributes. I think in that regard, they've also got this sort of interesting problem where they did this super hard thing that requires all this process power built over many years and is super hard to replicate. and much less amenable to an AI just because it is so physical and because it's all these little things, many of which aren't even legible to an AI. It's like some guy who is in the system kind of knows that you have to do this. Because there's so much demand and because they haven't met it, they basically cause everyone in the world to seek to copy their moat. That's not a great fact, but I [30:00] More generally, stepping outside the AI build-out. If you think about businesses, we talked about the meta example. Just the idea of businesses that have physical proximity or physical scale might benefit if they can use intelligence to serve their customers better or have lower cost as a consequence. I want to go back through the natural compliments to intelligence. Just have you explain why it's a compliment to intelligence so that people can go away and think about, OK, what companies have this feature that might be interesting? Good example to go talk about physical proximity. How would you think about DoorDash or Uber or these businesses that have local density effects that matter a lot? Does intelligence being ubiquitous sort of naturally magnify all of the scale winners of this type where physical proximity is some sort of key advantage? Well, so if you're the low-cost producer of something... I'll give a sort of stylized example. So if everyone has an AI agent, then what happens is many decisions get offloaded to the AI agent because the AI agent helps you optimize, helps you pick up all the pennies and dollars and nickels and stuff that you were leading on the ground. So I think about consumer choices in a spectrum. On the one hand, it's maximizing my bank deposit. This is something where I think I can almost entirely defer to the highest judgment. As long as it's FDIC insured, I just want the highest rate. On the other hand, this end of the spectrum might be picking my favorite sweater. This is something where it's just all of my judgment. The AI can source me things to look at, but in the end, I need to decide it. I can't really delegate that judgment to it. And there's a spectrum. And so let's just pretend for a moment that garbage collection is something where I can defer the judgment. You might say, no, I really care about the guy and I trust the service and there's driving to my house or whatever. But let's just set all that aside for a second. So right now, if I'm in the waste collection business, I might be great at it. But it's still really hard to get people to bother to switch. I doubt you know what you pay for garbage collection off the top of your head. Many people do, but I doubt you do. And so what happens is you have to do all this work to get people to pay attention. to bother to switch garbage collection companies, even if you're offering a better service at a lower price. You could imagine that the air agent could talk to my agent and be like, hey, I'll save you 50 cents a month. And the agent will say, will the garbage go away? And I'm like, yes, the garbage will go away. And no problem, and I'll change. So the first order effect is that you move to the low-cost provider. But the general market effects are actually sort of interesting, too, because now let's say that there's a street. And before, it would be more efficient for society if all of the people on my street had the same garbage collector, because if I could drive up the street, pick up the trash, and he could offer all of us a lower price and make more money. But it's really hard for us to coordinate that because what happens is all of us are just lazy about switching garbage collectors. And so we all have one. And the guy with two guys on the street can't call everyone else up and say, hey, I've got two guys on the street. Everyone else has just one. I'm a low-cost producer. Let me roll this up. So a second order effect might be that The guy who has two people on my street can go to everyone else on the street and say, hey, I'll offer you a lower price on the garbage and everyone will switch. And so everyone moves the low-cost producer who in turn gets to be the lower-cost producer. That actually creates, in some sense, a better business where the low cost guy wins a bunch more volume at much lower margins. As you study them, a lot of companies have mixed effects that point in different directions because of these things. What do you think the impact of agents is going to be on your world? What will be your role? How will it be different in five years? Will the best agent be able to build a portfolio of Carvana and Hilton Grand Vacations and Meta in the same way that you would because you trained it how you think? Well, hopefully better. In terms of what I do, I think that in the long term... will probably get competed out. Right. Basic pricing risk as the thing you do. Yeah. Yeah. When computer-aided chess came along, there was this long span of time. where freestyle chess was better, where the human working with the computer program was able to outcompete either the human or the computer by itself. And this is just because humans thought about things differently and computers thought about things differently and they were synergistic. And I think in investing for a while, we're in that era now. We're in that era. And that's because there are things that I understand in the world in ways that the model just doesn't. by virtue of being a person and having lived a life. And there's information I have about the world that just isn't in the world. models, because as much information is on the internet, not everything is actually on the internet. While that's true, which is going to, I think, be a while. the ability of me to use models to assist me and then make judgments. We'll make better judgments working together. You could imagine a circumstance, though, in the future where that gap diminishes to the point where it doesn't matter anymore. And then I don't think models will be able to predict the future. As we talked about, the future is in some ways very intrinsically unpredictable. But the model's distribution of outcomes may be the best that you can do. And that adding a human to the forecast would be kind of like adding my judgment to whether or not it's going to rain in seven days. You just kind of would be like, Cliff, I don't really care what you think. I'm using the forecast. This is going to be better. However you fudge it, it's just going to make it worse. That day may come instead of an interesting moment there. There's a lot to investing. Think about judging management or something, but maybe the models will be able to judge management. They'll be able to interview them. They'll be able to figure it out. And then It gets really interesting. How do you raise capital? You'll be like, you know, I'd like to run an amusement park. And models say, sure, here are six locations that have a 51% chance of an amusement park working. And as long as you're really committed to it, and I've assessed you and you [35:00] a 51.2% chance of succeeding because you're actually slightly below average. You're like, okay, well, what can I do better than average? And I say, well, you know, have you thought about running a go-karting park? Fortunately, I don't think it's going to affect my career that much because I think it'll take a while. What about replacing price as the organizing principle and signal? that coordinates economic activity. Price is a miracle because it's a signal wrapped in an incentive. What does that mean? Price... is an incentive. It causes you to want to buy more or less of something because it goes up or down. And it's also a signal because it says something about the relative value of something and the scarcity of something and the cost to produce something. It's a miracle of a mechanism to coordinate things. And I think you'd be really hard pressed to find a better system, even with unlimited intelligence. It'd be pretty hard. Maybe another way of asking the question is kind of like investing. If ubiquitous intelligence gets so much better and so much cheaper, And in some way, it can replicate what you do, and you're really good at pricing risk. A friend of mine, Dan, was saying there's three levels to this. There's I put in a prompt and I get something back. That's level one. Level two is I set up a watchtower. My instinct every night runs like a set of tasks and it happens in the background on a schedule. And Level 3, which produces the most use of intelligence, is just always on proactive figuring out its own way to use its own intelligence. And that just goes on forever and it's unlimited in how much it can do. If I think about that level three applying to everything, what things should be priced, efficiencies, optimization, all this kind of stuff, does that add something interesting to the price of the price? sharpens markets. There's a lot of stickiness to prices. It's a well-known macroeconomic phenomenon, this price stickiness. There's a lot of lazily priced stuff. Think about deposits and such. I mean, it's a really competitive world out there, but it's actually not as competitive as in theory it could be. So I think... If everyone has agents who are working on their behalf and constantly optimizing, then I think the price of garbage collection will move around a lot more. As all the input costs to garbage collection evolve, and every garbage collecting company is constantly rerunning its financials and trying to figure out its true marginal cost and bid every customer where it could serve for more than marginal cost. And every agent is switching to the lowest cost provider. You won't just pay $60 a month for 12 months and then have the guy come back to you and say it's a 5% increase. It's going to be 63 next year. And you don't even notice they're just picking up your garbage. your garbage collection will vary by day. And, One week, it'll be whatever. There was all this interesting, much more accurate, much more rapidly moving prices, I would imagine. Have you seen anything interesting in capital markets innovation? There was this awesome economist story talking about the behemoth that is American capital markets and how good it is at adapting to new stuff and creating new instruments, new ways of financing things. and that that's been as much a part of the American story as the technology innovation. There's this amazing interrelationship between Wall Street and Silicon Valley would be one way of thinking about it. Anything interesting percolating for you in markets, in capital markets, in ways of financing things, things like this? Well, I mean, it's fun to take a moment to pause and maybe congratulate ourselves as capital markets for what we're doing with AI. People were worried about this decline in American dynamism. I think we just redeployed. What is it, 3% of GDP right into this highly speculative venture, basically on the turn of a dime? It's just this remarkable thing. Of course, who knows? Maybe it should be two, maybe it should be six. But it is amazing that we can redeploy resources at this scale. this quickly and there's no one else in the world that's been able to do it. And certainly... the inability to move investment in productive resources, can be a binding constraint. on productivity growth, even with, especially with rapid technological progress. And so at least with regards to capital, and obviously we talk about capital, that the capital investments are tied to people doing different things. Americans are doing different things today. Every day, people are waking up and going to someplace in Pennsylvania and building a data center. And that wouldn't be happening if it wasn't for this whole infrastructure we have that's somehow been able to move this enormous amount of capital. It's incredible. As deep innovation, I think the fundamental idea that people are taking risk on speculative ventures and slicing up the pie in different ways is irreducible. I think people come up with lots of different names for what are essentially the same things over time. I don't think that's deeply different, but it is incredible. I should also note that I'm sure we'll discover all kinds of fraud and bad behavior and stuff eventually when the water goes out. But I think given the magnitude of this, I suspect it's so clean. It's amazing. If you read about the railroad build out or something, the amount of waste and fraud, invariably there is something. But there's a lot of people trying to basically do the right thing here. And the governance parts that we just take for granted, it's incredible. America's capital markets and the capital markets in general, I think, are just incredibly underappreciated. The idea that any person can buy $100,000 of stock and get a portfolio of well-run, well-motivated teams, building companies that are [40:00] mostly the right things, then do it basically instantly and basically no cost, and then sell it as soon as they're done is amazing. And then the fact that all those companies and other ways people invest can then allocate society's resources really well in the scheme of things is just incredible. The whole system is just very underrated. Why has America so dominated this, do you think? My pet theory, and everyone has their own theories, is some sort of interplay between genetics, cultural evolution, and institutional evolution. So basically, the people who came to the United States were the subset of people who were most interested in taking risk because going to the United States was very risky. And I think that risk-taking has a meaningful genetic component. And so I think that tendency is just more common amongst Americans. When you're in a place where risk-taking is more the norm, there's also a whole bunch of other things that come from that. It's more the norm. And then that leads to buildups of cultural institutional things that reinforce that. It's understood that people can take risk and fail. That leads to bankruptcy laws that make sense That leads to labor laws that support that That leads to venture capital that supports that And all of those things reinforce that And then that leads to success stories get this whole soup But that my pet theory. So if there's a moon base at some point, you're going to be bullish on that community's ability to do really well economically. Yeah, actually, that's a great way to put it. The United States has been persistently innovative in a very amazing way. So if there's a moon base, I wouldn't be shocked if it's ours, even though the Chinese are brilliant and numerous and the Europeans exist. When you see rates go up, you see that great long-term chart of rates, let's say the 10-year rate in the US. And it just looks like this forever, missing our whole lives. And I was going, oh, comment on that chart. Well, I mean, there's so much. And I'm certainly no macroeconomics expert, nor I spend too much time thinking about it, but I'll make a couple of observations. One, obviously, is just in general, when you're a society and you're building it, you're making a giant investment in the future, means you're going to have to consume less today. Fundamentally, all of the resources going into this AI build out are about the future. Are about the future. And fundamentally, the plumber building that today is not making consumption today. He could be making today's present consumption. Instead, he's making a data center which is future consumption So one piece of a very complex puzzle that I thought a bit about is this idea that rates are higher because we inducing people to save for the future And the way you do that is you pay them People would kill me if I didn ask about your portfolio a little bit, which is so interesting to me. It's 80% Carvana, vacations, 5% meta is basically your portfolio. Rough justice, it's like five things. Sure. And one thing is 80%. So if you think about your portfolio and those companies, which obviously you know extremely well, and you were advising them on how to think about using AI to transform their businesses. What's that speech look like? Well, it starts by making sure that everyone is vibe coding. Starting at the board and the CEO, everybody needs to have hackathons. And of course, I brought this to Carvan and they said, yeah, we call them pizza parties. I don't mean the developers. I mean, everybody needs to have hackathons. And it's not so much that you're going to build hugely useful software necessarily, that comes out of these hackathons, but it's that you're gonna teach everybody what can be done. And that's going to change their expectations, I think, for what can be done. And then the second thing which I tell them is the CEO needs to basically turn to the head of IT and say something to the effect of, so I want my Grokbot, Metamuse, agent, whatever, to be able to access every system in the company on a read basis Everything I could look up if I asked somebody I want to be able to get that in a prompt And the guy going to say that really hard We have umpteen trillion problems with that And he going to say that my point I want that And that's a big process and it's system by system and all the rest. But if you can get to that, then that gets you read-only access. To the starting line. Yeah, to the starting line. That gets you read-only access. And then the next logical thing as a CEO needs to use this everywhere and everyone is going to get jealous. And you're going to say, oh, you can have it too, but we need to work out your permissions. And then that in theory percolates down and now you've got permission to read only access. And then invariably someone is going to say something like, I'd really like to put data into the system. And you're going to say, that's fantastic. We got to work out right permissions. And then someone's going to say, I'd love to change a system because we just did this vibe coding thing and I'm confident we could do it. And you're going to say, that's also fantastic. We need to work out a process by which we're going to manage changes to the system. And what I just described, sure, it's four bullet points. That's like years of work. It's impossibly hard. So the question is, are you in a business where you have time for that to happen before someone who starts- Or the jump ball, someone else that starts from go. Exactly. And so it's good to have a really good business. It seems like one of the conclusions from that is it's the best time ever to start companies, both because of capital market conditions that- [45:00] People will fund... at fairly low cost of capital, highly experimental things. but also just because with these tools, if you start without the baggage, you can go so far so fast. Do you think that's roughly right? I mean, yeah, it's the best time to start a company until tomorrow. So I agree. It's amazing. I'd also just say that just thinking from my perspective. People are such a... Pain in the butt. If you're running a company, there's such a pain in the butt. And the idea that you have all these agents and stuff, it makes it so you can do so much more. with so many fewer people. Of course, one of the challenges always is if you can easily do something, then so can everybody else. So there has to be some wedge by which the thing you build... produces enduring value. Does it have a path where you end up with some sort of lock-in or proprietary data in this world where all the agents are happily changing providers every day. What gives you any enduring economics? But then, of course, then there's the further thing is that obviously if you do the thing where it seems like there's an obvious path to enduring economics – then that's a bad thing to do because everyone is trying to do that. So it's very competitive. And the really great outcomes are the ones that make no sense. You shouldn't do it at all. But then it turns out there is a path or the world turns. Your finance team isn't losing money on big mistakes. It's leaking through a thousand tiny decisions nobody's watching. Ramp puts guardrails on spending before it happens. Real-time limits, automatic rules, zero firefighting. Try it at ramp.com slash invest. As your business grows, Vanta scales with you, automating compliance and giving you a single source of truth for security and risk. Learn more at vanta.com slash invest. away the complexity across portfolio accounting, reconciliation, reporting, trading, compliance, and more, all at scale. Schedule a demo at Ridgeline.ai. Every investment firm is unique, and generic AI chatbots don't understand your process. Rogo is built by finance professionals. It's connected to your data, learns your process, and quickly produces client-ready outputs. Check them out at rogo.com slash invest. The best AI and software companies from OpenAI to Cursor use WorkOS to become enterprise ready overnight, not in months. Visit WorkOS.com to skip the unglamorous infrastructure work and focus on your product. What do you think are the most underrated sources of future competitive advantage then? If ultimately... Good businesses need sustainable edge, sustainable competitive advantage to have great profits. Anything underrated? Underrated. We talked about attention and data and physical proximity. These are examples of things that might be super or more valuable in the future than they are today because of cheap intelligence. Is there any type of it? Maybe it's regulatory capture. Are there any types of edge that you think are underrated? I get the sense that a lot of these are appropriately weighted. I guess one interesting question I've had, two of these, one is, if you think of corporate margins in aggregate, do they go up or down? Like this. Right. Do they go up or down? And I think everything we're talking about, are you down? There's many businesses where they make margin because more consumer surplus, more consumer service everywhere, everywhere. What's interesting is if you go back and read a lot of the research reports from the 90s when the Internet was coming along, they said the same thing. So the base rate here might not be great. Maybe we're wrong. But I think the main reason they were often wrong was this aggregation economics, the Van Thompson aggregation economics. So maybe absent that, maybe it wouldn't have been so wrong. And I do think that a lot of aggregation theory breaks down because that basically relies on the idea that consumers only look at the first thing on Google. But it turns out if your agent looks at everything unsearched, then being ranked first doesn't matter. And all the core things like quality and price and everything else, service become more salient. But the other thing is that maybe another downstream effect of this is you'll be able to build more complicated businesses. Right now, a lot of industries are more horizontal in structure because the complexity of running your slice is very high. And that makes it hard to do what's above and below you and not have undue complexity. With AI agents and basically cheap and ubiquitous code and intelligence, you could build the whole vertical stack. People are trying to do this, I think, in other countries, but you can have a full stack pet care business that does everything from make pet food all the way up through pet insurance and pet hospitals and pet vets and all these things that historically were multiple layers. But now, because you can solve the coordination and reduce the principal agent problems, it's a big deal. By having AI do it, maybe those sorts of businesses are possible. And then maybe those businesses are higher margin in the end, more efficient and more consumer surplus. Carvana is an example of that built before this was easy to do, but maybe people can do it in other verticals. In which case, maybe margins go up and there's consumer surplus. The other one is one other thing there is just how this affects wealth and income distribution. I can tell two stories. On the one hand, I think a lot of people have talked about this idea that it concentrates income distribution. You've got all these people who made money on various labs and whatever data NVIDIA. And this idea that there are increasing returns to small groups of people But the counter argument is if [50:00] We have cheap and ubiquitous intelligence. a couple of things happen. One is there's all these rent takers in the economy. Think you're the software company that charges bowling alleys, And those guys all get smoked, which sucks for them, but it does lower the cost of bowling. And so then that increases living standards for everyone who bowls. And that's flattening because my spending on bowling is small proportion to my income, but people who make less money spend more proportion to their income on bowling. And so that's a flattening effect on income. if all these little rent... pieces go away. the decline in corporate profits, all the money that the average American is giving to Bank of America goes away from Bank of America, the capitalists lose. the average consumer. wins sets leveling. The other piece is there's a whole professional class who basically you could define their role as intelligence. Intelligence. They take big corpuses of information and apply them to certain circumstances in a fairly straightforward way. And if those all get commoditized, one, everyone gets the best ubiquitous, which is one, good for everybody, but two, particularly good for people who can't afford the best right now or any. But then three, those individuals, they earned these extra rents because they were... smarter. They then have to drop to their next best alternative. And in theory, then that might flatten the curve. Now, the counterargument to that might be something like, well, yes, but returns to initiative and ambition and intelligence could go up. Because now I can do so much more because these models are like a bicycle for the mind on steroids. They give people who are smart and have initiative the ability to accomplish so much more. Maybe that actually causes increasing returns to that subset of people and it steepens the curve. Truman asked for a one-handed economist. I seem to very distinctly have two hands on everything. I think that's a big, interesting question. They're super interesting questions. And I find that a lot of situations I look at, there's a lot of cross-currents. We've started the conversation with the AI build. On the one hand, I can see all these incredible ways to use AI. And I use a gajillion tokens. And I can see how I could use a gajillion more. And obviously, I'm in the far right end of the distribution. Everyone uses as many tokens as me. We are so short. It's going to be this huge thing that's going on forever. And everyone who's in the build is going to make a fortune. On the other hand, the amount I get per token is like going to the moon. And it turns out I'm a pretty unusual guy in a lot of senses. And people are going to adapt more slowly. And the cost per token is going to collapse. And so maybe actually we're totally overbuilt. I can tell both stories with equal certainty. I have a good friend who has been running this very cool experiment where If you break investing down into picking what to own, And then how much to own of each thing that you've picked. Overly simplified, but interesting way of thinking about it. He's been running this experiment where he continues to pick the things to own. but then runs his own portfolio construction as he would have in the past in one lane. And in a second lane is just completely handing that off to an AI harness thing that he's built and says, you do all the portfolio construction, but stick with the stuff that I've picked. And what he's found so far is that that is drastically outperforming his version of portfolio construction, which itself is quite sophisticated. And the simple argument that he would make is, well, AIs are just really good at optimization. And they've read every paper on portfolio construction and all this kind of stuff and bet sizing and Kelly Criterion. They know it all. And that in the future, perhaps we make that transition where the investor's job is just picking what to own and not how much of each to own. And I ask you this question, what you think about this idea and the future, because you have such an unusual portfolio where. Again, 80% of it is in one specific asset. And my guess is if you were in lane two, AI would weight your portfolio very differently than you have chosen to do it. I'm underweight. So I'm just I'm dying to know what you think about this idea. That's clever. Underweight Carvana. Should be 90. Time scales probably are interesting on that to think about, too, because I think about how AI is going to encroach into this business. I think it's going to start at shorter time scales and work its way up to longer time scales. Because one of the challenges is that there's no verification loop. your friend's ability to say that, is based on results over, I mean, certainly not 20 year results. I strongly suspect that the march of AI into markets, and I mean, it already started. I don't know how Renaissance Technologies did it, but I assume... Seems like they might have discovered the Transformer for everybody else. It's like, I'm not kidding. It wouldn't shock me if... People who invest over shorter timescales, starting with seconds and working their way out to minutes, hours, weeks, months. find themselves delegating more and more to models. Makes perfect sense. And that you could definitely imagine bifurcating that between trying to enrich the model. I think other people do things like this already. And this is just a more sophisticated version, but you can have a fully quantitative strategy and then you can enrich it with what you think is qualitative human designed analysis. And then you can have [55:00] quantitative strategy to incorporate the qualitative. I think many of these multi-strat do these sorts of things. And what he's doing is just a slightly different flavor of that. Makes perfect sense. There are so many ways to be successful and models have just added more dimensions of freedom. But what about your own portfolio or more about how you feel comfortable having a position? Forget the company that it is. Sure. That's that big in the face of so much research that would suggest that's a crazy way to Build a portfolio. A lot of that research, I think, is more looking at price changes in terms of securities prices. and less focused on thinking about it the way someone who owns a company might think about it. I mean, we're talking about Carvana. We're not, for my sense, that Carvana's infrastructure and brand and everything else is just so differentiated that there are few scenarios where it works out. really badly. then I wouldn't feel so comfortable owning so much of it. It might just be as simple as that. That portfolio approach would have worked absolutely. Certainly, if you discovered Walmart or Home Depot or Fastenal or something back in the day, you wouldn't have taken full advantage of it. And that's kind of my sense with Carvana. If you read history, of course, all the people who did really great tended to be very concentrated. But of course, the history is literally... The driver should buy a sale. Yeah, or the history of the deal didn't. I also think that a lot of that literature is what most people who are looking for average or slightly better than average results should do. And it's totally the right thing to do. But that's not what I'm looking for. I like Carvana a lot. "'It wouldn't upset me, Annie, if—' Stock went up a bunch and I felt compelled to diversify it. That would also be okay. It's just more of a trade-off between expected returns and knowingly taking an unusual risk. What do you think about education? You and I have kids of roughly similar ages. Comment a little bit on how you're thinking about their educations, the institutions that have driven education in this country that have been really important in that whole path dependent story you've described. What do you think the present and future of education looks like or should look like? I've been talking to the head of my kid's school about this, and more and more it seems to be that his thinking... and my thinking are heading in this direction of the way you're educating kids is getting more bimodal in terms of what you're looking for. Let me step way back. What is the benefit of intelligence? personal intelligence, like your own intelligence, in a world where there's infinite intelligence for models. You might say, oh, it's useless. Doesn't matter. Someone with a 50 IQ can navigate the world just as well as someone with a 150 IQ. And I don't think that's quite right. And intuitively doesn't feel right. But I think to nail down where it is, is that the more intelligence you have and the more crystallized intelligence you have, which would be sort of education. the better you are at communicating with a model and telling a model what you want. I would have a very tough time having a deep conversation with a model to create leading quantum physics experiments because I don't know anything about quantum physics. And the model could try very hard to explain things to me, but in the end it would be explaining things in Crayon and Nerf. And it would just be very hard to get to the leading edge and get the best out of what the model has to offer. Whereas in an area where I do have some expertise, the model and I can communicate effectively and I can direct the model to my will much better. And so to direct the model, to do your will, which I think is going to be an important skill in the future. You do have to understand something about the world. And smarts help for that. And education helps for that. You can't ask it to assume it's a stochastic process unless you know what a stochastic process is. It could teach you what one is, but that's a lot of friction in figuring this all out. You don't have to work out the math, but you have to know what it is. And so when I've been thinking about education, there's this idea that you're teaching your kids a bunch of foundational skills that build on one another. Where I think in the end, the sum total of those, they have to do them without models, not because they're going to do them without models in the future, but because if they don't know how to do multiplication, it's going to be really hard to understand matrix multiplication. And if they don't know what matrix multiplication is at a high enough level. They don't have to be able to do it really well, but they have to be able to understand what the model is doing when it's doing it. And that's so they can direct the model. And so – Increasingly, I'm of the view that we should spend a bunch of time with kids basically separated from these models. Teaching them basic skills that we think are important and what those are is a really hard thing. But the idea being that, yes, model can do it, but you need to know it so that you can tell the model what to do. And then on the flip side, some English teacher is like, I don't like models because the kids just cheat on their Shakespeare essays. I'm like, you're not being ambitious enough. The kids are cheating on a Shakespeare essay. Give them something harder to do. Give them something a hundred times harder and more ambitious and have them use models to their fullest extent. Have them bring the work product back and then make sure that they didn't just give a prompt to the model, do the homework and make sure they own it. Ask them why it's like this. Force them to justify it. And I think that's another really important thing because kids need to be good at using them. And they also, frankly, just need to be taught that they can do anything. You and I get to talk every so often on the phone every time I wish it was recorded so that other people could hear it because I learned so much and... I think in this particular conversation today, which is sort of a version of a call we had several months ago, It's just such a valuable set of food for thought about. [60:00] the future of intelligence and what that means for markets and companies, and is a little bit, not in opposition to, but A different perspective than the most common one that you're hearing which is unlimited build out forever, short everything forever, no cap on the demand for intelligence. And I've loved talking to you about it both on the phone and today. So thank you so much for your time. Thank you. If you enjoyed this episode, visit Colossus.com. You'll find every episode of this podcast complete with hand edited transcripts. You can also subscribe to Colossus, our quarterly print, digital and private audio publication featuring in-depth profiles of slash subscribe. Patrick O'Shaughnessy is the CEO of Positive Sum. All opinions expressed by Patrick and podcast guests are solely their own opinions and do not reflect the opinion of Positive Sum. This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions. Clients of Positive Sum may maintain positions in the securities discussed in this podcast. To learn more, visit PSUM.VC. Outro Music. You know how small advantages compound over time? That's true in investing and just as true in how you run your company. Your spending system is your capital allocation strategy. Ramp makes it smarter by default. Better data, better decisions, better economics over time. See how at ramp.com slash invest. As your business grows, Vanta scales with you, automating compliance and giving you a single source of truth for security and risk. Learn more at vanta.com slash invest. The best AI and software companies from OpenAI to Cursor to Perplexity use WorkOS to become enterprise ready overnight. not in months. Visit WorkOS.com to skip the unglamorous infrastructure work and focus on your product. Ridgeline is redefining asset management technology as a true partner, not just a software vendor. They've helped firms 5x in scale, enabling faster growth, smarter operations, and a competitive edge. Visit RidgelineApps.com to see what they can unlock for your firm. Every investment firm is unique and generic AI doesn't understand your process. Rogo does. It's an AI platform built specifically for Wall Street, connected to your real outputs. Check them out at rogo.ai slash invest.