Eric Vishria - A Decade of Lessons Investing in Software & Hardware - [Invest Like the Best, EP.486]

频道: Invest Like the Best with Patrick O'Shaughnessy
视频: https://colossus.com/episode/sandcastles-and-silicon/
原文语言: en
统计: 共 70 轮 · Patrick O'Shaughnessy 12 · Eric Vishria 58

[0:00] Patrick O'Shaughnessy

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 customers grew revenue 3.2 times faster than the average American business.Visa, Vercel, Cursor, Stripe, Notion, 11Lab, Shopify, and 70,000 other businesses all runon Ramp.Mine does too, and so should yours.Learn more at ramp.com slash invest.OpenAI, Cursor, Anthropic, Perplexity, and Vercel all have something in common.They all use WorkOS.To achieve enterprise adoption at scale, you have to deliver on core capabilities like SSO,SCIM, RBAC, and Audit Logs.Instead of spending months building these mission-critical capabilities yourself, you can just use WorkOSAPIs to gain all of them on day zero.That's why so many of the top AI teams you hear about already 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.Felix by Rogo is a personal finance agent that turns a single prompt into finished,client-ready work using your firm's own templates, context, and standards.Send Felix an email like,take these comments and turn them for me,

〔广告〕Ramp 是唯一一个专为让你的财务团队更精简、更快、更好而打造的平台，平均每年为企业节省 5% 的开支，让你可以专注于增长。Ramp 的客户收入增速是美国企业平均水平的 3.2 倍。Visa、Vercel、Cursor、Stripe、Notion、ElevenLabs、Shopify，以及另外 70,000 家企业都跑在 Ramp 上。我自己的公司也在用，你也该用。访问 ramp.com/invest 了解更多。〔广告〕OpenAI、Cursor、Anthropic、Perplexity 和 Vercel 有一个共同点：它们都在用 WorkOS。要实现规模化的企业级采用，你必须交付 SSO（单点登录）、SCIM（账号自动同步）、RBAC（基于角色的权限控制）和审计日志这些核心能力。与其花好几个月自己造这些关键能力，不如直接用 WorkOS 的 API，第 0 天就全部拿到。这就是为什么你听说过的这么多顶尖 AI 团队都已经跑在 WorkOS 上。WorkOS 是成为「企业就绪」最快的路径，让你专注在最重要的事情上——你的产品。访问 WorkOS.com 开始使用。〔广告〕Rogo 出品的 Felix 是一个个人金融智能体，它能把一句提示词变成可以直接交给客户的成品，用的是你所在机构自己的模板、上下文和标准。给 Felix 发一封邮件，比如「把这些批注帮我处理成稿」，

[1:11] Patrick O'Shaughnessy

or update my tracker with the context of these emails.And Felix sends back finished PowerPoint decks, Excel models, and sourced research.Felix works the way your team already does,delivering work quickly and accurately around the clock.Learn more at rogo.ai slash Felix.Hello and welcome, everyone.

或者「用这些邮件里的上下文更新我的追踪表」。然后 Felix 会把做好的 PowerPoint 演示稿、Excel 模型和带出处的研究发回来。Felix 按你团队本来的工作方式干活，全天候地又快又准交付成果。访问 rogo.ai/felix 了解更多。〔正片〕大家好，欢迎收听。

[1:30] Patrick O'Shaughnessy

I'm Patrick O'Shaughnessy, and this is Invest Like the Best.This show is an open-ended exploration of markets, ideas, stories, and strategiesthat will help you better invest both your time and your money.If you enjoy these conversations and want to go deeper,check out Colossus, our quarterly publication with in-depth profilesof the people shaping business and investing.You can find Colossus along with all of our podcasts at Colossus.com.Patrick O'Shaughnessy is the CEO of Positive Sum.All opinions expressed by Patrick and podcast guests are solely their own opinionsand do not reflect the opinion of Positive Sum.This podcast is for informational purposes onlyand 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.I love asking you and all your partners this every time we hang out,

我是 Patrick O'Shaughnessy，这里是《Invest Like the Best》。本节目是对市场、想法、故事和策略的一场开放式探索，帮助你把时间和金钱都投得更好。如果你喜欢这些对话、想更深入，可以看看我们的季刊 Colossus，里面是对那些正在塑造商业与投资的人的深度特写。Colossus 和我们所有播客都能在 Colossus.com 找到。Patrick O'Shaughnessy 是 Positive Sum 的 CEO。Patrick 和播客嘉宾表达的所有观点都仅代表他们个人，不代表 Positive Sum 的观点。本播客仅供参考，不应作为投资决策的依据。Positive Sum 的客户可能持有本播客中讨论的证券的仓位。了解更多请访问 PSUM.VC。〔对话开始〕我每次跟你们见面都特别爱问这个问题：

[2:24] Patrick O'Shaughnessy

which is, okay, you've got these singular investments.You don't do that many investments each per year.And then the ones that go on to work, so far Sierra and Fireworks certainly are.You get to learn so much about the world through the lens of the company.So I'd actually love to do both of those, maybe starting with Fireworks.What do you know or what have you learned about the world and how it's reordering itselfbased on watching the world through the lens of Fireworksthat would maybe be surprising or interesting?

就是——好，你们做的都是那种独一无二的投资，每人每年做的笔数也不多。而那些后来真跑出来的公司，目前来看 Sierra 和 Fireworks 显然都跑出来了——你能透过一家公司这面镜头，学到关于这个世界的太多东西。所以这两家我都想聊，也许先从 Fireworks 开始。透过 Fireworks 这面镜头看世界，你知道了什么、学到了什么关于这个世界以及它正在如何重新排序的事，是可能让人意外或者觉得有意思的？

[2:49] Eric Vishria

One really interesting thing is these models are big.These are 2 trillion, 3 trillion, 4 trillion parameter models.It turns out running those models is damn hard.And running them efficiently is like super hard.The way to see this, and everybody can see this,which is everybody from AWS to Azure to GCP to the NeoCloudsto the Fireworks-based and togethers of the world, like all of them,they all run these stock open source modelsthat are available in their developer pools and everything else.And that's what it is.The performance difference for a Fireworks versus a cloud provider is like 5x.And that is just the speed performance.Then you add on top of that throughput, which is not visible externally.It's only visible if you know the economics of these businesses.And you're like, wait a minute.This is the same open source model with the same NVIDIA hardware.And there's a 5x performance difference and a multiple x throughput difference.And the way to just simply understand that is these companies are payingthe margins of the cloud providers and running on top and making money.How can that be?

有一件事特别有意思：这些模型很大，是 2 万亿、3 万亿、4 万亿参数的模型。事实证明，把这些模型跑起来难得要死，而要跑得高效，那更是难上加难。怎么看出这一点？其实所有人都看得见：从 AWS 到 Azure 到 GCP，到那些新型云厂商（NeoCloud，指专门做 AI 算力出租的新一代云公司），再到 Fireworks、Baseten、Together 这一类公司，全都一样——它们跑的都是同样的开源模型，都摆在各自的开发者池子里。就这么回事。可 Fireworks 和一家云厂商之间的性能差距能有 5 倍。而这还只是速度这一项。你再加上吞吐量（throughput，单位时间能处理多少请求），这个从外部是看不见的，只有你懂这些生意的经济账才看得见。然后你就会想：等一下，同样的开源模型、同样的 NVIDIA 硬件，居然有 5 倍的性能差距和好几倍的吞吐量差距。最简单的理解方式是：这些公司是在付着云厂商的毛利、跑在云厂商上面，还能赚钱。这怎么可能？

[4:06] Eric Vishria

My big takeaway on it was, wow, this stuff is actually really hard to run.It's just really hard to run.And there's a lot of expertise involved in doing that.And it's this very specific expertise that exists.And it is the kind of thing that when you look at it as an investor from the outside,and we should talk about early days of AWS, but when you look at it from the outside,it's like, this is a commodity.This is just as like pass through resale game.Yeah.Scale game, like whatever.And you're like, oh, wait a minute.No, it turns out it isn't.It isn't at all.Do you think that's just a moment in time thing?

我最大的收获是：哇，这东西真的很难跑起来，就是非常难跑。这里面涉及大量的专业能力，而且是非常具体的那种专业能力，是真实存在的一种手艺。这种事情，当你作为一个投资人从外面看——我们待会儿该聊聊 AWS 的早期——从外面看的时候，你会觉得：这就是个大宗商品生意嘛，就是个转手倒卖资源的活儿。（Patrick）对。（Eric）规模游戏，随便你怎么说。然后你才发现：哦，等等，原来不是。压根就不是。（Patrick）你觉得这只是一时的现象吗？

[4:35] Patrick O'Shaughnessy

And I'd love to just hear you riff on like cloud.You watch cloud very carefully and closely.You know a lot about it.And the adoption curve there versus how people use these things and the nature of those twobusinesses in comparison.It's tempting to say there will be one or two scale winners like there typically havebeen in a commodity market.Totally.Where cost to serve is everything and scale drives cost to serve down.And like, that's the whole story.I think the AWS example is so good.Okay.So 2006, you have S3 and EC2 watch, right?

我特别想听你聊聊云。你一直非常仔细、非常近距离地观察云，你对它了解很多。云的采用曲线，跟人们使用这些新东西的方式，以及这两门生意的本质，能不能对比一下？人们很容易得出结论说：最后会像典型的大宗商品市场那样，只剩一两个规模赢家。（Eric）完全同意。（Patrick）在那种市场里，服务成本就是一切，而规模会把服务成本压下去，故事就这样结束了。（Eric）我觉得 AWS 这个例子太好了。好，2006 年，S3 和 EC2 发布，对吧？

[5:03] Eric Vishria

Their compute platform and their storage platform is the first two AWS offerings in 2006.And you kind of start talking about it in late 2006 or whatever.2006, 2007.I think the 2007 annual letter, Bezos talks a lot about AWS and why it's important, whyit's interesting and everything else.And the investor reaction is just not good.I think if you put 30 of the smartest investors at that time in a room and ask them, what's theprobability that this AWS business is a good business with durable long-term margins andlike super interesting and everything not commodity, I think you would have gone zero for 30 withreally smart people that you and I know who were around at that time in 07.Fast forward from 07 to 2014.I joined the venture business in 2014.And a really common narrative in 2014 was, oh my God, AWS is going to eat everything.There's no enterprise opportunity left.It's going to eat databases and infrastructure, but it's going to eat the apps too.And they're going to offer it the cheapest and best.And we all have these like amazing SaaS and software businesses that we were involved withor investors.And part of the reason people loved them was they were annuities and they ran at 85% gross

它们的计算平台和存储平台，是 2006 年 AWS 最早的两个产品。大家大概是 2006 年底开始讨论它，2006、2007 年。我记得 2007 年那封年度股东信里，贝索斯用了很大篇幅讲 AWS，讲它为什么重要、为什么有意思等等。而投资者的反应就是很糟。我觉得如果你在当时把 30 个最聪明的投资人关在一个房间里问：AWS 这门生意成为一门好生意、拥有可持续的长期利润率、非常有意思、完全不是大宗商品的概率有多大？我觉得你会 30 个人全军覆没，而且是你我都认识的、那个年代非常聪明的人。从 07 年快进到 2014 年——我是 2014 年进的风投行业。2014 年一个非常流行的说法是：天啊，AWS 要把所有东西都吃掉，企业级机会已经没了。它要吃掉数据库和基础设施，连应用层也要吃掉，而且会提供最便宜、最好的产品。而我们手上都有这些超棒的 SaaS 和软件生意，我们要么在里面干活、要么是投资人。大家喜欢它们的一个原因是：它们的收入像年金一样稳定，跑着 85% 的毛利率。

[6:11] Eric Vishria

margins and everything else.And so it was just like, oh my God, AWS, Amazon can offer things at 8% gross margin.And like, they're just going to crush their margin.Yeah, your margin, my opportunity, like the whole thing, whole narrative.And think about from like 2014 to now in enterprise, of course, you have Snowflake,direct competitor to Amazon Redshift, ran on Amazon.You're out Amazon-ing Amazon on Amazon.But it's not just Snowflake.You had Confluent and Elastic and Mongo, Databricks, all of these companies.Amazing.That's the infrastructure layer.Then you have the whole app layer.In the app layer, think about offerings that they offered at the beginning.Datadoc, $100 billion company today.They had a competitive offer and they did it.And of course, there was tons and tons of roadkill.There was tons of roadkill.They did run over a bunch of stuff.But even then in 2014, the thesis was the view that AWS was going to eat everything wasmassively wrong, not because of all the examples I just mentioned.It was massively wrong because Azure and GCP were irrelevant then.And fast forward to 2026 and they're unbelievable businesses.Is AWS the biggest?

所以当时就是：天啊，AWS，亚马逊可以按 8% 的毛利率提供服务，他们会把我们的利润率直接打穿。（Patrick）「你的利润率就是我的机会」，整套叙事都是这个。（Eric）但想想从 2014 年到现在的企业软件：当然有 Snowflake，亚马逊 Redshift 的直接竞争对手，而且还跑在亚马逊上。（Patrick）在亚马逊上、用亚马逊的打法打赢亚马逊。（Eric）而且不只是 Snowflake，还有 Confluent、Elastic、Mongo、Databricks，所有这些公司。（Patrick）太惊人了。（Eric）这是基础设施层。然后还有整个应用层。在应用层，想想亚马逊一开始推出的那些服务。Datadog，今天是一家 1000 亿美元的公司，亚马逊出过竞品，也确实做了。当然路上也确实有大量「被碾死的」（roadkill）。（Patrick）确实碾死了一堆。（Eric）他们确实碾过去了一批东西。但即便如此，2014 年那个「AWS 会吃掉一切」的判断错得离谱，而且不是因为我刚说的那些例子。它错得离谱是因为：当时 Azure 和 GCP 根本不值一提，而快进到 2026 年，它们都是好得不可思议的生意。（Patrick）AWS 还是最大的吗？

[7:21] Eric Vishria

I think it's like a 40, 30, 20 split.You ended up with an oligopoly of that.And even outside of those big three, you have Cloudflare, which is a cloud provider in adifferent sort, which is another $100 billion company.So you have these smaller players that emerged as $100 billion companies outside of it.So what's the takeaway?

我记得大概是 40、30、20 这么个格局。最后形成的是一个寡头垄断。而且就算在这三巨头之外，你还有 Cloudflare——另一种形态的云厂商，同样是一家 1000 亿美元的公司。所以在三巨头之外，还冒出了这些成长为 1000 亿美元级别的「小玩家」。（Patrick）那么结论是什么？

[7:40] Eric Vishria

The takeaway to me is, oh, there's just a bunch of zero-sum thinking and not realizinglike how big, what if it all works?What if it all works?It all works.And of course, getting the relative winner right matters.And there was roadkill.And so it's all those things still matters.I'm not saying spray and pray.I'm not saying that at all.But I'm just saying the market was so big that one vendor could not scale and consumeit all.And they just couldn't consume the whole industry.It's different now and like all these things.But just that whole notion right now with what you and I are seeing in AI sure feelslike it rhymes.Anthropics are going to do everything.Right.Really?

对我来说结论是：当时充斥着大量的零和思维，没有意识到这个市场能有多大——「如果一切都成真了呢？」如果一切都成真了呢？结果就是一切都成真了。当然，押对相对赢家仍然重要，路上确实有被碾死的，这些都还是重要的。我不是说撒胡椒面式地乱投，完全不是那个意思。我只是说：市场大到一家厂商根本没法规模化地把它全吃下去，他们就是吞不下整个行业。今天情况又不一样了等等这些。但眼下你我在 AI 里看到的这一整套说法，感觉真的很押韵：Anthropic 什么都要做。（Patrick）对。（Eric）真的吗？

[8:19] Eric Vishria

Is that really right?To me, that view that it's just like, oh, this one company is going to eat it all doesn'thold.And I would tell you that scaling, while Cloud scaled very quickly, Cloud did not scaleanywhere close to as quick as what's happening right now.For that company to actually scale and deliver it, scaling in this case requires a ton ofinfrastructure buildup from energy, power, shell, chips, memory, obviously the algorithmsand everything else on top.It feels to me like we're going to end up with an oligopoly of winners.I really believe they will be these like $100 billion crazy smaller winners.Just feels like the same thing is kind of happening.Can you talk about it also from the demand side and compare it to how you watched Cloudget adopted in enterprise versus how you're seeing AI get adopted now?

真的是这样吗？在我看来，「这一家公司会把一切都吃掉」这个观点站不住脚。而且我要告诉你，虽然云当年扩张得很快，但云的扩张速度跟现在正在发生的事情根本不在一个量级。要让那一家公司真正扩张起来并交付出来，这一轮的规模化需要海量的基础设施投入：能源、电力、数据中心壳体（shell）、芯片、内存，当然还有跑在上面的算法等等。我感觉我们最后会得到一个寡头格局的赢家群体。我真心相信会出现那种 1000 亿美元级别的、疯狂的「较小赢家」。感觉就是同一件事又在发生一遍。（Patrick）能不能也从需求侧讲讲？对比一下你当年看到云在企业里被采用的过程，和你现在看到 AI 被采用的过程。

[9:10] Eric Vishria

So if you go back to like 2010, 2011, now you're like three, four years into AWS beingan offering.Enterprises were super skeptical.Traditional enterprise, blue chip enterprise.You had digital natives out here.You had new companies forming that were using the cloud.I think kind of famously Snapchat was built on GCP.And I think at one point in this era, maybe 2012, 2013, 2014, Snapchat was like 40% of GCP.Those kinds of things were happening.That's like cursor being 30% of all these things.100%.Same thing.Same exact thing happened.Yeah.Same exact thing happened.Enterprises were very skeptical, I think, of cloud until by 2014, 2015, 2016 is like, oh,yeah, yeah.Then I think the big banks and financial services and insurance companies and more conservativeblue chip enterprise were like, oh, wait a minute.This is actually different.And we're going to have to pay attention.And it became an issue in recruiting for them because they couldn't get the best developersbecause best developers wanted to work on the easiest platforms.And all of these kinds of things happened.So the difference now feels profound to me because while it is 100% the case that blue chip enterpriseAI is not well absorbed and well adopted and not the same thing as going to cursor and walking

回到 2010、2011 年，那时 AWS 作为一个产品已经推出三四年了。企业客户极度怀疑——传统企业、蓝筹企业。而另一边你有数字原生公司，有一批新成立的公司在用云。我记得比较有名的是 Snapchat 建在 GCP 上。而且在某个时点，可能是 2012、2013、2014 年，Snapchat 大概占了 GCP 的 40%。当时就在发生这种事。（Patrick）这就像 Cursor 占了这些东西的 30%。（Eric）一模一样。同一件事。（Patrick）对，一模一样的事情又发生了。（Eric）企业客户对云一直非常怀疑，直到 2014、2015、2016 年才变成「好吧好吧」。然后那些大银行、金融服务公司、保险公司这些更保守的蓝筹企业才反应过来：等一下，这东西是真的不一样，我们必须重视。而且这甚至变成了他们的招聘问题，因为他们招不到最好的开发者——最好的开发者想在最好用的平台上干活。所有这些事情都发生了。所以现在的差别让我觉得很深刻：虽然百分之百地说，蓝筹企业对 AI 的吸收和采用还很不到位——你去 Cursor 的办公室转一圈，

[10:24] Eric Vishria

the halls of cursor versus walking the halls of a big New York financial services firm interms of their AI use.They want it.They want to figure it out.They're running experiments.They are spending against it.They're trying to figure it out, talking about it.They're not being dismissive about it.And I think they view it probably as more of an opportunity than they did the cloud interms of the potential impact on their business.I think they probably view it more as a threat than they did the cloud.And maybe they just also learn lessons from the cloud in terms of what's possible here.I feel that they will continue to try to figure it out and absorb it.Having said all that, I do think one of the more interesting things,if you go from Silicon Valley out into the world and talk to these companies, these enterprisecompanies, and you realize what the pace of adoption is and what the barriers to adoptionare and everything else, there's a ton of opportunity to help the enterprises get there.To be one of the things that I tell the companies I work on is, hey, let's be their AI Sherpa.If we're in that position to be their AI Sherpa, where we're crossing both worlds,that's very valuable.And I think that'll continue to work.

跟你去一家纽约大型金融服务公司的走廊转一圈，从 AI 使用程度上看完全是两回事。但他们想要这个东西，他们想搞明白，他们在做实验，在为此花钱，在努力弄懂，在讨论它，他们没有嗤之以鼻。而且我觉得，就对自身生意的潜在影响而言，他们把 AI 更多地看成一个机会，比当年看云更积极；同时大概也更把它看成一个威胁，比当年看云更警惕。也许他们也从云那一轮学到了教训，知道这里有什么可能性。我觉得他们会继续努力搞懂它、吸收它。话虽如此，我觉得更有意思的一点是：如果你从硅谷走出去，到真实世界里跟这些企业客户聊，你会意识到采用的速度是什么样、采用的障碍在哪里等等——帮企业跨过这道坎的机会非常巨大。我对我合作的公司常说的一句话是：嘿，我们来当他们的「AI 夏尔巴向导」。如果我们能站在那个位置，同时横跨两个世界，那是非常有价值的。我觉得这条路会继续走得通。

[11:37] Patrick O'Shaughnessy

What is Sierra teaching you about the adoption of this stuff?It's really interesting contrast to Fireworks, where Fireworks is the sort of infrastructureprovider.Sierra is feeling its way through what are really cool things that we can do for end consumers,enable companies to do for end consumers, starting with customer service.But I know with Horizon now going beyond that, again, same question as for Fireworks.What do you know that the world doesn't fully appreciate because of how you've seen that?

Sierra 又教会了你什么关于这波技术采用的事？它跟 Fireworks 形成很有意思的对照——Fireworks 是那种基础设施提供方，而 Sierra 是在摸索我们能为终端消费者做哪些真正酷的事、能让企业为终端消费者做哪些事，从客服开始。但我知道现在有了 Horizon，它已经走得更远了。同样的问题：因为你亲眼看到了这些，你知道了什么是外界还没有充分意识到的？

[12:01] Eric Vishria

My partner, Peter and Brett, I think this is their third company working together.So it's like a 20-year relationship, which is just like an amazing and great place to start.And one of the things that I think makes Brett so special and the team is they're technologists,but they actually have lived in enterprise world for a long time and really understand it andeverything else. And I think actually he's personified this whole idea of, hey, let's betheir AI Sherpa. Let's start with customer service. We're very automatable and like be theirAI Sherpa. And now we have these long running agents with Horizon that can do more and more stuff.And one of the things that I think is most interesting about this to me is, yes,it's an application company on the surface, but they're doing real AI work and they're experimentingwith the models and they're building agents. And they're very close to the metal of the modelsand the capability and the harnesses. And they're understanding the jagged edge of AI capabilities,which is very different than the human smooth arc that we understand intuitively. They understandthat jagged edge of capability and they build around it. You saw the evolution of cursor from the

我的合伙人 Peter，还有 Bret（Bret Taylor），我记得这是他们俩第三次一起做公司了，所以是二十年的交情，这本身就是一个非常棒的起点。我觉得让 Bret 和这个团队这么特别的一点是：他们是技术人，但同时在企业软件的世界里摸爬滚打了很久，是真的懂这一行。而且我觉得他其实把「当客户的 AI 夏尔巴向导」这件事人格化了：我们从客服开始——这块非常适合自动化——去当他们的 AI 向导。现在我们有了 Horizon 这种能长时间运行的智能体，能做的事越来越多。这里面我觉得最有意思的一点是：是的，它表面上是一家应用公司，但他们在做真正的 AI 工作，他们在拿模型做实验、在造智能体。他们离模型的底层非常近，离模型能力和 harness（外层脚手架，指包住模型的那层调度与工具代码）非常近，他们理解 AI 能力的「锯齿边缘」（jagged edge，指 AI 的能力不是平滑的一条线，而是这里超强、那里突然很蠢的锯齿状）——这跟我们直觉里熟悉的、人类那条平滑的能力曲线非常不一样。他们理解那条锯齿状的能力边缘，并围绕它去搭产品。你看 Cursor 的演进：从

[13:08] Eric Vishria

IDE to like tab autocomplete to agentic work over and over and over again. They were obsoleting theirwork from six months ago. And it's what Brett Taylor calls like sandcastles. We used to be buildingcastles. Now we're building sandcastles that get launched away. And that's, if you don't thinksoftware that way. Yeah. You have to embrace it because if you were an artisan and you were like,hey, I built these perfect foundation and castle and there's bricks and it was like perfect. AndI really care about this and it's going to be here for a hundred years. You're just not going tomake it. So as you get emergent new properties in the models, which is, you know,every four weeks, you get new capabilities. They understand that jagged edge. They understandthat application of that jagged edge or that valley to their customer base. They're fillingin those gaps and translating it. You can't just be superficially applying things. I think it'sactually a complete inversion of how product development used to work to how product developmenthappens now. And product development, you'd also say like, hey, product manager, you trained,you know, the product manager should understand the technology, but like really shouldn't be

IDE，到 tab 自动补全，再到智能体式的工作，一轮又一轮。他们不断把自己六个月前的成果作废掉。这就是 Bret Taylor 说的「沙堡」：我们过去是在造城堡，现在造的是会被浪冲走的沙堡。如果你不这么看待软件——（Patrick）对。（Eric）你必须接受这一点。因为如果你是个匠人，心想：我打了完美的地基、砌了完美的城堡，每一块砖都完美，我很在乎它，它要在这儿立一百年——那你根本活不下来。因为随着模型涌现出新的性质——差不多每四周就有新能力出来——他们理解那条锯齿边缘，理解那条边缘（或者说那些凹谷）对自己客户群意味着什么，去把缺口填上、把它翻译过去。你不能只是浮于表面地套用。我觉得这其实是产品开发方式的一次彻底反转——从过去的产品开发方式，到今天的产品开发方式。过去的产品开发会说：嘿，产品经理，你受过训练，产品经理应该懂技术，但不该

[14:10] Eric Vishria

thinking about implementation and shouldn't be specifying implementation and shouldn't be doingthis and doing that. And the product manager's job is to really understand the customer andtranslate that problem to the engineer so that the engineers build a solution like that wastraditional product management. Good luck doing that now. That's a horrible way to do it. You can't doit that way. You actually really need to understand the nuances of model capabilities, what they'regreat at and where they fail. And you need to understand the customer problem and put thosetogether and bridge those gaps to build valuable solutions. I think that's one of the things thatMichael and team at Cursor did so well from the very beginning is they really understood the jaggedcapability and built a product that allowed that translation from developer to that jagged capabilityand kept on iterating on that as the edge changed. It's kind of ironic that all that sounds likethe returns to being technical are going up, even as models supposedly are taking away technical edge.Well, I think there's two things. Yeah, 100%. And actually, I think there's this weird thing wherethere was this whole discussion about the traditional roles of product manager and designer and engineer,

去想实现方案、不该去规定实现方案、不该干这个不该干那个。产品经理的工作是真正理解客户，把问题翻译给工程师，让工程师去造解决方案——这是传统的产品管理。现在你要还这么干，祝你好运。这是一种很糟糕的做法，你不能那么干。你必须真正理解模型能力的细微之处：它们擅长什么、在哪里会失败。同时你要理解客户的问题，把这两者拼起来、把中间的缺口桥接上，才能造出有价值的解决方案。我觉得 Cursor 的 Michael（Michael Truell）和团队从一开始就做得特别好的一点，就是他们真的理解那条锯齿状的能力边缘，造了一个能把开发者翻译到那条锯齿边缘上的产品，并且随着边缘变化不断迭代。（Patrick）挺讽刺的是，这一切听起来像是「懂技术的回报正在上升」，尽管模型据说正在抹平技术优势。（Eric）我觉得有两件事。对，百分之百。而且我觉得还有个很奇怪的现象：过去大家一直在讨论产品经理、设计师、工程师这些传统角色的分工，

[15:14] Eric Vishria

whatever it is. And I think what there really are, are people who understand customer problems,people who have taste, and people who understand the jagged edge of AI capabilities and are curious about.Those are the three things. If you've got those three, you're going to do well.If you've got those three things, you're going to do great. And it doesn't really matter if youare an engineer or a product manager or designer, but if you have taste and understanding of customerproblems and understanding the jagged edge, you're going to do well.Vanta automates security and compliance for over 16,000 fast-moving companies like Ramp,

不管是什么。而我觉得真正存在的其实是三种人：理解客户问题的人、有品味的人、以及理解 AI 能力锯齿边缘并且对它充满好奇的人。就这三样。如果你三样都有，你就会做得很好。三样都占，你会做得非常出色。你到底是工程师、产品经理还是设计师，其实并不重要；但如果你有品味、懂客户问题、懂那条锯齿边缘，你就会做得很好。〔广告〕Vanta 为超过 16,000 家快速成长的公司（比如 Ramp、

[15:51] Patrick O'Shaughnessy

Cursor, and Harvey, keeping them audit-ready around the clock. It's the number one agentic trust platform.And it now helps companies like yours watch for the risks that show up between audits,across your vendors, your AI tools, and your whole environment. Every new tool your team signs up for,every vendor that turns on AI features, is an opportunity for something to go wrong.And most security programs weren't built for AI's pace of growth. The Vanta agent works like a 24-7GRC engineer in the background, finding issues, drafting fixes for you, and cutting vendorassessment time by up to 50%. Whether you're a fast-growing startup or a global enterprise,Vanta helps you earn and prove trust. Invest like the best listeners get a special offerfor $1,000 off at vanta.com slash invest. Ridgeline is the first end-to-end system of record withembedded AI for investment management firms, running portfolio accounting, reconciliation,reporting, trading, and compliance on one unified platform. Firms are moving off legacy technologyand onto Ridgeline because of how far ahead Ridgeline's AI features are compared to anything elsein investment management software. Which is why I believe that firms that come out ahead in the AI era

Cursor 和 Harvey）自动化安全与合规，让它们全天候处于随时可审计的状态。它是排名第一的智能体信任平台。现在它还能帮你这样的公司盯住两次审计之间冒出来的风险——跨你的供应商、你的 AI 工具、你的整个环境。你团队新注册的每一个工具、每一家开启 AI 功能的供应商，都是一次出事的机会。而大多数安全体系不是为 AI 这种增长速度设计的。Vanta 的智能体像一个 7×24 小时的 GRC（治理、风险与合规）工程师在后台工作，发现问题、替你起草修复方案，把供应商评估的时间最多缩短 50%。不论你是快速成长的初创公司还是全球性企业，Vanta 都能帮你赢得并证明信任。《Invest Like the Best》的听众可以在 vanta.com/invest 拿到 1,000 美元的专属优惠。〔广告〕Ridgeline 是第一个面向投资管理机构、端到端且内嵌 AI 的记录系统，把组合会计、对账、报告、交易和合规跑在一个统一平台上。各家机构正在从老旧技术迁移到 Ridgeline，因为跟投资管理软件里的任何其他产品比，Ridgeline 的 AI 能力遥遥领先。这也是为什么我相信，在 AI 时代跑出来的机构，会是那些跑在 Ridgeline 统一平台上的机构。

[16:59] Patrick O'Shaughnessy

will be the ones running on Ridgeline's unified platform. If you're serious about your firm's AIstrategy, Ridgeline should be part of that conversation. You can request a demo at ridgeline.ai.When we first did this so many years ago now, it's just crazy. It'd be fun to revisit some of the ideasthat we talked about the first time with SaaS. But you use this term that I've used ever since,which you call it the competitive frontier, meaning like the things that will determinethe winners and the losers. My partner Bree has this great idea that's stuck in my head,which is that everything is a jump ball right now. I'm really curious, in addition to this idea ofsandcastles versus real castles, what else you're seeing amongst the people that are becomingcompetitive winners? Are the traits different for winners now, personality-wise or business strategywise or business model wise versus what you learned in the SaaS era? I'm very dismissive of this ideathat people are going to vibe code their own shit and whatever. That isn't the issue at all withSaaS companies. The issue with SaaS companies is the competitive frontier completely shifted andeverything they thought they were building against and it would make them win are not what's going to

如果你认真对待自家机构的 AI 战略，Ridgeline 应该出现在那场讨论里。你可以在 ridgeline.ai 申请演示。〔正片〕我们第一次做这期节目已经是好多年前了，真是疯狂。重访一下我们第一次聊 SaaS 时的那些想法会很有意思。你当时用了一个词，我从那以后一直在用，叫「竞争前沿」（competitive frontier），意思是那些真正决定谁赢谁输的东西。我的合伙人 Bree 有一个我一直记着的说法：现在所有东西都是「争球」（jump ball，篮球跳球开局，双方都可能抢到）。我很好奇，除了沙堡与真城堡这个比喻，在那些正在成为竞争赢家的人身上，你还看到了什么？现在赢家的特质——不管是性格上、业务战略上还是商业模式上——跟你在 SaaS 时代学到的相比，有什么不一样？（Eric）我很不认同「大家会自己 vibe code 出自己那套东西」这种说法，那根本不是 SaaS 公司的问题所在。SaaS 公司的问题在于：竞争前沿彻底移位了，他们过去以为自己在为之努力、能让自己赢的那些东西，已经不是决定胜负的东西了。

[18:05] Eric Vishria

make them win. Let's take databases. Databases have been a phenomenal area for software for a long,long time. Great margins. Why? You had app developers that would build against these specific databaseinterfaces that existed for that database. You would have more and more data over time. Migratingan app from one database to another database was a giant project that was like very, very difficultto do. So these were unbelievably sticky businesses that you could generate a ton of margin in, right?

我们拿数据库举例。数据库长期以来一直是软件里一个了不起的领域，毛利率极好。为什么？因为应用开发者是针对某个数据库特有的接口来写代码的。随着时间推移，数据越积越多。把一个应用从一个数据库迁移到另一个数据库，是个极其困难的大工程，非常非常难做。所以这些生意黏性大得不可思议，你能在里面赚到巨额毛利，对吧？

[18:34] Eric Vishria

Of course you got Oracle and SQL Server and like, and a whole slew of smaller players like that didreally, really well in databases. Well, let's think about that in the context of AI. Now you don't havea developer building against the database interface. You have cloud or codex building against the databaseinterface. One. Two, the beauty of database interfaces is they're very, very well specified.Well, it turns out AI is very good, very good at things that are very, very well specified.Three, agents don't get tired of monotonous work of translating one specification to another.So it turns out that now all of a sudden database migration, which used to be the like number onething you would not do in software is like kind of trivial. Yeah. Yeah. It's just like, yeah,puts money against it, move it. So what changed? Well, what changes is the criteria to be an amazingdatabase company changed. It isn't that we don't need databases or that everyone's going to buildtheir own database or whatever. That isn't what's going to happen. What's going to happen is thecriteria changed. So now you're going to have a ton more applications that start, obviously,as we're seeing everywhere. People are going to experiment a lot more because it's much cheaper

当然就有了 Oracle、SQL Server，还有一大堆在数据库上做得非常非常好的中小玩家。那么把这件事放到 AI 的语境里想一想。现在，不再是开发者针对数据库接口写代码了，而是 Claude 或者 Codex 在针对数据库接口写代码。这是第一点。第二，数据库接口最美妙的地方在于它们的规格定义极其清晰。而事实证明，AI 特别特别擅长规格定义得非常清晰的事情。第三，智能体不会因为「把一种规格翻译成另一种规格」这种单调工作而感到疲倦。所以突然之间，数据库迁移——这件过去在软件里排第一位的「打死也别干」的事——变得几乎不值一提了。（Patrick）对，对。（Eric）就是砸点钱进去，搬走。那么变的是什么？变的是「成为一家了不起的数据库公司」的标准变了。不是说我们不需要数据库了，也不是说每个人都要自己造数据库，那不会发生。会发生的是评判标准变了。所以现在会有多得多的应用被启动起来——这一点我们到处都看得到；人们会做多得多的实验，因为实验变便宜了，

[19:42] Eric Vishria

to experiment. It's much cheaper to start a new application. So now you need databases thatscale from basically zero usage to if it works all the way through, like that matters a lot more.Your cost matters a lot more. You want to be able to spin these things up, spin them down,tear them apart over and over again. So the iteration speed goes up and what you need on database.And ultimately, I think the cost becomes an arbiter of this. So the cost and then this zero to infinityscaling and transportability and everything else around that become like the arbiters of who wins and whodoesn't. That's really different than, hey, I specced this database for our user thing and I procured alicense and I ran it on this kind of hardware and everything else. There have been elements,of course, of these things over time. But I think that just criteria changed. And I think one of thebig messages to these SaaS companies a few years ago was you have a choice. Get to AI or be worth threetimes revenue. That was a hard message to hear. You were just like, get to AI or three times revenue.Those are your choices. We say three times revenue now because a lot of these public SaaS companiesare trading at six times or whatever. But keep in mind in 21, they were going for 30 times. Like

启动一个新应用便宜太多了。所以现在你需要的数据库，要能从基本为零的用量、一路扩展到「万一它跑起来了」的规模，这一点重要得多了。成本也重要得多了。你要能把它拉起来、关掉、拆掉，反反复复。所以迭代速度上去了，对数据库的要求也变了。而最终我觉得成本会成为胜负的裁决者。成本，加上这种从 0 到无穷的伸缩能力、可迁移性等等，成了谁赢谁输的裁决标准。这跟过去「我为我们的用户系统选型了这个数据库、采购了许可证、跑在这种硬件上」完全不是一回事。当然过去也有过这些因素的影子，但我觉得就是标准变了。而我觉得几年前对这些 SaaS 公司最重要的一句话是：你有个选择——要么走到 AI 那边去，要么就值三倍收入（3x PS）。这话很难听。你就是：要么 AI，要么三倍收入，这就是你的选项。我们现在说三倍，是因为很多上市 SaaS 公司在按六倍左右交易。但别忘了 21 年的时候它们是 30 倍。

[20:53] Eric Vishria

everybody, you're like three times. Wow. I've grown 4X since then. This is a whole multiple compressionis a bitch. I've grown 4X. The multiple has gone down by a factor of six. So I'm worth less,even though I've grown 4X after years and I've gotten to break even all these things. So like that wasthe first message. But I think that even it became really visceral to me where you're in these meetingsand you're like, hey, every single day that you are hitting your plan, you are destroying equity value.Think about that. Our whole careers, we learned you lay out a plan, you execute against it relentlesslyand violently. You hit your plan or you exceed your plan and you keep on building that. And that's howyou build equity value. That's what the whole management teams learned. That's what the CEOslearned. Like pre-AI, this is like everything learned. And now you're in here just every dayyou hit that plan. You're fucking up. You're fucking up. You're destroying value. And the pointof saying that to them was to set them free. Another articulation of this from my friend,Anne Lee Skates, was just the CEOs who were going through this transitory period and had a businessthat was at hundreds of millions. And they thought that, you know, they're all ready.

每个人都是：三倍？哇。我从那时起已经涨了 4 倍了。这种估值倍数压缩真是要命：我涨了 4 倍，倍数掉了 6 倍，所以哪怕过了好几年、我做到了盈亏平衡等等，我反而更不值钱了。这是第一层信息。但我觉得真正让我感受强烈的是：你坐在这些会议里，然后说：嘿，你每多达成一天的计划，你就在多摧毁一天的股权价值。想想这件事。我们整个职业生涯学到的都是：制定一个计划，然后凶狠地、毫不留情地执行它，达成计划或超额完成，然后不断累积——这就是创造股权价值的方式，这是所有管理团队学到的，是所有 CEO 学到的，在 AI 之前这就是全部。而现在你坐在这儿，每一天你达成计划，你就是在搞砸，你就是在摧毁价值。对他们说这句话的目的，是把他们解放出来。我朋友 Anne Lee Skates 有另一种说法：那些正处在转型期的 CEO，生意做到了好几亿美元，他们以为自己都准备好了。

[22:01] Eric Vishria

They were working on their business from 8 a.m. to 5 p.m. and then trying to do AI from 5 to 8in the evenings. And what they needed to be doing was the inverse. And it's so hard to do thatbecause of all of the training and all of the muscle memory and all the inertia and everythingthat we learned about the hill that we were climbing. Like we're all hill climbing in a way.The success model was set the plan, execute the plan, build value, compound value. And it's like,oh no, no, no, no, stop that. You got to completely invert. This is a very, very long way to get back toyour question of what the profile or what the mentality of the winners are right now.But if we look at Brendan from McCore or Lynn from Fireworks or Max or Brett, any of these people,they are so nimble about what the eval is. What are they optimizing against? They are so nimble on allof it. And if you look at every one of those companies, the evolution of the business, it's justthe business is constantly evolving. And they've done such an excellent job at that. I think that isvery, very different than the way I was taught. What's your sense of the disorienting nature ofmodel progress? You're one step removed from that as an investor versus as a technical founder with your

他们早上 8 点到下午 5 点在经营现有业务，然后晚上 5 点到 8 点去搞 AI。而他们该做的其实是反过来。这非常难，因为所有的训练、所有的肌肉记忆、所有的惯性，以及我们关于「爬这座山」学到的一切。我们某种意义上都在爬山，成功模型就是：定计划、执行计划、创造价值、复利累积。结果现在是：不不不不，停，你得彻底反过来。这是一段非常非常长的铺垫，回到你的问题——现在赢家的画像或者心态是什么。但如果我们看 Mercor 的 Brendan、Fireworks 的 Lin、Max、或者 Bret，这些人里的任何一个，他们对「评价标准（eval）是什么」都极其灵活：他们到底在优化什么？他们在所有事情上都极其灵活。你去看这些公司里的每一家，生意的演化路径就是：这门生意在不停地演化。而他们在这件事上做得非常出色。我觉得这跟我当年被教的方式非常非常不一样。（Patrick）模型进步带来的那种失重感，你怎么看？作为投资人，你比那种手放在机器上的技术型创始人隔了一层。

[23:22] Eric Vishria

hands on the metal. Are we behaving differently than you would have three years ago or somethingbecause of the pace? Anytime that I'm talking to a founder about a problem in their company or whatthey're doing or move they're making or anything else, which is what I spend 80% of my day doing,I'm very, this is how we used to do it. This is what we would typically do. This would be the typicalreadout on this old school readout of why this candidate is better than this candidate.Let's reevaluate that in the context of today. Let's reevaluate that in the context of an unstabletechnology substrate. Let's reevaluate that in the context of a business model that's growing thisway versus that way. I've really started to question every assumption and every lesson that I learnedbefore, which of it translates and which of it doesn't. That's a huge difference. I'll give you areally concrete example, which has been very disruptive inside of these scaling AI companies.So these scaling AI companies, we've had a lot of leaders come from the prior generation with greatexperience and everything else come to these scaling AI companies and completely flame out. And you seeit across the industry. And so the question is why? These are some of the best leaders from four or

（Patrick 续）因为这个节奏，我们的行为方式跟三年前相比是不是不一样了？（Eric）任何时候我跟创始人聊他们公司里的某个问题、他们在做的事、他们要走的一步棋或者别的什么——这占了我一天 80% 的时间——我都非常习惯说「我们过去是这么做的」「我们通常会这么做」「这种老派的评估会这样说明为什么这个候选人比那个候选人好」。那就把它放到今天的语境里重新评估一遍。放到一个不稳定的技术基座（technology substrate）的语境里重新评估。放到一个用这种方式而不是那种方式增长的商业模式语境里重新评估。我真的开始质疑我以前学到的每一个假设、每一条经验：哪些还成立，哪些不成立。这是个巨大的差别。我给你举个非常具体的例子，这件事在这些高速扩张的 AI 公司内部造成了很大的破坏。这些高速扩张的 AI 公司里，我们招了很多来自上一代、经验丰富的领导者进来，结果彻底折戟。整个行业到处都在发生。那么问题是：为什么？这些是四五年前最优秀的一批领导者，

[24:35] Eric Vishria

five years ago. They had all the lessons. They learned it all. They're excellent, but somehow it's nottranslating. And there's some impedance mismatch between the AI founders potentially, the needs ofthe business and what these people are bringing. And I saw it really abruptly with a particular salesleader who we hired, who's like, hey, we can't hit any of these things because the way that softwaresales has been taught forever is a quota capacity model. You have a quota capacity model. Each rep doesthis. In the early days of a company, the quotas are 1.2, 1.5 million. Maybe the ISRs are at 750 or850. And then over time, it scales up and enterprise gets to 2.5 million. And that's how all these financialmodels are built. You start with the quota capacity model. You take a discount on attainment. This iswhat we can do. Boom, boom, boom. Fundamentally, without realizing it, everybody was implementingsomething that was based on pushing demand, not pulling demand. And for so many of these companies,they're operating here, these customers, and there's a new AI-enabled product that comes along,and it's just fucking magic. So these companies are selling magic. Well, it turns out if you'reselling magic and you're the first one there- Sell a lot more than 2 million. You're going to sell a lot

他们学到了所有该学的经验，他们非常优秀，但不知怎么就是转化不过来。AI 创始人、生意本身的需求，和这些人带来的东西之间，存在某种阻抗失配（impedance mismatch）。我在一位销售负责人身上看得特别清楚——我们招进来的，他说：嘿，这些指标我们一个都完不成。因为软件销售一直以来教的都是配额产能模型（quota capacity model）：你有一个配额产能模型，每个销售代表做这么多。公司早期，配额是 120 万、150 万美元；内部销售代表（ISR）可能是 75 万或 85 万。然后随着时间推移往上走，企业级销售做到 250 万美元。所有这些财务模型都是这么搭的：你从配额产能模型开始，再打一个达成率折扣，这就是我们能做到的，一二三，砰砰砰。而根本上，所有人在不自觉地实施的，是一套基于「推动需求」（pushing demand）而不是「承接需求」（pulling demand）的东西。而对这么多公司来说，客户就在那儿，突然来了一个 AI 赋能的新产品，而它他妈的就是魔法。所以这些公司卖的是魔法。而事实证明，如果你卖的是魔法、而且你是第一个到场的——（Patrick）那卖的会远远超过 200 万。（Eric）你会卖出远超 200 万。

[25:50] Eric Vishria

more than 2 million. The whole notion of a quota capacity model and that being working,it's not that it doesn't matter. It matters kind of, but it's definitely not the first order thingor constraint. So you have these execs come over with this, like, here's our math, and here's theterritory assignment, and here's what we would do. And first you do West Coast, and then you do EastCoast, and then you do Central, like all of these things. Oh, wait, no, no, it doesn't work like thatat all. One of my big things that I started to realize is as I'm interviewing these folks and talking tothem is just like, hey, you need to check everything at the door. Check it all. Which is probablygood practice anyway, but check all the baggage. Check everything that you learned and just learnthis from first principles. How is it working? What are really the bottlenecks on delivery? Whatare the bottlenecks on demand? Because it turns out that in a lot of these companies, you have repsdoing 10 or 20, 30 million. I saw 50 recently. It turns out that's different. What is like the bestsalesperson that you've seen that's doing it in a de novo way doing? Honestly, the best salesperson inany of these companies is a founder. And what they're doing is bridging the jagged edge to what

配额产能模型这套东西还成不成立？不是说它完全不重要，它多少还是重要的，但它绝对不是第一性的因素或约束。于是你有这些高管带着「这是我们的算法、这是地盘划分、我们要这么干：先做西海岸，再做东海岸，然后中部」这一整套过来。结果是：哦不不，根本不是那么回事。我开始意识到的一件大事是，在我面试这些人、跟他们聊的时候，我会说：嘿，你得把所有东西都放在门外，全都放下。这大概本来就是好习惯，但把所有包袱都放下，把你学到的一切都放下，从第一性原理重新学这件事：它到底怎么运转？交付的真正瓶颈是什么？需求的瓶颈是什么？因为事实证明，在很多这样的公司里，你会有销售代表做到 1000 万、2000 万、3000 万美元。我最近还见到 5000 万的。事实证明这不一样。（Patrick）你见过的、用一种全新方式做销售的最好的销售是什么样的？（Eric）说实话，这些公司里最好的销售就是创始人。他们做的事就是把锯齿边缘

[26:52] Eric Vishria

the customer's capability is. And that's it. It sounds so simple, but it's not. But that's whatthey're doing. The market is just so big. Just like we were talking about with cloud. I think the biggestmistake everybody made was they just undersized the market. And it turns out the market's just really,really big. And this market is fair. It feels like right in this moment in time, actually just thismorning, you and I are in this great group chat together where the discussion is the demand forintelligence. Seems kind of unlimited. And it seems like the smarter the thing gets, the more demandthere is. Maybe the bottleneck is just capital. The world just feels like it needs to take a breath.The RSI concept, if you apply it across technology, doesn't need to breathe. The agents don't gettired. But the world feels kind of like, oh, man, it sure would be nice to have three months just todigest this a little bit. And for capital to form and evaluate its prospects and the scale is gettingso big. Does it feel to you like this can just keep going? Or are we just going to get tapped outof money that can be invested in these things when it seems like we could consume like any amount ofmoney to build any amount of stuff and serve any amount of inference? I'm not a macro economist.

和客户的能力对接起来，就这么简单。听起来太简单了，但其实并不简单。可他们做的就是这个。市场实在太大了，就像我们刚才聊云的时候说的。我觉得所有人犯的最大错误就是把市场估小了，事实证明这个市场真的真的非常大。（Patrick）这个市场感觉——就在此时此刻，其实就在今天早上，你和我在同一个很棒的群聊里，讨论的话题是「对智能的需求」，好像是无限的。而且好像这东西越聪明，需求就越大。也许瓶颈只是资本。这个世界感觉需要喘口气。RSI（recursive self-improvement，递归自我改进：AI 自己改进自己的能力）这个概念如果套用到整个技术上，它是不需要喘息的，智能体也不会累。但这个世界的感觉是：哎，要是能有三个月的时间稍微消化一下就好了，让资本形成、让它评估自己的前景，而规模又变得这么大。你觉得这件事能一直这么走下去吗？还是说我们最终会把能投进这些东西的钱耗尽——尽管看起来我们能消耗任意数量的钱、去建任意数量的东西、去服务任意数量的推理？（Eric）我不是宏观经济学家。

[27:56] Eric Vishria

I am worried about energy. If you think about the models as translating compute into intelligence,quite simply, what do models do? They very effectively translate compute into intelligence.What's the demand for intelligence? Well, it seems like a lot. So then it follows that we will continueto have more and more demand on compute. I'm saying compute broadly, not chips, not storage,not whatever. But then like, what do we need for compute? We need energy, a lot of energy.There's all this topic of distillation and Chinese open source and all these things. Butthe bigger thing to me is that I think China's bringing on 10 times as much energy next year as we arein the US. If energy is what you need for compute and is the bottleneck, and there's unlimited demandfor intelligence, then it stands to reason that if we have a lot less energy, then we will have a lotless intelligence or a lot less tokens or a lot more expensive tokens. If we have a lot more expensivetokens than supply demand, you're going to end up with less. And that seems very bad. To me,I think the energy bottleneck, however, that is, it'll manifest in 20 different ways. Gas turbinesgo up and down, natural gas get up and down, and solar, whatever, all these different things,

我担心的是能源。如果你把模型看作是把算力翻译成智能的机器——很简单，模型是干什么的？它们非常高效地把算力转换成智能。那对智能的需求有多大？看起来非常大。那就顺理成章：我们对算力的需求会越来越大。我说的算力是广义的，不是单指芯片、存储或者别的什么。那算力需要什么？需要能源，需要大量能源。现在有很多关于蒸馏（distillation，用大模型去训练小模型）、中国开源模型这些话题。但对我来说更大的事情是：我认为中国明年新增的能源是美国的 10 倍。如果能源是算力所需、而且是瓶颈，同时对智能的需求又是无限的，那么合乎逻辑的结论就是：如果我们的能源少很多，我们就会有少很多的智能、少很多的 token，或者贵很多的 token。而如果 token 贵很多，按供需关系，你最后拿到的就更少。这看起来非常糟糕。对我来说，能源瓶颈——不管它以什么形式出现，它会以 20 种不同的方式表现出来：燃气轮机产能起起落落、天然气起起落落、太阳能，随便什么，

[29:12] Eric Vishria

rare arts. That to me is probably more concerning. And if I'm thinking about it from a regulatoryperspective or government perspective, and I think the administration is doing some things aroundthis, fostering investment and development of all energy, solar, nuclear, gas, like whatever,do it all. We should do it all. And it'll work itself out. This is one of these things where,yes, one will be relatively better than the other. And I don't know, and I'm not smart enough topredict which one's which, but it'll all work. Speaking of compute, I would love to hear theCerebra story. I haven't heard you tell the full version of this. The reason I'm asking about it isI'm deeply interested in compute. I have bigger investments in compute,compute. And I'm fascinated by it. It's just like the most magical thing to watch happen. It'smind boggling when you get close to one of these things, what humans have been able to doon these chips and in these systems. I think you invested in 2016 or thereabouts. I think it wasyour first foray into like extremely difficult hardware type investment. It's so hard. And now theworld is full of opportunities like this. Whereas back then it was a one-off. Teach me everything

还有稀土。这些对我来说是更值得担心的。如果从监管或者政府的角度想，我觉得本届政府在这方面做了一些事：鼓励对所有形式能源的投资和开发——太阳能、核能、天然气，随便什么，全都干。我们应该全都干，最后它会自己理顺。这属于那类事情：是的，其中一种最后会相对更好，我不知道是哪种，我也没聪明到能预测哪种是哪种，但它整体上会成的。（Patrick）说到算力，我特别想听听 Cerebras 的故事，我还没听你讲过完整版。我问这个是因为我对算力极其感兴趣，我在算力上有更大的投资，而且我对它着迷。看着它发生就是最神奇的事情。当你近距离接触其中一台机器，你会觉得人类在这些芯片上、在这些系统里做到的事情简直匪夷所思。我记得你是 2016 年前后投的，我猜那是你第一次涉足极其困难的硬件类投资。（Eric）太难了。（Patrick）而现在世界上到处都是这样的机会，当年那还是个孤例。把你通过 Cerebras 学到的关于硬件投资的一切都教给我。

[30:14] Eric Vishria

you've learned about hardware investing through Cerebrus. Mostly it's really hard.It's an amazing example to me about the naivete required. The company came in in 2016. It wasfive founders and a deck. I did not want to go to the pitch, but it's my job. Because I'm like,why are we going to go to hardware investment? Like, it's crazy. It's been 10 years since we'dmade a semi-investment. I think basically the team was excellent. And then the kind of first slide wasjust like, GPUs actually suck for deep learning. They just happen to be a hundred times betterthan CPUs. You have to remember, this is pre-transformer. OpenAI is this weird researchlab at this time. NVIDIA was worth like $40 billion, not $4 trillion. The TPU hadn't been announced.None of that. So this is early. But the whole idea, as soon as you said it, I was like, oh shit,of course. Of course. Like why? I had spent the last 18 months trying to figure out applications ofdeep learning and looking at the security thing and looking at this medical imaging thing and likeall this other stuff. Thinking like, hey, there must be something here that's going to be reallytransformed by this stuff. Anyways, we go through this whole journey. We end up investing, which was

主要就是：真的很难。对我来说这是一个关于「需要多少天真」的绝佳案例。这家公司是 2016 年找上门的，五个创始人加一份 PPT。我本来不想去见这个 pitch，但这是我的工作。因为我心想：我们干嘛要去看硬件投资？这太疯狂了。我们上一次做半导体投资已经是十年前的事了。基本上是这个团队非常优秀。然后第一页 PPT 就写着：GPU 其实并不适合深度学习，它们只是碰巧比 CPU 好一百倍而已。你得记住，这是 Transformer 出现之前。当时 OpenAI 还是一个奇怪的研究实验室，NVIDIA 值 400 亿美元，不是 4 万亿。TPU 还没发布，什么都还没有。所以这非常早。但整个想法，他一说出来我就想：哦操，那当然了。当然啊。为什么？因为我在那之前的 18 个月一直在琢磨深度学习的应用场景，看了安全方向的项目、看了医学影像的项目，还有一堆别的东西，心里想着：嘿，这里面一定有什么东西会被这玩意儿彻底改变。总之我们走完了整个流程，最后投了，这很棒。

[31:22] Eric Vishria

amazing. We first met on Wednesday, part of meeting on Monday, had a bunch of meetings in between.Just built a lot of conviction that this was a great swing. And I'll tell you what I understood.And I really just understood so little, but basically there are three things that we know how to do tospeed up deep learning and hardware. Still till this day. Increase the number of course,increase the communication between cores, bring the memory closer to compute. Those are the threethings. That's it. Those are only three dimensions that we know in hardware. My articulation of whatthey said to me, honestly, all I understood was let's just take all three of those things to theirlogical maximum. You have a wafer scale chip. At that time, you would have 450,000 cores on it.You'd have something like 20 gig of SRAM on the chip. So you never have to go off chip to get tomemory. And because they were all in the same wafer, the communication between cores ismaximized. So this is the best you could do on that process. And I think for the first chip wasseven enemy or something. You're like, okay, that's it. That's what we do. And it turns outthat in software, if you have that logical block diagram of like why it works and everything else,

我们第一次见面是周三，周一开合伙人会，中间又开了一堆会。我们建立了很强的信念，认为这是一次很棒的挥棒。我告诉你我当时理解了什么——我其实理解得非常少，但基本上：在硬件上加速深度学习，我们知道的办法只有三件事，到今天也还是这三件。增加核（core）的数量、增加核与核之间的通信、把内存搬得离计算更近。就这三件，这是我们在硬件上知道的仅有的三个维度。用我自己的话复述他们跟我说的：说白了我当时只听懂了「我们把这三件事全都推到逻辑上的极限」。于是你得到的是一块晶圆级（wafer scale）芯片。当时那上面有 45 万个核，芯片上有大约 20 GB 的 SRAM，所以你永远不用出芯片去取内存。而因为它们都在同一片晶圆上，核与核之间的通信也被最大化了。所以在那个制程上这是你能做到的极致，我记得第一款芯片是 7 纳米还是什么。你就想：好，就这样，我们就干这个。而事实证明，在软件里，如果你有那张「为什么这样能行」的逻辑框图，

[32:27] Eric Vishria

you're kind of 80% of the way there. And it's a matter of go-to-market execution.In hardware, you're like 2% of the way there. There's things like physics, entire supply chainof vendors. There's of course TSMC, which everyone knows, but it's not just TSMC. There's 30 othervendors that matter and putting all this stuff together and everything else. I didn't know anyof that. Fast forward from 2016 to like 2019, I think they got their first parts back. You gothrough this like bring up and then it's like, bring up. Oh yes, we got a part back. And then it'slike, yeah, then you got to go through like, wait, bring up. And there's like 14 steps ofbring up and everything else. And then by like 2020, we had our like first thing that works.Then it's just this march of actually getting it to work. One of the lessons that I've learned onthis stuff is basically you go through all these SIMs and everything else in hardware and semis inparticular. That basically is your roof line. The best it's ever going to be is what that is.And then every bit of software and reality and compilers and kernels takes away from that roofline. You might start at 10% of the roof line. Once you bring it up, these guys are grinding for

你差不多已经走完 80% 的路了，剩下的是市场执行。而在硬件里，你只走了 2%。这里面有物理定律，有整条供应链的厂商。当然有大家都知道的台积电（TSMC），但不只是台积电，还有 30 家其他关键厂商，要把这一切拼起来等等。这些我当时一点都不懂。从 2016 年快进到 2019 年左右，我记得他们拿回了第一批芯片。然后你要经历所谓的「点亮」（bring up）：哦，太好了，我们拿到芯片了。然后是——对，然后你还得经历，等等，还要点亮。点亮有大概 14 个步骤等等。然后大概到 2020 年，我们才有了第一个真正能用的东西。接下来就是漫长的行军，把它真正跑起来。我在这件事上学到的一课是：在硬件、尤其是半导体里，你会先跑一堆仿真（SIM）之类的东西，那基本上就是你的「屋脊线」（roof line，理论性能上限）——它这辈子最好也就是那个数。然后每一点软件、现实、编译器、算子内核（kernel）都会从那条屋脊线上往下扣。你可能一开始只有屋脊线的 10%，点亮之后，这帮人要苦干

[33:36] Eric Vishria

months and years to like get closer and closer and closer to the roof line. It's really different.It's really hard. I'm astonishingly bullish if I kind of rewind. Part of the reason we made theinvestment was if you looked at the four prior generations of compute in my lifetime. So you hadCPUs, you had graphics, then networking, the mobile. There was a new workload each time. Soyou had multipurpose compute and it led to the CPU. You had massive parallelism led to the graphicsprocessor. Graphic processor offered massive parallelism, which led to graphics. Then withnetworking, you need really low latency chips. And so they had low latency chips. And then with mobile,you needed really power efficient chips. And in each case, we ended up with a new $100 billioncompany. The first question going back to 2016 was, is AI that big a new workload? Because there'vebeen many, many other attempts for specialized chips for other things that really just didn'tend up mattering. There's some fine outcomes, but they just didn't really end up mattering,et cetera. Right, exactly. And so it's just like, well, okay, well, you need something that's a really,really big workload. Okay, so that's one. We had a lot of conviction on that. And then two,

好几个月甚至好几年，才能一点一点越来越接近那条屋脊线。这真的很不一样，真的很难。如果把时间倒回去看，我极度看好。我们做这笔投资的部分原因是：如果你看我这辈子经历的前四代计算——先是 CPU，然后是图形，然后是网络，然后是移动——每一次都有一个新的工作负载。你有多用途计算，催生了 CPU；你有大规模并行的需求，催生了图形处理器——图形处理器提供大规模并行能力，用于图形。然后是网络，你需要极低延迟的芯片，于是有了低延迟芯片。再然后是移动，你需要极其省电的芯片。而每一次，我们都得到了一家 1000 亿美元的公司。所以回到 2016 年，第一个问题是：AI 是不是一个足够大的新工作负载？因为历史上有过非常非常多为别的东西做专用芯片的尝试，最后都没能真正成气候。（Patrick）有一些结果还不错，但确实没真正成气候之类的。（Eric）对，正是。所以就是：好吧，你需要一个真的真的非常大的工作负载。好，这是第一点，我们在这点上有很强的信念。然后第二点，

[34:42] Eric Vishria

was the nature of the workload, did it introduce a new constraint or problem in it? What I learnedwas basically the AI workload benefited from the parallelism of GPUs massively, but GPUs didn't solvethe core-to-core communication, basically the layers of the network problem. You're like, okay,wait a minute. There is a new constraint, which is communication's communication-bound problem.Then you're like, okay, is AI a new giant workload that is going to have specialized chips?

是这个工作负载的性质本身：它有没有引入一个新的约束或者新问题？我当时学到的是：AI 工作负载从 GPU 的并行能力里获益巨大，但 GPU 并没有解决核与核之间的通信问题，也就是神经网络层与层之间的问题。你就会想：等一下，这里确实有一个新的约束，是一个受通信带宽约束的问题。然后你再问：AI 是不是一个会催生专用芯片的、巨大的新工作负载？

[35:11] Eric Vishria

Everything that I just said was the entire, everything I knew at that time. That's obviouslyplayed out. And in each prior generation, we got Intel, we got NVIDIA, we got Broadcom and Vago,we got Qualcomm and ARM in each of these generations. And there will be these giantwinners, standalone winners. Obviously, the TBU itself is a winner, training is a winner.You've had Grok and Cerebrus, Etch. It'll keep getting fought out, but I think that will end upbeing big. And actually, I think there's a new sixth one that's coming, generation, which is,and I'm really excited. We've made an investment that's unannounced in this, but I think that forthe first time in a long time, there's actually room for a new CPU approach. The thing that'shappening right now, and you see this reflected in all the semi-stocks and everything else, is the LLMs,which are running on accelerators and GPUs, generate code. The code runs on CPUs. And right now,it's running on classic CPUs we've had around forever. But there's a whole bunch of constraintson CPUs that have existed, and CPUs have dragged all this baggage forward that you might not need toanymore. And so I'm actually really excited about that possibly.The next category.

我刚说的这些，就是我当时知道的全部。后来显然应验了。而在之前的每一代里，我们得到了 Intel，得到了 NVIDIA，得到了 Broadcom 和 Avago，得到了 Qualcomm 和 ARM，每一代都有。而这一代也会出现这样的巨型赢家、独立的赢家。显然 TPU 本身是个赢家，训练是个赢家。你还有 Groq、Cerebras、Etched。这仗还会继续打，但我觉得最后会打出很大的东西。而且我其实觉得，还有第六代正在到来，我对此非常兴奋。我们在这方面做了一笔还没公开的投资：我认为很久以来第一次，真的有空间做一种新的 CPU 路线了。现在正在发生的事——你在所有半导体股票上都能看到这个反映——是跑在加速器和 GPU 上的大模型在生成代码，而代码是跑在 CPU 上的。而现在，代码跑的是我们用了几十年的经典 CPU。但 CPU 上存在一大堆历史约束，CPU 一路拖着这些包袱走到今天，而现在你可能不需要再背着它们了。所以我其实对这件事非常兴奋。（Patrick）下一个品类。

[36:24] Eric Vishria

The next category.Does the experience with Cerebrus make you want to do a lot more investing in companies?Fuck no.But why not? In 2019, we're sitting in a board meeting, and this thing is melty. It's likefucking melty. Okay? And we've raised $500 million or something. And it's like, wait, what? It'smelting or burning or something? And we're looking at it. And I'm like, holy shit. I realize like $500million isn't that much in today's era. But I was just like, we're going to lose all this money.This is not going to work. And what that team did, insane. They're built differently. And I have somuch respect and thanks to them for what they've done. There are efforts that make you really proudto be a venture capitalist because you're funding something that makes a difference and matters.I'm an investor because that's a means to work with companies, not because I fundamentally loveinvesting or something like that. I like working with the companies. That's my favorite part of it.Working with teams like that and companies like that is so special on these giant, ambitious efforts.And I said this well before 2018 or whatever, is whether Cerebrus worked or didn't, I think it wasan effort that was worth venture capital. That was the kind of thing you should do. You should try to

（Eric）下一个品类。（Patrick）Cerebras 的经历会不会让你想做更多这类（硬件）公司的投资？（Eric）他妈的不会。（Patrick）为什么不呢？（Eric）2019 年，我们坐在董事会上，这玩意儿在融化，真他妈在融化，好吧？而我们已经融了 5 亿美元还是多少。我们就想：等等，什么？它在熔化还是在烧起来？我们盯着它看，我心想：我的天。我知道 5 亿美元放在今天这个时代不算多，但我当时就是觉得：我们要把这笔钱全赔光了，这事儿成不了。而那个团队后来做出来的事情，太疯狂了。他们的构造跟常人不一样。我对他们做成的事情充满敬意和感激。有些努力会让你真心为自己是个风险投资人而骄傲，因为你在资助一件真正有意义、能带来改变的事情。我做投资人，是因为这是与公司一起工作的手段，而不是因为我本质上热爱投资之类的。我喜欢跟公司一起工作，那是我最喜欢的部分。跟那样的团队、那样的公司，在这种宏大、雄心勃勃的事情上一起工作，是非常特别的。而且我早在 2018 年之前就说过：不管 Cerebras 成不成，我认为这是一件值得风险资本去做的事，这就是那种你应该去做的事。

[37:38] Eric Vishria

build something that people have tried for 50 years and have been unable to, but now we think we coulddo and there's a reason and application for it and everything else. I love that. I do like those kinds ofthings, joking aside, and we do have a robotics company and then this CPU project that we'retalking about. I think these kinds of things are actually really fun and interesting and good useof vendor capital, but they definitely aren't easy. The productive naivete that you described,Rer, is probably a virtue that you didn't know more than you knew, otherwise you wouldn't have done it.This is certainly my experience with etch. Like in any of these fields, you ask experts,they're going to tell you, don't do it. It sucks. It's too hard. Base rate's too low.Young people can't deal with 47 reasons. Is there anywhere where that is just a bridge toofar? That could be like bio or something like this, where you just are unwilling to investif you're naive. You know what's so funny is the hard thing about investing is most of the time,all of these stereotypical statements are correct. They're not correct three times out of four.They're correct 19 times out of 20, maybe 99 times out of a hundred. Like they are correct.

去造一件人们尝试了 50 年都没做成、但现在我们认为可以做成、而且有理由、有应用场景的东西。我爱这个。玩笑归玩笑，我确实喜欢这类事情，我们还投了一家机器人公司，还有我们正在聊的这个 CPU 项目。我觉得这类事情其实非常有意思、非常好玩，是风险资本很好的用途，但它们绝对不容易。（Patrick）你刚才说的那种「有生产力的天真」，大概正是一种美德——正因为你知道的没那么多，否则你就不会做了。这肯定是我在 Etched 上的体验。在任何这类领域，你去问专家，他们都会告诉你：别干，这行太烂了，太难了，基准成功率太低了，年轻人搞不定，能给你举出 47 个理由。有没有哪个地方是「这一步实在太远了」？比如生物之类的，你在天真的状态下就是不愿意投？（Eric）好笑的是，投资难就难在：绝大多数时候，所有这些刻板印象式的说法都是对的。它们不是四次里对三次，而是二十次里对十九次，甚至一百次里对九十九次。它们就是对的。

[38:39] Eric Vishria

The thing that Bruce, one of our founders always says, what could go right? We have to ask ourselves,what could go right? And do we see that path? Yeah. Young people can't build chips or we shouldn'tdo another semi company or you shouldn't do this or like whatever. All of that stuff is actuallytotally right. Except when it isn't. Apparently there's a saying that someone said to one of mypartners, which was, if it doesn't work, it'll be for all of the reasons that your partner said.If it does work, it will be because those reasons didn't matter. It's just such a good example ofthis whole thing, which is, yeah, most of the time when we lay out the reasons that a company won'twork, it's right. But then sometimes they just don't matter. You and I are both interested inrobotics. It's not controversial that if robotics works, it might dwarf what we're currently livingthrough. What do you think has to be true for it to work? Obviously it's exciting. I want a robot in myhouse, folding my laundry. It sounds great. It's kind of one of these classics. It's always 10 yearsaway and it's been that way for a long time. What do you see happening? What has to happen for thisto actually be a thing in the near to medium term? We've had classics robotics forever and they're

我们的创始人之一 Bruce（Bruce Dunlevie）总说的一句话是：什么可能会做对？我们必须问自己：什么可能会做对？我们看得到那条路径吗？（Patrick）对。（Eric）「年轻人造不了芯片」「我们不该再做一家半导体公司」「你不该做这个」之类的，所有这些说法其实完全正确——除了它们不正确的那些时候。据说有人对我的一位合伙人说过一句话：如果它失败了，一定是因为你合伙人说过的所有那些理由；如果它成功了，那是因为那些理由都不重要。这就是这件事最好的例子：是的，大多数时候我们列出一家公司做不成的理由，都是对的；但有时候，那些理由就是不重要。（Patrick）你我都对机器人感兴趣。有个不太有争议的说法是：如果机器人成了，它的规模可能会让我们现在经历的这一切都相形见绌。你觉得要让它成，必须有什么条件成立？显然它很激动人心，我想要一个机器人在我家里叠衣服，听起来很棒。但它属于那种经典的「永远还差十年」，而且这个状态已经持续很久了。你看到了什么？要让这件事在中短期内真正成立，必须发生什么？（Eric）传统机器人我们一直都有，

[39:47] Eric Vishria

all over the place and they're on assembly lines and manufacturing lines and all those things whereyou're doing repetitive tasks in controlled environments. Repetitive tasks in controlledenvironments is like more or less solved and like that, you know, that'll continue to happen.But having tweaked tasks in real world environments, that's where you need the AI. That's where you needthe AI plus the robots. The trick with it all is you need a model that can do that. Of course,the first problem is there is no internet scale data to bootstrap the whole thing, right? LLMs are allbootstrapped on the internet, which is a ton of human knowledge. And the equivalent for that,for robots doesn't exist. And people are trying different things with videos and simulations andteleoperation. So like there's a lot of different ways. But if you kind of think about, take teleopas an example, how much you need to teleop robots to get to an internet scale data. This first step islike getting a good set of data to bootstrap the model. One of the, I think, insights that you can haveis with the internet. There's like a lot of slop data. Even before AI generated all this stuff, there wasalso a bunch of junk data. And some data was more valuable than others. Like you may value, okay, certain

到处都是，在装配线上、生产线上，那些在受控环境里做重复任务的场景。受控环境里的重复任务基本上算是解决了，这块会继续推进。但在真实世界环境里做有变化的任务，那才是你需要 AI 的地方，是你需要 AI 加机器人的地方。整件事的诀窍在于：你需要一个能做到这件事的模型。当然，第一个问题是：没有互联网规模的数据来给整件事做冷启动，对吧？大模型都是靠互联网冷启动的，那是海量的人类知识。而机器人没有对应的东西。人们在尝试不同的路子：视频、仿真、遥操作（teleoperation，人远程操控机器人来采集数据）等等，路子很多。但你想想，拿遥操作举例，你要遥操作多少机器人才能凑到互联网规模的数据。第一步就是拿到一批好数据来给模型冷启动。我觉得可以有的一个洞察是：互联网上有大量的垃圾数据（slop）。就算在 AI 生成这些东西之前，本来也有一堆垃圾数据，而有些数据比另一些更有价值。比如你会觉得 Reddit 上的某些内容比另一些更有价值，

[41:08] Eric Vishria

things on Reddit more valuable than other things. You might value Wikipedia more than like other forums. Youmight value GitHub more than other things. And all of the model companies did that, right? They prioritizeddata that was more valuable and less valuable and ran that through the model. I think one of the mostinteresting things that these AI robotics companies are doing are saying, okay, well, let's just goafter the high value data to start. If we go out for the high value data, then we can kind of bootstrapthis model. Now, once you do that, then can you, through the pre-training process, get to a placewhere you can have very small auxiliary examples of data that you add in post-training and all of a suddenit works for that. That's the magic we have with LLMs is you have this giant pre-trained base,and then you add a little magic on it in RL and post-training and you teach it a new thing thatit wasn't in the pre-training set and you kind of go from there. And I think the exact same thinghas happened in robotics. We're investors in Sunday robotics, which is going out for this exact pipeline.It's a cool company.It's a cool company and they're doing household robots. But the key part before you get to household

你可能觉得维基百科比其他论坛更有价值，你可能觉得 GitHub 比别的东西更有价值。所有模型公司都这么做，对吧？他们把数据按价值高低排序，再喂进模型。我觉得这些 AI 机器人公司在做的最有意思的一件事是：好，那我们就先去搞高价值的数据。如果我们冲着高价值数据去，就能给这个模型做冷启动。然后，一旦你做到了这一步，你能不能通过预训练走到一个位置：只要在后训练里加入很少量的辅助样本，它突然就会做那件事了。这就是我们在大模型上拥有的魔法：你有一个巨大的预训练底座，然后在上面用强化学习（RL）和后训练加一点魔法，教它一件预训练集里根本没有的新事情，然后就这么走下去。我觉得机器人里正在发生的是完全一样的事。我们投了 Sunday Robotics，他们走的正是这条流水线。（Patrick）这公司挺酷的。（Eric）挺酷的公司，他们做的是家用机器人。但在做到「家用」之前，关键的部分

[42:13] Eric Vishria

and all that stuff matters much less actually than can you get this training pipeline to workand how do you do that? And one of the lessons I learned and I look back, I try to learn fromhistory because it doesn't repeat, but it rhymes. And if I look at autonomous vehicles as an example,which is they're robots, basically. They're AI plus robots. You look at Waymo and you look at Tesla.They were both designed vertically integrated in their own way. You have a Tesla,you had a fleet of Teslas that everybody owned. It was collecting data on the Teslas and it was usedto train the models to drive the Teslas. The same thing with Waymo. I think part of the lessons is theygather very, very high quality data. They did their pre-training and then they worked off of that.That was a simplification to allow you to get to a complete product or a complete solution,which of course is going to continue to improve and ultimately will generalize. I'm sure it'll generalizein some ways. You'll be able to strap it on any car and everything else. So I think the same thing ishappening in robotics where you have companies like Sunday and others who are using techniques wherethey're vertically integrating the robot and the model, the data collection around the robot.

其实比这些重要得多：你能不能让这条训练流水线跑起来，以及你怎么做到。我学到的一课，回头看，是我尽量从历史里学习——因为历史不重复，但会押韵。如果我拿自动驾驶汽车举例，它们本质上就是机器人，是 AI 加机器人。你看 Waymo，你看 Tesla，它们都以各自的方式做了垂直整合。Tesla 是：路上有一整支所有人都拥有的 Tesla 车队，车队在采集数据，数据被用来训练驾驶 Tesla 的模型。Waymo 也是一样。我觉得部分教训是：它们采集的是非常非常高质量的数据，做完预训练，然后在这个基础上往前推。这是一种简化，让你能做出一个完整的产品或者完整的解决方案，而它当然会持续改进，最终会泛化。我相信它总会以某种方式泛化，你能把它装到任何一辆车上等等。所以我觉得机器人领域正在发生同样的事：像 Sunday 这样的公司在用垂直整合的打法，把机器人和模型、以及围绕机器人的数据采集整合在一起。

[43:28] Eric Vishria

Sunday uses gloves that are designed with the robot hands. So they're perfect. So you get very gooddata transferability from one to another. You do this pre-training and you have a great pre-trainingdata set. You have a good model. And then you start adding these examples and URL and post-train on topand you get cool emergent behavior. Do you have a most visceral moment?

Sunday 用的手套是配合机器人的手来设计的，所以两者完全对应。这样你从人到机器人的数据可迁移性非常好。你用它做预训练，就有了一个很好的预训练数据集，有了一个好模型。然后你开始加入这些样本、做强化学习和后训练，就会得到很酷的涌现行为。（Patrick）有没有哪个瞬间让你感受最强烈？

[43:50] Eric Vishria

When we invested in Sunday the first time we saw them, partner Peter arranged a demo. We went down intothe basement at Stanford lab and they had like this totally janky cardboard glove thing. You see a fewevolutions of it. The last time we saw a demo, we went down in the basement of their now officebuilding and there was like a dozen robots just folding arbitrary lines. It wasn't in the demo.It was just trial and error. And then they have people like taking the clothes and then measuringthem to make sure that they were folded properly and creating a rigorous baseline and evaluationcriteria. And I was like, oh my God, this is happening. Do you ever worry about how to pick the rightcustomer for these companies? Every one of these things folds laundry, which I don't think anyonelikes folding laundry. So it seems like a good use case, but it feels like we don't actuallyunderstand the demand for what these things will be used to do. Do you ever worry about that?

我们投 Sunday 的时候，第一次见他们，合伙人 Peter 安排了一次演示。我们下到斯坦福实验室的地下室，他们拿出来的是一个特别山寨的纸板手套之类的东西。后来你会看到它的几次迭代。最近一次看演示，我们下到他们现在办公楼的地下室，那里有大概十几个机器人在叠各种任意形状的衣物。这不是演示里排练好的，就是不断试错。而且他们还有人把叠好的衣服拿走去量，确认是不是叠得规整，建立一套严格的基线和评估标准。我当时就想：我的天，这事儿真的要成了。（Patrick）你会不会担心怎么给这些公司挑对客户？每一家都在叠衣服，我觉得没人喜欢叠衣服，所以看起来是个不错的用例，但感觉我们其实并不真的了解这些东西未来会被用来干什么、需求在哪里。你会担心这个吗？

[44:46] Eric Vishria

We're sort of building solutions that will then be in searches of problems?I don't. And I'll explain why. And it's certainly informed by watching the LLM evolution. Some of thepeople who were involved in the early LLM at OpenAI really understood that code was going to beimportant. And obviously, you know, the Anthropic team had this perspective that you could get to RSIif you've got code generation going and automating AI research and whatnot. If you think about it,the first use cases were very much language oriented. They were very much essay writing and editing andmarketing. I think the first application that really took off was Jasper, which was justwriting marketing copy. I think it's going to evolve a lot, basically. I think laundry iskind of a good task because it's arbitrary, it's complex, it requires dexterous manipulation.It isn't time sensitive. If it takes three times longer, so be it. Who cares? It doesn't matter.Just let it run all day. So I think it has like some of those properties. I don't think the task isactually that important. What I think is much more important is, are you pre-training an amazing modelthan being able to post-train on top of it and get that flywheel going? If you get that flywheel going,

（Patrick）我们是不是在造一堆解决方案，然后再去找问题？（Eric）我不担心，我说说为什么。这肯定是被我观察大模型演化的过程影响的。OpenAI 早期做大模型的一些人真的意识到代码会很重要。而且显然 Anthropic 团队有这个观点：如果你把代码生成搞定、把 AI 研究自动化，就能走到 RSI（递归自我改进）。但你想想，最早的用例其实非常语言导向：写文章、编辑、做营销。我记得第一个真正跑起来的应用是 Jasper，就是写营销文案的。我觉得它会演化很多。我觉得叠衣服其实是个不错的任务，因为它是任意形态的、复杂的、需要灵巧操作的，而且它不是时间敏感的：慢三倍又怎样？无所谓，让它跑一整天都行。所以它具备这些属性。我不认为任务本身有那么重要。我认为重要得多的是：你有没有在预训练一个很棒的模型，能不能在它上面做后训练、把飞轮转起来。如果那个飞轮转起来了，

[45:56] Patrick O'Shaughnessy

then the task capability will just keep multiplying. Maybe this question will beannoying or slightly uncomfortable for you. But if you ask basically every founder and critically otherinvestors of your type, almost everyone, if I ask like who's the best board partner, will say you.You come up way more often than anyone else that I've come across. And I'm curious why you thinkthat is. What it is that you're doing that other people, the incentives are there to do a great jobas a board partner. What do you think you're doing on the boards of these companies, partner with thefounders, that's actually different from other really talented investors who are also nominally doingthe same job, but don't come up nearly as often when asked that question?

任务能力就会不断倍增。（Patrick）这个问题可能有点烦人，或者让你稍微不太舒服。但如果你去问几乎每一位创始人，还有关键的、跟你同类型的其他投资人——几乎所有人——「谁是最好的董事会伙伴」，他们都会说是你。你被提到的频率远高于我遇到过的任何其他人。我很好奇你觉得这是为什么。你做了什么是别人没做的？做好董事会伙伴这件事，激励其实是齐备的。你在这些公司的董事会上、在跟创始人搭档这件事上，到底做了什么，是那些同样非常有才华、名义上也在做同一份工作、但被提到的次数远没有那么多的投资人所没做的？

[46:37] Eric Vishria

One of the things that I've realized is we each are attracted to different types of entrepreneurswhere we have chemistry. I'm an investor second and I try to be a partner first.We'll see companies come in. We had one come in yesterday and it's what I would call an investmentgrade opportunity. You can invest, it probably works, you make money, it's good. And an investor would dothat. A partner wouldn't because that's not sufficient for a partner. Unless you have realchemistry with that person where you feel like you're going to be able to work together reallyeffectively and I'm going to learn a ton from them and they're going to learn something from me.Together, we're going to just feed each other's loops. Unless you feel that way, you can't be apartner. And so you pass on that. That's a really important fit element to me.It kind of starts with this mutual selection, actually, weirdly. They want to partner with usand I want to partner with them. I'm really looking forward to working with them together.And I'll give you a really good example of where this comes into play for me.If I take Saji and Benchling, Benchling is life sciences, SaaS, companies absolutely crushed,done really, really well. And then of course, you have this biotech crash and everything else.

我意识到的一件事是：我们每个人都会被不同类型的创业者吸引，跟不同的人产生化学反应。我首先是伙伴，其次才是投资人。会有公司找上门来——昨天就来了一家——那是我称之为「投资级机会」的东西：你可以投，大概率能成，能赚钱，是个好项目。一个投资人会投。但一个伙伴不会，因为这对伙伴来说不够。除非你跟这个人有真正的化学反应，你觉得你们能非常有效地一起工作，我能从他们身上学到很多，他们也能从我这儿学到点什么，我们能互相给对方的循环加料——除非你有这种感觉，否则你没法做伙伴。所以你就放弃了。这对我来说是个非常重要的匹配要素。奇怪的是，它其实是从一种双向选择开始的：他们想跟我们搭档，我也想跟他们搭档，我真心期待跟他们一起工作。我给你举一个特别好的例子。拿 Benchling 的 Saji 来说，Benchling 做的是生命科学 SaaS，这家公司做得非常非常好。然后当然，生物科技崩盘了等等。

[47:52] Eric Vishria

And the company became grindy. The company had never had any churn for the longest time,such that even on their reports for every SaaS company, you have this like, okay,gross ARR added, churn line, net ARR added, like everybody does the same thing. They never had achurn line, never reported it for the first of like six years that I worked with the company.So then they got seven years of churn in like 12 months. Turns out life sucks when you get sevenyears of churn in 12 months. Through that grind and through it all and related to the whole,wait a minute, the goalpost moved. We have to do something different, everything else.They kept thinking about how to apply AI for these biotech and pharma customers,which they're very close to. How can we make it better for them? How can we apply these modelsin their world in a way that they're excited about and continue to iterate? And it was grindy.And I was there for it. You're excited to work with that person. Cause like one, of course youthink it's a really special opportunity and there's a way out, there's a path and we can find it.But two, because of the joy of the game, the relationship and like everything else,like that's part of it. It's very different. I've seen different models and lots of different

公司变得很磨人。这家公司在很长时间里从来没有过流失（churn，指客户流失掉的收入），以至于在他们的报表上——每家 SaaS 公司都有这套：新增 ARR、流失、净新增 ARR，大家都一样——他们从来没有流失这一行，我跟这家公司合作的前六年他们就没报过这一项。结果他们在大约 12 个月里吃到了七年的流失。事实证明，12 个月吃七年的流失，日子会很难过。在那段磨人的过程里，也跟「等一下，球门被挪了，我们必须做点不一样的事」这整件事相关：他们一直在想怎么为这些生物科技和制药客户应用 AI——他们离这些客户非常近——怎么让客户过得更好？怎么用一种客户会兴奋的方式把这些模型用到他们的世界里，并且不断迭代？那个过程很磨人，而我全程在场陪着。你会很期待跟那样的人一起工作。因为一来，你当然觉得这是个非常特别的机会，是有出路的，路径存在，我们能找到它；二来，是因为这场游戏本身的乐趣、这段关系等等，那也是其中一部分。这非常不一样。我见过很多不同的模式，

[49:01] Eric Vishria

models of venture capital work. Moritz was a writer. Doerr was a sales guy. Gurley was an engineer.Peter's a career venture capitalist. They're all different. Working with them. It's like when I callthese people, I learn something and they push back on me and then I ask them questions. And what I'verealized is like so much of my job is they know the answer. They know what they want to do. Theyknow the answer. And it's maybe asking questions of them to maybe help solidify their conviction orsolidify their articulation of what they want to do and why. And you just keep doing that through thatprocess. Hopefully we get 1% better a few times a year. We make a 1% better decision, 2% betterdecision a few times a year. And if you do that over a decade, that compounds to real results.One of the questions that I ask myself before making an investment is there are all these peoplethat I care about through my life, like you carriers. Could I talk one of them into going to this companyand honestly, intellectually, honestly to myself, explain to them why this could be their life'swork? And if I can't do that, I should not invest. It just means that the project doesn't line up inthat way. And so as long as we have one of those things, it doesn't matter that much what it is

很多不同的风险投资模式都能成。Moritz（Michael Moritz）是写作出身，Doerr（John Doerr）是销售出身，Gurley（Bill Gurley）是工程师出身，Peter 是职业风险投资人。他们各不相同。跟他们共事——我给这些人打电话时，总能学到东西，他们会反驳我，然后我再问他们问题。我意识到的是：我的工作很大程度上在于，他们自己知道答案，他们知道自己想做什么，他们知道答案。而我要做的也许是向他们提问，帮他们把信念固化下来，或者把「他们想做什么、为什么」的表达固化下来。你就一直在这个过程里这么做。希望我们一年能有那么几次做到好上 1%，做出一个好 1% 的决定、好 2% 的决定，一年就那么几次。如果你坚持十年，这会复利成真正的结果。我在投资前会问自己的一个问题是：我这辈子有很多我在乎的人，那些一路共事过来的人。我能不能说服其中一个去这家公司，并且诚实地、对自己在智识上诚实地，向他们解释为什么这可能是他们的毕生事业？如果我做不到，我就不该投。这只是意味着这个项目在那个维度上对不齐。所以只要我们有这样一件事，做的具体是什么其实没那么重要，

[50:22] Eric Vishria

to me. It's just, it's important. It could make a big dent. And if it can make a big dent and it'sa special person, I'd love to work on it. Are there any other questions like that that you askyourself before investing? That's a particularly good one. The other question, if this person callsme at 9 PM on a Saturday night, will I pick up the phone? Calls that the green button test?

对我来说重要的是它要紧、它可能砸出一个大坑。如果它能砸出一个大坑、而且对方是个特别的人，我就很想跟他一起干。（Patrick）在投资前你还会问自己别的类似的问题吗？这个问题特别好。（Eric）另一个问题是：如果这个人在周六晚上 9 点给我打电话，我会不会接？（Patrick）你管这叫「绿色按钮测试」？

[50:41] Eric Vishria

Yeah, yeah, yeah. It's a chemistry thing and they have to feel the same way, obviously. One of theother ones is if it's right, does it matter? Which is different than this like first one,but it's, there's so many things that we could be right on as a business. I looked at one last weekand I told an entrepreneur, I was like, I really think that you can build an amazing company hereand you just shouldn't raise venture capital. But there's like so many things that you can be righton, but they ultimately just don't matter. Like nobody cares. That's a better way to say it. If we'reright, we'll let many care. If they won't care, then you're just not going to build enough equity value.That's another useful one.One of the coolest things that's happening right now is all of that that you just describedhas higher stakes and more leverage attached to it, which is manifested most simply in more dollarsand higher prices. You and I have talked about this notion of what high multiple on invested capitalinvesting is like and what it has been like and what it's moving into. You did something recently,which was you raised a growth fund for the first time in a long time. I think that is related tothis concept. These companies need more capital. The prices are higher. The outcomes are bigger.

对对对。这是个化学反应的问题，而且显然对方也得有同样的感觉。还有一个问题是：如果我们判断对了，这件事重要吗？这跟第一个问题不一样。因为有太多东西我们可以在商业判断上做对，但它最终就是不重要。我上周看了一个项目，我跟创业者说：我真的觉得你能在这儿造一家很棒的公司，而你就是不该拿风险投资。有太多事情你可以判断对，但它们最终就是不重要，没人在乎——这么说更准确。如果我们判断对了，会有很多人在乎吗？如果他们不在乎，那你就创造不出足够的股权价值。这也是个有用的问题。（Patrick）现在最酷的一件事是：你刚才描述的这一切，都被挂上了更高的赌注和更大的杠杆，最简单的表现就是更多的钱和更高的价格。你我聊过「高投入资本回报倍数（MOIC）的投资」是什么样子、过去是什么样子、以及它正在变成什么样子。你最近做了一件事：你们很久以来第一次募了一支成长基金。我觉得这跟这个概念有关：这些公司需要更多资本，价格更高了，结果也更大了。

[51:43] Patrick O'Shaughnessy

Maybe we can earn the same multiple on a billion dollar entry price that we could on a $50 millionentry price 10 years ago or whatever. Can you talk through that evolution, talk through thepartnerships, like way of thinking about it and talking about it, what you believe to be truethat results in this decision to do this?

也许我们在 10 亿美元的入场价上，能赚到跟十年前 5000 万美元入场价一样的倍数。能不能讲讲这个演变，讲讲合伙人之间是怎么思考、怎么讨论的，以及你相信什么是真的，才导致了这个决定？

[51:58] Eric Vishria

Just go back to like why did LPs starting with Swenson and everyone else, why did they startinvesting in venture capital? And fundamentally, it wasn't because they thought they could beat theNASDAQ or the index by like three percentage points a year, five or whatever. It's becausethere are situations where venture capital could drive these insane multiples on invested capital.From a financial perspective, that's what they're seeking.For the longest time for most of the industry's history, two things were synonymous, early stageinvesting and high cash on cash multiples. The way to get high cash on cash multiples was to do earlystage. That's it. Those two circles in the Venn diagram like almost perfectly overlap.The thing that's changed is recently, relatively recently in the last few years, because outcomes havegotten so much bigger and these markets are bigger and everything else, the circle of high cash oncash multiple opportunities is bigger than just early stage. And it's not so, so big that there'sa gazillion new companies in there that you can generate 100 Xs on. That's not true. But there arecertainly many outside of early stage where you can generate high returns. That's it. I think that's

回到最初：为什么 LP（出资人）们——从 Swensen（David Swensen，耶鲁捐赠基金）开始的所有人——为什么他们开始投风险投资？根本上，不是因为他们认为自己能每年跑赢纳斯达克或者指数三个百分点、五个百分点之类的，而是因为存在这样一些情形：风险投资能带来疯狂的投入资本回报倍数。从财务角度看，这才是他们在找的东西。在这个行业历史的绝大部分时间里，有两件事是同义词：早期投资，和高现金回报倍数。要拿到高现金回报倍数，唯一的办法就是做早期，就这样。这两个圈在文氏图里几乎完全重合。而变化在于：最近几年，因为结果变得大得多、市场也大得多等等，「高现金回报倍数机会」这个圈，已经大过「早期」这个圈了。它也没有大到里面有无数新公司能让你做出 100 倍，那不是真的。但在早期之外，确实有不少地方可以拿到高回报。就这样。我觉得这就是

[53:12] Eric Vishria

what we want to go after. You could argue we're a few years late. I think I'd take that criticism.But I think that opportunity exists on a go-forward basis. We should go do it. Everything else that werepresent, which is the high conviction, high commitment partnership, that has to still be there.As part of that discussion, what were like the other sides of the debate, such as maybe we wouldhave said the same thing in 99 and halfway through 2020. As markets get exciting, the possibilities,we all do this extrapolation error. What were the counter arguments to like, let's, despite all that,still not do it. I think the biggest counter argument that really made this time right versustwo years ago or whatever was you need the team that can do it. It's just a different mentality.There are differences in how you evaluate and think about things. All the other stuff or there's,you know, why not change and stay within your circle of competency and all those things are true.But that was like the biggest one. We had several examples over the last couple of yearswhere we had, I think, the right intuition on a company or an opportunity. We didn't do it.Because it was outside the box. Because it was outside the box. That was obviously dumb. And I think

我们想去做的事。你可以说我们晚了几年，这个批评我接受。但我觉得这个机会在往前看的时间里是存在的，我们就该去做。而我们代表的其他一切——高信念、高投入的伙伴关系——必须依然在场。（Patrick）在那场讨论里，反方的论点是什么？比如说，也许我们在 99 年、或者 2020 年过半的时候也会说同样的话。市场变得令人兴奋的时候，我们都会犯这种外推的错误。有哪些反驳意见是「即便如此，我们还是不该做」？（Eric）我觉得最大的反方论点、也是让这次比两年前更合适的原因，是你需要一支能干这件事的团队。这是一种完全不同的心态，在评估和思考方式上有差别。其他那些理由——为什么不该变、该待在自己的能力圈里之类的——都成立。但最大的那一条就是团队。过去这几年我们有好几个例子：我们对某家公司或者某个机会有正确的直觉，但我们没做。（Patrick）因为它在框框外面。（Eric）因为它在框框外面。那显然很蠢。而我觉得

[54:21] Eric Vishria

it is quite different than a lot of, than she does. We are really chasing these very rare specialcompanies that have like very high cash on cash opportunities where we think there just can berunaway successes and we can invest in them. Is there any lesson to be pulled from the many,let's call it 20 to 100 X's that you personally have observed? I mean, it's such a crazy amountof return. Obviously, it doesn't pencil in the beginning. You can't make something pencil if itwas that clear. The price would be different. What have the 20 to 100 X's taught you in aggregate,if anything? Work with really special people. You want to work really hard. You want to work smartand get lucky. And you need it all. It really all has to come together. There's a lot of thingsthat are timing dependent. You have no control over as a company. And you take the Cerebrus exampleas a good one, which is, this is our second. We took it public this time in May, but we tried totake it public in 2024. And it would have been taken public at a much, much lower valuation. Andit didn't work out because of CFIUS and all this stuff. So timing matters. The advancement of that 18months made all the difference in the world. For a bunch of things that were honestly outside of

这跟很多同行的做法很不一样。我们真正追逐的是那些极其稀有、极其特别的公司，它们有非常高的现金回报机会，我们认为那里会出现失控式的成功，而我们可以投进去。（Patrick）从你亲眼见过的那么多——姑且说 20 倍到 100 倍的案例里，能提炼出什么教训吗？这是个疯狂的回报量级。显然它在一开始是算不出来的，如果那么清楚，价格就不会是那个价格了。这些 20 到 100 倍的案例加总起来教会了你什么，如果有的话？（Eric）跟真正特别的人一起工作。你得非常努力地工作，你得聪明地工作，还得走运。这些你全都需要，它真的必须全部凑到一起。有很多事情是依赖时机的，作为公司你完全无法控制。拿 Cerebras 举个好例子：这是我们第二次尝试，这次是今年 5 月上市的，但我们 2024 年就试过一次上市。那次上市的估值会低得多得多，而且因为 CFIUS（美国外国投资委员会）审查这些事没成。所以时机很重要。往后推的这 18 个月带来了天壤之别，

[55:29] Eric Vishria

our control, there were some things that were in our control, getting inference running andeverything else. But there was a lot of stuff that was outside of our control. These are allthe classic things. We just got to focus on what you can control. That's one of the things that'sreally different than software companies. With software companies, aside from like building onAWS or whatever, you pretty much own your whole stack. And so you're really fully in control ofyour destiny in that way. With hardware companies, you don't. There's an entire supply chain and allthis other stuff. HBM's a thing.DRAM's a thing. And TSMC's a thing. And a lot of those cross geopolitical borders. And so geopoliticsgets involved. And that makes that complicated. That's a really big difference. And so you gotto get lucky on the timing and macro and other stuff. But I think it really starts with workingwith these crazy people with unbounded opportunities. And if you work with these crazy special people onunbounded opportunities, you get lucky from time to time. You're bound to. When I was 20 years old,I was working at an investment bank. Ben Horwitz, Mark and Ben had started LoudCloud. It was still instealth. And Ben gave me an offer to be his assistant. I was talking to this associate who

而其中很多事情说实话是我们控制不了的。有些事在我们控制之内，比如把推理跑起来等等，但有一大堆是在控制之外的。这些都是老生常谈：我们只能专注于自己能控制的事。这也是跟软件公司非常不一样的一点：软件公司除了建在 AWS 之类的东西上，你基本上拥有整个技术栈，所以在那个意义上你完全掌握自己的命运。而硬件公司不是，有一整条供应链和一大堆别的东西。HBM（高带宽内存）是个变量，DRAM 是个变量，台积电是个变量，而这里面很多都跨越地缘政治边界，于是地缘政治也掺和进来，把事情搞复杂。这是个非常大的差别。所以你在时机、宏观和别的事情上得走运。但我觉得根本还是要跟这些疯狂的人、在无边界的机会上一起工作。如果你跟这些疯狂而特别的人在无边界的机会上共事，你时不时就会走运，这是必然的。我 20 岁的时候在一家投资银行工作。Ben Horowitz——Marc 和 Ben——刚创办了 LoudCloud，当时还在隐身状态，Ben 给了我一个当他助理的 offer。我当时在跟一位 associate 聊天，

[56:31] Eric Vishria

seemed like this elder at the time. He's probably 25. He said this thing to me, which stuck with me.He was like, do you golf? It's like classic banking question. Do you golf? No, I don'tfucking golf. But he's like, with golf, you keep on practicing, keep getting the ball in a three-parlike close to the pen, close to the pen, close to the pen, close to the pen. You keep getting theball close to the pen and you keep practicing. That's hard work. That's like working smart. That's whatyou want to keep doing. Getting the hole in one, that's luck. I kind of love that framing I usewith my kids, actually. What it says is like, yeah, there's luck involved. And there really isluck involved. But there is actually a way to increase your luck. And the way to increase yourluck is get a lot of balls close to the pen. Eventually one will drop. I like that. Each ofus invests in one to two companies a year. I think in my 12 years, I've invested in 18 companies total.Crazy.Which is a relatively small number. So it's a very high conviction and very high commitment.I have a lot of skin in the game. I believe in these companies. If you keep on workingwith these very special people in these opportunities, magic can happen.

他在当时看起来像个长者，其实大概 25 岁。他跟我说了一句话，我一直记着。他说：你打高尔夫吗？这是投行的经典问题。你打高尔夫吗？我说：不，我他妈不打高尔夫。但他说：打高尔夫，你不断练习，在三杆洞上不断把球打到离旗杆很近的地方，一次比一次近，你不断把球打到旗杆附近，不断练习——这是苦功，这是聪明地工作，这是你该一直做的事。而一杆进洞，那是运气。我挺喜欢这个框架，我现在还拿它跟我的孩子讲。它说的是：是的，运气是有的，而且确实有运气成分。但其实有办法提高你的运气，办法就是把很多球打到旗杆附近，最终总会掉进去一个。（Patrick）我喜欢这个。（Eric）我们每个人一年投一到两家公司。我在 12 年里一共投了 18 家公司。（Patrick）太少了。（Eric）这是个相对很小的数字。所以这是非常高信念、非常高投入的。我有大量的切身利益在里面，我相信这些公司。如果你持续跟这些非常特别的人、在这些机会上一起工作，魔法是会发生的。

[57:33] Eric Vishria

What have you learned about the best reasons and conditions for going public?Bedford does these Monday night dinners. And so we had a CDA last night, multi-hundred billiondollar private company. And we had this whole conversation. So it's kind of fresh.I think that ultimately when you go public, you have a range of new opportunities in what you can do.Public trust, because there's some transparency that comes with being a public company. You obviouslyhave a currency that you can do things with that ends up being there. You have an unbelievable abilityto raise capital, which I think is why the labs will ultimately go. Although I think the trust thingis actually a really important element of why they should go. And it's beneficial to the world andto America if they do go public. It's like-See what's going on.Yeah. See what's going on. Everyone can see it. I think that's like a really beneficial setup.There's another element of it, which is what does a collegiate athlete want to do?

（Patrick）关于「什么时候、在什么条件下上市最好」，你学到了什么？（Eric）Benchmark 有周一晚上的例行晚宴。昨晚我们请来了一位 CEO，一家估值好几千亿美元的私营公司，我们就这件事聊了一整场，所以还很新鲜。我觉得最终，当你上市之后，你在能做的事情上会有一整片新的机会。公众信任是一块，因为上市公司会带来一定的透明度。你显然还会有一种「货币」（可用于收购的股票）拿去做事。你有不可思议的融资能力，我觉得这也是那些实验室最终会上市的原因。不过我觉得「信任」这件事其实是他们该上市的一个非常重要的理由，而且如果他们上市，对世界和对美国都是有益的。（Patrick）能看见到底在发生什么。（Eric）对，看见在发生什么，所有人都能看见。我觉得那是个非常有益的设置。还有另一个层面：一个大学运动员想要什么？

[58:28] Eric Vishria

Go pro. They want to play at a higher level. Is it harder? Yeah, it's harder. Is the competitiontougher? Yeah, the competition's tougher. They move faster. They're tougher. They're bigger.They're stronger. The stakes are bigger. The stage is bigger. The scrutiny is bigger. All of that'strue. It's kind of the same thing with companies. There are a handful, and it really is a handful,three, four, whatever, that can get to this tremendous scale without going public becausethings have gone through their execution and excellence.Lots of free cash flow.They have lots of free cash flow, and they've done really well over a really long time.And I think that's fantastic. Good for them. But in general, for everyone else, get out there.The other thing that I would tell you is there are windows for a particular type of company.The SaaS companies that went public in 2021, a whole boat of them have struggled, and it's beentough in the public markets because their stock's ripped to this multiple compression issue.They were trading at 30 times. They've 4X'd in size, but now they're trading at six times. Itturns out you're under still. That's a tough place to be. I will also tell you that there's 500

转职业。他们想在更高的水平上比赛。更难吗？更难。竞争更激烈吗？更激烈。对手更快、更强悍、更高大、更有力。赌注更大，舞台更大，被审视的程度也更高。这些都是真的。公司也是一样。确实有一小撮公司——真的就是一小撮，三家、四家——能在不上市的情况下做到极大的规模，因为他们在执行和卓越上做到了。（Patrick）大量自由现金流。（Eric）他们有大量自由现金流，而且在很长时间里做得非常好。我觉得这很棒，为他们高兴。但一般来说，对其他所有人：出去（上市）吧。另外我要告诉你的是：对某一类公司来说，是有窗口期的。2021 年上市的那批 SaaS 公司，一大船人都很挣扎，在公开市场很难受，因为股价被这个估值倍数压缩问题打穿了。他们当时按 30 倍交易，现在规模翻了 4 倍，但只按 6 倍交易了。结果你还是往下走的。那是个很难受的位置。我还要告诉你，有 500

[59:32] Eric Vishria

something, probably SaaS companies that are between 100 million and 500 million that are private.What happens? Those employees never got a chance to sell. Those employees don't have annual tenders.Those employees don't have an opportunity to exit. Those investors don't have an opportunity to exit.They're stuck. I don't think they're all going away. And as I said, I don't think they're allgetting vibe coded and everything else. But ultimately, the AI natives with their growth rateshave sucked all the oxygen out of the room and all the interest, and the window was missed.That's tough. What are the biggest debates right now inside of the partnership? I always love cominghere and talking to you guys when there's something interesting going on because you debate. It'shealthy. It can be really fun to watch, and I learn a lot from it. What are those debates today?

多家、大概率是 SaaS 公司，收入在 1 亿到 5 亿美元之间，还是私营的。会发生什么？那些员工从来没有机会卖股票，那些员工没有年度老股转让（tender），那些员工没有退出的机会，那些投资人也没有退出机会。他们被困住了。我不认为它们会全部消失，而且像我说的，我不认为它们会全被 vibe coding 干掉。但最终，AI 原生公司凭着它们的增长率，把房间里的氧气和所有注意力都吸走了，而窗口错过了。这很难受。（Patrick）现在合伙人内部最大的争论是什么？我特别喜欢在有意思的事情发生时来找你们聊，因为你们会辩论，这很健康，看着也很有意思，我能从中学到很多。今天的争论是什么？

[1:00:18] Eric Vishria

There's a ton of debate around in this AI, infra, apps, for foundational models, infra, apps,ecosystem. Where does value accrue, and how does it accrue, and where are the moats,and how do we think about that? But also the business model innovation. I think one of thethings people don't understand about why SaaS did so well versus traditional software was it wasn'tjust that it was a better delivery model and everything else. There was actual business modelinnovation on it. You really did have this subscription element that ended up being fantasticfor both the company and the customers. It was a win-win situation. That same thing actually existsin AI and selling by outcome and that piece of it. But then wrapped up into that debate, discussionis how much value just accrues to the labs. Yes. How much of the value just accrues to the semis?

在 AI 这个「基础模型—基础设施—应用」的生态里有大量争论：价值在哪里沉淀？怎么沉淀？护城河在哪里？我们该怎么看这件事？还有商业模式创新。我觉得关于「SaaS 为什么比传统软件做得好得多」，人们没理解的一点是：不只是交付模式更好等等，而是它确实有商业模式创新。你确实有了订阅这个要素，而它对公司和客户来说都极好，是个双赢局面。同样的事情其实在 AI 里也存在：按结果（outcome）收费这一块。但这场争论里还裹着一个讨论：有多少价值只会沉淀到实验室（模型公司）那里？（Patrick）对。（Eric）有多少价值只会沉淀到半导体那里？

[1:01:08] Eric Vishria

That's a real discussion. What do you think?Like, I am of the view that it all works. It's a very weird thing. Will the CSPs do well?Yes.Yes. Not all of them, but will some of these neoclads do well? Yes. Will the fireworks of theworld do well? Yes. Will NVIDIA do well? Yes. Will these chip startups do well? Some set of them? Yes.Are we going to have edge inference on our phones? Yes. Are we going to have near edge inference on pops?

（Patrick）这是个真实的争论。你怎么看？（Eric）我的观点是：这些都会成。这是件很奇怪的事。云服务商（CSP）会做得好吗？（Patrick）会。（Eric）会，不是全部，但有些新型云厂商会做得好吗？会。Fireworks 这类公司会做得好吗？会。NVIDIA 会做得好吗？会。这些芯片创业公司里会有一部分做得好吗？会。我们会在手机上做端侧推理吗？会。我们会在网络接入点（PoP）上做近端推理吗？

[1:01:38] Eric Vishria

Yes. Are we going to have big models and data centers? Yes. There's so much zero-sum thinking,which is just like, okay, how do we cut up this pie and they're going to eat this much? And like,oh no, no, no, no, no. Anthropic or whomever is going to eat 98% of the value and they're going todo all the drug discovery. And I was like, come on. No, that's not what's going to happen.When I say like, I think everything's going to work and I listed off all these other things,it's really important to understand that doesn't mean that every company that's doing every one ofthose things is going to work. It actually means quite the opposite of that. Most companies in each ofthose areas are not going to work. And it's actually more important than ever to have realdifferentiation to like really take each of these thoughts to their logical extreme and understand,wait a minute, you got to go all the way on these things and really be differentiated on it.Is there anything that you have your eye on, whether it's in the funding market,in the technology world, anything at all that you really are watching carefully?

会。我们会有大模型跑在数据中心里吗？会。现在有太多零和思维了：好，我们怎么切这块饼？他们要吃掉这么多？不不不不不。「Anthropic 或者随便谁会吃掉 98% 的价值，他们会把新药研发全干了。」我心想：拜托，不，不会那样的。当我说「我觉得一切都会成」并且列出这么多东西时，有一点非常重要要理解：这不意味着做这些事情的每一家公司都会成，它其实意味着恰恰相反——这些领域里的大多数公司都不会成。而且现在比以往任何时候都更需要真正的差异化，把每一条思路推到逻辑上的极致，理解到：等一下，你必须在这些事情上走到底，真正做到差异化。（Patrick）有没有什么是你正在特别关注的？不管是融资市场、技术世界，任何你在密切观察的东西？

[1:02:29] Eric Vishria

It's actually funny to me. Some of the people are so, so smart and yet they're in this tech worldwhere they're like reaching these deterministic, almost conclusions of like mass unemployment andall of these different things. I'm going to give you a really concrete example, which I think isjust so good. Take Jeff Hinton in radiology. So I think it was 2016 where he was like, we should stoptraining radiologists. AI is going to do it all better. Jeff Hinton's three orders of magnitudesmarter than I am. Could not have been more wrong, but the actual thing that led him to make thatstatement or that conclusion was a hundred percent correct. If you look at these radiology images,we should be able to train AI to do a better job reading these things than humans. And that'sprobably true. And actually I think the studies and areas have shown that to be true. And we havean investment in a company called New Lantern, which is approaching this. But the big hurdle and thebig thing that it articulated was like, wait a minute, first off, all of the aggregated trainingdata set doesn't exist anywhere. What you see is companies going after like chest CTs or like veryspecific elements, but your typical radiologist looks at a whole variety of things every single day from

有件事我觉得挺好笑的：有些人极其聪明，但他们身处这个科技世界里，会得出一些近乎确定性的结论，比如大规模失业等等。我给你举一个我觉得特别好的具体例子：Geoffrey Hinton 和放射科。我记得是 2016 年，他说我们应该停止培养放射科医生，AI 会把这活儿干得更好。Hinton 比我聪明三个数量级。他错得不能再错了。但导致他做出那个陈述、那个结论的东西，百分之百是正确的：如果你看那些放射影像，我们应该能训练出比人类读得更好的 AI。这大概是真的，而且我觉得相关研究和实践也确实证明了这一点。我们投了一家叫 New Lantern 的公司，就在做这个方向。但最大的障碍、它暴露出来的最大的事情是：等一下，首先，那个聚合起来的训练数据集根本不存在于任何地方。你看到的是一些公司去攻胸部 CT 这样非常具体的细分，但一个典型的放射科医生每天要看各种各样的东西，从

[1:03:41] Eric Vishria

x-rays to CTs to MRIs of all parts of the body and everything else. And so in AI climbing in specificareas like chest CTs is very marginally helpful because it's only doing that one thing, which could beone of 20 things or 40 scans that they read that day. Problem number one, you don't have the data,just like we talked about in robotics and everything else to train the AIAC. Problem number two, the wholehealthcare industry is oriented around reimbursing doctors for making readouts. How is that going towork? And there's liability associated with that. And there's repercussions of getting something wrongor missing something. And there's medical malpractice and everything else. How are we going to avoid that?

X 光到 CT 到全身各部位的 MRI 等等。所以 AI 在胸部 CT 这样的具体领域爬坡，边际帮助其实很小，因为它只干那一件事，而那可能只是他们那天读的 20 种检查、40 个扫描里的一种。这是问题一：你没有数据，就像我们刚聊机器人时说的那样，没法训练那个 AI。问题二是：整个医疗行业是围绕「给医生出具诊断报告付费」组织起来的，这怎么办？而且这里面有法律责任，出错或漏诊是有后果的，有医疗事故责任等等。我们要怎么规避这些？

[1:04:19] Eric Vishria

And how are we going to get around that? Problem number two, real world stickiness. We're going toend up with this application where AI really does help radiologists. It helps radiologists get moreand more higher, higher throughput because the AI can do some parts and the radiologists do some partsand they're checking each other and everything else. And you do kind of weirdly end up in thiscopilot situation for some time. And then you're going to slowly have the AI read more and more of thescans and build up and build up and build up. But the actual duration to get from here to there isgoing to take a long time. In the ensuing time, we need more radiologists, not less because, oh,by the way, everyone's getting more imaging than they used to get because the cost of imaging isgoing down in a Javon's Paradox kind of way. My point on it is you have someone very, very smart whoreally understands the capabilities, really understands what's happening, has the right data,but by not thinking of that data in the real world application comes to the wrong conclusion.And that's how I think of the unemployment thing. I think it's just, it's almost the exact samesetup. Eric, I love talking about markets and companies with you. An absolute blast. Thanks for

我们要怎么绕过去？问题二：真实世界的黏性。最后我们会得到这样一个应用形态：AI 确实在帮助放射科医生，帮他们把吞吐量做得越来越高，因为 AI 能做一部分、医生做一部分，互相检查等等。而你会有点奇怪地在相当长一段时间里停留在这种「副驾驶」（copilot）状态。然后 AI 会慢慢读越来越多的片子，一点点累积上去。但真正从这里走到那里所需要的时间会很长。而在这个过程中，我们需要的放射科医生是更多、不是更少——因为顺带一提，现在每个人做的影像检查都比过去多了，因为影像的成本在下降，有点像杰文斯悖论（Jevons Paradox：某种资源用起来更便宜后，总消耗量反而上升）那样。我想说的是：你有一个非常非常聪明的人，真的理解这些能力、真的理解正在发生什么、手里也有正确的数据，但因为没有把那些数据放到真实世界的应用场景里去想，最后得出了错误的结论。我觉得「失业」这件事也是一样，几乎是完全一样的套路。（Patrick）Eric，跟你聊市场和公司太开心了，绝对是一场享受。谢谢你抽时间。（Eric）谢谢。

[1:05:24] Patrick O'Shaughnessy

the time. Thank you. If you enjoyed this episode, visit Colossus.com. You'll find every episode ofthis podcast complete with hand edited transcripts. You can also subscribe to Colossus, our quarterly print,digital and private audio publication featuring in-depth profiles of the founders, investorsand companies that we admire most. Learn more at Colossus.com slash subscribe.You know how small advantages compound over time. That's true in investing and just as true in how

〔片尾〕如果你喜欢这一期，请访问 Colossus.com。你会在那里找到本播客的每一期，都配有人工编辑的文字稿。你也可以订阅 Colossus——我们的季度印刷版、数字版和私享音频出版物，收录我们最欣赏的创始人、投资人和公司的深度特写。了解更多请访问 Colossus.com/subscribe。〔广告〕你知道微小的优势会随时间复利，这在投资里成立，在你经营公司的方式里同样成立。

[1:06:10] Patrick O'Shaughnessy

you run your company. Your spending system is your capital allocation strategy. Ramp makes it smarterby default, better data, better decisions, better economics over time. See how at ramp.com slashinvest. As your business grows, Vanta scales with you, automating compliance and giving you a singlesource of truth for security and risk. Learn more at vanta.com slash invest. The best AI and softwarecompanies from open AI 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 yourproduct. Ridgeline is redefining asset management technology as a true partner, not just a softwarevendor. They've helped firms 5x in scale, enabling faster growth, smarter operations and a competitiveedge. Visit RidgelineApps.com to see what they can unlock for your firm. Every investment firm is uniqueand generic AI doesn't understand your process. Rogo does. It's an AI platform built specifically forWall Street connected to your data, understanding your process and producing real outputs. Check themout at rogo.ai slash invest.

你的支出体系就是你的资本配置策略。Ramp 让它默认更聪明：更好的数据、更好的决策、随时间更好的经济效益。去 ramp.com/invest 看看效果。〔广告〕随着你的业务成长，Vanta 会跟着你一起扩展，自动化合规，为安全与风险提供唯一的事实来源。了解更多请访问 vanta.com/invest。〔广告〕从 OpenAI 到 Cursor 到 Perplexity，最好的 AI 和软件公司都在用 WorkOS 一夜之间变得「企业就绪」，而不是花几个月。访问 WorkOS.com，跳过那些不光鲜的基础设施活儿，专注在你的产品上。〔广告〕Ridgeline 正在把资产管理技术重新定义为一个真正的伙伴，而不只是一个软件供应商。他们帮助多家机构把规模做大 5 倍，实现更快的增长、更聪明的运营和竞争优势。访问 RidgelineApps.com 看看他们能为你的机构解锁什么。〔广告〕每一家投资机构都是独特的，通用 AI 不理解你的流程，而 Rogo 理解。这是一个专为华尔街打造的 AI 平台，连接你的数据、理解你的流程、产出真正的成果。去 rogo.ai/invest 了解他们。