本集简介
双语字幕
仅展示文本字幕，不包含中文音频；想边听边看，请使用 Bayt 播客 App。
我觉得基本上有两条路值得讨论。
I think there's basically two paths that are meaningful to talk about.
一条是实验室赢下一切，另一条是实验室没能赢下一切。
One of them is that the labs win everything, and then the other one is the labs don't win everything.
而且这两边你都能找到非常有力的论据。
And you can make very strong arguments on either side of that.
如果十年前我给你十亿美元，会发生什么？
What would happen ten years ago if I gave you a billion dollars?
你会怎么做？
What would you do?
雇一大堆人来搞定
Hire a ton of people to get through the
下一个话题。
next topic.
然后你
And you
会搞砸的
would blow up
整个商业情报部门都会乱套，整件事会变得一团糟。
the whole BI, and the whole thing would be like a total mess.
对吧？
Right?
所以现在我们知道该怎么花这笔钱了。
And so now we actually know what to do with that money.
在人类历史、在工程发展的历史上，我们从未有过这样的情况：二十个人——我觉得对——
In the history of humanity, in the history of engineering efforts, we've never been able to have 20 people, I don't think Mhmm.
能够高效地动用二十亿美元。
Being able to productively use $2,000,000,000.
这到底意味着什么呢？
Like, what does that even mean?
要在这么小的团队和这么短的时间内，把这么多钱真正用起来，这怎么可能呢？
Put that much money to work with that small of a team and that small of a timeline.
你可以把钱投到这些东西上，它们往往会转化为使用量。
You can put capital into these things and tends to turn into usage.
比如说你投10美元来做这件事。
Let's say you put in $10 to do this.
我不知道你能否拿回9美元，但在这个行业的历史上，我们从来没能做到投10美元还能拿回任何东西。
I don't know if you get $9 back on the other side of that, but what we've never been able to do in the history of this industry is put in $10 and get anything back.
但现在真的是投入10美元，然后很快就能得到一些回报。
But now it really is $10 in and then some amount out pretty directly.
AI不仅仅在改变公司能构建什么。
AI isn't just changing what companies can build.
它还在改变构建公司本身的经济模式。
It's changing the economics of building companies in the first place.
Martin Casado与Theo Jaffee和Sophia Dew一起参加MTS节目，解释为什么AI创造了一种不寻常的现象：小团队能够高效地投入巨额资本，并将这些资本转化为能力、使用量和增长，速度比以往的技术周期都要快。
Martin Casado joins Theo Jaffee and Sophia Dew on MTS to explain why AI has created something unusual, small teams that can productively put enormous amounts of capital to work and turn that capital into capability, usage, and growth faster than we've seen in previous technology cycles.
他们讨论了前沿实验室最终是否会吞噬AI堆栈的其他部分，开源和应用在其中的位置，以及模型路由如何揭示市场可能的演变。
They debate whether frontier labs ultimately eat the rest of the AI stack, where open-source and applications fit in, and what model routing tells us about how the market could evolve.
马丁还解释了为什么他更关注新技术栈中的战略控制点，而不是今天的利润和护城河，他称之为自己职业生涯中见过的最大财富释放。
Martin also explains why he's less interested in today's margins and moats than identifying the strategic control points of a new technology stack in what he calls the biggest wealth unlock he's seen in his career.
大家好，欢迎收听MTS。
Hello, everyone, and welcome to MTS.
今天，我们邀请到了马丁·卡萨多，他是安德森·霍洛维茨的普通合伙人，负责公司的基础设施业务。
Today, we are joined by Martin Casado, who's a general partner at Andreessen Horowitz and leads the firm's infrastructure practice.
上周，SpaceX完成了对Cursor的600亿美元收购，Stripe也同意收购OpenRouter，这两家都是a16z投资的重要公司。
In the last week, SpaceX closed its 60,000,000,000 acquisition of Cursor, and Stripe agreed to acquire open router, two major a16z backed companies.
所以今天，我们来聊聊这些交易告诉我们，在人工智能时代，价值到底在哪里积累。
So today, we'll talk about what those deals tell us about where value is actually accruing in the age of AI.
马丁，欢迎来到MTS。
Martin, welcome to MTS.
非常高兴能来这里。
Super happy to be here.
谢谢你们的邀请。
Thanks for having me.
我们非常高兴你能来。
We're super excited to have you.
是的。
Yeah.
这真是不可思议的一周。
It's been what an incredible week.
我知道。
I know.
这是有史以来
Of one of the weeks of
最精彩的几周之一。
all time.
那真是相当不错。
That's been pretty well.
嗯，是的，这是十年磨一剑，但一切都在几天内发生了。
Well, yeah, ten years in the making, but it all happened in a couple of days.
对。
Yeah.
对。
Yeah.
所以我想我们可以先从一点背景故事开始。
So I guess we we could start with a little backstory.
你拿了计算机科学的硕士和博士学位，然后在30岁左右创立了Nasira？
You got a master's and PhD in computer science and founded Nasira when you were 30ish?
30岁左右。
30ish.
对。
Yeah.
30岁左右。
30ish.
对。
Yeah.
对。
Yeah.
谁还在乎年龄呢？
Who's counting?
对。
Yeah.
三十岁左右。
The 30ish.
对。
Yeah.
对。
Yeah.
但现在，很多顶尖创始人，包括Cursor的创始人，大概二十岁就辍学了。
But these days, a lot of top founders, including the founders of Cursor, will drop out of college at, like, 20 Yeah.
为了创业。
In order to start companies.
所以，是的。
So Yeah.
对。
Yeah.
你觉得哪条路更好？
Which which path do you think is better?
你知道吗，听我说。
You know, listen.
我其实在本科和研究生之间休学了一段时间，然后去国家实验室工作了。
I I actually took time off between undergrad and grad school, and I I I went and worked for a national lab.
我确实觉得，实际经验很重要，这样就不完全是学术性的。
I actually think it's probably a good idea to get on the ground experience so that it's not entirely academic.
不过我要说的一点是，很多AI相关的工作都是很深入的研究。
I mean, the one caveat I will say is a lot of the AI stuff is pretty deep research.
嗯。
Mhmm.
我确实觉得，一方面，现在对高等教育有很多反对的声音，不读大学好像挺时髦的。
I do think, like, you know, in in one hand, you have this whole kind of outcry against higher education, and it's kind of very in vogue not to do it.
但另一方面，它又从来没像现在这么重要过。
On the other hand, it's kinda never been more relevant too.
所以我觉得我们处在一个有点矛盾的时代，这让我觉得你应该完全按照自己的感觉来。
And so I think we're a bit of a schizophrenic period, which suggests to me you should do exactly what you're feeling is right for you.
我觉得两条路都完全合理。
I think both paths are totally legitimate.
对。
Yeah.
我是说，现在有一系列新的实验室，一方面有那种刚起步的初创公司，联合创始人只有22岁。
I mean, there's a there's, like, a spectrum of neo labs where on the one hand, you have, like, flapping airplanes where the cofounders are, like, 22 years old.
对。
Yeah.
另一方面，有杰夫·迪恩和奥里奥尔·维尼奥尔斯的新实验室，我一时想不起名字了。
And then on the other hand, you have Jeff Dean and Oriole Vignoles' new lab, which I'm I'm forgetting the name.
我一时想不起来了。
It's slipping my mind.
但显然，这些人都是非常资深的，有几十年的经验。
But, obviously, these are, like, incredibly senior people with, like, decades of experience.
所以，几十年的经验到底有多重要呢？
So, like, how how much does decades of experience matter in
在我的生活里？
my life?
好的。
Okay.
我们常常忘记一件事，那就是以前创始人大多是博士。
So one thing that we forget is, like, the PhD used to be the rarity for founders.
对吧？
Right?
以前的情况是，你如果从博士或研究生项目退学，就会去创业，比如著名的谢尔盖和拉里。
Like, it used to be like you would start a company if you dropped out of PhD or you dropped out of grad school, like very famously Sergei and Larry.
对。
Mhmm.
这种情况就是这样的。
This is the case for.
我们以前总说，如果你拿到了博士学位，那你其实算是失败的，因为你显然没有足够好的想法去创业。
And we'd always say, like, you're kind of failing if you got the PhD because clearly you didn't have a good enough idea to do it.
但现在我们确实看到很多非常优秀的博士创业者，对吧？
We're actually seeing a lot of really good PhD founders now, right?
比如，Olegodsey的创始人就有博士学位。
Like, Olegodsey has a PhD.
我们做过研究，当我在读研究生的时候，乔治·弗雷泽，Fivetran的CEO，也有博士学位。
We did research when I was, you know, when I was in I was in grad school, George Fraser, CEO of Fivetran also has a PhD.
所以我觉得，我们现在看到的博士创业者会比历史上任何时候都多。
So I would say we're gonna see more PhD founders now than ever before in the history of the industry.
我们一直都有年轻的科技创业者，对吧，比如马克·扎克伯格。
We've always had the young tech founder, right, Mark Zuckerberg.
所以我觉得，就连在X上的观感，也感觉是偏向更年轻、经验更少的创始人。
And so I I would actually think that even the optics on X feels like it's going towards, like, the younger, less experienced founder.
我觉得加权来看，趋势可能确实是往那个方向走的。
I would say weighted, it's actually probably the trend is more in that direction.
现在似乎最重要的是你能融到多少钱，因为不再受限于谁能开发出最好的软件了。
It seems like the main thing right now that matters is how much you can raise because you're not bottlenecked anymore on who could build the best software.
现在最大的竞争是谁能融到最多的钱，通常也是最能竞争的公司类型。
Right now, the big competition is who can raise the most money are typically the types of companies who can compete the most.
那你是不是看到，那些能融到最多钱的公司，就是最有竞争力的？
So are you seeing that for the companies who are able to raise the most, are the most competitive?
我的意思是，当一切尘埃落定，我们回头看这一波浪潮的时候。
I mean, I think that when it's all said and done and we look back on this, you know, this wave.
我觉得最大的变化是，你确实可以把很多钱用在刀刃上。
I think the biggest change is the fact that you can actually put a lot of money to good use.
比如，十年前如果给你十亿美元，你根本没法用。
Like, you'd never be able to like, so what would happen ten years ago if I gave you a billion dollars?
你会怎么做？
What would you do?
雇一大堆人。
Hire a ton of people.
然后你就完蛋了。
And you would blow up.
整个事情会变得一团糟。
The whole thing would be like a total mess.
对吧？
Right?
所以现在我们确实知道该怎么花这笔钱了。
And so now we actually know what to do with that money.
所以我觉得现在这已经变成了一场拼资金规模的游戏。
So I I do think that now it's become a scale up capital game.
我觉得那些融了很多钱的公司，不一定就会搞砸。
I think the the companies that do raise a lot of money, they don't necessarily blow up.
他们知道怎么用这笔钱。
They know how to use it.
而且我觉得这确实在改变我们实际运作中的很多情况。
And I think that that actually is changing a lot of what's happening on the ground for us.
所以以前常见的路径是，如果太早拿到太多钱，可能会出问题，但现在你不再看到这种情况了，是吗？
So the typical path before is if you get too much money too early, it could be an issue, but you're not seeing that anymore being the case?
嗯，是的。
Well, yeah.
所以这有点像工程物理学里的定律，对吧，你筹了一大笔钱，然后试着招一堆人来开发产品。
So there's this kind of a law of like engineering physics, right, where you, you know, you raise a bunch of money and then you try and hire a bunch of people to build a product.
这就是“神话月”这个说法的由来。
And this is where the mythical month came from.
而且你真的没法用这种方式加快时间表。
And you just can't really speed up timelines that way.
但最终你会增加烧钱速度，增加很多复杂性，而且路线图也会变得很分散。
But then you end up, like, increasing burn and adding a lot of complexity and, like, the roadmap gets very diluted.
然后因此就会产生很多组织上的复杂性。
And then you have all, like, the the organizational, you know, complexity that happens as a result.
所以，听我说，一些大型模型项目其实是由非常非常小的团队完成的。
And so, I mean, listen, some of these major model efforts were done with very, very small teams.
比如说，我不说具体是哪个模型，但有一个非常著名的多模态模型，很多人都在用，可能是最流行的模型之一，就是由大约20人的团队打造的。
Like, you know, I'm not gonna say the exact model, but one of the very famous models, you know, multimodal model that many people use is probably one of the most popular models was built with a team of about 20 people.
而且我估计成本大概超过20亿美元。
And I would say the cost of that was probably $2,000,000,000 plus.
对吧？
Right?
这在人类历史上，在工程项目的历史上都是闻所未闻的。我觉得以前从来没有过20个人能高效地花掉20亿美元的情况。
And so this is unheard in the history of humanity, in the history of engineering efforts, we've never been able to have 20 people, I don't think, being able to productively use $2,000,000,000.
这到底意味着什么？
Like, what does that even mean?
用这么小的团队，在这么短的时间内，投入这么多钱。
Put that much money to work with that small of a team and that small of a timeline.
所以我知道我们喜欢从技术复杂性和新能力的角度来看待这波浪潮。
And so I know we like to look at this wave in the context of like technical sophistication, you know, new capabilities.
但我实际上认为，其中一个主要的故事是，我们能够在短时间内将大量资金有效地投入到我们试图解决的任何问题上。
But I actually think like one of the major stories is the fact that we're able to apply large amounts of money productively in short amount of times to whatever problem that we're trying to solve.
你会说这个时代和风险投资与之前的任何时代都完全不同吗？
Would would you say that this age and venture is just completely different from any age before it?
完全同意。
Totally.
完全同意。
Totally.
我的意思是，你知道，风险投资界一直有一种奇怪的说法，就是它其实算不上一个真正的资产类别。
I mean, I mean, it's just you know, again, there's always been this weird meme in venture, which is it's kind of not much of an asset class.
只有少数几家公司，你无法在其中投入大量资金，这对我来说一直很奇怪，因为那些人的整个工作本不应该是一种零和思维。
There's only a few companies you can't deploy a lot of money in it, which has often been very weird to me to have, like, this zero sum thinking from people whose entire job shouldn't be zero sum thinking.
对吧？
Right?
我们正在学到的一件事是，你确实可以在私募市场吸收大量资金，投入进去并获得回报。
And the one thing that we're learning is you actually can take large amounts of capital in private markets and deploy it and get the returns.
而且，你知道，我可以举两个例子。
And, you know, you can just give two examples.
其中一个就是，公司上市的时间更晚了，这样积累了很多价值。
One of them is like companies are staying private longer and accruing a lot of value that way.
这是第一个。
So that's one.
第二个就是，这些模型消耗大量资金，然后把资金转化为增长。
And then the second one is like, you know, these models consume a lot of money and they turn that money into growth.
所以那种认为风险投资必须限制资金规模、我们投得太多了的过时观点，现在完全站不住脚了。
And so this kind of almost archaic view that venture has to be limited in capital, we're overfunding it, absolutely is not the case now.
听我说，我觉得你可以从中得出两个结论。
Listen, I think there's two conclusions you can kind of draw from that.
一个是，出现了新的技术浪潮，可以吸收更多资本。
One of that is, well, there's a new technology wave that can consume more capital.
比如，AI能消耗更多资本，所以这是技术浪潮的结果。
Like, AI can consume more capital, and so it's a result of technology wave.
我碰巧认为因果关系是反过来的，也就是说，如果你把更多资金投入私人市场，它实际上会扩大总可寻址市场，公司就不需要那么快上市，也能做更多私人的事情。
I I happen to think causality goes the other way, which is if you put more money privately into the private markets, it actually grows the TAM and companies don't need to go public as quickly and companies can do more privately.
所以我真的认为，因果关系是，流入私人市场的钱越多，市场就会增长得越大，然后更多的回报就会进入私人市场。
So I actually think that the causality is the more money goes to private markets, the larger the markets are going to grow, and then the more returns go in the private markets.
但有趣的是，我们看到这两种情况同时发生了。
But but just interestingly, we're seeing both things happen happen at once.
那么，a16z第一次意识到，哇，AI这件事会变得特别重要，是什么时候？
So when did a16z realize for the first time, like, wow, this AI thing is gonna be a really big deal.
我们得投入巨额资金在基础设施和其他方面。
We're gonna need to spend, like, enormous amounts of money investing in infrastructure and in other things.
我们第一次是什么时候……
When did we first I
嗯，听我说。
mean, listen.
我觉得我们算是技术最前沿的风险投资基金之一，我们基本上就是假设长期来看一切都会成功。
I think that we're the most kind of one of the most tech forward venture funds, and we just kind of assume everything is gonna work in the long run.
所以我还记得，你知道，当OpenAI还是个实验项目的时候，我们有过非常认真的讨论，问自己是不是应该那时候就全力押注AI。
And so I remember when, you know, when OpenAI was still an experiment and we had very legitimate conversations about should we be going all in on AI then?
那大概是GPT真正变得重要之前的七年左右。
And this was like seven years before, like, say, GPT, like, actually was pretty significant.
所以，你知道，我们一直在投资AI。
And so, you know, we've always been investing in AI.
我们投资了AI的多个浪潮。
We've invested in multiple waves of AI.
比如我在2016年加入公司的时候，那会儿是AI的无人机和自动驾驶浪潮。
Like when I joined the firm in, you know, say 2016, that was kind of like the drone AV wave of AI.
对于这波生成式AI浪潮，我觉得我们是最早积极参与的基金之一。
For this generative wave, I think we're one of the earliest that we're very active in it.
比如，我们投了很多早期的基座模型公司。
Like, we did a lot of the early foundation model companies.
对吧？
Right?
我们投了11 labs。
We're in 11 labs.
我们投了OpenAI。
We're in OpenAI.
当然，我们也投了Cursor。
Of course, we're in cursor.
我们投了Ideogram。
We're in ideogram.
我们投了BFL。
We're in BFL.
我们投了Mistral。
We're in Mistral.
所以我们确实做了很多非常积极的投资部署。
So we just did a lot of very, very active deployment.
但我要说的是，部署的意图其实在那之前就有了。
But I will say, like, like, the intent to deploy was even before that.
再说一次，我觉得我们对技术浪潮的判断往往比实际进展要超前不少，这有时候管用，有时候不管用。
Now, again, I would say that, like, our conviction on technology waves tends to actually proceed quite a bit, the the, like, on the ground gains, which can, like, sometimes work and sometimes not work.
但在这种情况下，这个判断确实挺有先见之明的。
But in this in in in this case, it was actually quite prescient.
对。
Yeah.
有意思的是，以前的观点是非常零和的。
It's interesting because the old view was very zero sum.
就是，既然OpenAI已经那么好了，为什么还要投资其他模型呢？
It's what's the point of investing in another model when OpenAI is pretty good.
所以，这些都不具备防御性。
So None of it's defensible.
完全正确。
Exactly.
收入方面。
Of revenues.
顺便说一句，这就是为什么我知道基于财务的投资者和早期投资者之间的区别，前者应该关注增长市场。
By the way, every single way so this is why I know the difference between, like, like, a finance kind of, you know, based investor that should be in a growth market and an early stage investor.
所以如果你主要从资产负债表和财务角度考虑，你会担心利润率、客户流失率、收入质量这些问题，如果你认为这家公司没有战略价值，这些担忧都是合理的。
So if you think primarily in the balance sheet and in finance, you worry about things like margins, churn, revenue quality, which are all very legitimate things to worry about if you see no strategic value in the business.
所以如果你认为整个业务，忽略技术、战略价值，忽略别人能从中获利的能力，我只关注资产负债表，很多人都会犯这个错误，那么当然，如果你只看这个，你会有不同的意见，但这完全脱离了公司的战略价值。
So if you think like the entire business, forget the tech, the strategic value, forget the ability for somebody else to monetize it, I'm just only going to look at the balance sheet, which many, many people do and make that mistake, then sure, If you look at that, you'll have like difference of opinions, but that's entirely divorced to the strategic value of a company.
有些公司可能亏损，但它们会达到一个能够实现巨大价值转移的位置。
Like some companies may lose money, but they'll get to a position where they can command a tremendous transfer of value.
如果你不这样看，我觉得你就不可能成为早期投资者。
And if you don't look at it that way, I just don't think you can be an early stage investor.
你必须从这个角度来看：这是新架构中非常重要的一部分。
You have to look under the lens of this is a very important part of the new stack.
这是一个非常重要的控制点。
It's an very important control point.
而且要么有人会因为它的增值潜力而想拥有它，要么它自己就能增值，即使现在看它的财务状况，不像那种在公开市场上能高价交易的东西。
And either somebody will want to own that because of accrue value or it itself will be able to accrue value even if right now when you look at it, the financials don't look like whatever, something that would trade highly on a public market.
对。
Yeah.
所以与此相关，有人提出一个观点，说这些实验室会吞掉一切。
So relatedly, there is an argument that some people make where it's like, the labs are just gonna eat everything.
他们会有最聪明的模型。
They're gonna have the smartest models.
他们会占领每一个垂直领域。
They're gonna take over every vertical.
OpenAI和Anthropic现在都在做生物科技了。
OpenAI and Anthropic are doing biotech now.
他们会把礼来吃掉。
They're gonna eat Eli Lilly.
他们会把所有制药公司都吃掉。
They're gonna eat all the pharma companies.
显然，16z并不相信这一点。
Obviously, a 16 z doesn't believe in this.
嗯，我不确定我们是否不相信那个。
Well, I don't know if we don't believe in that.
我的意思是，我不太听那些。
I mean, I don't listen.
我们最近确实有个外出会议，其中一个重要的讨论点就是模型的未来会怎样。
We we actually had an off-site recently, and one of the big discussions points in, like, what is the future of models?
对。
Mhmm.
我觉得基本上有两条值得讨论的路径。
And I think there's basically two paths that are meaningful to talk about.
一条是实验室赢下所有，另一条是实验室不会赢下所有。
One of them is that the labs win everything, and then the other one is the labs don't win everything.
而且两边都能提出非常有力的论点。
And you can make very strong arguments on either side of that.
如果你愿意的话，我很乐意详细说说两边的论点。
If you want, I'm very happy to detail the arguments on either side.
让我先给你讲讲支持的那一方。
Let let me give you the argument for.
那么，为什么模型会赢下一切呢？
So so why will the models win everything?
首先，我们有三年多的数据，他们占了95%的市场份额。
Well, the first one is we've got three years of of data, and they own 95% of the market.
所以你看数据，就会觉得，嗯，好吧。
So you just look at the data, you're like, well, okay.
他们做得确实非常好。
So they're doing, you know, tremendously well.
他们占据了大部分收入，至少按美元加权来算是这样。
They have the the majority is certainly dollar weighted of revenue.
第二点是，他们筹集资金的能力简直令人难以置信。
The second one is that their ability to raise capital is just unbelievable.
然后他们能把这些资金直接转化为增长，这种情况我们很少见到。
And then they can turn that capital pretty directly into growth, which we've very rarely seen.
比如说，OpenAI 和 Anthropic 已经筹集了 2000 亿，或者 2200 亿美元吧。
So let's say open- and anthrapbook have raised 200,000,000,000 I don't know, $220,000,000,000.
这比整个下游生态系统的总和还要多，真是难以置信。
That's more than the entire downstream ecosystem combined, which is really unbelievable.
还有一个论点，就是说，你看，他们仅仅因为稍微领先一点，就能保持定价权。
There's another argument, that argument is like, listen, they've been able to maintain pricing power just by being an Epsilon higher.
所以处于前沿位置成本很高，但同时也能够带来巨大的定价权。
So being on the frontier is expensive, but also being on the frontier can command a tremendous amount of pricing power.
所以并不是说你必须比别人好很多。
And so it's not like you have to be way better.
你只要稍微好一点就行，因为这些模型在竞争均衡中会被使用。
You can just be a little bit better because these get used in competitive equilibria.
我们开始看到自我催化效应了。
We're starting to see auto catalytic effects.
我说的自动催化效应是什么意思呢？
What do I mean by auto catalytic effects?
就是说，如果你有足够先进的AI，你可以用那个AI来创造更多的AI，这也会提升你自己的能力。而且，如果你跟那些在实验室里深入研究这些的人聊聊，他们会觉得实际上他们离开源越来越远了，而不是越来越近。
Like if you have sophisticated enough AI, you can use that AI to create more AI and that improves I your unit also think if you talk to a lot of the people deep in these labs that are working on these, they think they're actually getting further away from open source than not.
但基准测试并不能反映这一点，因为AI基准测试本身就是一个AI完备性问题。
And the benchmarks don't reflect that because benchmarking in AI is kind of an AI complete problem.
就是说，你得实际用它去做事，才能知道它到底有多好。
Like, you just kind of have to use it to do the thing to know how good it is.
所以有很多数据都表明，听着，
So there's a lot of data that would suggest, listen.
他们有资金，他们有资本。
They have the cap they have the capital.
他们有，你知道的，团队和人才。
They have the, you know, the teams and the talent.
他们拥有最好的模型。
They have the best models.
明白吗？
Know?
他们会一骑绝尘的。
They're gonna run away with it.
对。
Yeah.
你说的那种自动催化效应是指什么？
By what kind analytic effects?
你是说递归式的自我改进吗？
Do you mean, like, recursive self improvement?
嗯，是的。
Well, I yeah.
这里我就有点较真了。
This is where I'm just like an annoying pedant.
对我来说，递归是有特定含义的。
So to me, recursive means something.
比如说，你知道，我的博士是计算机科学。
Like, I'm a you know, this is my PhD's in computer science.
我做过编程语言。
I've done programming language.
递归是指你拿一个东西，然后完整地复制一份。
Recursive is when you take something and you make another copy of that thing wholesale.
比如，如果我有一个编译器，能编译编译器，那就是递归。
Like, if I have a compiler, could a compiler, that would be recursive.
而自催化是指用这个东西来帮助你更快地制造它，作为一个工具。
Where auto catalytic is is use the thing to, like, you know, help you make that thing faster as a tool.
比如，我用电脑设计芯片，因为我用的是某种芯片设计软件，你会把这叫做递归自我改进吗？
Like, if I use a computer to design a chip because I'm using, like, whatever, like, some chip design software, would you call that recursive self improvement?
我会说不是。
I would say no.
我的意思是，以前没人用过这个词，
I mean, nobody's used that word before,
而且我们一直用的是字面意思。
and we've always the literal sense.
不是。
No.
我的意思是，其实任何意义上都不是，因为我们从来没用过那个词。
I mean, well, not in any sense because we've never used that word.
比如说，我们都用软件来开发软件，但我们从来不叫它RSI。
Like, let's say, we we we all use software to build software, we never we don't call it RSI.
对。
Mhmm.
而且从来都没叫过。
And we never have.
所以我觉得我们今天看到的更多是自动催化效应，这种效应我们一直都有。
And so I feel like what we're seeing today is much more auto catalytic effects, which we've had forever.
但不管怎么说，我觉得这其实是博斯特罗姆留下的影响。
But for whatever reason, I think this is a legacy of Bostrom.
我们叫它RSI，也就是递归自我改进。
We call it RSI, recursive self improvement.
话虽如此，确实有对递归自我改进的关注，这很好，它是更广泛现象的一个子集，对我们所有人来说理解这个现象非常重要，那就是自催化效应。
That said, there's definitely focus on recursive self improvement, and that's great, it's a subset of the broader phenomenon, which is very important for all of us to understand, which is auto catalytic effects.
那就给我举个例子吧。
So just give me an example.
我用AI来创建一个非常好的GPU内核，比人类程序员做得更快，这提升了我运行或构建AI的能力，我会说这就是自催化。
I use AI to create a really good GPU kernel that's faster than a human beacon do, and that improves my ability to, like, run or build AI, I would say that's auto catalytic.
对。
Mhmm.
我认为这对我们来说在经济和行业收敛性方面非常重要，但很难说清楚。
I would think that's very important for us on the economics, the convergence properties of the industry, but it's hard for our say.
那么，反对的观点怎么说呢，就是实验室不会拥有一切？
So what about the arguments against then that the labs will not own everything?
哦，顺便说一句，另一个支持的观点是，他们拥有所有供应。是的。
Oh, another one for it, by way, is they they own all the supply Yeah.
在供应有限的情况下。
In a supply limit.
好的。
Okay.
那么反对的理由呢。
So the arguments against.
反对的理由是，应用领域正在迅速扩大。
The arguments against is the surface area is really expanding.
我的意思是，真正有效的其实主要是代码和语言推理。
I mean, really what's working are primarily code and language reasoning.
所以还有很多其他领域，实际上可能需要服务部门，可能需要更多与客户的联系，这对单一组织来说很难做到。
And so there's a bunch more domains where actually probably you need a services arm, probably you need more connection with the customers, and this would be very hard for a single organization to do.
第二点是，有很多开源模型表现很好，我们最近也看到了一些例子。
Second one is there is a wealth of open source models that are doing very well, and we've seen some recent examples of that.
而且围绕这些模型，服务它们的生态系统也在不断成熟。
And there's a maturing ecosystem around them to serve those.
我觉得这些实验室能领先这么多，很大程度上是因为他们能轻松获得便宜的资金，这几乎肯定会趋于合理化。
I think a lot of the reason the labs are so far ahead is they have very cheap access to capital, which will almost certainly rationalize.
你想，如果我给你免费的钱，说你就值个几万亿美元，那你当然可以用这些钱去发展。
Like, listen, if I give you free money and I say you're worth like whatever trillions of dollars, like, of course, you can use that to grow.
但到了某个时候，我可能会觉得，你知道，你大概值，嗯，不一样的价值了，因为我们已经把公司合理化了，然后你再想发展就难了。
But at some point in time, I may be like, you know, like, you're probably worth, you know, you know, something different because we've rationalized, like, the company, and then it'll be harder for you to grow.
我觉得现在他们是在利用供应紧张的优势，因为他们能大批量买到GPU。
I think that right now they're taking advantage of the supply constraints cause they can actually buy GPUs in bulk.
等供应多了，他们就没这个优势了。
When you have more supply, they won't be able to do that.
所以我觉得现在的形势对大型实验室非常有利，主要是资金和供应方面的优势。
So I think actually the landscape right now is very much in favor of the large labs primarily from funding and supply.
一旦这些优势合理化，不断扩大的应用领域很可能会分散开来，这在行业历史上几乎总是会发生，到时候就会对其他人有利。
And once those rationalize, the growing surface area is likely to fragment, which almost always happens in the history of this industry for sure, then I think that will play in favor of everybody else.
如果让我猜的话，这完全是马丁在瞎猜，我一点都不知道。
If I were to guess, this is Martin totally guessing, no idea.
如果让我猜的话，我觉得供应紧张大概会在2028年左右缓解。
If I were to guess, I'd say supply constraints will ease in twenty twenty eight ish.
我觉得大实验室按金额算的话，未来可能会占据80%的市场份额，因为历史上大公司通常都是这样。
I think that the the big labs will probably dollar weighted, get 80% of the market going forward because that's historically what we've seen for large incumbents.
但按token数量算的话，60%会来自长尾和开源。
But I think token weighted 60% will be long tail and open source.
还有一点关于这些的，抱歉我有点说多了。
And one more thing about all of this, I'm sorry to kind of rant on it.
我就是觉得这一点特别重要。
I just think it's such an important point.
我们看到越来越多的价值流向了实际的应用。
We're seeing increasing value go to the actual apps.
这些应用做得非常好。
The apps are doing incredibly well.
所以我觉得它们会开始蚕食一部分利润份额。
And so I think they will start to actually erode on some of that margin share.
你不觉得它们跟那些大实验室是竞争关系，但我确实认为未来我们会获取越来越多的价值。
And you don't view them as like necessarily competitive with the labs, But I do think that we're gonna capture more and more of the value going forward.
但这还得等上几年。
But it'll be a couple years.
我觉得OpenRouter就是一个很好的例子，能预示未来我们长期会怎么使用这些模型。
I mean, open router is a good example of what the future might actually predict on how we'll be using these models long term.
对。
Yeah.
所以，听我说。
So listen.
OpenRouter其实有很多种角色。
So open router is open router is many thing.
它本质上是一个API路由器。
I mean, it's an API router.
你可以这么理解：你可能有Anthropic的密钥，有OpenAI的密钥，还有Grok的密钥，但其他所有模型，你都通过OpenRouter来用。
But one thing you could think of it as is is, you know, maybe you have your Anthropic key, maybe you have your OpenAI key, maybe you have your Grock key, but for all the other models, you use OpenRotter.
对。
Yeah.
对吧？
Right?
基本上它就是一个双边市场，让你能访问各种模型。
And it's like, it's it's basically a two sided marketplace to give you access to a bunch of models.
它是一个统一的可见性平台。
It's a single place for visibility.
它是一个统一的分析平台。
It's a single place for analytics.
它是一个统一的模型访问入口。
It's a single place for access to these models.
它就在token流通的路径上。
It's on the token path.
在这个领域里，它算是品牌垄断的领导者。
It is the leader in kind of a brand monopoly in the space.
从品牌角度来看，它在行业里是非常知名的品牌。
From a brand standpoint, it's very well known brand in the space.
而且，你知道，我觉得在这些现象中，当有几个主导者时，总得有人去整合那些长尾市场，我觉得这肯定会改变未来模型市场的格局。
And, you know, I think with all of these phenomenons where you have a couple of incumbents, somebody has to aggregate kind of the longer tail marketplace, which I think will impact the dynamics of the model market going forward for sure.
但是你觉得智能路由会越来越受欢迎吗？
But are you seeing smart routing just growing in popularity?
所以也许未来的应用，你根本不知道实际用的是哪个模型。
And so maybe the future of applications is you don't even know what model is actually being used.
这个问题问得好。
You know, it's a good question.
我对此也反复思考过。
I go back and forth on this.
实际上，我们发现这些模型的用户粘性比人们想象的要强得多。
Like, we've actually learned that these models are a lot stickier than people assumed.
大家都说换就换，但实际上并不经常发生。
Like, everybody talks about just swapping them out, but it actually doesn't happen very often.
我觉得这很大程度上其实跟采购的机制有关。
And I think this a lot of this could be actually like procurement dynamics.
比如，我已经从OpenAI买了一大堆积分。
Like, bought a bunch of credits from OpenAI.
那我为什么要换掉它们呢？
Like, why would I swap them out?
所以你可以从两个角度来看智能路由。
So you can think about, like, smart routing in in one of two contexts.
一个是从质量的角度来看。
One of them, you can be like, you know, like, almost from, like, a quality standpoint.
比如，选对路由的话，回答问题的效果会更好。
Like, you'll answer the question better on the right router.
我觉得这是一个非常难的问题，甚至可能是个AI完全问题。
I think it's a very, very hard problem and maybe an AI complete problem.
而且目前来看，这种情况其实不常发生。
And right now, I think, you know, that that doesn't happen very often.
你可以从另一个角度来看，那就是成本效益，也就是说，我要把成本降到最低。
You can think about it in a second way, which is basically cost performance, which is like, I will minimize the cost.
而且因为token太贵了，这对很多人来说其实是优先考虑的事，而且我们看到了很好的效果。
And because tokens are so expensive, that's actually kind of a high priority for many, and that we're seeing great results.
对吧？
Right?
如果你看过，比如Cursor Auto和Cursor Router做得很好，OpenRouter也做得很好。
If you've seen, like, example, cursor auto and cursor router does a great job, open router does a great job.
所以我建议，如果你考虑智能路由，我觉得应该从这样的角度去想：对于给定的任务，如何在帕累托前沿上根据成本选择正确的模型，而不是说我要选一个能正确回答问题的模型。
And so I I would encourage, you know, if you think about smart routing, I think in the context of, you know, how do I, for the given task, choose the right model on the Pareto frontier relative to cost as opposed to, like, I'm somehow gonna choose the right model that's gonna answer the question in the right way.
对。
Yeah.
模型路由确实是个非常难的技术问题。
It does seem like model routing is, like, a very difficult technical problem.
我觉得这是AI完全问题。
Think it's AI complete.
对。
Yeah.
说它是AI完备问题，好吧。
By by AI complete okay.
那我们想象一下，你在试着回答一个问题：宇宙中最聪明的东西需要回答什么问题？
So let's imagine you're answering the trying to answer the question, what question does the smartest thing in the universe need to answer?
就是说，你需要宇宙中最聪明的东西来回答那个问题。
Like, think you need the smartest thing in the universe to answer that question.
你明白我的意思吗？
You see what I'm saying?
就是说，能回答那个问题的，其实只有最小的，抱歉，是最聪明的模型，那这种情况下，就直接交给最聪明的模型好了。
Like, only the only thing that can answer that question is actually like the smallest, you know sorry, the the smartest model, and then in which case, just give it to the smartest.
所以这就变成了一个成本优化的问题？
So it just becomes a cost optimization type of question?
问题？
Question?
展开剩余字幕（还有 412 条）
听我说。
Well, listen.
好吧。
So okay.
其实有很多不同的看法。
So there there's there's many opinions.
我觉得我们还没有答案。
I don't think we know the answer to this.
而且你经常听到这种说法。
Like, and you hear this often.
很明显，前沿模型变得越来越不均衡，这不是核心的技术能力问题。
Clearly, the frontier models are getting very jagged, meaning and this isn't this isn't like a core technical capability.
而是创造它们的人决定让它们在某个方面特别强，比如前端、三维或者语言。
It's like the decisions of the people that that that created it decided to make it very good and say front end or very good in three d or very good in language.
所以并不是说这个模型比那个更聪明。
And so what about it's not like this one's smarter than this one.
就像这个模型特别擅长这类任务，而那个模型特别擅长那类任务。
It's like this one's very good at this type of thing versus, you know, very good at this type of thing.
那当然。
That for sure.
我觉得我们有很多专业模型，很多，你懂的。
I think we have a lot of specialist models, a lot of you know?
然后一个模型路由器会决定，这个在代码方面更好，这个在语言方面更好，这个在前端方面更好，这个，你懂的，随便什么。
And then a model router will just decide this is better in code, this is better in language, this is better in front end, this is, you know, whatever.
所以，这种情况会发生。
So, like, that will happen.
但我觉得现在，如果你看看模型路由和那些应用公司部署带来的收益，我提到过Cursor，但我们还有很多应用公司，我觉得主要的收益其实往往是保持高质量的同时降低成本。
But I think today, if you look at, like, model routing and the gains, as deployed by the application companies, I mentioned Cursor, but we've got a huge portfolio of application companies, I think the primary gains actually tend to be keeping the quality bar high while decreasing the cost.
那么，OpenRouter的价值在多大程度上在于它最终能够实现这种AI完整的模型路由呢？
So what like, to what extent is, like, the value of OpenRouter that it will eventually be able to do this kind of, like, AI complete model routing?
我听着呢。
I listen.
我觉得你需要一个双边市场。
I I think that you need a a two sided marketplace.
在很多方面，你可以把OpenRouter的一个产品看作需求方。
And in many ways you can view one of the products of open routers, the demand.
所以我可以去找下一个模型提供商说，嘿，听着，我们一直有数百万用户，我们可以把这些用户带给你。
So I can go to the next model provider and say, hey, listen, we've got millions of users all the time and we can bring that to you.
这是很常见的事情。
And this is a very common thing.
然后你可以对所有开发者说，嘿，听着，下一个模型发布的时候，它会出现在这里，你可以在你现有的系统里访问它，等等。
And then you can go to all the developers say, Hey, listen, the next time a model comes out, it'll be on here and you'll have access to it in your existing system and so forth.
所以我认为，双边市场本身就有巨大的价值，跟这些都没关系。
And so I think there's a tremendous amount of value in the two sided marketplace independent of all of that.
话虽如此，显然随着时间的推移，做这种套利的能力会越来越强。
That said, clearly over time, there's gonna be more and more ability to do this sort of arbitrage.
我觉得不管出于什么原因，我们现在有点过于关注这个了。
I think for whatever reason, we tend to kind of over rotate on that now.
这个问题很难回答，因为我们甚至不知道该怎么给这些模型做基准测试。
And it's a hard question to answer just because we don't even know how to benchmark these models to begin with.
定价是动态的。
The pricing is dynamic.
而且经常有补贴。
It's often subsidized.
然后，说实话，下一个大模型出来的时候，反正它往往都是帕累托最优的。
Know, and then and then and then, honestly, like, the next big model comes out and it tends to be Pareto efficient and everything anyways.
所以，他们最后可能都会路由到一个模型，比如OPUS 4.5。
And so, like, then you all they would route to, like, one model like OPUS four five anyway.
所以我觉得现在，实际的这个路由部分，你知道，收益是不确定的。
So I think right now, like, the actual routing piece is, you know, the gains are uncertain.
然而，这是一个非常非常受欢迎、非常成功的互联网产品和品牌，我觉得它更像是一个双边市场。
And yet, this is a very, very popular, very successful, you know, Internet property and brand, and I think it's more of the two sided marketplace.
对。
Yeah.
成本是被补贴的。
The costs are subsidized.
那这种情况什么时候会结束？
When will that stop?
我的意思是，现在这些人拿的都是免费的钱。
I mean, right now, these guys get free money.
我也会这么用。
Would use it too.
对吧？
Right?
比如说，如果我能筹到的钱是整个下游生态系统的三倍还多，是的。
Like, you know, if if I could raise, you know, three times more money than the entire downstream ecosystem Yeah.
加起来。
Combined.
我的意思是，这其实是很经典的情况。
I mean, you know, like, this is a very classic thing.
当然，你会为了获取个人用户而进行补贴。
Of course, you'd subsidize in order to get kind of single users.
而且顺便说一句，这个业务的主要利润其实都来自企业客户，对吧？
And by the way, all the margin tends to be made in the enterprise anyways, right, for this business.
那你为什么不做呢？
And so why wouldn't you do that?
而且，说实话，运营企业，市场营销真的很难。
And by the way, this is by the run businesses, marketing is hard, man.
你知道，真的很难。
Like, you know, like, it's really hard.
无论是现场营销还是内容营销，你永远不知道投入一块钱能带来什么回报。
Field marketing, content marketing, you never know if I put a dollar in what it's gonna get me.
比如办活动，可能会带来一些潜在客户，但这些客户能转化吗？
Like, events, like, it'll maybe get leads, will those leads qualify?
我完全不知道。
I have no idea.
我打算做这个内容。
I'm gonna do this content.
它确实有那种氛围价值，但一直以来，除了非常成熟的组织，很难花一块钱营销费就能说，这能带来这么多销售合格的线索。
Like, it's got this atmospheric value, but it's always been hard to spend a marketing dollar and be like, this is gonna be this many sales qualified leads, except for very mature organizations.
AI 的一个了不起之处在于，你可以花一块钱就能获得用户，因为对 token 的需求是无限的。
A remarkable thing about AI is you can spend a dollar and get users because there's unlimited demands for tokens.
所以在很多方面，它正在颠覆整个营销。
And so in many ways, it's kind of disrupting the entire marketing.
这就是我们看到的。
And so this is what we're seeing.
我们会想，如果我是公司，想吸引漏斗顶端的用户，为什么不补贴呢？
We're saying, like, if I'm a company and I want to attract users for the top of funnel, why wouldn't I subsidize?
对。
Yeah.
我的意思是，如果你在每月 200 美元的 iAnthropic 计划里用了不少 token，他们其实在那上面亏了不少钱。
I mean, like, if you use a decent amount of tokens on your $200 a month opening iAnthropic plan, like, they're losing a decent amount of money off of that.
好吧。
Well, okay.
所以，再说一遍，这些事情有点复杂。
So, again, these these things are these things are a little complicated.
我的意思是，他们只有在想这么做的时候才会这样，因为这是一个可以调节的旋钮。
I mean, are they only if they want to because this is a knob.
对吧？
Right?
而且通常你只在最顶尖的5%用户身上亏钱。
And so and and it tends to be you only lose money on, like, the top 5% users.
对于那类顶尖5%的用户，总是像打地鼠一样，按下葫芦浮起瓢。
And there's always this whack a mole with those type of top 5% users.
对吧？
Right?
比如，这就是为什么你会看到一些规定，比如一个人不能拥有多个账户。
Like, that's why you see things like you can't have multiple you know, like, a single person can't have multiple accounts.
所以他们一直在改这些条款，就是为了控制那百分之五的人。
So they keep changing those terms to, like, keep the five percent under control.
那些人是最吵的。
Those are the loudest ones.
但其他人呢，其实你亏不了多少钱。
But the rest of them, you know, like, it tends to be you're not losing that much money.
但总的来说，这些经常是我参与的董事会级别的讨论，你会说，好吧，听着，我们有这个免费层级。
But in aggregate, these are often board level discussions that I'm involved in, you're like, okay, listen, we've got this free tier.
有个调节开关。
There's a knob.
我们要不要用这个开关来扩大漏斗顶部，增加使用量？
Do we want to use the knob to grow top of funnel and to grow use?
那样的话，我们会走向负利润率，或者我们用这个开关来更偏向利润率增长。
And that way we'll go to negative margins or we do and then use the knob to like kind of go more towards margin gain.
实际上，需求太强了，所以这基本上是个很简单的商业决策，而过去你根本不知道怎么把钱花在营销上。
And it's actually like this stuff is the demand is so strong that, you know, it's the it's kind of a very simple business decision where in the past, you literally don't know how to, like, apply money to marketing.
你就是没试过。
You just haven't.
对。
Yeah.
顺便说一句，我见过一些非常有意思的事情。
By the way, I have seen some very interesting things.
比如说，中国有一些非常精明的操作，他们会利用那些单一服务层级来套利。
So for example, there are these very sophisticated operations out of China that will use the single service tiers and arbitrage them.
他们的做法是这样的：他们会注册一个200美元的套餐。
The way they do it is like so they will sign up to like a $200 plan.
然后大概三天内就把所有额度用光，接着取消订阅，还能拿到剩下27天的按比例退款，尽管额度已经用完了。
They'll use all the tokens in, like, three days, and then they'll cancel, and they'll get prorated for the the the twenty seven days even though they used all the tokens.
然后他们就用这个来给别人提供服务，本质上就是在套利这些套餐。
And then they'll use that to basically provide people with, like a service where they're arbitraging these.
所以围绕这些套餐的洗白市场，其实非常非常成熟。
And so like the the market around like laundering these these plans is actually very, very sophisticated.
所以我们看到很多大平台之间的猫鼠游戏。
And so we're seeing like a lot of cat and mouse between the big left.
这就像新的路由器，在这些订阅模式之间被套利。
It's like the new router is just arbitraged between these subscription models.
不。
No.
说真的。
Seriously.
他们就是这么干的。
That's what they do.
他们确实这么做。
They do it.
他们会注册。
They'll sign up.
然后把额度用完，再取消订阅。
They'll drain it all, then they'll cancel it.
他们会把成本赚回来，然后用这个来提供比如20美元的服务，对吧，基本上就是给某人提供完整的Opus代币之类的。
They'll they'll recoup the cost, and then they kind of use that to offer like a $20 service, right, to to to somebody for basically kind of full, you know, Opus tokens or whatever.
有道理。
Fair.
那你是不是在说，更多的资本几乎直接等同于更强的能力？
So is one of the things you're arguing is that more capital is almost directly correlated to more capability?
能力这个词，我不是想较真。
Capability again, I don't I don't mean to pedantic.
更多的资本，因为需求太高了，你可以把钱投到这些东西上，这往往会转化为使用量。
More capital because the demand is so high, you can put capital into these things and that tends to turn into usage.
而且似乎也往往会转化为某个方向的能力。
It also seems to tend to turn into capability in some direction.
比如我在训练一个模型，我想让它擅长X，我可以创建一个X的强化学习环境，或者我可以花钱雇人来回答X的问题，我就能把资本转化为擅长那个X。
Like if I'm training a model and I want it to be good at X, I can create an RL environment of X or I can go pay someone to, like, answer questions for x, and I can turn capital into being good at that x.
问题是你不知道什么会变差。
The problem is is you don't know what gets worse.
所以，我觉得我们现在差不多到了这个阶段，开始看到更多的提升，也开始看到更多的权衡取舍。
So, like, I think we're kind of at this point where you are starting to see, like, more jackedness and you are starting to see more trade offs.
而我不知道的是，比如说你投入一块钱，或者比如说你投入十块钱来做这件事。
And what I don't know is, like, let's say you put a dollar and you get like, let's say you put in $10 to do this.
我不知道你另一边能不能拿回九块钱，但在这个行业的历史上，我们从来没能做到投入十块钱还能拿回任何东西。
I don't know if you get $9 back on the other side of that, but what we've never been able to do in the history of this industry is put in $10 and get anything back.
对。
Yeah.
以前真的是投入十块钱，然后就是不停地搞工程、搞工程、搞工程，等上两
It was literally put in $10, engineer, engineer, engineer, wait two
年，手指头。
years fingers.
很可能把事情搞砸。
Probably screw stuff up.
很可能会失败。
It'll probably fail.
但也许，另一方面，你会得到一个可以变现的产品。
But maybe maybe on the other side, you'll have a product that you can monetize.
但现在真的是投入10美元，然后直接得到一些回报。
But now it really is $10 in and then some amount out pretty directly.
对。
Yeah.
对。
Yeah.
太棒了。
Incredible.
这就像新的CMO现在变成了CFO。
It's like the new CMO is turning into a CFO now.
对。
Yeah.
我一直是这么说的。
Been putting it that way.
确实有。
There is.
说实话，你可以，我是说，真的，就像很多以前需要更多艺术性的事情，比如工程和市场营销。
Literally, you can be I mean, really, like like many of the the things that required a lot more art, like engineering and marketing.
嗯。
Mhmm.
这两个，你知道，我，我跟这些公司合作太紧密了，以前经常有关于市场营销艺术的长时间讨论，比如，你做付费广告吗？
These are the two think, you know, again, I I just I worked so close to these companies that it used to be there are these long discussions about the art of marketing and what like, you do paid?
比如，你做内容吗？
Like, do you do content?
你做活动吗？
Do you do events?
你做社交媒体吗？
Do you do social?
就像，你怎么做这些？
Like like, how do you do this?
现在基本上就是，我们该多补贴还是少补贴？
And now it's literally like, should we subsidize more or less?
所以你说得对。
So you're right.
这正在变成金融问题。
It's becoming finance.
工程方面也是同样的道理。
Engineering is the same type of way.
以前呢，这很复杂，我们在工程上花了很多时间。
Like it used to be like, okay, like, I mean, it's such a complex thing and we spend so much time in the engineering.
现在很多问题归结为：我们能不能筹到资金买GPU来实现我们的目标？
Now a lot of it reduces down to like, can we raise the capital for the GPUs to do what we want?
能还是不能？
Yes or no?
哇。
Wow.
确实。
Very
所以每个新角色都只是穿着风衣的CFO。
So every new role will just be the role of a CFO in a trench coat.
不。
No.
我不这么认为。
I don't think so.
但金钱与创新、增长和需求之间的关系从未如此紧密。
But the relationship between money and innovation and growth and demand has never been closer.
顺便说一句，我真的觉得这就是这个行业正在发生的重大变化。
And by the way, I really think that is the, like, the the big thing that's going on here in the industry.
对。
Yeah.
所以我们聊到了OpenRouter。
So we talked about OpenRouter.
这周另一笔大收购是Cursor Oh。
The other big acquisition of the week was Cursor Oh.
我记得，如果我没记错的话，这是有史以来对风投支持的初创公司最大的一笔收购。
Which was, I believe, I'm getting this right, the biggest acquisition of a venture backed startup ever.
我觉得除了马斯克跟X那种自己人交易之外，应该是的。
I think so outside of Elon's kind of self dealing with x.
对。
Right.
是的。
Yeah.
不过确实。
But yeah.
不过确实。
But yeah.
所以对于独立公司来说，这应该是史上最大的一笔私募并购。
So for, like, an independent thing, think it's the largest private m and a ever.
对。
Yeah.
那么，Cursor对SpaceX的价值主要来自哪里？
So where does the bulk of the value of Cursor to SpaceX come from?
是团队吗？
Is it the team?
哦，不是。
Oh, no.
我是说，Cursor是一家非常出色的公司。
Mean, Cursor is a phenomenal business.
他们拥有所有的数据。
They have all the data.
而Elon拥有所有的算力。
Elon has all of the compute.
明白。
Mhmm.
你知道，他们过去发布过非常出色的模型。
You know, they've they've released phenomenal models in the past.
很多人意识到，编程某种程度上是通往一些人所说的通用人工智能的道路，但更广泛地说，是通往更广泛的计算机使用或智能的道路。
Many people realize that coding is kind of the path to some will say AGI, but say broader computer use or intelligence more broadly.
所以我实际上认为两家公司都带来了很多价值。
And so I actually think both companies bring a ton to the table.
如果你要把它简化成很简单的东西，就像，你知道，一个有数据，一个有算力，一个有分发渠道，而另一个有足够的资源，让你保持在最前沿。
And if you're going to reduce it to something very simple, it's like, you know, one has the data, one has the compute, one has the distribution, you know, and the other has, like, enough resources that you would need to be on frontier.
因为记住，前沿是一场资本游戏，而这当然是埃隆非常擅长筹集的东西。
Because remember, the frontier is a capital game, and, of course, this is something Elon is phenomenal at scaring up.
你觉得他们600亿美元的估值是独立的价值，还是主要是因为SpaceX？
Do you think their 60,000,000,000 value would have been their value independently, or was that mainly because of SpaceX?
嗯，对谁来说？
Well, to who?
就是，就是，它的价值。
To just to to like, its value.
对，对，对，对，对，对SpaceX来说肯定是这样。
To to to to to to SpaceX for sure.
但如果没有SpaceX，它的价值还会是那样吗？
But would that have been its value independently outside of SpaceX?
你是说对投资者来说？
To like an investor?
是的。
Yeah.
是的。
Yeah.
当然。
For sure.
我确定，我确定，因为我们很容易就能以那个估值融到资。
I I I know I know for sure because we could easily have raised at that at that value.
对。
Yeah.
话虽这么说，私人投资者确实很疯狂，他们好像能看到未来之类的。
Now that that said, private investors are crazy, and they kinda see the future and everything else like that.
那它在公开市场会表现如何呢？
How would it have done the public market?
我不知道。
I don't know.
我甚至没法猜测这类事情，但可以肯定的是，私下的独立价值就是那个量级的。
I can't even speculate about those types of things, but for sure, the private independent value was of that order.
而且有很多非公开的证据可以证明这一点，但我不能谈。
And there's a lot of kind of like, you know, nonpublic proof points to that, which I can't talk about.
但没错，我确实这么认为。
But yes, I do think so.
再说，那门生意当时做得非常好。
And again, the business was doing phenomenally.
比如，如果你看到他们提到的那些数字，我是说，那是我在投资十年、在硅谷待了二十年里见过的最快的增长。
Like, if you saw the numbers that that they talked about, I mean, is the fastest growth, certainly I've ever seen in ten years of investing in and twenty years in the Valley.
对。
Yeah.
但是，如果能成为SpaceX的一部分，拥有他们所有的算力、基础设施，还有SpaceX整个工程团队的支持，那不是更好吗？
But would it not be even better to be, like, part of SpaceX with all of their compute and all of their infrastructure and, you know, the the entire engineering force of SpaceX behind them?
这似乎会让Cursor现在更有价值。
It seems like that that would make Cursor more valuable now.
当然。
Of course.
对。
Yeah.
对。
Yeah.
我同意。
I agree.
我同意。
I agree.
对。
Yeah.
对。
Yeah.
哦，等等。
Oh, wait.
你是在建议Cursor收购SpaceX吗？
Are you suggesting that Cursor buy SpaceX?
那可能再过几年就会发生。
That might've happened in a couple years.
对。
Yeah.
几年前的话。
Few few more years ago.
不。
No.
你看。
Look.
听我说。
Listen.
我觉得OpenRouter和Stripe挺有道理的。
I I think think OpenRouter and Stripe make a lot of sense.
这两家公司都把业务看作市场。
Both companies kind of view things as markets.
但你知道，一种观点是，token就是价值。
But, you know, one view is like, you know, know, tokens is value.
另一种呢，支付就是价值。
The other, you know, payments is value.
这类公司在创始人层面有很多共识。
There's a lot of alignment at the founder level for these types of businesses.
Cursor和SpaceX的情况也非常相似。
Very, very similar thing for Cursor and SpaceX.
对吧？
Right?
都是非常注重工程的文化。
Very engineering focused cultures.
你知道，他们都相信代码是通往AGI的道路，通用计算机使用也是通往AGI的道路。
You know, both believe that code is the path and general computer use is the path to AGI.
而且，我实际上觉得埃隆喜欢这种类型的团队，就是那种非常拼、行动非常快的团队。
And, you know, you know, and I I actually think Elon likes these types of of teams that are kind of, like, you know, very, very scrappy, move very, very fast.
你甚至写过一篇关于这个的文章。
You even wrote wrote a post about this.
这是我见过的除埃隆公司之外最快的团队。
This is the fastest ever anything team I've ever seen outside of an Elon company.
所以我觉得他们之间有很多共同点。
And so I would say there's a lot of alignment.
有很多协同效应。
There's a lot of synergy.
我觉得这两家公司合并后，双方都能变得更强，但我还是觉得这600亿美元花得值。
I think the combined entities are greater on both sides for these acquisitions, but I still think it was worth $60,000,000,000.
他们是怎么做到的？
How did they do it?
就是说，Cursor的文化里有哪些方面让他们能这么快地迭代？
Like, what is what are the aspects of Cursor's culture that make them capable of iterating so quickly?
这个问题问得好。
You know, that's a good question.
我，我，不，不。
I I I no.
在很多方面，我合作过的这些AI公司都特别注重研究。
They in in in many ways, a lot of these AI AI companies that I I work with are so research heavy.
如果你不是在研究新的模型架构，就很难真正引起核心团队的兴趣。
If you're not working on a new model architecture, like, it's kind of hard to actually interest the core teams.
他们一直把重点放在主要的事情上，也就是Cursor，那就是改变软件工程的方式，他们从根本上相信这是一个产品问题，而不一定是一个模型架构问题。
And they always kept the main thing the main thing, cursor, which is like changing how you write software engineering, making and they believed fundamentally it was a product problem, not like necessarily like a model architecture problem.
模型很重要，但这是个产品问题。
The model is very, important, but it was a product problem.
所以呢，他们招人招得特别好。
And so, you know, they hired incredibly well.
他们文化也建设得特别好。
They set culture incredibly well.
我觉得创始人大概花40%的时间在招人和建设文化上。
I would say the founders probably spend 40% of their time hiring and setting culture.
他们非常专注于工程，而不是研究和模型架构。
They're very focused on engineering and not like research and model architecture.
我觉得研究和模型架构非常重要，但它的生命周期不同，需要不同的环境，而当时他们并不是专注于那个。
I think like research and model architecture is incredibly important, but it's got a different life cycle and it requires a different type of environment, and that's not what they were focused on at the time.
当然，现在他们有了资源，就更专注于这个了。
They are more so now, of course, now that they've got the resources.
然后你听我说，在这个领域，这难道不疯狂吗？
And then listen, I mean, like, is this wild in this space?
实际上，真正专注于产品的公司并不多。
There's actually not a lot of companies that are actually focused on product.
对吧？
Right?
他们更倾向于做服务。
They'll focus on services.
比如，他们会说我们是某个领域的Palantir，这挺好的。
Like, there'd be like, we are the Palantir of X, which is great.
这里面有很多价值。
There's a lot of value there.
很多做微调和后训练的公司，其实更偏向服务型。
A lot of the fine tuning post training companies are kind of more service oriented companies.
还有很多研究型公司。
And there's a lot of research companies.
我的意思是，Neolab是个研究公司，但他们说自己是产品公司。
I mean, Neolab is a research company, but they're like, we are a product company.
我觉得这确实是这个时代的一个关键区别。
And I think that's actually a key differentiator in this era.
对。
Yeah.
挺酷的。
It's pretty cool.
我们之前在MTS上请了Cursor团队的人来分享他们的内部文化，他们一直在内部为自己开发产品，然后自己先试用。
We had people from the Cursor team actually here on MTS, and they're sharing a little bit about their internal culture, and they're constantly building products internally for themselves and then testing it on themselves.
他们自己的，他们自己的。
Their own their own.
对。
Yeah.
他们就是自己的用户。
They're they're their own users.
我也很喜欢这一点。
I mean, also, I love that.
我很喜欢这一点，他们基本上都是开发者，想打造一个自己会用的产品。
I love that that, like, literally, they're all like, well, we're all developers who wanna build a product that we will use.
而且他们真的把这一点牢记在心。
And they really, really took that one to heart.
对。
Yeah.
我觉得我最喜欢的例子是设计主管Rio Lou。
I think my my absolute favorite example of this was Rio Lou, who was the head of design.
太好了。
Great.
对。
Yeah.
是的。
Yeah.
是的。
Yeah.
他太棒了。
He was amazing.
他做了一个叫Rio OS的项目。
He did this project called Rio OS.
对，就是那个。
Which is Yeah.
我很喜欢。
I loved it.
就是，
Like,
那种
that kind
有点像复古Mac的外观。
of was, like, like, retro Mac looking as such.
有点像复古Mac OS模拟器那种。
Like a retro Mac OS emulator of sorts.
我确实觉得这挺有意思的。
I I I do think it's funny.
所以，你知道，我用一种很随性的方式做编程，然后做复古电子游戏。
So I, you know, I do vibe coding in a silly VC way, and I build retro video games.
我感觉自己就像那种在车库里搞爱好的老家伙。
And I feel like the kind of, like, you know, older person in the garage doing a hobby.
以前可能是玩火车模型什么的，现在变成了做复古电子游戏或者类似的。
It used to be like train sets or whatever, and now it's like building retro video games or retro or whatever.
你看到很多这样的人冒出来。
You see an awful lot of that coming out.
对。
Yeah.
那这两笔交易，一个是和那个，另一个是OpenRouter的，你觉得它们说明了价值总体上在往哪里走？
What did both the and open router deals tell you about where value is occurring in general?
所以，你知道，我要回到我之前说的，就是X和社交媒体上那些幸灾乐祸和批评，很多都是关于商业质量指标的。
So I I you know, I'm gonna go back to what I said before, which is so much of the general schadenfreude and criticism on x and the socials are around business quality metrics.
而且说实话，他们经常是错的。
And often they're wrong, by the way.
比如，那是个很好的业务，而且做得很好。
Like, was a great business and it was doing great.
但这忽略了重点，我们正处于一个变革浪潮中，新的技术栈组件正在被开发，它们具有战略性的独立价值，未来会有巨大的选择空间。
But it misses the point of, like, we're in a transformative wave where new pieces of the stack are being developed and they have strategic independent value that will have a tremendous amount of optionality going forward.
对吧？
Right?
我的意思是，路由器在令牌路径上，是长尾模型和新模型领域的领导者，而且它有一个双边市场。
I mean, router is on the token path that is the leader in the long tail of models and new models that has a two sided marketplace of those.
Cursor 是迄今为止领先的软件开发工具。
Cursor was by far the leading software dev tool.
所以对我来说，很明显你会拥有这些具有战略价值和战略控制点的资产。
And so to me, it's obvious that you're going to have these properties that are strategic value, strategic control points.
我不认为这更深入地说明了价值会在技术栈的哪个部分积累。
I don't think it says something more deeply about, like, where in the stack value is gonna accrue.
据我所知，价值正在各个技术层面累积。
As far as I can tell, by the way, values are accruing all layers of the stack.
我是说，英伟达做得很好。
I mean, NVIDIA is doing great.
模型公司也做得很好。
The model companies are doing great.
应用公司也做得很好。
App companies are doing great.
推理公司也做得很好。
The inference companies are doing great.
服务公司
The services
在谈论
companies talking about
媒体公司也做得很好。
the media companies are doing great.
听着，我觉得这是我整个职业生涯中最大的财富解锁机会。
Like, right now, listen, I think this is the biggest unlock of wealth I've in my entire career.
我从没想过在互联网之后还能看到另一个这样的机会。你知道，我觉得任何想创业或投资的人，我强烈建议不要过于担心零和思维、商业模式或防御性，而是真正思考在这个新世界里，什么才是战略上重要的？
I never thought I would see another one after the And, you know, I I I think anybody that wants to do a company or invest in the company, I would strongly recommend zero sum thinking or worrying about modes or defensibility too much in the near term and really think about, like, what is strategically important in this new world that's being created?
你觉得在这个正在形成的新世界里，目前最大的机会是什么？
What do you think are the top opportunities right now in this new world that's being created?
如果有人来找你寻求建议，比如，我该从哪里开始？
If someone were to come to you asking for advice, like, where do I start?
我该在哪里建设？
Where do I build?
你会敦促他们做什么？
What would you urge them to do?
你知道，我喜欢做风险投资家的一点是，我不需要回答这个问题。
You know, one thing I love about being a venture capitalist, I don't have to answer that.
我不需要知道答案。
Like, I don't have to know the answer to that.
不。
No.
说真的。
Seriously.
不。
No.
还是不。
And no.
不。
No.
这个问题问得真好。
It's it's such a great question.
大家以为作为风险投资人，我能看到未来。
People assume as a venture capitalist, like, I see the future.
我对此有理论，但实际上我没有。
I've got thesis on it, and I just don't.
我的意思是，作为创始人做过两家初创公司，其中一家就是那种，你知道的，整个事业。
I mean, like, having been a founder doing two startups, one startup was like, you know, it was the whole thing.
就是那种，你知道的，打造一个全球性的企业。
It was like, you know, build a global business.
在那里你不得不预测未来。
And there you have to predict the future.
在那里你必须像，你必须融入科技战略棋局中的神经系统。
And there you have to like, you have to be piped in the nervous system of the techno strategic chess game being played.
未来会是什么样子？
And what does the future look like?
现有企业会怎么做？
What are the incumbents gonna do?
技术会如何变化？
How's the technology gonna shift?
它会不会让我变得无关紧要？
Is it gonna make me irrelevant?
而且你就像是一直在胃溃疡一样，总是要琢磨什么才是热点。
And you it's like a walking ulcer all the time is like trying to figure out what's hot.
而作为风险投资人，你会想，你知道吗？
And as a venture capitalist, you're like, you know what?
有一整个生态系统的创业者都在承受这种压力，他们整天都在想这些。
There's a whole ecosystem of founders that are bearing that ulcer, and then that's what they think about.
我很乐意搭上他们的梦想，支持他们。
And I am very happy to borrow along their dreams and support them.
所以总的来说，我对人工智能非常看好，也很乐观。
And so I am generally very bullish and optimistic about AI.
我觉得这是一个巨大的突破。
I think this is a massive unlock.
如果这个领域有三四个厉害的创业者，而且是我了解的领域，我就不会去质疑。
If there are three or four strong founders in the space that I understand has to be a space I understand, I won't second guess it.
我的意思是，他们要考虑机会成本，还要冒着时间和家庭的风险，我就觉得这肯定是个好领域，然后我就会投资。
I mean, you know, they're gonna be taking the opportunity cost and risking their time and their, you know, family's time on this, I just assume it's a good space, and I'll and I'll invest.
对。
Yeah.
那你对a16z infra的招聘是怎么考虑的？
So how are you thinking about hiring at a16z infra?
你在找什么样的人？
Like, what kind of people are you looking for?
什么样的
What kind of
技能是你想要的？
skills are you looking for?
我们发现最适合我们的是那些有产品背景的人。
So we found so we found what works best for us are people that actually have product background of some sort.
我觉得这跟我们平时聊的内容有关。
I just think it's the nature of the conversations we have.
我觉得我们之所以能有那么多战略性的收购退出，是有原因的。
I think there's a reason why we have probably the most strategic acquisition exits.
我们有很多这样的案例，因为我觉得我们把这些东西都看作是战略资产，这意味着你必须理解产品市场契合度。
We've had a lot of those because I think we view a lot of these things as strategic assets, and that means you have to understand product market fit.
所以你必须了解市场、市场的演变，你必须了解产品，以及技术如何映射到产品上，还有那个接口。
So you have to understand the market, the evolution of the market, you have to understand the product, how the technology maps to the product and that interface.
所以那是一种非常基于产品的讨论，而且你往往会在你还没有足够的财务数据来真正判断业务结果并以此估值之前，就进行这种讨论。
So that's a very kind of like product based discussion, and you tend to do this like before you have enough financials to really read kind of the outcome of the business and value it that way.
我想说，和很多优秀的人共事过，我真的很喜欢和我的团队一起工作，但我们过去也雇过一些没有这种背景的人，比如他们说自己纯粹是金融背景出身。
And I will say having worked with many great people, actually, I I love working with my team, but we have in the past hired people that don't have that background, like they say they come strictly from finance.
即使他们能胜任工作，比如能联系客户，但他们就是缺乏那种敏感性，无法提出
And even though they can do the work, like, can call in the customers, they just don't have the sensitivity to come up
正确的答案，因为品味？
back with the right answers because taste?
那是什么？
What's that?
就是那种品味？
Like, the taste?
这几乎就像你必须知道该问什么、该看什么，才能判断他们会不会买，会不会有价值，会不会有长期价值。
It's almost like you have to know what to ask and what to look for to know if there's gonna you know, if they're gonna buy, if there's gonna be value, if there's gonna be long term value.
我的意思是，这就是产品经理的工作。
I mean, like, this is what a product manager's job is.
所以，不是每个和我们合作的人都是产品经理，但大多数人都在某种程度上参与过产品，所以他们明白那个界面有多重要。
So not everybody who works with is a product manager, but like most people have been involved in product in some sort or another and so that they understand how important that interface is.
顺便说一句，我觉得，即使说到我们合作的创始人类型，你知道，有些真的很有趣。
And by the way, I mean, I would say even when I when it comes to, like, the types of founders we work with, I mean, know, some it's so funny.
我觉得投资者有两种观点。
I feel like there's two views from investors.
有些非常注重产品。
Some are, like, very product focused.
对。
Like Mhmm.
我昨天和Ben Toros聊过。
I was talking to Ben Toros yesterday.
他说，回到最强的创始人，创始人就是一切，这倒是真的。
He's like, back to strongest founders, the founders, everything, which is true.
然后其他人非常关注市场，比如我董事会里的安迪·拉德克利夫。
And then other people are very market focused, like Andy Radcliffe, who's on my board.
他觉得市场就是一切。
He's like, the market is everything.
就是说，你需要一个好的创始人在一个好的市场里，你懂的，就这样。
Like, you need a good founder in a good market, you know, whatever.
而且我觉得对于我们所投资阶段的初创公司来说，我们主要的考察点就是创始人与市场的匹配度。
And I think for infrastructure for the stage we invest in, our primary form of inquiry is around founder market fit.
对吧？
Right?
有些创始人就是适合某些市场，而有些很优秀的创始人反而不适合。
There's just certain founders that will be good for certain markets, and there's great founders would not be good for them.
所以我们花了很多时间去理解这种匹配关系。
And, you know, we spend a lot of time trying to understand that interface.
而且我觉得，这就是为什么我们组建了这样的团队，
And I I think that that is why we build the the type of team that
我们确实是这样。
we do.
抱歉。
Sorry.
当然。
Definitely.
你对此有什么看法？
What's your perspective there?
你从哪里
Where do you
我对创始人市场匹配的看法？
My perspective on founder market fit?
还是你只是
Or just are you
更偏向于创始人本身很重要，就是
more pro just the founder matters, just
市场重要吗？
the market matters?
创始人市场匹配。
Founder market fit.
我觉得就是创始人和市场的交集。
I'm like literally the intersection of the founder and the market.
我觉得除非是极少数情况，比如特拉维斯·卡兰尼克那种。
Like, I don't think unless it you know, unless it's very, very rare, like Travis Kalanick.
你知道吗？
You know?
就像那些人，你知道吗？
Like, those you know?
这些人太厉害了，而且非常罕见。
These these guys are so amazing, and they're so rare.
你知道吗？
You know?
我觉得，一个人进入创业公司之前走过的路，会在某种程度上塑造他。
I I think that, you know, the path that someone took to getting to a startup kind of carves them in a certain way.
他们会有某些特别的敏感度。
They'll have certain sensitivities.
他们会有一些别人没有的、通过实践得来的知识。
They'll have certain earned knowledge that other people don't have.
而且我觉得，你要把这些东西，加上他们天生的能力，去对应市场的实际需求，这意味着你得做大量的市场调研。
And I think you want to take that and their kind of inherent skills and map it to like what the actual market needs, which means you have to do a lot of market work.
可以说，我们大部分的工作，并不是在某个时间点去分析某个具体的项目。
Would say the majority of the work that we do is not like analyzing a given deal at any point in time.
我们大部分的工作，是在没有具体公司的情况下，去分析市场，以便真正理解它，这样当我们遇到公司的时候，才能做出这些决定。
The majority of the work we're doing is analyzing the market, you know, in the absence of any given company to understand it so we can actually make these decisions once we meet the company.
嗯。
Mhmm.
确实。
Definitely.
那么，马丁，问你一个私人问题。
Well, Martin, personal question for you.
是什么激励着你？
What motivates you?
你希望朝着什么样的世界去努力？
What's the world you're trying to build towards?
总的来说，是什么激励着我？
What motivates me in general?
说实话，我就是个喜欢爬山的人。
Honestly, I'm just a hill climber.
这很有趣。
It's so funny.
当我卖掉我的公司的时候，你知道，我去了VMware。
When I was when I was selling my company, you know, I I went to this is to VMware.
我记得帕特·基辛格。
I remember Pat Gelsinger.
你知道，他当时就像是在面试我们。
You know, he was, like, you know, kind of interviewing us.
他是VMware的CEO，他当时就是在面试我们。
He was the CEO of VMware, and he was kind of interviewing us.
然后我和他有一对一的交流，他问我，马丁，是什么让你早上起床的？
And I had this one on one with him, he's like, Martin, kind of what what what would get you up in the morning?
我就想，这真是一个很深的哲学问题。
I'm like, you know, this is a really deep philosophical question.
我得好好想想。
I'm gonna have to think about that.
我明天告诉你。
I'll tell you tomorrow.
所以我回家后，就认真想了想这个问题。
So I went home, you know, and I thought about it.
我回来后告诉他们，我说，听着，听着，我是个喜欢爬山的人。
I came back in and I told them, I'm like, listen, listen, I am a hill climber.
我的初恋，我绝对的初恋，是科技和创业，还有创造性破坏和创新。
My first love, my absolute first love is technology and startups and and creative destruction and innovation.
我非常非常看好硅谷。
I am very, very long Silicon Valley.
我觉得我们拥有历史上最独特的文化，我喜欢成为其中的一部分。
I think we've got the most unique culture in history, and I love to be part of that.
所以只要有一座山等着我去爬，我就能在我热爱的地方做到。
So as long as there's a hill for me to climb it, I can do it in this place that I love.
我会非常开心。
I'll be incredibly happy.
对。
Yeah.
对。
Yeah.
马丁，能邀请到你真是太棒了。
Well, Martin, it's been amazing having you on.
我很喜欢。
I love it.
太好了。
So good.
恭喜你推出开放路由器。
Congratulations on open router.
恭喜你发布第一版。
Congratulations on version one.
这周真是不可思议。
Incredible, incredible week.
非常感谢。
Thank you so much.
这是我的荣幸。
Was a pleasure
能请你上MTS节目真是太棒了。
getting to have you on MTS.
超级有趣。
Super fun.
谢谢。
Thank you.
感谢收听本期a16z播客。
Thanks for listening to this episode of a 16 z podcast.
如果你喜欢这期节目，记得点赞、评论、订阅，给我们评分或写评论，并且分享给你的朋友和家人。
If you like this episode, be sure to like, comment, subscribe, leave us a rating or a review, and share it with your friends and family.
想听更多节目，可以去YouTube、Apple Podcast和Spotify。
For more episodes, go to YouTube, Apple Podcast, and Spotify.
在X上关注我们@a16z，并订阅我们的Substack，地址是a16z.substack.com。
Follow us on x at a sixteen z, and subscribe to our Substack at a sixteen z.substack.com.
再次感谢收听，我们下期再见。
Thanks again for listening, and I'll see you in the next episode.
本信息仅供教育用途，不构成购买、持有或出售任何投资或金融产品的建议。
This information is for educational purposes only and is not to buy, hold, or sell any investment or financial product.
本播客由第三方制作，可能包含付费推广广告、其他公司提及以及与a16z无关的个人。
This podcast has been produced by a third party and may include paid promotional advertisements, other company references, and individuals unaffiliated with a16z.
此类广告、公司和个人的内容并不代表Ah Capital Management LLC、a16z或其任何关联方的认可。
Such advertise ments, companies, and individuals are not endorsed by Ah Capital Management LLC, a16z, or any of its affiliates.
信息来源于发布之日被认为可靠的渠道，但a16z不保证其准确性。
Information is from sources deemed reliable on the date of publication, but a16z does not guarantee its accuracy.
关于 Bayt 播客
Bayt 提供中文+原文双语音频和字幕，帮助你打破语言障碍，轻松听懂全球优质播客。