20VC：Meta的Muse登顶第一。ChatGPT终于有了对手 | Menlo敲响AI泡沫警报 | Factory估值翻三倍至50亿美元 | Keith Rabois与Airwallex之争：谁对？| Crusoe完成39亿美元融资。数据中心交易是否过热？

所以这一切都说得通了。

So it all made sense.

现在还是早期阶段，后面会有很多谈判。

It's the early days, there will be much negotiation.

但这挺有意思的。

But it's interesting.

我的意思是，这里更大的教训是，六个月前谷歌和OpenAI那边有一堆关于各种智能体支付机制的讨论，对吧？

I mean, bigger lesson here is, you know, there was a whole bunch of Google OpenAI blathering on six months ago about various different agent payment mechanisms, right?

那都是聪明人聚在一起，把事情想清楚。

And it was all your smart people getting in a room and thinking things through.

但事实是，在科技领域，那些东西根本无关紧要。

And the truth is in tech, none of that shit ever matters.

真正重要的是有人把消费者的需求聚合起来，就像Instinct和Meta做的那样，他们开始猛敲API，然后现在所有人都得认真对待。

What really matters is someone aggregates consumer demand, like Instinct and Meta have done, they start pounding on the API, and then now everyone has to focus.

我敢打赌，就在我们说话的时候，Resy和OpenTable那边肯定有人在制定他们的智能体API政策。

I'm willing to bet someone at Resy, even as we speak, and OpenTable, is formulating their agent API policy.

我敢打赌亚马逊也在考虑这事。

I'm willing to bet at Amazon.

他们这周正在和Meta谈。

They're talking to Meta this week.

需求会促使他们赶紧把这些乱七八糟的事理顺。

Demand creates urgency to sort all this shit out.

他们都会输的。

They're all gonna lose.

我觉得这是那些不必要的记录系统的最后挣扎。

I think it's the last stand of the unnecessary system of record.

Resy会输。

Resi is gonna lose.

短期内亚马逊输得最少，因为他们作为商家，在销售上已经接近垄断了。

Amazon is the least losey in the short term because they're they're at the edge of a monopolist for what they sell as a merchant.

但最终智能体还是会绕过他们。

But the agents will bypass them ultimately.

你必须要做出决定了。

You're going to have to decide.

这些代理的立场非常明确。

The agents have very clear positions.

你说得对，Rory，这对Shopify有利。

And you're right, Rory, this benefits Shopify.

Shopify当然会支持这个，对吧？

Of course Shopify leans into this, right?

Stripe和PayPal当然也会支持。

Of course Stripe and PayPal lean into this.

这对他们有利，对吧？

It benefits them, right?

毫无疑问。

There's no question.

而亚马逊失去广告、失去追加销售、失去购物车页面，全都是负面影响。

And Amazon losing ads, losing upsell, losing shopping cart sites, all a negative.

但是，只要智能体能够绕过亚马逊，只要你能做到，它们不仅会这么做，而且会乐此不疲。

But to the extent an agent can route around Amazon, to the extent you can, the agents not only will, they will gleefully do so.

它们会非常乐意这么做。

They will gleefully do so.

你想让我直接给餐厅打电话吗，杰森？

What would you would you like me to call the restaurant directly, Jason?

哦，我发现了一个后端API，OpenTable仍然可以用，尽管它们在Resi上是独家合作的。

Oh, I found a back end API where OpenTable still works even though they're they're exclusive on Resi.

哦，我找到了另一种方式，通过DoorDash，这个接口还开着。

Oh, I found another way to do this on DoorDash that's left open.

这些智能体特别擅长发现那些暴露了不该暴露的接口的漏洞API。

The agents are wonderful at finding broke APIs that have other surfaces that shouldn't be exposed.

这可能是它们唯一的选择，但我相信所有那些记录系统、记录平台，都在严阵以待，与智能体对抗。

This is a It may be their only choice, but I believe all the systems of record, places of record, they're just battening down the hatches and they're fighting the agents.

总的来说，这对它们任何一方都不是好事。

And net net, it's not a positive for any of them.

这对它来说不是好事。

It's not a positive for it.

这对亚马逊来说不是好事。

This is not a positive for Amazon.

我明白你的意思。

I hear you.

正如杰森指出的，这个代理不知疲倦。

The agent, as Jason points out, is tireless.

这个代理确实有能力打电话访问所有17个网站。

The agent does have the ability to call to hit all 17 websites.

这只是计算量的问题。

It's only compute.

他们不在乎。

They don't care.

所以你说得对，

So you're right, a

很多

lot of

这些纯数字化的聚合业务，我觉得它们不会像你说的那样跌得那么惨，Jason，但肯定会有一些绕开像Resy这类需求系统的压力。

these digital only aggregation businesses, I don't think they'll fall as brutally as you say, Jason, but there definitely will be some end run pressure to go around the kind of demand systems like a Resy or something like that.

我不确定这会像你想的那么致命。

I'm not sure it's as fatal as you think.

然后

Then

我不是说我觉得……好吧，我说清楚点，因为我觉得你会同意我的看法。

I It's not that I think well, let me be clear, because I think you'll agree with me.

我不是说它们会被干掉。

It's not that I think they're gonna be killed.

我觉得它们会被打残。

I think they're gonna be maimed.

不管这是否意味着Resy的增长会下降这么多。

It doesn't matter if this means that Resy's growth falls this much.

如果这意味着会影响亚马逊广告收入的10%。

If this means it impacts 10% of Amazon's advertising revenue.

那就像Meta的股票涨了20%多一样。

That's just like Meta stocks up 20 some odd percent.

如果这导致亚马逊股价下跌20%，因为影响了利润率，那就是大事了。

If this drives Amazon stock down 20% because it impacts margins, it's a big deal.

我可以告诉你，整天跟我们的代理商聊天，他们可受不了这种破事。

I can tell you talking to our agents all day, they don't put up with this bullshit.

他们真的受不了，对吧？

They don't friggin' put up with it, right?

听着，我不想点名道姓说是哪个供应商，求你别逼我，但今天早上我们收到一个核心供应商的涨价通知，涨幅巨大，价格高得离谱。

Listen, I don't want to name the name or the vendor, please don't make me, but we got a raise this morning from one of our core vendors, massive price increase, massive pricing.

我们的代理商立刻就说，我想绕开它。

Immediately our agent said, I want to work around it.

这是我的想法，直接甩掉他们。

Here are my ideas, dump them.

而且它实际上自己制定了一个十二个月的迁移计划，准备离开这个供应商。

And it actually laid out a twelve month plan to migrate off this vendor on its own.

他们可不会容忍这种破事，对吧？

They will not tolerate this crap, right?

而且有很多产品你只能在亚马逊上买到。

And there are so many products you can only buy on Amazon.

它有很多优势，比如仓储和物流，但有些东西也可以在Shopify商店里买到。

There's, it has so many advantages as warehouse, it has fulfillment, but some of that stuff can be bought on a Shopify store.

这可能就够了。

It could just be enough.

这可能就够了。

It could just be enough.

这种打击对AI来说是个大事。

This maiming is a big deal for AI.

这会重创现有的那些公司，除非他们接受他们不想接受的东西。

It's gonna maim the existing folks unless they embrace things they don't want to embrace.

我有两点要说。

Two comments.

各说一个方面。

One each way.

第一点是，我确实认为亚马逊最大的防御优势在于它的实体基础设施和配送能力，但再往大了看。

One is yes, I do think Amazon's biggest defense will be its physical infrastructure and ability to deliver, but zooming.

但你的观点是对的。

But your point is the right one.

我记得有人说过，大概二十年前，现在可能二十五年前了，互联网刚出来的时候，说互联网讨厌低效。

I remember someone said, like twenty years ago when the Internet probably twenty five years ago now, Internet first came out, that the Internet abhors inefficiency.

换句话说，中间商会被挤压掉。

In other words, middlemen get pounded down.

想想那些旅游网站，像Expedia、Airbnb，它们都是把需求和供给拉得更近。

And if you think about things like the travel sites, Expedia, Airbnb, all of that is that, is getting demand and supply closer together.

我觉得你说的，Jason，是对的。

And I think what you're saying, Jason, is correct.

人工智能也是一样的。

AI is the same.

它会淘汰那些中间人，那些只是掌握信息并利用信息来报价的人。

It's going to pound down people who are in the middle, who just have information and are using that to make offers.

如果你只是做这些，那你的利润空间就会受到挤压。

If if that's all you're doing, there will be pressure at the margin.

因为如果你能降低搜索成本，这对人类来说是有价值的，但也许智能体自己就能做到。

Because if you're reducing search costs, that's valuable to me as a human, but maybe the agent can just do it itself.

所以我还是坚持关于命名的那个观点。

So I'll go with the naming comment.

哪怕只有5%或10%的影响，那也是很有意义的。

If it's even 5% or 10% impact, it's meaningful.

这就是关键所在。

That's the thing.

对亚马逊来说，关掉Perplexity很容易。

It's easy to shut off perplexity for Amazon.

它们就像小虫子一样，不值一提。

It's just they're just a gnat.

当所有人都用AI的时候，亚马逊就会面临有趣的选择。

Amazon will have interesting choices when it's everybody.

比如，你把所有人都屏蔽掉。

Like, you shut off everybody.

这在技术上是可行的。

It is technically possible.

但到了某个时候，事情就会变得复杂起来。

But at some point, it's gonna be complicated.

我们真的确定要屏蔽所有人吗？

Are we sure we wanna shut off everybody?

我们继续往下聊。

Moving swiftly on.

还有别的关于Muse的话题我没提到吗？

Anything else that I haven't touched with Muse?

没有。

No.

因为这其实不是本周的故事。

Because it's not really the story of the week.

我的意思是，这是上周的故事。

I mean, it's the story of last week.

我们要讲本周的故事，杰森会作为实际用户来给我们讲讲Jev。

We're gonna do the story of the week, and Jason's gonna be ready as a hands on users to tell us about Jev.

那才是本周的故事。

That's the story of the week.

我们怎么进行？

How do we go?

如果你想的话，我们可以聊聊Jev。

We can talk about Jev if you want.

这是下一个要讲的，而且就是那个故事。

It's next one up, and it's the story.

这是下一个话题，而且罗里这周的故事就是它。

It's the next one up, and fine is Rory's story of the week.

所以，对于不了解的人来说，ChatGPT 和 Bantaships Jev，这是一个决定而不是聊天的模型，它成了 Vercel 有史以来最快的一次发布。

So for those that don't know, ChatGPT and Bantaships Jev, it's a model that decides instead of chats, and it becomes Vercel's fastest ever launch.

杰森，我相信你已经玩过了。

Jason, I'm sure you've played with it.

你觉得怎么样？

What did you think?

杰森的评测。

Jason's review.

全新的环节。

New new segment.

嗯，首先，听我说。

Well, first of all, listen.

我可能不够聪明，没法深入探讨语义学。

I'm not I I maybe I might not be smart enough to get into semantics.

我甚至不觉得那是个模型。

I don't even think it's a model.

我觉得它是个由大语言模型驱动的分类器。

I think it's a classifier backed by an LLM.

我用过它。

I've used it.

它很棒。

It's awesome.

但我不确定它会不会像X上那些可能根本没用过的人说的那样具有颠覆性。

But I don't know that it'll be as disruptive as everyone on x talking about it that probably never used it said it is.

但它确实很棒。

But it is awesome.

对吧？

Okay?

Jason，你怎么看？

Jason, what do you

你用它来做什么？

use it for?

我主要用它来判断SaaS社区里哪些人应该互相认识。

What I use it for is deciding who in the SaaS community should meet each other.

它在这方面挺擅长的。

It's quite good at that.

Rory是GCI的CRO。

Rory is the CRO at GCI.

Harry是Perplexity的CRO。

Harry is the CRO at Perplexity.

他们该不该见面？

Should they meet?

其实这个问题没那么简单。

It's actually not that simple a question.

他们所处的阶段差别挺大的。

They're at pretty different stages.

他们住在不同的地方。

They live in different places.

他们该见面吗？

Should they meet?

我整个周六都在Jev上白费功夫，因为我得重新设计提示词的方式。

I spent all of Saturday nothing worked on on Jev because I had to redo the way I did prompts.

好吧，我当时觉得这算是失败了，对吧？

Okay, so I thought it was a failure, right?

然后我学到的是，你已经投入了时间。

Then what I learned is you've invested the time.

它能以比Anthropic或ChatGPT便宜一百倍的价格，在几毫秒内回答那个问题，实际上甚至更便宜。

It can answer that question in milliseconds for a hundredth the price of Anthropic or ChatGPT, actually even less.

它只能回答问题，而且只能回答一部分问题。

It can just answer a question and it can answer a subset of questions.

但我觉得，从数据标注等经验来看，有很多分类工作需要做，很多数据标注工作需要做，但它不能输出文本，也不能做任何特别复杂或需要推理的事情。

But I think as we've learned from data labeling and others, there's a lot of classification that needs to be done, a lot of data labeling that needs to be done, but it doesn't output text and it can't do anything particularly complicated or with reasoning.

所以这挺重要的。

So it's like a big deal.

针对这个特定的用例，我说的是谁应该见谁。

And for this particular use case, I'm talking about who should meet who.

它有颠覆性，但可能只占这个应用所有大语言模型调用的20%左右。

It is disruptive, but it's maybe like 20% of all the LLM calls associated with this application.

所以我非常喜欢。

So I love it.

我觉得每个人都会做类似的事情。

I think everyone will do some version of this.

我觉得，你知道，也许连XA和Parallel也应该做类似的事情。

I think, you know, maybe even XA and Parallel should do some version.

也许OpenAI和Chat也应该做。

Maybe OpenAI and chat.

每个人都应该做类似的事情，但它不会完全取代你的整个技术栈，也不会取代ChatGPT，因为它就是做不了那些事情。

Everyone should do a version of this, but it's not going take over 100% of your stack and it's not going to replace ChatGPT because it just doesn't do those things.

但这确实提醒了我们，天哪，我们在简单的事情上浪费了太多token，这些事根本不该让Astra来做。

But it does remind us that man, we effing waste a lot of tokens on simple stuff that Astra shouldn't be doing.

我百分之百同意你的看法。

I pretty much a 100% agree with you.

我的意思是，从宏观角度来看，是这样的。

I mean, the zoom out comment is this.

这是一种新模型。

It's a new model.

它和LLM不一样。

It's different than an LLM.

LLM返回的是文本，计算量非常大，在计算上又慢又贵。

An LLM returns text and is very computationally intensive, and in computing terms, slow and expensive.

而这个只是返回真或假，或者一个排名或分数，对吧？

This just returns true or false or a ranking or a score, right?

所以有三种答案类型，而且它超级快，超级便宜。

So there's three answer types, and it's super fast and super cheap.

他们甚至都不收输出token的费用，因为输出量少得几乎可以忽略不计。

They don't even charge for output tokens because they're so meaninglessly few.

所以它只是更快、更精准。

So it's just a faster, more precise.

这就是为什么他们在网站上把它称为系统一，这是丹尼尔·卡尼曼的《思考，快与慢》里的概念。

That's why it's called they refer to it in turn on the website as a system one, which is the Daniel Kahneman thinking fast, thinking slow.

而这就是思考快的那部分，快速给出答案。

And this is the thinking fast part, quick, fast answers.

但杰森的计算完全正确。

But Jason's math is exactly right.

不过我们周一讨论过，目前Anthropic、OpenAI和开源模型在LLM调用上的花费大约是一千亿美元。

But we're talking about this on Monday, the part of meeting is that there's roughly a $100,000,000,000 today being spent on LLM calls between Anthropic OpenAI and the open source models.

如果我们刚同意OpenAI会花350亿，亚马逊250亿，那这个数字可能会变。

Maybe that goes if we just agreed that OpenAI is gonna be doing 350, Amazon, 250.

所以五年后可能会达到五千亿美元。

So it's probably gonna go to $500,000,000,000 five years from now.

20%。

20%.

而且，杰森，我们得出的数字完全一样。

And, Jason, we had exactly the same number.

其中只有20%的情况是相关的，也就是说，现在你用复杂昂贵的模型去处理那些其实用简单方案就能解决的问题。

Only 20% of them are relevant to this, where it's today, you're using a complex expensive model for something that really needs a much simpler solution.

但1000亿的20%就是200亿。

But 20% of a $100,000,000,000 is 20,000,000,000.

嗯，我觉得钱不会那样流动。

Well, I don't think I don't think the dollar is gonna flow that way.

我觉得这些令牌

I think the tokens

别打断我，让我说完，真是的。

will don't think the dollar Let me finish, damn you.

接下来要说的是，因为你听起来就像我周一那组里的那些怀疑论者。

The next thing because you you got you sound like the cynics of my group on Monday.

这个问题我们之前已经讨论过了。

We had this I've already had this discussion once.

所以我能把它讲得这么顺。

That's why I can carry it off.

所以那是200亿，但你说得完全对。

So that's 20,000,000,000, but you're exactly right.

就像我一位合伙人说的，兄弟，你刚才说价格会降到原来的五分之一。

As one of my partners said, but, dude, you just said the prices are going to down by five x.

所以那200亿会变成40亿。

So that 20,000,000,000 is going to become 4,000,000,000.

你说得完全对。

You're exactly right.

这就像开源软件一样。

It's just like open source.

它会抢占整个大语言模型市场的很大一部分，用更好的产品、更低的价格，以五分之一的价格去竞争，最终结果就是200亿变成40亿。

It's gonna take a slug of the total addressable LLM marketplace, attack it with a better, cheaper product at one fifth the price, net result at 20,000,000,000 becomes 4,000,000,000.

但如果你是Jev，或者你是做类型安全的，你会说，嘿，那40亿就是我的了。

But if you're Jev or if you're a type safe, you're saying, hey, that 4,000,000,000 becomes mine.

这就是赌注所在。

And that's the bet.

同时，回到你那个命名概念，我很喜欢，Foundation的Jaya在Biggs发了一篇不错的文章。

Now, at the same time, going back to your naming concept, which I love, Jaya from Foundation did a nice post at Biggs.

这只是在总规模1000亿的支出中切走了一块，这些钱本来注定要流向OpenAI和Anthropic，现在却分流到了更便宜的低成本供应商那里。

This is just a slug of the total 100,000,000,000 spend that was kind of automatically destined to go to OpenAI and Anthropic and is now kind of sluicing off into a cheaper low cost provider.

这就是现在正在发生的事情。

That's what's happening here.

你说得对。

You're right.

这并不意味着基础模型的终结。

It's not the end of foundation model.

那些聪明的工作仍然会通过基础模型来完成。

The clever shit is still gonna be done using foundation models.

只是稍微伤了一下，大概百分之十到二十的收入现在会被别人以五分之一的价格拿走。

It's just a little maiming of ten, twenty percent of the revenue that's now gonna be done by someone else at one fifth the cost.

现在我们可以稍后再谈竞争问题，因为我觉得那才是真正的问题，但这就是目前的情况。

Now we can talk about competition in a second because I think that's a real issue, but that's what's going on here.

我觉得成本只有百分之一，但我同意所有

I just think it's a hundredth the cost, but I agree with all

你说得对。

of You're right.

百分之一，那这样的话，是的。

A 100 which which in which case yeah.

我们会看到ChatGPT被拆解，用户会说，哦，这个

We're gonna see, like, the unbundling of ChatGPT, and we're have users go, oh, well, this one

不会的。

would be No.

好

Good

问题不在于ChatGPT这个应用本身，而在于OpenAI的API和Anthropic的API。

The the the problem is not not ChatGPT, the app, but the OpenAI API and the Anthropic API.

你说得对。

You're right.

这是一个开发者产品。

This is a developer product.

你作为终端用户，我作为终端用户，我们不会用它。

You as an end user, I as an end user, don't use it.

你可以去试试用一下，但今天早上它已经满了，我试过了。

You can go on and try and use it, but it's full this morning, I tried.

但说实话，如果你听发布会的话，它非常专注于开发者。

But it's really, I mean, if you listen to the launch announcement, it's very focused on developers.

而且这个想法，挺有意思的，几乎又回到了Muse上，我觉得你可能会看到——我可能说得不对——人类的需求，尤其是消费者对AI的需求，会继续和软件开发者对AI的需求分化。

And the idea is, and this is an interesting thing, almost comes back to Muse, that I think you might see, I could be wrong on this, that the needs of humans, especially consumers for AI, are gonna continue to diverge from the needs from software developers for AI.

而这正是核心的软件开发者的东西。

And this is a core software developer thing.

你永远用不到它的，哈利。

You'll never need it, Harry.

但那些开发软件应用的人可能会发现，他们想要的核心智能中，有很大一部分是系统一的智能。

But someone who's building a software app might realize that a significant portion of the core intelligence they want is system one intelligence.

你就告诉我这是A还是B，比如动物、矿物还是植物。

Just tell me if this is an a or a b, you know, animal, mineral, or vegetable.

鸡蛋比Y更好吗？

Is eggs better than y?

给我个快速回答。

Give me a quick answer.

我不想听你啰嗦，然后像那种谄媚的LLM一样说‘好问题，罗里’。

I don't want to blather and have you tell me that's a great question, Rory, like a sycophantic LLM.

直接给我答案就行了。

Just give me the damn answer.

所以这不是给你这种用户用的，而是在开发者层面进行拆分，那里更可能发生这种情况。

So it's not for you, the user, but it's unbundling at the developer level where it's more likely to happen.

你说得对。

You're right.

我需要更高级的答案。

I require far more intellectual answers.

不。

No.

你只是需要一个打包好的产品。

You just require a bundled product.

等等。

Wait.

等等。

Wait.

其实，哈利，我觉得你真正喜欢的是LLM那种奉承你的部分，它们会说，哈利，你真聪明。

Actually, Harry, I think what you really like is the sycophantic part of the LLM when they tell you, you're so smart, Harry.

但回答你的问题，我整个周末都在用Jev，周六失败了，后来搞清楚了提示词和用法，周日就让它跑通了。

But to answer your question, what I did learn from using Jev all weekend long, failed on Saturday, figured out the prompts and the use case, got it to work on Sunday.

便宜到根本不用去衡量，快十倍才是真正的用途，但仅限于这些特定的场景。

So cheap it doesn't, you don't even to measure it 10 times faster is the use case, but only for these, for these, for these limited use cases.

我学到的另一点，可能听起来有点跑题，就是用一个框架去选模型，还得选对，真是累死人了。

What I did learn, may almost sound tangential is man, using a harness to pick a model and getting it right is effing exhausting.

以后只会越来越难，越来越难。

This is going to get even harder and harder and harder.

什么时候该用Jev？

When should I use Jev?

什么时候不该用？

When should I not use it?

你得跑那么多评估，那么多测试，因为我在Jev上做的事情太多了。

What about You have to run so many evals, so many tests because there's so many things that I do with Jev.

你知道在Jev上什么行不通吗？

You know what doesn't work in Jev?

在Harry、Rory和Jason之间，谁更适合加入二十VC当合伙人？

Between Harry, between Rory and Jason, who's the better person to join twenty VC as a partner?

它回答不了。

It can't answer.

你会发现它会搞错。

You'll find it's gonna get that wrong.

好的。

Okay.

所以，在那种问题和罗里和杰森是否应该见面喝咖啡之间，它会答对后者。

So between that and should Rory and Jason meet for coffee, it's gonna get that one right.

它不会瞎猜。

It's not gonna guess.

它实际上会知道，嘿，我们都在湾区。

It's actually going to know, hey, we're both in the Bay Area.

我们彼此认识。

We've known each other.

我们应该见面喝杯咖啡。

We should meet for coffee.

你得对每一个用例都做质量保证和测试，才能获得那些好处。

You have to QA and test every single use case to get those benefits.

这太累了。

It's exhausting.

然后我在Replit上被切换到了Autorouter，它会在Astra、Fable、开源的那些以及所有其他之间切换。

And then I got switched over on Replit to Autorouter where switches between Astra, Fable, the open source ones and everything.

现在我都不知道它用的是哪一个。

Now I don't even know which one it's using.

然后我注意到性能有所下降。

And then I saw some performance degradation.

所以我又不得不把所有东西都切回Astra。

So then I had to switch it all back to Astra.

你信任你的测试框架吗？

Do you trust your harness?

你能对Jev的10000次使用做质量保证吗？

Can you QA 10,000 uses of Jev?

这倒不是什么坏事，但这就像是DevOps的复兴之类的。

This isn't like a bad thing, but this is like the renaissance of like DevOps or something.

每个人都需要一个庞大的团队来优化它。

Like everyone's gonna need this massive team to optimize it.

而且随着Jev和Friends的出现，需求还在不断增加。

And the needs are going up with Jev and Friends.

你的团队也得扩大规模。

Like your team's gonna have to get more.

这不仅仅是基准测试和评估的问题。

It's not just benchmarks and evals.

我们得在无数种组合中运行这些24/7的任务，这固然是好事，但对我来说，这负担太重了。

Like we're gonna be running these 20 fourseven across countless permutations and it's good, but it's for me it's it's too much.

我真的撑不住了。

Like I can't do it anymore.

我再也选不了这些模型了。

I can't pick these models anymore.

我认输了。

I'm tapping out.

这些工具虽然好，但像Databricks这样的平台，用工具省钱确实容易。

These harnesses are only so like Databricks is it's easy to save money with the harness.

嘿，我们把80%的流量路由到开源模型，现在只做10%这样的。

Hey, we route 80% to open weights models, and now we're doing 10% like this.

这对CFO来说听起来很棒，但我发现这对我没用。

It sounds great to the CFO, but I found it doesn't work for me.

很多人说他们用Jev作为即时路由器。

And a lot of folks were saying they're using Jev as an instant router.

就像，Jev，Jev，你来决定。

It's like, Jev, Jev, you make the decision.

我们是这样做的。

Here's what we're doing.

你在几毫秒内选好模型。

You pick the model in milliseconds.

太好了。

Great.

但是，如果Jev有20%的时间会选错模型。

But if, but Jev is like wrong, like 20% of the time.

如果20%的时间选错模型，对我来说，编写一个关键功能时，整天都在处理满是bug的东西。

If you pick the wrong model for me coding a mission critical feature 20% of the time, I'm going to be trying to build this thing bug ridden all day long.

听着，聪明人总会找到解决办法的。

Listen, smart people will figure this out.

选模型变得越来越复杂，而不是更简单。

It's just getting more complicated to pick your model, not easier.

而且，这对投资者来说是好事，但让开发变得更复杂了。

And, it's it's good it's good for investors, but, it's, makes building more complicated.

如果你是Anthropic或OpenAI的人，他们内部看到Jev时会怎么想？

If you're Anthropic or OpenAI, what do you think the conversation is internally when they look at Jev?

OpenAI已经变得非常商业化。

OpenAI have become ruthlessly commercial.

他们的态度会是，管他呢。

Their take will be like, screw it.

如果有人要这么做，我们也应该这么做。

If someone's gonna do that, we should do that.

我们几周内就能推出类似的东西。

We can get something like this out in a few weeks.

我们也来竞争，推出一个低成本的产品，因为他们就是想赚钱。

Let's compete and have a low cost offering too, because they're trying to make a buck.

而Anthropic，他们想造神，这又不是神，所以他们何必费心呢？

And Anthropic, they're trying to build God, and this isn't God, so why would they even bother?

其实，Jev的发布视频说的是“不是神”，对吧？

In fact, the launch video for Jev was prod not God, right?

所以他们故意在说“不是神”。

So they're deliberately saying not God.

如果你跟Anthropic的人聊，他们的使命就是通用人工智能，也就是神。

And if you talk to Anthropic, the mission is AGI slash god.

所以他们会觉得，你不过是个微不足道的小角色。

So they're like, you are just a minor nothing thing.

我为什么要自降身价去碰那些俗气的商业呢？

Why would I even deign to sully my hands on vulgar commerce?

我觉得Anthropic和OpenAI做了一个非常战略性的决定。

I think that Anthropic and OpenAI have made a very strategic decision.

他们暂时不会进入这个市场。

They will not play in this market for now.

OpenAI和Anthropic各自都有两个最差的模型。

They have OpenAI and Anthropic each have the two worst models that exist.

它们叫OpenAI Mini和Haiku。

They're called OpenAI Mini and Haiku.

它们很糟糕。

They're terrible.

它们真的很糟糕。

They're terrible.

现在你问Claude，简单的工作流该用什么？

Now, you ask Claude, what should I use for simple workflows?

Haiku。

Haiku.

如果你问ChatGPT，它会告诉你Mini很棒。

And if you ask ChatGPT, it will tell you Mini is great.

它很便宜。

It's so cheap.

而且至少在我自己的小测试里，它们百分之百失败。

And at least my little evals, they fail a 100% of the time.

Rory和Jason应该喝杯咖啡吗？

Should Rory and Jason have coffee?

是的。

Yes.

送他们去伦敦。

Send them to London.

最差的回答，我是说，我有点夸张了。

Worst answer I mean, I'm exaggerating.

Mini和Haiku是我用过的最差的模型。

These are the worst models I've ever used, Mini and Haiku.

所以他们决定推出一些功能受限的模型，可以稍微宣传一下，但暂时不想涉足这些低利润的业务。

And so they've decided to launch crippled models that they can lightly promote, but they don't want to play in these low margin businesses for now.

他们本来可以做出这个的，对吧？

They could build this, right?

他们本来可以做的，我觉得他们一个小时或者一个周末就能做出一个Jev的版本，但他们故意限制了低端市场。

They could build, I think they could build a version of Jev in an hour or over the weekend, but they've intentionally decided to cripple the low end of the market.

而且我觉得他们对于开放权重模型比预期更早接近他们这件事感到紧张。

And I think they're puckered about the open weights models that just got a little bit closer earlier than they planned.

因为那比放弃低端市场更具威胁。

Because that is more of a threat than abandoning the bottom of the market.

这只是我的看法，但我认为他们从来没有认真对待过低端市场。

This is just my view, but I don't think they've ever taken the bottom of the market seriously.

他们推出的那些入门级产品，我觉得没有哪个正经开发者会去用。

They have checked the box offerings that no serious developer uses, I don't think.

你去Claude上，如果能切换到haiku，就问它一个问题试试。

Go on Claude and switch to haiku if you can and ask it a question.

它什么都记不住，2023年之后的事一概不知，而且它连东西在哪儿都搞不清楚。

You it it it doesn't remember anything since 2023, and it doesn't know know where anything is.

而且这

And that

说白了，这就是Jev的机会所在。

in a nutshell, it is the Jev opportunity.

虽然不是完全一一对应，但你说得对。

It's not quite one for one, but, yeah, you're right.

我稍微跑个题，不过大家确实喜欢我们聊Venture。

I'm I'm going slightly off on a tangent on this, but people do like it when we talk about Venture.

Jev那轮种子轮融了4000万美元。

It was a $40,000,000 seed round for Jev.

而且，你知道吗，就在这之前我刚跟团队聊过，他们说，哥们，我们现在再也看不到低于两千万美元的首轮融资了。

And, you know, I just had a conversation with my team right before this, and they're like, dude, we never see anything less than a $20,000,000 first raise for anything anymore.

种子轮的最低标准，大概就是八到一千万，还得是从好公司出来的人。

The seed rounds, minimum are, like, eight to 10 with someone spinning out of a good company.

我们是不是正在见证传统种子轮的消失？

Are we seeing the evaporating of traditional seed?

你觉得安德森为什么刚搞了个大学？

Why do you think Andreessen just did a university?

他们得走到更早的阶段，甚至在公司成立之前。

They need to go pre inception.

你觉得我在开玩笑？

You think I'm kidding?

不。

No.

我是说，终于有人明白了，彼得·蒂尔十八年前在蒂尔奖学金和宿舍基金就搞对了，那个基金差不多投出了Cursor。

Mean, finally someone figured out that Peter Thiel got this right eighteen years ago at the Thiel Fellowship and the dorm room fund, which sort of did cursor.

现在想做种子轮，就得去投更早的阶段。

You got to go pre inception if you want to do seed now.

早期阶段？早期阶段太贵了。

Inception Inception's too expensive.

你得去投更早的阶段。

You got to go pre inception.

但安德森·霍洛维茨今天确实推出了霍洛维茨·安德森大学，投入四千万美元做更早期投资，对吧？

But Andreessen Horowitz did launch Horowitz Andreessen University today with $40,000,000 to do pre inception investing, right?

就是把泰尔奖学金放大来做。

To do the Thiel Fellowship on steroids.

我觉得这是对你那个观点的非常理性的回应。

I think it's a very rational response to your point.

但我确实认为彼得·泰尔比任何人都更早看到这个领域，并且付诸实践了。

But I do think Peter Thiel saw this space before anybody did and executed.

他唯一不想做的就是把它规模化。

The only thing he didn't want to do was scale it up.

他不想像霍洛维茨和德里森那样建一所蒂尔·拉博伊斯大学。

He didn't want to build a Thiel Rabois University like Horowitz and Dreeson.

他就是不想扩大规模，本来可以扩大的。

He just didn't want to scale it It could have been.

我是说，蒂尔奖学金里有最优秀的人。

I mean, they have the best people in the Thiel fellowship.

现在他们也能招到最优秀的人，他们确实能招到最优秀的人。

They get the best today too, they get the best people.

你知道吗？

Do you know what?

我觉得他们现在不一定能招到最优秀的人了。

I I I don't think they do anymore.

实际上，我觉得Z世代奖学金现在能招到最优秀的人。

Think Z fellows get the best today, actually.

好的。

Okay.

也许吧。

Maybe.

Z Fellows 招到的人真是好得不得了。

Z fellows gets unbelievably good people.

说到底，还是看想法，对吧？

At the end of day, idea, right?

这类项目的问题在于，如果投入的人不够多，就做不大。

The problem with these things is if you don't put enough people into them, they don't scale.

这就是为什么 Z Fellows 可能比 Thiel Fellowship 更有意思，而 Horowitz 和 Dreesen 跟 Marching 差不多，因为你就是往里面投入更多资源，以前你靠一个博客就能运营基金。

That's why Z Fellows maybe is more interesting than the Thiel Fellowship, and why Horowitz and Dreesen is more if it's the same as Marching, because you're just putting more resources and things that, you you used to be able to run a fund with a blog.

就像那样就能给你带来你需要的所有项目机会，连续投出五个独角兽，但你现在得把这些东西扩大规模，你懂吧？

Like that would give you all the deal flow in the world you needed for like five unicorns in a row, but you gotta scale this stuff up, you know?

现在你都在跟教授一起录播客了。

Now you're on a podcast with a professor.

对，以前是。

Yeah, used

光靠一个播客和一个人躲在壁橱里就能搞定。

be able to just do with a podcast, and one dude in a in a in a closet and do just fine.

不过我们还没回答我的问题呢，就是——

We haven't answered my question, though, which

我想回到你的观点，哈里。

is I like will go back to your point, Harry.

你说得对。

You are correct.

一旦你超越了个人想法，你说得对。

Once you go beyond the vigil to even an individual an idea, you're right.

你会看到更大的支票。

You're seeing bigger checks.

你会看到20个。

You you are seeing 20 look.

我觉得TypeScript那个，可能还更大。

Think the TypeScript was I think it was even bigger.

我可能说错了。

I could be wrong.

我记得差不多是40，但我也可能记错了。

I think I thought it's almost 40, but I could be wrong.

不过确实，现在从底层到上层，大家写的支票金额都更大了。

But yes, people are writing bigger checks now up and down the stack.

或者如果你想做传统的种子轮投资，对吧，很多时候你得接受更低的股权比例，当然也不是所有情况都这样。

Or if you wanna do traditional seed checks, right, you have to tolerate far low lower ownership in many cases, not all cases.

你还是可以自己找项目，对吧？

You can still hunt your own deals, right?

然后算一下账，你得找到能到250亿美元估值的项目。

And then when you do the math, you got to hunt $25,000,000,000 outcomes.

如果我投四分之一或者三分之一到Jev，如果Jev退出时估值超过250亿美元，扣除稀释后，我还是能拿到我需要的100倍或者50倍回报，对吧？

If I'm going to do four into Jev or three into Jev, if Jev exits north of 25,000,000,000 after dilution, I can still do my 100X, right, that I need to do or my 50x, right?

但种子轮低股权比例是可行的，不过天哪，你得需要巨大的回报才行。

But the low ownership for seed works, but man, you need the big outcomes.

还有一件事，我觉得大家总是忽略的，就是那些特别枯燥的经济学常识，但名义GDP，也就是通货膨胀加上增长，从2010年到现在大概涨了两倍半。

And then the other thing just to take into account that I always feel everyone forgets, right, is discuss on really dorky economics common, but nominal GDP, in other words, just inflation, but inflation plus growth from 2010 to the day is about two and a half x.

所以，你知道，这意味着2010年的一亿美元，今天就相当于两亿五千万美元。

So, you know, what that means is a $100,000,000 in 2010 is $250,000,000 today.

你知道，就你花钱的能力而言，其实就是调动资源的能力，因为很明显，软件行业的工资至少跟上了名义GDP的增长。

You know, just in terms of your ability to command with money is an ability to command resources, because obviously, software salaries have at least kept up with nominal GDP growth.

所以，如果你在2010年写五张三百万美元的支票，我们就算四百万一张吧，那你在2026年就应该写一千万美元的支票。

So what it means is if you were writing five $3,000,000 checks in 2010, let's do $4,000,000 checks, you should be writing $10,000,000 checks in 2026.

这还只是数学，其他什么都没变。

That's just math before anything else has changed.

然后在此基础上，你再考虑到2010年的时候，世界还没那么糟糕，而2026年，科技领域的一切看起来都棒极了。

Then on top of that, you take into account the fact that in 2010, the world wasn't a shitter, and in 2026, everything in tech looks amazing.

你眨眼之间就能到两千万。

You get to 20,000,000 before you blink.

那你是不是认为现在其实是更好的时机？因为你的支票金额要求涨了两倍半到三倍，但后端的结果规模，当我们看那个光标时，涨幅远远超过两倍半到三倍。

Would you actually argue then that now is a better time because your check size requirements have gone up two and a half to three x, but the outcome sizes on the back end have gone up significantly more than two and a half to three x when we look at the cursor.

不是。

No.

我完全不这么认为，因为你混淆了一点：今天看到的那些成果，其实来自于当年那些更小的支票。

I wouldn't make that argument at all because what you're confusing for the it it the The outcomes today happened from the checks that were smaller.

换句话说，你的意思是，嘿，五年前写的支票，现在回报惊人。

In other words, because what you're saying is, hey, the outcomes today on checks written five years ago are amazing.

所以今天这些大得多的支票，也会同样惊人。

Therefore, the checks today, which are much bigger, will be amazing too.

这背后你其实是在假设，今天250亿美元的成果，三年后会变成500亿美元。

Implicitly in that you're assuming, you know, that the $25,000,000,000 outcome today becomes the $50,000,000,000 outcome three years from now.

如果真是这样，那你说得对。

And if that's the case, then you're correct.

同样的通胀率也适用于退出时的估值。

I'm the same rate of inflation applies to the exit scenarios.

对，但我觉得，退出时的估值不只是跟着名义GDP涨，它们涨得更多。

Yeah, but I take it, but the exits didn't just go up by nominal GDP, they had by even more than that.

换句话说，你已经看到了。

In other words, you've seen it.

这就是为什么事情总是很难拆开来看，名义GDP在涨，但股市的涨幅远远超过了它。

That's why it's always hard to disaggregate things, you have normal GDP go up, but then the stock market has massively outpaced that.

而风险投资的退出规模更是远远超过了股市。

And venture exits have even more massively outpaced that again.

所以，退出规模确实有一个长期的趋势性增长。

So there is a long term secular trend and the size of exits going up.

但现在可能还叠加了一个估值提升的因素，这个因素可能会持续。

But there's probably an overlay of a valuation lift right now that might persist.

这里我得特别提一下Menlo的Venky，他写了一篇很棒的文章，昨天在风投圈里传开了，讲的就是在周期顶部怎么玩这个游戏。

And here I gotta give a shout out to Venky from Menlo who did a really nice piece that, you know, went around all the venture community yesterday just about, you know, playing the game at the top of the cycle.

你怎么考虑这件事，当音乐停止的时候，确保自己有个椅子坐。

And how do you think about it when, you know, when the music stops, be sure you have a chair.

这篇文章值得一读。

It's worth reading.

我明白你的意思。

I get you.

但恕我直言，Menlo在大多数轮次里出的价都是所有人中最高的。

But with the greatest respect, Menlo have paid the highest price of everyone on most rounds.

所以他们才在放音乐啊。

That's why they're playing the music.

正因为如此，但我觉得不是这样。

That's why but I think no.

我觉得他会说，他会这么说。

I think that what he would say is I think he would say that.

他会说，我们非常明智地看。

He would say, we very wisely look.

我觉得那真是一篇很好的分析。

I I think it was a really good analysis.

我不是故意要岔开话题，但基本上现在有两家主要玩家了。

I didn't mean to take us off, but he's basically there's two players now.

有些玩家是在用赢来的钱在玩，他把自己归到了那一类。

There's players who are playing with the house money, he put themselves in that category.

他们在Anthropic上做得太好了，可能有点得意，接下来可能会很激进。

They've done so well on Anthropic, and they're probably feeling a little happy, and they're probably gonna be aggressive.

然后他说，还有一些玩家没有参与早期轮次，现在正在拼命追赶。

And then he said there's some players who didn't do that didn't do those early rounds and are now playing to catch up.

他真正想说的核心观点是，在那些兴高采烈的人和那些慌得要命的人之间，有很多人写支票的时候心里都带着恐惧，怕错过机会。

And the point he was making, the macro point he was making is between the giddily happy people and the and the terrifyingly desperate people, there's a lot of people writing checks with fear in their heart of miss of FOMO.

这就回到了你刚才说的那个点。

And it gets back to your comment.

某种程度上，规模变大有一部分是数学上说得通的。

Some part that's why it's always some part of the increase in size is justified by math.

但说实话，我觉得有一部分是因为怕错过下一个Cursor，纯粹是FOMO在驱动。

But let's be honest, I think some part of it is justified by FOMO and the fear of missing the next cursor.

这倒也没什么，但这种东西说变就变。

And that's fine, but that's the kind of thing that can change on a dime.

这就是关键所在。

And that was the insight.

你反思这一点时，会改变你的投资方式吗？

Does that change how you invest when you reflect on that?

罗里，我应该从中得到什么启示？

What should I take from that, Rory?

正在听的LP和GP们应该从中得到什么？

What should LPs listening, GPs listening take from this?

问得好。

It's a good question.

再说一次，我总是处于中间立场。

And again, I'm always in the middle.

我觉得有时候积极进取确实会有回报，但你不想让自己在那么多高价交易上投入过深，以至于一旦经济下行，你就撑不住。

I think you have to say at times leaning in can pay out, but you don't want to find yourself so lent in on so many deals that are so high priced at such a, you know, on sort that if a downturn comes, get, you know, you just can't survive it.

投资往往不是基于规则的事情，不是只做A，而是通常要做一些A，也要做一些B，关键在于组合的平衡。

As often is the case, investing is not a rules based business, do A and only A, it's typically do some A, but some B and the mix is everything.

我觉得对我们来说，就是要保持稳定的节奏，做的事情大体不变，同时要认识到，你知道，时机真的很难把握。

I think for us, it's a consistent pace, broadly the same stuff and recognize that, you know, it's super hard to time that.

但至少你得清楚自己承担的风险，这是最基本的。

But be cognizant of the risk you're taking as the bare minimum you have to do.

我这么说不是不礼貌啊。

Is that I'm I'm not being rude to say.

我是说，

I'm I'm going for

你在这儿，但说真的，你冲我来吧。

you here, but fucking you go for me.

要清楚自己承担的风险。

Be cognizant of the risks you're taking.

真的吗？

Seriously?

说实话，我觉得那段话挺吹牛的。

Like, I thought that piece was a blowhard piece, if I'm honest.

小心音乐停下来。

Be careful of the music stopping.

要留心。

Be cognizant.

再说，你们已经付了

Again, you guys have paid

钱。

up money.

这就是下一轮的目的。

That's what the next one's for.

为什么需要小心音乐停下来？

Why do need be careful the music stopping?

你们每十八个月就募集一次基金。

You raise a fund every eighteen months.

我们都有机会重来一次。

We all can get a mulligan.

如果我这次没有分成，那就没有分成。

If I don't make any carry, don't make any carry.

我下次再赚回来。

I make it on the next one.

没关系。

It's alright.

无所谓。

Doesn't matter.

但是，我，我，我刚想到一点。

But the but I I I just got a point.

我觉得那不是

I don't think that that's

我不是说那就是问题所在，但我的意思是，从那里能得到的收获并不多。

I'm not saying that is it, but what I'm saying is there's just there's not that much to take away from that.

要注意音乐停下来。

Be cognizant of the music stopping.

谢谢。

Thanks.

我同意你的看法。

I'm with you.

关于哈利，我同意你的看法。

I'm with you on Harry.

这有点居高临下的感觉。

It it it it it's a little condescending in its own way.

哦，谢谢。

Oh, thank you.

我没意识到人工智能淘金热有一天会结束。

I wasn't aware that the AI Gold Rush might end someday.

说得好。

Good point.

我整个周末都沉浸在Jev里了。

I so lost in Jev all weekend.

我没抓住重点。

I missed the point.

你说得对。

You're right.

我没抓住重点。

I missed the point.

来自那些赚得盆满钵满的人，我为他们感到高兴。

From the guys who made out like Bandits, which I'm thrilled of.

他们是非常棒的人。

They're they're awesome people.

他们可能还行吧。

They may okay.

看这个。

Watch this.

他们之所以能大赚一笔，是因为在大家还犹豫不决的时候，他们果断出手了。

The the big they made out by Bandits by being aggressive at a time when people were still uncertain.

所以，那些交易的价格，虽然绝对值很高，但相对于实际进展，其实反映了很大的不确定性，对吧？

And therefore, price of those deals, while high in absolute terms, reflected a fair amount of uncertainty relative to traction, right?

同样的交易今天做，价格可能要高上三四倍。

The same deal today would probably be three or four X higher.

这里的关键点是，如果你只对自己说，四年前那笔交易成功是因为那样的风险回报比，而今天再做同样的交易，因为定价的原因，回报可能只有四分之一，那你至少应该停下来想一想，而不是盲目地说，X成功了，所以所有看起来像X的交易也都会成功。

And it's simply a point, if all you say to yourself is that deal that worked four years ago had that risk return profile, and underwriting that same deal today probably means I'm gonna get four times less return because of where pricing is, you should at least pause and think rather than blindly saying x worked, Therefore, all the other deals that just look just like x will work also.

我从来没觉得这有什么明显的。

I never thought it fucking obvious.

哈里，总的来说，我最喜欢的一句话，我要引用一下，你冲我来吧。

Harry, in general, one of my favorite quotes and I'm gonna take you you're coming for me.

我也会冲你来的。

I'll come for you.

巴鲁克，我想我之前引用过这句话。

Barney Baruch, I think I quoted this before.

如果引用过了，我很抱歉，但他是二三十年代华尔街的金融家。

And I'm sorry if I did, but he was a Wall Street financier in the twenties and thirties.

他说，如果你每天照镜子，对自己说二加二等于四，你大概能避免很多错误。

He just said, if every day you look in the mirror and you say to yourself, two and two makes four, you probably could avoid a lot of mistakes.

就是明摆着的事。

Stating the obvious.

通常让你栽跟头的，都是那些一直很明显的事，而你偏偏选择视而不见。

Typically the things that bite you in the ass are things that are actual obvious all along, and you just chose to ignore them.

所以，这确实很明显，但你也得把它考虑进去。

So, it is obvious, but, you know, you have to take into account.

那你对此怎么处理？

What do you do about that?

比如说，我给你举个例子。

Like, I'll give you an example.

2021年，我做了一笔投资。

In 2021, I made one investment.

我做了种子轮，刚刚以23亿的估值完成。

I did the seed rounded owner, just closed at 2,300,000,000.

对我来说，这是一笔相当大的投资。

It's a pretty large position for me.

好的。

Okay.

我们经历过高潮和低谷，创始团队很棒，2021年全年就做了一笔交易，那也是我唯一做的一笔。

Now we've had ups and downs, great founder team, one deal in all of 2021, only deal I did.

你现在也可以做同样的事情。

You could do the same thing right now.

你可以说，未来四年，我只打算做一两笔交易。

You could say it for the next four years, I'm going to do like one or two deals.

就像，我必须感觉到市场有点不对劲。

Like, I have to have a little dislocation in the force.

必须是这样。

It's gotta be this.

必须是那样。

It's gotta be that.

而且，这就是你如何应对那些椅子即将结束的方式。

And that that's how you take that the the chairs are gonna end.

否则，你就得部署基金。

Otherwise, you you just gotta it's you gotta deploy the fund.

那是别人的钱。

It's other people's money.

你得让它运转起来。

You gotta put it to work.

或者，可能有一个非常合理的假设，嘿。

Or there could be a really rational assumption that, hey.

音乐很快就会停止，我们将经历StackRank，从中获得流动性，并且现在就要主动卖出头寸。

The music's gonna stop pretty soon, and we're gonna go through StackRank where we can get liquidity from and really be proactive in selling positions now.

好的。

Cool.

嗯，那是一种情况。

Well, that's one thing.

这个问题挺有意思的。

And that's that's an interesting question.

但假设你想保守一点，

But let's say you wanted to be let's say you wanted to be conservative.

你会怎么做？

What do you do?

投资一家估值五千万美元、年增长60%的公司，指望它能重新加速，估值达到营收的三倍，然后卖给Bending Spoons？

Invest in the company at $50,000,000 growing 60 percent a year and hope it reaccelerates and is worth three times revenue and sells to Bending Spoons?

我是说，有什么备选方案吗？

I mean, what's the plan B?

Bending Spoons每年看一千个项目，最后也就做两到四个。

Like, Bending Spoons looks at 1,000 deals a year and still only does two or four.

那些更保守的公司，以低于AI行业的增速复利增长四十年，我该把它们卖给谁？

Where am I gonna sell these more conservative companies that are compound for at at at sub AI rates for forty years?

我真不知道谁会买这些公司。

I just don't know where the who's gonna buy them.

如果这些资产有市场，那就这样吧。

If there is a market for these assets, so be it.

我喜欢那些安静复利的公司。

I love the quiet compounders.

只是现在没有市场了。

There's just no market for them anymore.

说实话，我不太同意说安静复利的公司到了一定规模就永远没有市场。

I'm not sure I agree that there always won't be a market for quiet compounders above a certain scale, to be clear.

但我觉得利润

But I think profit

可能会回来

Could come

几百什么？

a couple 100 what?

可能会回来，但现在感觉市场很薄。

Could come back, but right now it feels pretty thin.