20VC：英伟达盛宴：收购Poolside并投资Mercor和Perplexity | Anthropic假设30万亿美元收入 & OpenAI确认IPO | 为何客户服务、国防和机器人领域存在泡沫

远不止这个数。

Way more.

你觉得是

Do you think it's

哦，我知道远不止，因为Serge那边已经做到三到三点五了。

Oh, I know it's way more because Serge are doing three, three and a half.

Mercor做到二点五。

Mercor are doing two and a half.

Handshake做到一。

Handshake are doing one.

Micro One做到零点五。

Micro One are doing half.

但你说得对，这个问题。

But the question you're right.

所以我的观点是，这取决于今年这一千亿美元的开支是否增长到四千亿或五千亿，如果是的话，那你说得对。

So my point is, so it's predicated on if that $100,000,000,000 of spend this year goes to 400 or $500,000,000,000 then you're right.

那5%就是250亿美元，分给四五个公司。

Then 5% of that is $25,000,000,000 divided four or five ways.

我觉得你拿不到前沿模型市值的10%，但每年还是能有200亿美元的稳定收入支出。

I don't think you get 10% of the market cap of the frontier models, but you still get a healthy $20,000,000,000 a year revenue spend.

再分成三份。

Divide that up three ways.

然后就像Jason说的，其实最关键的问题就是Jason问的那个，就是你给这个收入定多少倍估值？

Then as Jason said, it actually turns out the most important question is the one Jason asked, which is what multiple do you attach to that?

你是按AI的倍数来估，还是按低毛利率的倍数来估？

Do you attach the AI multiple or do you attach the lower gross margin multiple?

这才是真正决定结果的因素。

That's really what will swing it.

它把成本大幅补贴了。

The radically subsidized its cost.

我忘了Cursor最初的价格，好像是每月200美元不限量使用。

I forget what cursor cost initially, $200 a month for unlimited use.

对。

Right.

但它没有，它没有，它没有任何办法来守住那个优势。

But it didn't have, but it didn't, it had no way to defend that.

对。

Right.

所以后来它不得不限制使用，然后不得不停止这项服务，然后不得不开发自己的模型，然后不得不做，不得不做。

And so then it had to cap it for and then it had to stop doing it and then it had to develop its own models and then it had to do it had to do it.

但风险投资的逻辑是，这些孩子是史上最聪明的一批。

But the VC ism is these are some of the smartest kids ever.

他们有很强的市场地位。

They have a strong market position.

他们正顺风顺水。

The wind is at their sails.

这些孩子会想出办法的。

The kids will figure it out.

他们知道你能获得高毛利，而且基本上他们也做到了，还达到了600亿美元。所以这对像Mercor这样我们曾经取笑过的模型来说，是一种解脱，对吧？

They know you get a gross margins and more or less they did, and they got to the $60,000,000,000 And so it's kind of freeing for models like Mercor that we made fun of, right?

好吧，这是一个商品化的低利润业务。

Okay, this is a commodity, low margin business.

但这些年轻人正在摸索出办法，老兄。

But the kids are figuring it out, man.

这些年轻人正在摸索出办法。

The kids are figuring it out.

所以全力以赴吧。

So guns a blazing.

对。

Yeah.

不，我觉得这里面确实有点生存的意味。

No, I think though there is a bit of survival.

首先，我同意你的看法。

First of all, I agree with you.

而且我觉得，作为风险投资人，当你投资一个毛利率有问题的项目时，对于某些赌注，理性地说，毛利率会随着时间慢慢好转。

And I think it gets to the when as a venture investor, you're investing in something that has trouble in gross margins, it is rational to say for certain bets, the gross margins will come right over time.

你可以拿Kurse作为例子。

And you add Kurse as an example of that.

为了完整性起见，我不知道为什么我今天这么悲观，但并不是每个负毛利率的公司都能成功。

Just in the interest of completeness, I don't know why I'm the Debbie Downer today, not every negative gross margin company makes it.

最终，我认为你想要的，高毛利率的公司才是最好的投资。

And in the end, I think you want and strong gross margin companies are the best investments.

所以不要只看那些从负毛利率开始最后成功的案例。

So don't just look at the sample of deals where you started with negative gross margin and it all worked.

我能想到很多交易，包括我们自己做的，一开始毛利率很差，最后还是很差，就是判断错了。

I can think of plenty of deals, including some we've done, where you started with shitty gross margins and you ended with shitty gross margins and you were just wrong.

Jason，我觉得你说得对，负毛利率并不是不做交易的理由，显然不是。

I think, Jason, to your point, the thing is having negative gross margins not a reason, clearly not a reason not to do a deal.

如果财务数据各方面都很完美，他们就不需要风险投资了，因为他们已经盈利了。

Mean, if every part of the financials were pristine, they wouldn't need venture capital because they'd be profitable.

问题在于，在哪些情况下，承销大规模改进是合理的？

It's a question of, in which cases is it rational to underwrite massive improvement?

对于基础模型本身来说，情况一直如此。

And it has been for the foundation models themselves.

我的意思是，Anthropic 在一年内从负 91% 的毛利率变成了正 30%。

I mean, Anthropic went from negative 91 to positive 30 in a year.

正如你所说，Jason，对于 Cursor 来说，情况也是如此。

It has been, as you say, Jason, for Kursor.

看看像 Mercor 这样的公司，长期维持的利润率会是什么样子，会很有趣。

It'd be interesting to see what the sustaining long term margins for something like Mercor are.

我不知道他们是否具备像 Cursor 那样的能力来改善这些利润率，因为培训公司只有三到五个大客户，而 Cursor 有成千上万的客户。

I don't know if they have the same dynamics in terms of the ability to approve those margins that, say, Cursor did because the training companies only have three to five big customers, whereas the Cursor has literally hundreds of thousands.

所以我不太确信那些利润率会像 Cursor 那样好，但你说得对。

So I'm not as convinced those margins go as well as the cursor ones, but you're right.

Cursor 确实打破了利润率陷阱。

The cursor beat the margin trap.

还有Rory，就像我在周一例会上对我的合伙人和下属说的，如果一切顺利会怎样？

And Rory, and as I tell my partners and subordinates on Monday meetings, what if it all goes well?

如果一切比我们预期的还要好呢？

What if it all goes better than we expected?

如果一切顺利，我还会挑战我的团队。

What if it all goes and I challenge my team.

如果一切都顺利呢？

What if it all goes right?

对于这些赢家来说，入场价格并不重要。

Entry price for any of these winners doesn't matter.

如果一切都顺利呢，各位？

What if it all goes what if it all goes right, guys?

这是我创造的术语，我注意到业内很多同事都在模仿。

That's And I've coined that term, I've noticed many many of my colleagues in the industry have copied it.

如果一切都顺利呢？

What if it all goes right?

你懂我的意思吗？

You know what I mean?

这又回到了我们常说的那句话：乐观者赚钱，悲观者正确。

It goes back to the optimists make money and pessimists are right, that we often say.

我之所以这么说，是因为我承认自己天生就是个悲观主义者。

And the only reason I say it is because I admit that I can be naturally a pessimist.

所以我其实在努力重新训练自己的思维模式，让自己更多地去那样想。

So I actually I'm trying to learn to retrain the model to do that more.

Jason，我很喜欢“如果一切都顺利呢？”这句话。

And Jason, I love the sentence of what if it all goes right?

前两位已经说过了。

The first two have said it.

我知道。

The I know.

我当然知道。

I know that, of course.

但回到Poolside，还有一个附带观点：即使事情并不完全顺利，但如果你身处一个巨大的市场呢？

But the ancillary point, going back to Poolside, is it turns out even what if it all doesn't quite go right, but you're in a great market?

事实证明，那也没关系。

It turns out that can be okay too.

只要足够多的事情顺利，大家都能过得好。

If enough goes right, you can all do fine.

关于Aaron Srinivas和Perplexity，一直有很多冷嘲热讽和怀疑。

There's been a lot of cynicism, skepticism around Aaron Srinivas and Perplexity.

这家公司确实因为某种原因，人们就是不喜欢它。

It's a company that actually people like to dislike for whatever reason.

我们和他做过节目。

We've done shows with him.

我们是这家公司的小投资者，所以实际上英伟达撑起了我大部分的投资组合

We're a small investor in the company, so NVIDIA is actually propping up most of my portfolio

这周。

this week.

对。

Yeah.

谢谢。

Thank you.

我们的朋友在OpenAI，OpenAI从来都不无聊。

Our friends at OpenAI, never a dull day at OpenAI.

首席财务官Sarah Fry告诉员工，OpenAI将在2027年上市。

CFO Sarah Fry told employees at OpenAI, we will be a public company in 2027.

这时候AI交易就崩了。

This is when AI trades cracked.

我不知道这是不是有趣的时机。

I don't know if it's interesting timing.

说实话，这差不多是我们预料到的。

It's kind of what we expected, to be quite honest.

他们受到来自Anthropic的压力，据报道Anthropic将在未来几个月内上市。

They've got pressure on them from Anthropic who obviously are going public in reportedly the next few months.

莎拉·弗莱对团队说的话有什么让你意外的吗？

Is there anything surprising here about Sarah Fry's statement to the team?

我觉得他们别无选择，只能发表那些声明，我来告诉你为什么。

I think they had no choice but to make those statements, and I'll tell you why.

你看他们第一和第二季度的数据，我现在记不太清了，大概是五十多亿到六十多亿，这相当于季度环比增长18%，如果按复利算，年化增长率会略低于100%。

If you look at q one and q two for them, now I can't remember, it's like 5 point something billion to six point something billion, which was a q on q growth rate of 18%, which would have turned into an annualized growth rate of slightly under 100 depending on compounding.

这意味着他们从去年的125亿美元GAAP收入，今年大概会降到不到30亿。

And it would have mean that when they went from 12,500,000,000 last year of GAAP revenue to roughly probably under 30 this year.

如果第二季度的数字能保持，那显然会让他们远远落后于Anthropic，后者年中已经达到600亿美元的年化运行率。

If that Q2 number was sustained, would put them obviously, A, would put them so far behind Anthropic at 60,000,000,000 run rate midyear.

再说，我们还没看到Anthropic的GAAP数据，但显然他们更大，增长也更快。

And again, we haven't seen gap numbers for Anthropic, but clearly bigger and clearly growing faster.

这对OpenAI来说会非常糟糕，因为任何人一算这笔账就会想，哦，再这样下去两年，你就没戏了。

That it would be terribly bad for OpenAI because anyone would run that math and go, Oh, two more years of this and you're in relevance.

还有Perplexity呢。

You have Perplexity too.

然后第二件事没有发生，但实际发生的是，像博通、英伟达这些公司，原本预期能向OpenAI卖出两千亿美元的芯片，可能会突然想，如果他们增长不快，可能就不需要两千亿美元的芯片了。

And then the second thing that didn't happen, but what happened is all those people like Broadcom, NVIDIA, were expecting to sell 200,000,000,000 worth of chips to OpenAI might suddenly go, maybe if they're not growing quickly, they won't need $200,000,000,000 worth of chips.

所以，如果你只看第一季度的数据，你可能会得出这样的结论。

So that's kind of what if all you had was the Q1 numbers, that was a conclusion you could draw.

我不是说这个结论一定正确。

I'm not saying it'd be correct.

如果OpenAI在第三季度初确实在加速增长，他们别无选择，只能向外界公布这个消息，否则大家都会往最坏的方向想。

If in fact OpenAI is massively accelerating in early Q3, they had no choice but to share that information with the world because otherwise everyone would assume the worst.

他们不会坐视不管，说，好吧，三个月后我再给你第三季度的更新。

What they're not going to do is sit on their thumbs and say, Well, I'll give you a Q3 update in three months.

与此同时，你就只能干等着。

Meanwhile, you should just sweat it out.

所以我确实感觉到他们在非常努力地传达一个信息：第二季度是异常，第三季度在爆发。这从他们对内部团队的讲话和对外发布的内容中都能看出来。

So I really detected a very concerted attempt to tell a Q2 is an anomaly, Q3 is exploding story, you could see it in the comments that they made to their internal team, you can see in the stuff that's coming out.

而且一直有一个大规模重新加速的故事在流传。

And there's been a whole massive reacceleration story.

再说一次，除非你在GAAP数据里看到，否则很难确定。

And again, I know nothing except until you see it in GAAP numbers, it's hard to be sure on it.

但在我看来，推动那个议程的原因是你别无选择，否则你就会被甩在后面。

But to me, the reason for pushing that agenda was you had no choice because otherwise, you were just going be left behind.

如果你能翻倍增长，那已经很了不起了。

If you're going 2x, it's amazing.

在120亿美元的基础上翻倍增长，这确实很惊人。

Growing 2x at $12,000,000,000 is amazing.

但如果你的竞争对手在600亿美元的基础上以8倍或10倍的速度增长，而这种情况持续下去，两年后你的相对市场份额就只有20%或30%。

But if your competitor is growing 10x or 8x at $60,000,000,000 you're staring at relative market share of 20% or 30% in two years if that continues.

所以，Q2的趋势绝对不可能不受质疑，否则OpenAI就无法保持对Anthropic的紧密第二的位置，而他们现在正是这样。

So, there's simply no way the Q2 trend could stand unchallenged and still leave OpenAI as a credible close number two, which is where they are now, to Anthropic.

这就是为什么消息会泄露，因为这对他们来说是生死攸关的。

That's why it leaked because it's existential for them.

说实话，当我看到GAAP收入增长18%的数字时，我惊呆了，因为我回想起我之前说过的话。

I mean, I was stunned when I saw the 18% GAAP revenue numbers, because I go back to my comment.

从某个层面来说，如果我的公司每个季度增长18%，一年翻一倍，哪怕规模只有那个的十分之一，我也会高兴得不得了。

At one level, if I had a company growing 18% Q on Q doubling a year at, forget it, one tenth that size, two and a half, I'd be ecstatic.

但是，相对于预期来说，这样的增长速度会让所有人极度失望，包括那些原本计划以更高增长率向他们出售芯片的人。

But that level of growth relative to expectations would have disappointed massively everyone, including all the people planning to sell them chips to them for a much higher growth rate.

如果Anthropic以2万亿美元的估值上市，OpenAI会以什么价格上市呢？

If Anthropic goes out at 2,000,000,000,000, what price does OpenAI go out at?

我完全不知道，Harry。

I have no clue, Harry.

但最重要的一点是，现在它的估值会比对手低。

But the most important point you have to say is this, it will be lower than the other guys now.

从宏观层面看，就像成绩单一样，数学里有个概念，我记不清具体叫什么了，就是无法精确打分，但可以排序。

At the high level, in terms of report card, there's a concept in math, I can't remember what it is, where you can't do accurate grading but you can rank things.

大局是排名已经变了，你现在是第二名。

The big picture fact is the ranking has changed, you're now number two.

所以，除非你大幅改变发展轨迹，否则你上市的时间会更晚，价格也会比对手低。

So you will go out later then on a lower price than the other guys, unless you change the trajectory massively.

那是什么，我不知道。

What that is, I don't know.

我只是觉得到年底看看OpenAI在平台和企业方面的定位会很有趣，因为年初的时候只有两个选择，对吧？

I just think it'll be interesting by the end of the year to see where OpenAI is positioned in the platform, the enterprise versus Anthropic, because at the beginning of the year there were two choices, right?

大部分情况下就是Anthropic和OpenAI，你有一个默认的选择。

There was Anthropic and OpenAI for the most part, and you had a default choice.

然后人们想要有多模型。

Then people wanted to be to have multi models.

他们想要至少两个，但你通常会使用更便宜的版本。

They wanted to have at least two, but you'd often use the cheaper version.

你可能会用Sonnet和Opus之类的。

You might use Sonnet and Opus or whatever it was.

现在我们肯定想要不同的LLM。

Now we definitely want different LLMs.

每个人都希望自己的技术栈里有多个LLM，不管怎样。

Everyone wants multiple LLMs in their stack, one way or the other.

但如果Anthropic在平台方面排名第一，这很明显，数据摆在那里，Perry说得对。

But if Anthropic is number one in platform, which it clearly is, you can't argue with the numbers, Perrys point.

现在第二名可能是，可能有很多选择，可能有五个选择。

Now the number two could be it could be there could be there's five choices.

所以OpenAI的位置，某种程度上，我们会看到变化很大，各位，但不管数据如何，它可能会越来越弱，因为第二名有很多选择。

So OpenAI position in a sense, and we'll see there's so much change, guys, but it could be no matter what the numbers are, it could be getting perpetually weaker because there's so many choices for number two.

选择太多了。

There's so many choices.

有很多开放权重模型，性能很接近。

There's so many open weight models where the performance is close.

显然我们讨论过OpenRouter。

And obviously we've talked about open router.

我们也讨论过其他路由器。

We've talked about other routers.

如果你想，可以用78个模型，但当第二名有七个选择时，这很难。

You can use 78 models if you want, but it's tough when there's seven choices for number two.

就是更难了。

It's just tougher.

就是更难了。

It's just tougher.

对。

Right.

尤其是如果你是个高端产品。

Especially if you're a premium product.

对。

Right.

那你是说，这对OpenAI的企业价值影响其实要小得多，因为……

And you're saying that the material impact on the EV of OpenAI then is considerably less because of the

我觉得他们压力更大，因为作为第二名，来自开源和开放权重模型的竞争压力比以前更大。

I think there's more pressure on them merely because bad being number two, there's more pressure on you from from open weights and open source competitors than there was otherwise.

你就是在争夺第二名这个位置。

You're just battling out for that second spot.

你没法在价格上竞争，但可以在品牌和安全性上竞争。

You can't compete on price and you can compete on brand and security.

但说真的，你至少得让自己嵌入到每个工作流程里，对吧？

But man, you want to just you want to at least be the be plugged into to every workflow, right?

你至少希望这样，如果你能抓住一个开放的机会，那会很好。

You want to at least have that would be the nice thing if you had caught an open eye.

然后你就得到一个不错的寡头竞争销售团队。

And you just get a nice oligopol bake off sales team.

然后你从Salesforce招一堆人，他们进来搞寡头竞争那套，放个PPT，就讲我们和他们。

Then you hire a bunch of folks from Salesforce that walk in for the oligopol thing and they put up a PowerPoint slide and it's just us versus them.

他们知道怎么卖这个。

They know how to sell that.

就是Anthropic对我们。

It's Anthropic versus us.

你不需要那个疯狂的人，你需要的是政府信任的人。

You don't want the crazy guy who, you want the guy the government trusts.

那就是我们。

That's us.

但是有11个或12个竞争对手在开源平台上跑推理，天哪，这对第二名来说竞争太激烈了。

But with 11 competitors or 12 competitors running on open source, inference on new platforms, man, it's just it's a hyper competitive world for number two.

我觉得，即使他们是那样，我也不认为他们会因为所有开源的东西而被逼到那个地步。

I mean, I think that even if they are I don't think they get pushed into that compared with all the open source.

我觉得没那么严重，但我同意你的看法。

I don't think it's quite as dire as that, but I do agree.

确实有来自下面的压力。

There was a push from underneath from that.

但我觉得回到你问的那个问题，所谓的

But I think to your question on what, quote unquote

罗伊，罗伊，罗伊，抱歉。

Roy Roy Roy sorry.

打断一下。

Just interrupt.

接着刚才那个话题说。

Just going on that.

你知道Vercel开放他们的数据时，开源权重占比68%还在上升吗？

Do you know when Vercel obviously opened their data, 68% open weights increasing?

我的意思是，这显示了浪潮正朝着Jason说的方向移动。

I mean, it shows the tidal wave moving towards what Jason's saying.

同意。

Agreed.

是的，我同意。

I yes.

我认为绝大多数处理的token将是开放权重的。

I think that the vast predominance of tokens process will be open weight.

而且显然，正如我一直强调的，大部分收入仍将来自前沿顶尖模型，因为它们能带来比单纯推理价格更高的价值。

And obviously, as I keep reiterating, the significant majority of revenue will still be frontier state of the art because it can command more value than just the price of inference.

所以我想在某种程度上，Jason，你说得对，因为如果有五六家开放权重公司——我不认为它们是同行——但如果有五六家开放权重公司拉低所有人的毛利率，你基本上就是在试图从封闭前沿模型中抢走毛利率，尤其是如果其中一家开放权重公司现在由NVIDIA资助，那就消除了所有关于中国的旧评论，对吧？

And so I suppose at some level, Jason, it's correct because if there's five or six and I don't see them as being peers, but I think if there's five or six open weight companies dragging down everyone's gross, you're basically trying to steal gross margin away from the closed frontier models, especially if one of those open weight companies is now funded by NVIDIA, so it gets rid of all the old Chinese comments, right?

杰森说得对。

Jason is right.

在那种情况下，一个成本低、功能几乎相当的竞争对手想进入你的行业，你肯定宁愿当第一而不是第二，因为第一可以说，你们就得买我们的，因为我们是第一。

In a situation where a low cost competitor with nearly equivalent functionality is attempting to enter your industry, you would far prefer to be number one than number two, Because number one can say, You got to just buy us because we're number one.

但你说得对，杰森。

But you're right, Jason.

第二就得说，请也买我们的，别买那些便宜货。

Number two has to say, Please buy us as well and don't buy the cheapo guys.

我收回我之前的话，同意杰森的看法。

I'm recanting my position and agreeing with Jason.

在一个你开创的行业里当第二已经够糟了，他们现在就是这样，但要是当第二，还得应付一堆小喽啰，那真是烦透了。

It would be okay bad enough to be number two in an industry you invented, which is where they are now, but to be number two with a whole bunch of ankle biters on top that you have to deal with, that's a total pain in the ass.

这就回到你的问题了，哈里，我来回答你关于2027年他们以什么价格上市的问题。

Which gets back to your question, Harry, I will answer your question on what price do they go out in 2027.

他们2027年能拿到什么价格就什么价格，因为他们等不了了。

They'll go out whatever price they get in 2027 because they can't wait any longer.

就是这么简单。

It's as simple as that.

如果Anthropic今年按他们现在说的规模上市，我无法想象OpenAI会说，我们要埋头苦干，然后2028年再上市，但他们必须上市。

If Anthropic goes out this year at the scale they're talking about now, I can't imagine a world of OpenAI says, We're going to hunker down and get flow, and then we'll go out in 2028, but they have to go out.

所以从某种程度上说，这又是一个命运不由你掌控的评论。

So to some extent, again, it's a destiny outside your control comment.

如果Anthropic估值两万亿，那他们可能拿到一万五千亿。

If Anthropic trades at two, then they might get one and a half.

如果Anthropic估值一万亿，他们可能拿到七千亿，但他们还是会做必须做的事。

If Anthropic trades at one, they might get 700, but they'll do what they have to do.

意思是，有一句很棒的话，那是什么来着？

Mean, there's a great quote in that what was it?

快点。

Come on.

那本书，《风暴》，塞巴斯蒂安·荣格的书，超棒的书。

The book, Storm, Sebastian Junger's book, super book.

但当那些人终于意识到自己要在海上丧命时，对吧？

But when the guys finally realized they're going to die at sea, right?

书里说，如果危险可以描述为没有选择，那他们当时就处于危险之中。

And it said something like, if danger can be described as the absence of choice, they were now in danger.

我觉得对OpenAI来说，你现在看到的是，没有选择的情况开始堆积起来。

And I think for OpenAI, what you're seeing here is the absence of choice is starting to pile up.

其他公司领先了。

The other guys are ahead.

其他公司盈利了。

The other guys are profitable.

没错，CFO说他们不会写唯一的叙事，但在某种程度上，你的叙事正在被别人书写。

Correctly, the CFO says they're not going to write the only narrative, but to some extent, your narrative is getting written.

你必须盈利，因为其他公司盈利了。

You've got to get profitable because the other guys are profitable.

你必须上市，因为其他公司盈利了。

You've to get public because the other guys are profitable.

所以他们现在的选择比一年前少了很多，这也就意味着，2027年我们会上市。

So they have less options than they did a year ago by a lot, and that will translate as in 2027, we're gonna go public.

如果那一年市场稍微低迷，我们就接受现实，继续前进。

And if the markets are slightly down in this year, we'll take our medicine and keep moving.

杰森，你喜欢那本《风暴》吗？

Don't you love that book, The Storm, Jason?

我非常喜欢。

I I loved it.

那是我最喜欢的书。

My my favorite.

作者很厉害。

Super writer.

虽然有点悲伤，但那句话一直让我印象深刻。

Kinda sad, obviously, but, wow, That line always struck with me.

没有选择，这就是你身处危险的标志。

Absence of choices, that's how you know you're in danger.

我觉得，听我说，我们可以进入下一个话题了。

I think, listen, we can move on to the next topic.

另外，我在听Rory讲IPO的时候，还有一件事，可能不太重要，我们可以跳过，但我今天越来越搞不清楚OpenAI的差异化使命到底是什么。

The other thing, when I'm listening to Rory in the IPO, the other thing, maybe it's minor, we can move on, but I'm getting confused today what OpenAI's differentiated mission is.

为什么是OpenAI？

Why OpenAI?

我的意思是，我们都可以看看评估结果，读读RAMP和Rippling的报告，然后把这看作只是一个LLM在支付市场最高价，对吧？

I mean, we can all look at evals and we can read what RAMP says and what Rippling's report says, and we can view this as just an LLM paying top of market, right?

每个工程师付八位数薪水。

Paying 8 figures per engineer.

但这些公司都有非常具体的使命，对吧？

But these were companies on very specific missions, right?

我们开始这个播客的时候，我都不知道OpenAI的使命是什么。

When we started this pod, I don't know what OpenAI's mission is.

我觉得Dario现在很多时候都疯了，显然今天在面试一些人，问他们如果一切归零，是否还愿意加入Anthropic。

I think Dario is nutso a lot of the time right now, apparently interviewing folks today, asking them if they'd happy joining Anthropic if it all went to zero.

这个问题问得挺有意思的，不过确实有点疯狂。

Cool question, actually, but nutso.

我觉得山姆现在作为CEO，亲和力强多了。

I think Sam is a much more approachable CEO now.

他熬过了那些关于奥特曼的骗局风波，现在穿着毛衣，显得更讨人喜欢了。

He got through all the the the scam Altman crap and he's got the sweaters out and he's the more likable guy.

但我真不知道，如今OpenAI有什么与众不同的使命，能让我作为员工或股东愿意为之奋斗。

But I don't know what is the differentiated mission of OpenAI today that I would rally around as an employee, as a shareholder?

还是说，它只是基础设施加一些软件的一部分？

Or is it just a piece of just a piece of infrastructure plus some software?

我就是不明白，跟Anthropic或者现在这些强劲的竞争对手相比，它的使命有什么特别之处。

I just don't know what is special about the mission vis a vis Anthropic or now all these strong competitors.

我真的不知道。

I just don't know.

还是说，它只是一个评估标准？

Or is it just an eval?

就只是评估表上的三行字吗？

Is it just three lines on an eval?

说实话，我真的不知道。

I honestly don't know.

而这些组织从一开始就是非常以使命为导向的。

And these have been very mission based organizations from inception.

对吧？

Right?

这是我们这辈子见过的最多的。

The most we've ever seen in our lifetimes.

对吧？

Right?

这些使命，这些疯狂的使命。

This mission these crazy missions.

对我来说，最令人惊讶的是ChatGPT拥有的消费者品牌，以及它在受众中的渗透率，这是其他大语言模型无法比拟的。对世界上大多数普通民众来说，AI就是ChatGPT。

For me, the the astonishing thing is the consumer brand that ChatGPT has and the penetration it has in audiences that no other LLM has, to most of actual the general population in large majorities of the world, AI is ChatGPT.

我真的很佩服山姆，他没有说我们要成为下一个谷歌。

I am in awe that Sam is not going, we are the next Google.

我们的业务会是广告，而且我们会前所未有地看到杰文斯悖论，当我们有一个真正与消费者合作的消费硬件设备，并且你看到使用量的时候。

Our business is gonna be advertising, and we're gonna see Jevan's paradox like never before when we have a consumer hardware device that actually partners with consumers, and you see usage

但那是计划。

But that was the plan.

他只是被超越了。

He just got outraced.

他只是被甩在了后面。

He just got lapped.

那是个好计划。

It was a good plan.

而且Sora也在其中，还有很酷的视频。

And Sora was in it and cool videos.

只是对于有限的算力来说，这不是最高的投资回报率。

It just was not the highest ROI for limited compute.

只是那不是最好的用途。

It just wasn't the best use of sentence.

在那之前，大家都是在瞎扯，但Jason说到了点子上。

Up until then, it was all babbling stuff yourselves, guys, but Jason nailed it.

结果发现，我是说，OpenAI这个名字，以及ChatGPT这个名字，就是Evan联想到AI的代名词，对吧？

It turns out I mean, OpenAI is the name, and ChatGPT is the name Evan associates with AI, right?

他们拥有巨大的消费市场份额。

They have massive consumer market share.

而且在某个时候，你凭直觉会告诉自己，这会变成一门相当大的生意。

And at some point, intuitively, you say to yourself, that turns into a pretty big business.

至于具体有多大，我们可以之后再讨论。

We can circle back on how much.

但Jason是对的。

But Jason's right.

结果发现，我再说一遍，结果发现那并不是算力的最高回报用途。

It turns out, again, I repeat, it turns out that it wasn't the highest ROI used for compute.

而且我觉得，有时候你回头看结果，才会发现，哦，其实只有一句话是重要的。

And I believe this is sometimes when you look back at outcomes, you realize, oh, only one sentence matters.

如果你真的把这句话听进去了，你就已经发财了。

And if you just internalize that sentence, you've been rich.

我常举的例子是，如果你在雅虎董事会待了十年，这十年里你一直喊‘一切都在于搜索’，你就能帮他们赚一千亿美元。

The example I always use is, if you've been on the board of Yahoo, and for ten years, all you've done is screamed, it's all about search, you could have made them a $100,000,000,000.

现在，Anthropic 得到的启示就是‘一切都在于代码’。

Today, what Anthropic got is, it's all about code.

就是这句话。

That's the sentence.

这是唯一重要的一句话。

That's the only sentence that matter.

因为具体到 Jason 说的，编程是采用最快的市场，是回报率最高的市场，是金矿。

Because to make it concrete in what Jason said, coding is the fastest adopting market, it's the highest ROI market, it's the mother load.

就这么简单，OpenAI 专注在这里，而 Anthropic 专注在那里。

It's as simple as OpenAI was focused here and Anthropic focused there.

但结果可能会更糟。

But it could end up being even worse.

我们看财报就知道了。

We'll see in the financials.

但OpenAI和Anthropic的消费者业务的问题是，它们对重度用户来说补贴力度很大，对吧？

But the problem with the consumer businesses for OpenAI and Anthropic is they're for power users are massively subsidized, right?

你花200美元在Anthropic，或者100多美元在OpenAI，就能拿到价值8000到12000美元的token。

You can spend $200 on Anthropic or 100 and some odd dollars on OpenAI and get 8 to $12,000 worth of tokens.

在ChatGPT还只是平台概念验证的时候，这没问题。

And that's fine for when ChatGPT was a proof of concept for a platform.

我们总说OpenAI是家消费者公司，但它一开始并不是。

Like we talk about OpenAI being a consumer company, but it's not where it started.

ChatGPT只是个概念验证应用。

ChatGPT was just a proof of concept app.

对。

Right.

克劳德也是同样的道理，但对Anthropic来说效果要好得多。

And Claude was just the same, but it worked much better for Anthropic.

对。

Right.

Anthropic可以在某些消费者身上亏个几千美元，这不会对他们造成太大影响。

Anthropic can lose, you know, a couple thousand dollars on some consumers and it won't impact them.

但对OpenAI来说就更难了。

But it's tougher for OpenAI.

这是个糟糕的生意，把价值一万美元的代币以两百美元卖出去，这可能是我们这辈子见过的最差的商业模式之一。

It's a crappy business that consumers selling $10,000 worth of tokens for $200 is one of the worst business models of our lifetimes.

对吧？

Right?

如果只有这个业务，这些公司早就完蛋了。

If that was the only business, these guys would be dead in the water.

这生意确实挺糟糕的。

It's a pretty bad business.

我的意思是，你看，另一方面，我想说谷歌是地球上最好的企业之一，因为服务成本很低。

I mean, look, on the other hand, I'm just gonna argue Google is one of the best businesses on the planet because the cost to serve is low.

而且很可能，我就直说了，我不想偏离对OpenAI又爱又恨的话题，但很可能在未来五年，如果你能继续成为AI领域的主导消费品牌，随着服务成本下降，随着你管理好服务成本，随着你建立起广告业务，我完全可以看到一个业务规模达到谷歌消费业务正负50%的企业，也许在未来十年内。

And it may well be in the just to put it out, I kinda I don't wanna kinda veer from, you know, OpenAI love them, hate them to, yeah, it may well be over the next five years that if you can continue to be the dominant consumer brand in AI, as the cost to serve goes down, as you manage that cost to serve, as you build an advertising business, I could totally see a business plus or minus 50% to the same size as the Google consumer business, maybe over the next decade.

所以，这可不是小事。

So, not like it's nothing.

只是说，在编程采用方面的S曲线非常高。

It's just that the S curve in the adoption sense for coding was super high.

ChatGPT在消费者采用方面的S曲线非常高，但不幸的是，消费者的付费意愿几乎为零，相对为零，而在编程方面，付费意愿很高。

The S curve in adoption for consumers was super high for chat GPT, but unfortunately, the propensity to pay was almost zero, relatively zero, whereas on the coding side, the propensity to pay is high.

我就要直说了，因为就在我们刚才聊天的时候，我看到了这件事成真，正好可以拿来吐槽另一边。

And I'm just going to say it because I just saw the thing come true as as we were talking here to dump on the other side.

我就说Anthropic在《华尔街日报》上，他们相信他们的TAM是30万亿美元。

I just say Anthropic in the Wall Street Journal, they believe their TAM is 30,000,000,000,000.

然后我对自己说，哦，我记得每个人在做得特别好的时候都会有点自欺欺人。

And then I say to myself, oh, I remember everyone when they're doing really well gets slightly delusional.

你的市场总规模，你知道的，就是整个美国的GDP。

Your TAM is, you know, the entire US GDP.

多谢多谢，拉里。

Thanks thanks a bunch, Larry.

知道了。

Good to know.

这是每年这个时候都会出现的那些夸大其词的言论之一。

It it's one of those overreaching statements that you get at this time of the year.

韦恩，他需要这个数字足够大。

Wayne, he needs it to be a pretty big number.

他只有2%。

He's got 2%.

所以

So

那是假的。

That's fake.

如果你声称的市场规模跟美国经济一样大，那标准确实很高。

If you're claiming a TAM that's the size of The US economy, yeah, that's a high bar.

我们就按这个来吧。

Let's just go with that.

要不要往上一层，聊聊公开市场上很多核心AI公司的实际表现？

Do wanna go up a layer into public's and actual performance of a lot of the core AI names falling off?

自四月以来最差的表现，蒸发了8200亿美元的价值，或者我们可以回到Hugging Face可能被收购的话题，以及这对一个中立平台突然可能变得有偏见意味着什么。

Worst run since April are raising 820,000,000,000 of value, or we can go back down to Hugging Face potentially being bought and what that does in terms of a neutral platform suddenly becoming potentially biased.

咱们别聊Hugging Face了，因为我不够聪明，理解不了为什么有人会花130亿美元买它。

Let's not do Hugging Face because I'm not smart enough to understand why anyone would pay 13,000,000,000 for it.

我就是搞不懂。

I I just don't get it.

我就是不够聪明，

I'm just not smart enough

理解不了。

to understand

为什么它值130亿，但，我猜吧。

why it's worth 13,000,000,000, but, I I guess.

这其实有点过度了，坦白说，和Poolside还有OpenRider那事儿一样，我觉得这些都连在一起。

It's kind of a muchness with, frankly, the poolside and the OpenRider thing, which is everyone's I I think it all goes together.

大家都在看一个世界，我可能觉得有点不切实际，但OpenAI和Anthropic声称的TAM比整个美国GDP还大。

Everyone's looking at a world where, I might think it's delusional, but OpenAI and Anthropic are claiming TAMs that are larger than the entire US GDP.

如果我是美国一家IT公司的老板，我会说，让我搞清楚，他们声称要拿走一切。

And if I'm running an IT company in The US, I'm saying, let me get this straight, they're claiming they're going to take everything.

天哪，我得赶紧弄点东西，所以我想在不是封闭源前沿模型的模型上保持相关性。

Shit, I better get me something, and therefore I want to be relevant in models that aren't closed source frontier models.

所以你就看到了企业拥有自己模型的整个趋势，从开放权重模型开始，然后采用，而Hugging Face就是你能访问大量这些模型的地方。

So you get in this whole trend for enterprise having their own models, starting with open weight models and then adopting and Hugging Face is the place where you can access loads of those models.

目前收入相对较少，但估值却高达150亿美元。

Revenue was relatively light at the moment, well to a $15,000,000,000 outcome.

大概1.5亿美元吧。但如果你想想，如果一家IT公司想建立对封闭权重前沿模型的制衡，他们可能想拥有的战略资产，这绝对是个超级有趣的资产。

Think it's roughly $150,000,000 But if you think of strategic assets that an IT company might want to own if they were trying to build a counterbalance to the closed weight frontier models, this would be a super interesting asset.

不过，不知道这个价格能不能卖出去。

Now, don't know if it'll sell at that price.

我完全搞不懂。

I can't make head or tail.

我同意你的看法，Jess。

I'm with you, Jess.

这个价格我完全摸不着头脑。

I can't make head or tail the price.

但如果你想想资产，如果你是微软、IBM，你会很想拥有它来保持竞争力，这绝对是一个，绝对是一个。

But if you think about assets, if you were Microsoft, IBM, you'd love to own to be relevant, this would be one, absolutely one.

我不太明白，但凭直觉，现在，这是OpenRider 2，现在正是从开放权重浪潮中获益的好时机。

I don't get it other than it seems intuitively to me if right now, and this is OpenRider two, right now is the moment in time to benefit from the lift of open weights.

对吧？

Right?

需求太强劲了。

The demand is so strong.

就像伊洛德·吉尔说的，现在手上有AI资产就赶紧卖。

Just like Ilod Gill said, sell if you have an AI asset right now.

对吧？

Right?

我觉得如果你有产品正受益于向开放权重的转型，那现在就是最好的出售时机，前后差不多九十天。

I think even better if you have a product that's benefiting from the transition to open weights, there can't be a better time to sell than plus or minus ninety days from today.

这就像是一个阶段性的转变，你的数据会看起来非常亮眼，就像他们说的，OpenAI在1.5亿美元的时候，增长率是15%。

It's just a phase transition and your numbers are going to look amazing for like they said OpenAI was growing 15% at 150,000,000.

对。

Right.

有消息说，Stripe收购它的时候，如果那15%的增长率能加速并扩大规模，那真是惊人。

The information said when Stripe bought it and breathtaking if that 15% accelerates and scales.

对。

Right.

但也可能不会。

But it might not.

对。

Right.

现在这一刻，突然之间，你知道，开放权重和这些模型从实验变成了主流，卖了吧，宝贝。

This is the moment when all of a sudden, every, you know, open weights and these models went from experiments to mainstream sell, baby.

如果你能拿到70亿、100亿美元以上的估值，我可能会卖。

If you can get north at $7,000,000,000 $10,000,000,000 I'd probably sell.

我的意思是，就算只有15倍，我也可能会卖。

Mean, even if I only got 15x, I'd probably sell.

这不会持续下去的。

It's not going to last.

这个转型不会再是转型了。

This transition's not going to be a transition anymore.

我同意。

I agree.

很有可能，记住，你得知道Hugging Face的创始人也有超越财务回报的使命目标，所以他们可能选择不卖，因为他们可能对此有顾虑。

It may well be that, remember, you have to know the founders of Hugging Face also have mission objectives beyond financial enrichment, so they may choose not to sell because they may have angst about that.

但是，从估值高峰的角度来看，这一切始于萨蒂亚的评论，说每个企业都需要拥有自己的知识，不能把它交给前沿模型，对此你只想说，萨蒂亚，这不是废话吗。

But yes, from a valuation peak perspective, it started with the Satya comments on every enterprise needs to have its own knowledge and not give it up to the frontier, to which you want to say, No shit, Satya.

好吧，谢谢你三年来资助OpenAI。

Well, thanks for funding OpenAI for three years.

但是，是的，IT界的每个人都醒悟了，意识到这两个前沿模型可能会抢走他们大量的市场空间，每个人都在说，我们最好有个不同的故事，企业们这么说，Palantir也这么说。

But yes, everyone in IT has woken up and realized that these two frontier models could steal a lot of their TAM, and everyone is saying, We better have a different story, and the enterprises are saying it, Palantir is saying it.

你说得对，杰森。

You're right, Jason.

如果你是开放权重模型的赋能技术，现在就是巅峰时刻。

If you are an enabling technology for open weight models, now is peak moment.

关于Hugging Face的事，我不觉得现在出手还早。

On the Hugging Face thing, I don't think it's early.

对于罗里的观点和使命问题，我想说一点小事。

For what it's worth to Rory's point and the mission thing, I would say one small thing.

如果有人出于任何原因收购了Hugging Face，交易条件必须是你不去动它，因为一旦你动了，就会破坏它。

If someone does buy Hugging Face for whatever reason, the deal has got to be you don't touch it because if you touch it, you break it.

对。

Right.

如果你推广它，那就会变成一个大得多的TMPN挑战。

If you promote, it's a much bigger version of the TMPN challenge.

如果它变成OpenAI的商业产品，TBN就没有价值了。

If it becomes an OpenAI commercial, TBN has no value.

我知道可能只有我们会把TBN和Hugging Face相提并论。

I know we're probably the only people that are going to compare TBN to Hugging Face.

但如果你动这个有一万个模型的平台，哪怕只是在顶部放个小广告，你也会毁了它。

But if you mess with this marketplace for 10,000 models, you know, even if even if you put a little ad at the top, you destroy it.

所以被收购从来都不是什么好事。

So it's never fun to get acquired.

但我几乎可以肯定，如果有人真的花30亿，更别说130亿，他们95%会在24到36个月内不去动它。

But I can I'm almost confident if anyone actually spends 3,000,000,000, let alone 13,000,000,000, they're gonna 95% leave it alone for twenty four to thirty six months.

对吧？

Right?

为什么Hugging Face可能会像TBPN那样遭殃？

Why Hugging Face could suffer like TBPN?

然后展开说说。

And then lay it out.

这太冷门了，因为Jason说得对，除了我们，没人再关注这个了。

It's I I I it's too obscure because true Jason correctly says no one other than us is tracking that anymore.

你知道吗，我最喜欢的公开消息是Cat和Griffin的Citadel把Leopold Aschenbrenner那本预售书的80%都撤了。

You know, my favorite news from the public was Cat and Griffin's Citadel unwinds 80% of Leopold Aschenbrenner's for sold book.

我就想，天哪，千万别跟Cat Griffin对着干。

I'm like, man, never fight with Cat Griffin.

他总能赢。

Man will come out on top.

已经卖了80%了。

80% sold already.

再说，我觉得这也不意外，但是。

Again, I don't think it's surprising, but.

不是。

No.

并不是。

It wasn't.

我的意思是，是的，他们不是那种长期持有这类资产的机构。

I mean, yes, they're not in the business of holding those kind of assets long term.

你知道，他们是做市商和短线交易者，而这是一笔很棒的短线交易。

You know, they're a market maker and a short term trader, and this was a great short term trade.

没错。

Exactly.

有趣的是，大约一两个月前，他好像开始增加他们的投资选择，进行更长期、多月的持有。

It's funny because about a month or two ago, he had kind of did something about starting to add to their kind of stable of investment options, longer term, multi month holds.

显然，那只是一个概念性的想法。

And clearly, that was a conceptual idea.

但事实证明，当你以低于市场价10%的价格买入一堆资产，然后市场又因为资产从弱手转移到强手而额外上涨5%或10%时，正确的做法就是拿钱走人。

But it turns out when you buy a bunch of stuff at 10% below market and then that market jumps an extra 5% or 10% just because you've put the assets out of weak hands into strong hands, then the correct response is to take a money and run.

看，这确实很棒。

Look, it's great.

只是你不可能每个月都这么做。

It's just you can't do that every month.

所以对我来说，这没那么了不起。

So to me, it's not that impressive.

这确实非常了不起。

It's incredibly impressive.

只是不太有意思，因为你只需要有足够的资产负债表和胆量去等待，就像沃伦·巴菲特每隔几年做的那样，抓住这些特殊机会。

It's just not interesting because you just have to have the balance sheet and the Cajones to wait so that every couple of years, Warren Buffett used to do, you could pounce on one of these special situations.

对吧？

Right?

对。

Yeah.

每隔几年，就会有人对公开市场中的杠杆运作方式感到困惑。

Every few years, someone gets confused about how leverage works in the public markets.

他们搞砸了，而你已经准备好定价和买入。

They screw up and you're ready to price and buy.

而且，没错，在你那门依然赚钱的好生意之上，每隔三年，就有某个傻瓜白送你三四十亿美元，你礼貌地收下，然后投到迈阿密的房地产里。

And, yes, on top of your nice business, which is still earning good money, every three years, some idiot gives you 3 or 4,000,000,000 of free money, and you politely take it and put it into real estate in Miami.

当肯真好。

It's good to be Ken.

对。

Right.

完全正确，杰森。

Exactly right, Jason.

我就是这么理解的。

That's how I read it.

里面没什么让人意外的。

There's nothing surprising in there.

而且，顺便说一句，这又回到了英伟达以及他们所有的投资和供应商融资上，因为在两种情况里，无论是情境意识还是英伟达，关键在于，当你处理资金杠杆时，你不仅要在长期上判断正确，还得在每一步都判断正确。

And, you know, by the way, it does get to the it's kind of I'm gonna circle back to NVIDIA and all their investment and vendor financing because in both ways, both of situational awareness and NVIDIA, the the here is when you're dealing with money leverage, you don't just have to be right in the long term, you also have to be right every step along the way.

如果你没有杠杆，你只需要长期看对并持有就行了。

If you don't have leverage, all you have to do is be right in the long term and hold.

这很可能是一个长期看对的局面，但当你加上四比一的杠杆，你就必须每一步都走对。

And it's probably a situation where it was right in the long term, but when you put four to one leverage on it, you have to be right every step along the way.

供应商融资也是一样的道理。

And the same is true about vendor financing.

只要你做的是卖芯片的生意，你只需要最终人们想买芯片就行了。

As long as you're in the business of selling chips, all you have to do is write in the end, people want to buy chips.

如果你选择用这些芯片做抵押放贷，那你基本上就是说你必须一直都对。

If you choose to lend against those chips, then you're basically saying you've to be right all along.

公司明年必须增长，明年就得还债。

The company's got to grow next year, they've to pay their debt back next year.

所以杠杆就会带来这样的影响。

So leverage does that.

它提高了看对带来的回报，也提高了时刻看对的重要性。

It raises the return from being right and raises the importance of being right all the time.

而情境意识正好是另一面。

And situational awareness just got the other side of that.

因为记住，他在生意上并不追求每次都判断正确。

Because remember, he doesn't try in his business to be right all the time.

他并不是在做五年期的押注。

He's not trying to make five year bets.

他就像是在说，股票今天比昨天贵了10%。

He's like, stocks are worth 10% more today than yesterday.

我们应该卖掉。

We should sell.

继续往下说。

Moving right along.

关于股票的另一面，对哈里来说，我知道这是显而易见的，但如果我们把COSPI看作某种AI加杠杆，对吧，还有韩国交易所的风险，它今年仍然上涨了56.46%。

The other thing on the other side of the stocks for what for Harry, I know this is captain obvious, but if we look at COSPI as sort of like AI and steroids, right, with risk or Korean exchange, It's still up 56.46% for the year.

我不是日内交易者。

I'm not a day trader.

我会打开我的高盛和摩根士丹利账户，看看它们的表现如何。

I pull up my Goldman and Morgan Stanley accounts and look how they're doing.

我仍然觉得自己在公开市场股票上像个天才，因为总体上涨46%已经很不错了。

I'm still feeling pretty like I'm some genius in my public market stock because overall plus 46 is pretty good.

只是，天哪，当我看到图表时，利奥好像被困在了一把匕首里。

It's just, boy, whatever Leo got just trapped in a dagger when I look at the chart.

对吧？

Right?

抱歉。

Sorry.

你可以笑话我，告诉我我应该知道里面有什么。

You can laugh at me and tell me I should know what is in

科斯比。

Cosby.

韩国存储器，三星。

Korea memory, Samsung.

现在就是个四舍五入的误差。

A rounding error now.

韩国现在就是两家内存供应商，外加一堆其他业务。

Korea now equals two memory providers with a bunch of other stuff attached.

而且这个市场波动很大，就像打了兴奋剂一样。

And it's a very volatile market, so it's on steroids.

但纳斯达克就像科技股打了兴奋剂，而COSPI就像所有AI相关的东西都打了兴奋剂。

But it's still just like Nasdaq is tech on steroids, COSP is like all the components of AI on steroids.

对。

Right.

六月份最高点是9000，然后砰，可怜的Leo，7月29号一下子跌到5600。

And the peak was 9000 in June and then boom, poor Leo, the dagger 5,600 in July 29.

那家伙遭遇了世代级别的亏损，如果亏损再大一点，可能都会拖垮我们的金融体系，不过你知道的，从那以后已经反弹了百分之二十多。

The guy had a generational loss that If it were even bigger, might have brought down our financial ecosystem, but it's, you know, it's rebounded 20 something percent since then.

今年已经涨了56%。

It's up 56 of the year.

一年涨了56.46%，这还有什么好抱怨的。

Cry me a river if it's up 56.46% a year.

我的意思是，除非你是做日内交易的，不然一年涨56%你肯定高兴。

I mean, you got to be a day trader or whatever to to not love up being up 56% a year.

这挺好的。

It's Okay.

但所有头条都在说，哦，Cosby今天跌了6%。

But all these headlines are like, oh, Cosby is down 6% today.

就是波动太大了。

And it just it's just hyper volatile.

增长和利润率，我们在半导体行业从没见过这么高的利润率。

The growth and margins, we've never seen margins like this in semiconductors.

所以波动和预期，我记得《华尔街日报》有篇文章说，现在韩国最抢手的单身汉是三星的工程师。

So the volatility and expectations there was an article I think was in the Wall Street Journal where in Korea now that the most eligible bachelors are Samsung engineers.

大家都想要他们。

They want it.

所有人都想嫁给做内存的人。

They all everyone wants to marry a memory guy.

这是这个国家历史上第一次，做内存的人能成为全国最抢手的单身汉。

It's the first time in the history of the nation when being a memory guy was, like, made you one of the most eligible bachelors in the country.

你没看到吗？现在英伟达有50%的员工身价超过2500万美元了。

Did you not see that, like, 50% of NVIDIA employees are now worth over $25,000,000?

对。

Yeah.

这几乎是必然的。

Pretty inevitable.

我走在街上都能看到，虽然没什么可买的。

I see it walking down the block when nothing's for sale.

对。

Yeah.

嗯，我觉得这不一样，我是说，我们可以聊聊，但我觉得整体上AI的泡沫和疯狂，确实是这样。

Well, I think it's a diff I mean, we could talk about I just think overall AI inflation and craziness is yeah.

就是这样，没办法。

It it is what it is.

那边发生了什么？

What happens there?

这是持续存在的新常态，还是暂时的通胀现象？

Is that like a persistent continued new world, or is that a temporary moment of inflation?

现在Dogpatch那边普通公寓的租金都涨到一个月一万美元了。

Rents in the mediocre apartments just in Dogpatch are $10,000 a month now.

就是YC旁边那条街上的普通公寓，Avalon那栋。

Mediocre apartments down the street from from YC, the Avalon.

我在YC之前就在Dogpatch工作过，那时候那里又破旧又有趣。

I used to work in Dogpatch pre YC, and it was gritty and fun.

我记得他们建Avalon的时候，说实话你根本不想住那儿。

And I remember when they built this Avalon and, you know, you didn't really wanna live there.

当时是新的，现在租金超过一万美元一个月，还得排队申请，而且不允许在里面创业。

It was new, and now it's over $10,000 a month and you got to wait, you got to apply and you're not allowed to run your startup out of it.

你得签一份文件，承诺不能在里面做生意。

You have to sign a document that you won't run a business out of it.

离YC就两个街区。

Two blocks from from YC.

如果Avalon的一居室月租要一万美元，那你得赚多少才能觉得自己有钱？

If it's $10,000 a month to rent a, you know, one bedroom at the Avalon, how much do you have to make to feel rich?

得赚很多。

A lot.

光是在Dogpatch的Avalon租个公寓，一个月就要一万二。

That's $1.20 just in a rent to not to just have an apartment at the Avalon in Dogpatch.

在加州，税前你得赚两万四才能付得起这个房租。

You need $2.40 k in California pretax to pay the rent.

你可能得赚四万八才能觉得自己过得不错。

You probably need $4.80 to feel good about yourself.

对吧？

Right?

不过对我来说，坐在伦敦看这件事真的很有意思，因为钱不在这里。

It's just so interesting for me sitting in London, though, because the money's not here.

我听到你说了，也明白你的意思。

I hear you and I hear you say that.

是的，确实有一些像我这样在风投行业的幸运儿，很高兴能做我们正在做的事，但钱真的不在这里。

And yes, there are some fortunate people like me in Venture who are thrilled to be doing what we're doing, but it's just not here.

财富的分布几乎为零。

That dispersion of wealth is just nil.

对。

Yeah.

因为这是科技财富，而且都集中在……你看。

Because it's it's tech wealth, and it's all con I mean, look.

加州不仅仅是在各方面都领先。

Saw California didn't just outperform everywhere else.

它拿走了总资金的四分之三。

It got three quarters of the total dollars.

不过呢，这个数据其实被扭曲了，因为Anthropic和OpenAI加起来可能拿走了总资金的60%。

Now that's eventually, that's skewed by the fact that Anthropic and OpenAI together got probably 60% of the total dollars.

我脑子里算了一下，其他人总共才拿到15%。

I'm doing the math in my head, and everyone else got 15.

但确实，这是一大笔钱涌进了一个非常小的地方。别忘了，旧金山的人口只有七十七万八千人。

But yes, this is a wall of money flowing into a very small area where Reminder, the population of San Francisco is seven seven and eighty thousand people.

那是个非常小的小镇。

It's a teeny tiny town.

伦敦有八百九十万人口。

London's eight nine million.

整个湾区也只有七百万。

The whole Bay Area is only 7,000,000.

这笔钱就像一堵墙一样砸进一个小地方，那是个三面环海的半岛，有点山，还有个小山谷叫硅谷。那边土地不多，而且很难盖房子。

This is a wall of money falling into a tiny place that's a peninsula with sea on three sides and a little bit of mountains and a tiny valley called Silicon Valley on the Fort Side, property is not plentiful and it's hard to build.

接下来会发生的就是房价要涨了。

What's going to happen is prices are going to go up.

大多数其他人都会被挤出市场。

Most everyone else is going to get priced out.

当他们被挤出去的时候，他们会非常恼火。

When they get priced out, they're going to get pissed off.

所以这种水平可能不会持续太久，因为我经历过99年、2000年和2007年。

So probably it doesn't last at this level because I've been around in 'ninety nine, 2000, and 2007.

总会有某种调整，也会有一些重置，但不会回到原来的水平，因为从来不会这样。

There will be some kind of correction, and there will be some kind of reset, but it's not going back to where it was because it never does.

它会一步步往上走。

It ratchets up.

快进五年，到那时，AI热潮已经被消化了，不像现在这么疯狂，但价格的基础水平已经上涨了，这意味着其他人在旧金山的生活成本会上升，你付给组织里任何人的成本，你付给任何与你打交道的人的成本，都必须上涨，因为这里的生活成本会更高。

Fast forward five years, at that point, the AI boom has been digested, it's not as crazy it is now, but the base level of prices has gone up and the cost what it means is that the cost for anyone else to live in San Francisco goes up, the cost that you pay, you know, anyone in your organization, the cost that you pay anyone that you interact with, all has to go up because the cost of living here is gonna be higher.

我真的很感激能和你们一起做这个节目，因为我从你们身上学到了很多，这也是我第一次看到这样的周期。

I genuinely appreciate doing the show with you guys so much because I I I learned from you, and it's the first time for me seeing cycles like this.

在这波AI浪潮中，未来五年，我们会脚下失稳吗？

Does the floor fall from our feet in this AI wave for the for the next five years?

我们是不是只会看到更多的钱，更多的上涨，更多的巨额退出？

Do we just continue to see more money, more up and to the right, more mega exits?

我觉得只是更集中了。

I think it's just more concentrated.

由于各种原因，我们需要更少的人来创造比以往更多的收入。

We need fewer people to generate more revenue than ever for a variety of reasons.

这会集中退出的规模。

And it's going to concentrate exit size.

这会集中财富。

It's going to concentrate wealth.

这会允许，你知道，Anthropic和OpenAI的薪水看起来疯狂，但现在有些正在正常化，因为如果你能用一半或三分之一的人做到，你确实不能付给他们两到三倍的工资。

It's going to allow, you know, the salaries that Anthropic and OpenAI seem crazy, but some of it's normalizing now because if you can do it with half the people or a third of the people, you really can't pay them two to three times as much.

对。

Right.

所以最终，当然，而且我们可以像贝比·鲁斯那样选日子，这也会结束。

So eventually, of course, and where we can pick the date like Babe Ruth, this this will end.

但是，你知道，我觉得你得相信我们才走了不到三分之一的路程。

But, you know, I I think you have to believe we're less than a third of the way through this cycle.

哦。

Oh.

就连那个小小的云服务周期都持续了九年。

Even the even that little tiny cloud thing lasted nine years.

我们这才刚刚开始呢。

We're just getting going here.

杰森，这些公司能赚到足够的钱，并且赚得够快，来维持云服务周期的运转吗？如果收入趋势……

Will the companies make enough money, Jason, fast enough to keep the cloud cycle going if the revenue trends

……停止，每年还要投入一千亿美元来维持这个生态系统的运转。

stop keep spending $100,000,000,000 a year to keep the ecosystem going.

说实话，那会有帮助。

Honestly, that will help.

让我试试，因为我也在琢磨这个问题，毕竟这确实是最关键的问题。

Let me try because I'm trying to figure this out too because, obviously, it is actually the only question.

我的想法是这样的。

My mental model is this.

在供应端，没人会退缩。

On the supply side, no one's gonna blink.

英伟达不会退缩。

NVIDIA isn't gonna blink.

超大规模云厂商也不会退缩。

The hyperscales aren't gonna blink.

只有Anthropic会退缩。

Only Anthropic is gonna blink.

对吧？

Right?

不会退缩。

Not gonna blink.

所以我觉得，虽然我们是周二录的这期，英伟达周三就发财报了。

That's why I think even though we're recording this on Tuesday, NVIDIA is reporting on Wednesday.

周四就会公布。

It's gonna appear on Thursday.

有可能，但我觉得英伟达明天不太可能说算力需求放缓。

It's possible, but I think highly unlikely that NVIDIA gets on tomorrow and says, compute demand to slow down.

那句话不会发生。

That sentence is not gonna happen.

所以，供应方面会继续下去，因为不太可能有人退缩，没人会退缩。

So to your it's gonna keep on going on the supply side because it's not likely to blink, no one's gonna blink.

所以唯一能阻止它的两件事就是，你知道，要么资金耗尽，要么需求耗尽。

So the only two things that stop it are, you know, you run out of capital or you run out of demand.

说到资金，我的直觉是，除非公开市场参与进来，否则你还没耗尽所有可用的钱，这也是为什么这两个大型IPO必须发生。

When you say capital, my gut is until the public markets get in on the game, you kind of haven't exhausted all the money that's there, which is why in many respects, these two big IPOs have to happen.

通常，金融危机发生在边际买家耗尽的时候，但AI领域还有大量未开发的需求，因为这些公司还没上市。

Typically, financial crashes go when you run out of marginal buyers, there's still a whole bunch of untapped demand to play in the AI game because these companies haven't gone public.

在资金方面，显然还有一次机会，那就是Anthropic和OpenAI上市，这会继续推动发展。

You've clearly, on the capital side, got one more turn of the crank, which is when Anthropic goes out and OpenAI goes out, that's going to keep it going.