AI泡沫如何破裂——与保罗·凯德罗斯基

不。

No.

我的意思是，这样来回讨论一下这些论点挺好的。

I mean, the the so just not This is good to go back and forth and talk through the arguments here.

论点基本上是，这项技术太新了，发展太快了，需要时间来搞清楚这些事情。

The argument would be that basically this technology is so new and it's moving so fast that it's gonna take time to figure these things out.

所以你不能在第一天，比如那个寓言出来，或者你家里有神话，就一下子把整个经济都吞下去。

And so you can't just like, you know, on day one that fable comes out or you have mythos in house, know, go ingest the entire economy.

必须

Have to

一步一步来。

do this like step by step.

一个例子就是Claw设计，Anthropic一直在关注这个设计，当然，Mike Krieger以前在那里负责产品，他是Figma的董事会成员，这就导致了整个问题，让Dylan Field成了Anthropic最大的批评者之一，Figma。所以，与其直接说我们要全面推行，不如一步一步来。

And a case in point is a claw design where like Anthropic has been watching the design, of course, Mike Krieger, who used to run product there, was on the board of Figma, and that's led to this whole issue and made like Dylan Field like one of Anthropic's biggest critics, the Figma And so instead of like going out and saying, we're just gonna do this wholesale, we'll we'll do it step by step.

我们会看看技术能做什么，看看人们怎么用它，看看市面上有什么其他解决方案，然后再去构建。

We'll we'll see what the technology is capable of, see how people are using using it, see what other solutions are out there, and then go ahead and build.

这就引出了代币通缩这个概念。在我看来，我们现在处于一个大家都认同的阶段：底层的模型正在变成大宗商品，拥有模型本身的价值，仅仅体现在你能否通过深度定制，让你的产品和自研模型之间形成那种别人无法复制的紧密协同。

And it just goes to this whole concept of the the the deflation of of tokens is like, you know, it seems to me that we're at this point where everybody agrees that these models underneath are commoditizing and owning the model is valuable only in the way only in your ability to customize your own products to have that, like, deep sync between your products and what the and the models you build that nobody else could have.

没错。

Sure.

这也是未来的发展方向。

And that's where this is going.

是的。

Yeah.

我觉得大体上确实是这样。

I think that I think that's broadly true.

不过我不太同意你说的，觉得大家都认为这些模型正在变成通用商品。

And I think I I don't necessarily agree with you that everyone believes these models are commoditizing.

对我来说，这感觉还是——

It still feels to me like That's

没错。

true.

好的。

Okay.

比如说，我不知道，iPhone 4 那个时代，人们会对新发布感到兴奋，然后大家又抱怨，说这没改变世界，又不是 AGI。

Like, I don't know, iPhone version four era when people get all excited about a new release, and then everyone whines because they say, well, this didn't change the world, and it's not, you know, AGI.

这有点像 iPhone 的第四代，人们还期待有突破，而在大型语言模型刚推出时确实有巨大突破。

That's kind of like the fourth iteration of the iPhone where people want to believe that there's breakthroughs still coming, and there were huge breakthroughs in the early days of launching large language models.

但现在，不仅是商品化，而且我有些数据经常给人看，显示一种趋同也在发生。

But now, it's not just commoditizing, but I've you know, some data that I often show people that shows a kind of convergence that's also happening.

所以不只是有平台期现象，而且不同模型之间的差异，最佳实践，已经在很大程度上趋同了，降低了各模型综合表现的差异，这意味着更换模型的成本低得多。

So it's not just that there's kind of a plateauing phenomenon going on, it's that the variance among models, the best practices across all of these different models has kind of has converged to a large degree, reducing the variance in terms of the composite performance of various models, which means that the opportunity cost for changing models is much lower.

所以如果机会成本低得多，除非他们能锁定我，否则我会有很大动力去不断套利，在它们之间来回切换，这就是为什么中国模型崛起，DeepSeek 突然表现这么好，还有 Qwen 和其他，以及 OpenCode 成为很多人可行的工具。

So if it's if the opportunity cost is much lower, unless they can lock me in, there's a huge incentive for me to constantly arbitrage and play back and forth across them, which is hence the rise of Chinese models, why DeepSeek is doing so well all of sudden a and why, you know, Quen is and others and why OpenCode is emerging as a, you know, viable tool for many people.

因为有种感觉，我并没有真正被锁定，趋同意味着模型差异虽然存在，但小到我分辨不出来，就像百事可乐现象，再说回投资，这表明竞争更多变成营销支出，另一种成本，以及价格。

Because there's this sense that I don't not really locked in at all, and the convergence just means that the model differences, while there, are so minimal as I can't tell the difference in a kind of PepsiCo phenomenon, which again, to cut to the investment chase, suggests that the competition then becomes much more about marketing expenditure, another form of costs, and, price.

所以这两点都是对的。

So both of those are Right.

从这类资产的投资回报来看，表现很差。

Poorly in terms of the investment returns for this asset class.

没错。

Correct.

所以我觉得我们俩的看法差不多，就是经济压力会迫使这些公司往高端市场走。

And so this is sort of I I think we're we're both seeing it in a similar way, which is that the economics is gonna force these companies to go upmarket.

百分之百同意。

At 100%.

这时候就有意思了。

Where things get interesting.

对。

Yeah.

而且我认为这会加速，假设IPO能成行，上市会加速这个过程，因为公开市场的投资者会看看这种叫代币的商品背后的经济逻辑，然后问：你们还有什么别的？

And I think that's gonna accelerate with these assuming the IPOs happen, that's going to accelerate with the IPOs because public markets investors will look at the underlying economics of this fund of the commodity called tokens and say, so what else you got?

对吧？

Right?

然后说，我们接下来要做什么？

And say, what are we gonna do next?

你们要进入哪些市场？

What markets are you gonna move into?

所以这会加大压力，促使他们向高端市场进军。

And so that's gonna increase the pressure to do this absorptive move up market.

然后你就会遇到这个问题，这也是我今年早些时候对SaaS末日论抱怨的一点，就是人们对公司为什么购买软件存在深深的误解。

And then you get into this problem that and this was my complaint early on about the SaaS pocalypse earlier this year is there's a deep misunderstanding about why companies buy software.

不是因为他们认为ServiceNow或Salesforce或其他什么公司是某种不可替代的大胆创新者。

It's not because they think ServiceNow or Salesforce or whoever is somehow, you know, bold innovators that could not be replaced.

不是的。

No.

是因为他们有问题要解决。

It's because they have a problem.

他们不想自己构建，而且他们需要有人可以起诉或责骂。

They don't wanna build it themselves, and they want someone to sue or shout at.

就是这样。

That's it.

这就是人们购买SaaS软件的原因。

That's why people buy SaaS software.

不是这样的，所以每当你开始为自己构建东西，或者认为公司会越来越多地使用这些前沿工具来为自己构建东西，或者反过来，前沿公司愿意被地球上所有人起诉和指责，因为他们为这些公司构建垂直应用，这种想法很快就会被打破，因为那是一门糟糕的生意。

It's not and so whenever you start building it for yourself, this notion that companies are going to increasingly use these frontier tools to build things for themselves or vice versa, that the frontier companies are willing to be sued and shouted at by everyone on Earth for building vertical apps for them, this will rapidly be disabused because it is a terrible business.

你不想处于那种不断服务客户的位置，这些客户使用你产品的主要目的是有人可以指责或起诉，这虽然是个夸张的说法，但确实误解了垂直软件存在的原因，以及那些公司为什么存在来服务这些垂直领域的人们。

You do not want to be in that position of continually having to service people whose main utility for your product is having someone to shout out or sue, which is, again, it's it's a gross exaggeration, but it's a misunderstanding of why verticalized software exists and why those companies exist to service the peoples in those verticals.

而天真地说前沿模型公司会轻率地向上市场进军，以服务他们新的公开投资者，这误解了这些市场最初存在的原因。

And to just naively say the frontier models companies will blithely race up market in, you know, in service of their new public investors is to misunderstand why those markets exist in the first place.

对。

Right.

我的意思是，也许他们不得不这样做。

I mean, maybe they'll have to though.

这就是问题所在。

That's that's the thing.

不。

No.

不。

No.

不。

No.

他们确实得这么做，但我的意思是，这不会容易，更重要的是，可能会非常痛苦和代价高昂。

They'll have to, but that's my point is is that that'll just it won't be easy, and even more importantly, it'll probably be very painful and costly.

所以，对。

And so Right.

我觉得，到那时候，你就会明白“小心你所求的”这句话了。

Be careful what you wish for, I think, is where you get to on that one.

好的。

Okay.

让我再提一个实验室的观点，然后……好。

Let me let me make one more of the labs arguments and then Sure.

我其实想深入聊聊你看到的一些弱点，还有我也看到的那些。

I actually wanna get into some more of the weaknesses that you see and that I see.

好的，另一个，你知道，你刚才有点暗示过这个，就是那些相信这次不一样的人。

Okay, the other, you know, you kinda winked at this time, you know, the people that believe this time is different.

这不是一个特别技术性的论点。

This is not like super, a super technical argument.

这更像是你可能会听到的那种一般性论点。

This is sort of like the general argument that you might hear.

这是安德森式的论点。

It's the Andreessonian argument.

是的。

Yes.

有人会说，你知道吗，保罗？

Would be somebody saying, you know what, Paul?

这次不一样。

This is different.

这些实验室造出了那种神奇的、能思考的计算机器。

You have these labs who've built magical, you know, thinking computer machines.

是的，他们投入了很多。

Yes, they're they're investing a lot.

但在科技领域，你要做的是打造一个资产，想办法通过计算机规模化，然后加价，人们会买，因为它比其他任何选择都好。

But in tech, what you do is you build an asset, you find a way to scale it through computers in some way, and you mark it up, and people will buy it because it beats any other alternative.

而且你最近看到的是，即使在过去的六七个月里，能力已经大幅提升。

And what you've seen recently is like even in the past let's say six, seven months, the capabilities have scaled dramatically.

现在你可以让这些计算机自己运行，它们能自己编程，而且做得不错，以至于它们不仅对工程有用，对各种工作都有用。

You've been able to like now leave these computers alone and they can code on their own and do a decent job to the point where like they're not just useful for engineering, they're useful for all types of work.

所以随着时间的推移，尽管已经投入了那么多，就像我们说的，今年到明年之间可能有2万亿美元，但你知道，最终会证明这些投入是值得的。

And so over time, you know, that there will despite the fact that so much has been invested, like we said, know, maybe $2,000,000,000,000 that are coming between this year and next.

对。

Mhmm.

那将具有巨大的经济价值，以至于这个行业必须成功。

That will be so economically useful that the business is going to have to work out.

这就是为什么大家都在往这个方向冲。

And there's this is sort of why people are rushing toward it.

你怎么看？

Your thoughts.

好。

Sure.

但再说一次，这是一个经典的逻辑谬误，就是把你想要证明的东西当成前提。

But again, this is this is this is a classic logical fallacy of assuming what you're trying to prove.

对吧？

Right?

所以你提前下结论说它必须成功，因为你需要它成功。

So you've you you raise ahead and say it has to work out because I need it to work out.

而且我会更深入地探讨这个“这次不一样”的问题。

And and I'll go more deeply into this whole question of of the this time is different thing.

“这次不一样”的推论是，在这些时刻之后，总会留下一些有用的东西。

The the corollary to the this time is different thing is all that there's always something useful left after these moments.

对吧？

Right?

然后，那种想法就是，在光纤泡沫几年之后，总会有一些重要的、有用的资产留下来，你知道，我们可以用这些光纤来做一些事情。

Then the idea that there's always some important some useful assets left after after the fiber bubble years later, we could, you know, we could use the fiber for things.

尽管有一半的铁路因为过度建设最终被废弃了，但铁路仍然非常有价值。

Even though half of railroads were eventually abandoned because of overbuilding, railroads are still hugely valuable.

这些都没错，但这有点不合逻辑。

That's all true, but it's kind of a non sequitur.

当然，这是真的。

Well, of course, it's true.

我们建造它，不是因为它没用。

We didn't build it because it was useless.

我们建造它，是因为它有用。

We built it because it was useful.

问题在于，大规模过度建设在更广泛的经济中会带来什么样的连锁后果？

The issue is what are the the what are the consequences of massive overbuilding in terms of spiraling consequences in the broader economy?

你知道吗，现在数据中心相关债务的最大买家，有些是保险公司。

You know, increasingly, some of the largest purchasers of of data center related debt or insurance companies.

我们知道，每当保险公司掺和进这种事情，会发生什么。

We know what happens whenever insurance companies get in the middle of this stuff.

我们在全球金融危机中已经见识过了。

We've seen it in the global financial crisis.

我们反复见识过。

We've seen it repeatedly.

所以，正确的问题不是，哦，拍拍别人的头说，一切都会好起来的，因为过去总是这样。

So that's the right question is not, you know, oh, you know, pat people on the head and say this is all gonna work out because it's always worked out in the past.

第一，嗯，过去确实总是这样。

One, well, it's always worked out in the past.

但至少四次差点把全球经济拖垮，这点值得注意。

It's nearly taken up the global economy at least four times, so that's worth noting.

另一个问题，我觉得更阴险，人们往往忽略，因为你会看到有人提到一个叫卡洛塔·佩雷斯的女人，她写了一本书叫《技术革命与金融资本》，在某种意义上，这本书是很多推动这些事情的乐观派人士的圣经。

And and the other the other issue is and this is, I think, the more insidious one that people miss because you'll see people refer to there's this woman named Carlotta Perez who wrote a a book called Technological Revolutions and Financial Capital, which in a sense is the bible for many of the most, I don't know, bullish partisans pushing some of this stuff.

他们会说，这就是必须发生的事。

And they'll say, well, this is what has to happen.

我们必须经历这种大规模的投资和浪费，但最终会好起来的。

It has to we have to have this kind of huge moment of spending and waste and everything else, but then it works out.

但这个论点有个问题。

Here's the problem with that argument.

在之前的案例中，人们并不知道这一点。

In prior episodes, people didn't know that.

这是一个非常重要的区别。

That's a really important distinction.

我们创造了一种反身性，现在我们用过去过度投资成功来为现在的过度投资辩护。

We've created this reflexivity where now we justify overspending on the basis of prior overspending having worked out.

但在之前的案例中，比如农村电气化，人们并不会说，这看起来可能不好，但铁路投资最终成功了。

Well, in prior episodes where that happened, people were not justifying the overspending by saying, say, in rural electrification, you know, this may look bad, but it worked out in railroads.

不。

No.

不。

No.

不。

No.

不。

No.

不。

No.

你不能这么玩。

You don't get to play that game.

我们当时没有那个。

We didn't have that.

所以现在的情况是，这已经变成了一个封闭的飞轮，因为我们在用以前没有的信息来为更大的过度建设辩护。

So now what's happened is is it's become a hermetically sealed fly almost a flywheel in a sense because we're justifying things on the basis of information that we that we didn't have in prior episodes and using that to justify an even larger overbuild.

这就是为什么说这次不一样，确实不一样，但不一样在非常危险的地方。

And that's why the notion that this time is different, it is different, but it's different in a really dangerous way.

好的。

Okay.

所以，我听到了。

So so I I I hear that.

我接受。

I accept that.

我想我试图摆在你面前的那个论点，其实有点不同。

The argument that I was trying to, like, put forth on your plate here is, you know, is is a little bit different, I think.

我想让你回应的那个论点是，保罗，这就是AGI，老兄。

Like the argument that I'm trying to get you to respond to is the Paul, it is AGI, man.

就像，这是……嗯，是的。

Like, this is Well, yes.

这是……是的。

This is Yes.

你知道的。

The you know.

那你对此有什么回应？

So what is your response on that front?

感受一下AGI。

Feel the AGI.

所以是的。

So yes.

那么问题就变成了，你愿意为AGI的看涨期权付多少钱？

So the question then that turns into this one of what would you pay for a call option on AGI?

这基本上就是，论点是你不可能过度支出，因为价值就像是在说，你愿意为永生期权付多少钱？

This is essentially the the the argument is you cannot possibly overspend because the value it's like saying, what would you spend for a call option on immortality?

嗯，从数学上讲，我应该愿意付出任何代价。

Well, mathematically, I should be willing to spend anything.

同样，AGI的看涨期权，没有哪个净现值是折现后大到不能接受的。

Similarly, a call option on AGI is is there's no discountable net present no discountable net present value that is too large.

所以一旦你走上这条路，一旦你接受这个前提，就是，你知道，为AGI或为永生加油，那么我又陷入了这个陷阱，嗯，当然，绝对是的。

So once you start down that path, once you accept that premise that this is, you know, fuel the AGI or fuel the the immortality, then again, I'm into this trap of, well, yeah, absolutely.

但问题是，我现在，我们现在都在跟着这个有点像邪教的想法走，我们都同意，你知道，你应该努力去认可的东西，我也应该假设，所以我们应该愿意付出任何代价。

But the problem is that I'm now we're now going along with this cultish idea that we both now agree that, you know, what you should be trying to approve, I should assume, and therefore, we should be willing to spend anything.

这就像变成了一个有毒的棋盘游戏。

And it's simply it's that becomes a like a toxic board game.

就像没有网的网球。

It's tennis without a net.

对吧？

Right?

一旦对话的另一方问“你愿意为永生期权付多少钱？”，我们就没法进行合理的对话了。

There's no way for us to have a reasonable conversation once the other side of the conversation is, what would you be willing to pay for a call option on immortality?

你愿意为AGI期权付多少钱？

What are you willing to pay for a call option on AGI?

所以不需要那个论点。

So you don't need that argument.

我们应该能够提出一个论点，说这项技术非常强大、非常重要、具有变革性，并且它会以某种方式改变世界，而不必像经典不可知论理论那样，把上帝塞进每个找不到好论据的缝隙里。

We should be able to make an argument and say that this technology is very powerful and very important and transformative, and here's the way it's going to change things without having to have It's like in classic sort of agnostic theory, this idea of inserting God into every gap in an argument where you can't find a good argument.

这就是一种用上帝填补空白的论证。

This is a God of the gaps argument.

我引入AGI，是因为这样我就能创造一个无法估价的、不可折扣的看涨期权。

I'm inserting AGI because now that allows me to create an undiscountable call option that I can't price.

因此，我愿意不惜一切代价。

Therefore, I should be willing to spend anything.

但我拒绝这种说法。

And I reject that.

你不觉得所有投入AGI或AI建设的资金，那些写支票的人被告知了AGI的论点，然后他们就被套住了，尽管你在讨论中指出的那些经济弱点，他们基本上还是在为AGI的看涨期权写支票。

Don't you think that all the money that's going towards this AGI or AI build out, The people writing the checks have been told the AGI argument and they're stuck or despite all of the economic weaknesses that you've pointed out in our discussion are are basically writing that check for a call option on AGI.

在某种程度上，我接触过的投资者都非常 cynical。

To a degree, investors that I've talked to are are very cynical.

他们很乐意在LP面前用这个说法，但他们自己并不相信。

So they are perfectly happy to use that in front of their own LPs, they but don't they don't believe that in house.

他们用非常冷静、精于计算的眼光看待这件事，并把它与其他类似的房地产项目进行比较。

They look at this very cynically and with very cold calculating eyes and compare it to other similar real estate projects.

这其实跟其他类型的项目融资差不多，比如水电站或者那些长期资本密集型项目。

It's really compared to other sorts of project financing from hydroelectric dams to long lived capital intensive projects.

这是它必须跨过的门槛。

That's the hurdle that it has to clear.

从推广的角度来说，确实可以这么说。

In terms of promoting it, sure.

我们可以把这些叫做AGI工厂。

We can call these AGI factories.

我最近跟新墨西哥州一个负责区域经济发展的人聊过，有个超大规模数据中心公司跑过去跟他们说，你们不想成为AGI工厂的一部分吗？

I was talking to a regional economic development official in New Mexico recently who had a hyperscaler show up and tell them, don't you wanna be part of AGI factories?

我当时就想，什么？

I was like, what?

这是在描绘这样的前景吗？

This is the picture being made?

因为你们是在为未来签约，现在你们能帮我们建工厂，而这些工厂将决定AGI的未来。

Because you're you're you're signing up for the future because now you can help us build the factories that dictate the future of AGI.

所以，你知道，你可能会提出的所有这些反对意见，比如他们想要的税收优惠，以及水、电、房地产等方面的东西，都不重要，因为想想我给你的这个看涨期权的规模，而且这是一张免死金牌。

And so, you know, all of these objections you might raise in terms of the kinds of tax evangelists that they want with respect to water and power and real estate and other things, doesn't matter because think about the scale of the call option I'm I'm offering you, and it's a it's a get out of jail free card.

而且这真的，我觉得，很不幸地让人反感。

And it's and it's really, I think, unfortunately offensive.

但不管怎样，这更多是营销手段而已。

But nevertheless, it's more marketing than anything else.

当我跟那些投入资金的最大投资者交谈时，他们会跟自己的有限合伙人用这套说辞，但在合伙人会议上他们不会这么做。

And when I when talk to the largest investors who are putting capital into this, they'll use it with their own LPs, but they don't do it in partner meetings.

有意思。

Interesting.

所以他们不这么做，是因为他们其实并不相信？

So they don't do it because they don't actually believe it?

不。

No.

他们不信。

They don't believe it.

不是他们不相信，我换个说法。

Not not that they don't believe it, I'll put it differently.

不是他们信或不信的问题。

It's not that they believe it or don't believe it.

他们就是懒得去关心，因为他们觉得自己能搞定。

They just couldn't be bothered caring because they think they can put it.

这无关紧要。

It's nonmaterial.

那他们为什么能接受呢？

So then why are they okay.

如果你跟这些人聊过，你肯定看到了这里面的明显问题。

If you you're speaking to these folks, you're you're seeing the the clear problems here.

他们心里是怎么给自己找理由的？

What is the justification that they make in their mind?

假设他们完全相信你说的，那么，我们说过你得换GPU，代币价格要降下来才能行得通，需求得是现在的100倍才行。

Let's say they take everything that you say and they of, they give it credence, The fact that, okay, we've talked about you have to replace the GPUs, Token prices are going down to make the make this work, the demand would have to be like a 100 x what it is.

嗯，其实更像是一百万倍，不过也行。

Well, more like a million x, but okay.

一百万倍。

A million x.

好。

Okay.

就这么说吧。

Let's just say that.

一百万倍。

A million x.

这不在考虑范围内吧？

Isn't in the checks?

因为

Because

对。

yeah.

不是。

No.

这有点疯狂。

It's it's a bit crazy.

但确实如此。

But yeah.

那他们为什么还在读支票？

So why are they still reading checks?

但你可能会说同样的话。

But you might say the same thing.

这又回到了金融危机期间哈罗德·普林斯的那句话。

It's back to the Harold Prince line back at the during the financial crisis.

只要音乐还在播放，我就继续跳舞。

As long as the music is playing, I keep dancing.

就是这个意思。

This is that.

对吧？

Right?

只要音乐还在播放，他们都会继续跳舞，因为从主权基金到私募股权和私人信贷，地球上最大的资本提供者根本没有动力退出，因为你会受到ELP的压力，他们会问：你为什么不参与这个？

This as long as the music is playing, they're all gonna keep dancing because there is absolutely no incentive as any of the largest capital providers on earth from sovereigns down to private equity and private credit to walk away because you get pressure from ELPs saying, why aren't you participating in this?

更糟的是，作为主权财富基金，我曾在这些机构内部待过，一旦你管理着数千亿美元，你开始看的不是机会的经济价值，而是支票的规模。

And then even worse as a sovereign, as a sovereign wealth fund, and I've been inside these folks, that once you're managing hundreds of billions of dollars, you start looking at opportunities not in terms of their economic value, but in terms of check size.

你会说，我需要写一张支票，比如一千亿美元，因为我不想写一千零一十亿美元的支票。

And you say, I need to write a check for, fill in the blank, a $100,000,000,000 because I do not wanna write a $101,000,000,000 checks.

所以这种奇怪的筛选就开始了，现在这些项目就像，看看我的朋友，沙特、卡塔尔。

So this weird filter starts happening where you now these projects are like, look at my friend, Saudi, Qatar.

我有一个项目非常适合你。

I have this project that's perfect for you.

你想写一张五百亿或一万亿美元的支票吗？

You wanna write $50,100,000,000,000 dollar checks?

除了这些巨大的数据中心园区，比如Meta在路易斯安那的项目，或者随便你选，地球上没有其他地方能让你写这么大的支票。

Nowhere else on earth can you write it other than these giant data center campuses like the the the Meta Project in Louisiana or take your pick.

所以一旦你开始用支票大小来筛选项目，世界就开始扭曲了，这就是为什么这些项目变得更有意思，因为真的没有别的东西能跟它们比。

And so once you become a a developer check size filter, the world starts to twist on its axis, and that's why these projects become even more interesting because there's just nothing else out there like them.

那我就这么说吧，你刚才说的那些在我听来都太疯狂了，居然有人是这么运作的。

So I will just respond by saying everything that you just said sounds crazy to me, that that is the way people operate.

哦，我可是在里面待过的。

Oh, I was inside.

我给你讲个有趣的故事。

I'll tell you a funny story.

我曾经在一个七亿美元的创投基金里，当时我们拒绝了五个很棒的项目。

I was inside of a $700,000,000 venture fund at one point that turned down five terrific projects.

这还只是很小的规模。

So this is at a very small scale.

现在你想想，如果是主权基金呢。

So think about it now as a sovereign.

因为那个创业者想要四百万美元，但基金想写一张两千万的支票，然后说，你知道，我写不了四百万的支票。

Because because the entrepreneur we want the entrepreneur wanted $4,000,000 and the fund wanted to write a $20,000,000 check and said, you know, I can't do a $4,000,000 check.

他们放弃了四个很棒的项目，我当时觉得这简直，用专业术语来说，就是疯了。

And they walked away from four terrific projects and I thought that is absolutely to use the technical term bat shit.

我当时以为再也不会见到这种情况了，结果现在我在全球一些最大的基金里反复看到。

And then I thought I'll never see that again, and I've now seen it repeatedly inside of the some of the largest funds on earth.

他们看项目的时候，只看一个标准：我能不能开出一张足够大的支票？

Looking at projects through the filter of, can I write a large enough checks?

我手头有这么多钱。

I have this much out.

我口袋里这么多资本在烧着？

I have this much capital burning all in my pocket?

所以这并不意味着我会随便给任何人。

So it doesn't mean that I'll just give it to any random.

我会给任何人任何项目写支票，但这确实改变了你筛选可行投资的方式，而且是以一种非常扭曲的方式。

I'll write checks to anybody for anything, but it does change the way that you filter the landscape of viable investments in a really and truly perverse way.

所以全球这些庞大的资本聚集点，就是问题的一部分。

So these point masses of capital worldwide are part of the issue.

对。

Right.

所以基本上总结一下，我们到目前为止聊了四十分钟关于投资这些AI项目的逻辑。

So to basically sum it up, we've talked for forty minutes so far about the logic of investing in these AI projects.

我们已经把所有逻辑都过了一遍，我也提出了所有的论点。

And we've gone through all the logic and I've made all the arguments.

你也提出了反驳的论点。

You've made the counter arguments.

基本上你说的归结起来就是，这只是群体思维和便利性的问题，这就是为什么所有这些钱都流向这个方向。

And basically what you're saying it boils down to is this is just a group think and convenience thing, which is why all this money is going this direction.

这占了很大一部分。

There's a huge component of that.

还有一点，我经常提出这个论点，美国历史上最大的泡沫通常要么与技术有关，要么与宽松信贷、政府政策有关。

There's also this and I've made this I make this argument all the time that the largest bubbles in US history usually had either to do with technology, loose credit, government policy.

对吧？

Right?

这些因素的某种组合。

Some combination of these things.

美国尤其擅长那些也涉及房地产的泡沫，所以我们可以把房地产也加进去。

The US in particular is very good at ones that also include real estate, so we can add that to the mix.

所以是科技、房地产、宽松信贷、政府政策。

So technology, real estate, loose credit, government policy.

这是美国历史上第一次同时处于这四者的交汇点。

This is the first moment in US history that sits at the intersection of all four of those.

所以，那些生活在各自泡沫中的人，觉得可以用自己的理由来为正在发生的事情辩护，这并不特别令人惊讶。

So it isn't it shouldn't be particularly surprising that we have people who live in each of those bubbles, who feel as if they can justify what's happening on their own bases.

所以，有房地产投资者把这看作一个房地产项目，他们会说，你看，看看这个。

So I have real estate investors who see this as a real estate project, and they're like, you know, look at it.

我经常做长周期、高资本支出的项目。

I do long duration, high CapEx projects all the time.

别告诉我该怎么做。

Don't tell me what to do.

我有搞技术的人告诉我，这是历史上最重要的技术。

I have technology people telling me this is the most important technology in history.

人们总是告诉我们，这些事最终都会成功，这就是安德鲁·塞利安的观点。

People always tell us that these things are gonna when they always work out, it's the Andrew Seylian argument.

而且，真正把这件事推向悬崖边缘的另一个因素是，这被视为与全球一些竞争对手的生死之战，比如中国。

And it's so and then the other the other piece that's pushing this to a real cliff is that it's also seen as a scene as an existential battle with some of our competitors around with first of competitors around the world like China.

所以这里有政府层面的因素，我们必须赢。

So there's this government component where we must win.

我们必须赢，因为对。

We must win because Right.

不赢的话，某种程度上就预示着未来的衰落。

To to to not win is to somehow, you know, foreshadow some future decline.

所以，站在这四个力量的交汇点上，这个想法极其重要，因为你面对的是四个非常有力的理由，其中一个就是那些手握巨额资本的大型基金，他们需要开出大额支票。

And so the notion the the idea of sitting at the intersection of those four forces is incredibly important because you're getting up with four very powerful justifications, just one of which is the large funds with point masses of capital who need to write large checks.

所以，在音乐停止之前，支票会不断涌来。

So the checks will keep coming until the music stops.

休息之后，我想聊聊什么会让音乐停下来，以及音乐停止后会发生什么。

On the other side of this break, I wanna talk about what would cause the music to stop and what happens when it stops playing.

我们稍后回来。

We'll be back right after this.

大家好。

Hi, everyone.

我是Alex Kantrowitz。

Alex Kantrowitz here.

我想向大家介绍一部我和Gravity合作制作的纪录片，探讨AI代理安全的未来。

I wanna tell you about a documentary I've made with Gravity to explore the future of AI agent security.

为了了解我们是否真的准备好迎接自主代理，我采访了MIT教授Ramesh Raskar、前白宫首席信息官Theresa Payton、Michelin集团首席数据和AI官Ambika Rajagopal，以及阿里巴巴前高管Sharon Guy。

To find out if we're truly ready for autonomous agents, I sat down with MIT professor Ramesh Raskar, former White House CIO Theresa Payton, Michelin's Group Chief Data and AI officer Ambika Rajagopal, and Sharon Guy, a former executive at Alibaba.

他们每个人都对这个不断发展的领域提供了独特的见解。

They each offer unique insights into this evolving landscape.

最后，我们与Gravity的首席执行官Rory Blundell进行了对话，讨论了未来的发展方向。

We conclude with Rory Blundell, CEO of Gravity, to discuss the path forward.

在Gravity的引领下，加入我们的旅程吧。

With Gravity leading the way, join us on this journey.

你可以在节目笔记中的链接观看完整纪录片。

You can watch the full documentary at the link in the show notes.

欢迎回到Big Technology播客，今天我们请到了投资人和分析师Paul Kedrosky。

And we're back here on Big Technology podcast with investor and analyst Paul Kedrosky.

你可以去paulkedrosky.com订阅他的精彩通讯。

You can go and sign up for his great newsletter at paulkedrosky.com.

好的，Paul。

Alright, Paul.

你知道，我们在休息前聊过，只要音乐不停，钱就会源源不断地来。

You know, we talked a little bit on the other on the on the first side of this this break or before the break about the money will keep coming until the music stops.

我想，要让音乐停止，肯定得发生什么大事。

I mean, I imagine it would take something dramatic for the music to stop playing.

你觉得会是什么引人注目的事件呢？

What do you think could be the compelling event?

所以我的观点是，人们容易陷入一个误区，认为事情要么是这样，要么是那样。

So the argument I make is, you know, people fall into this trap of saying it's gonna be this or it's gonna be that.

我觉得实际上从统计学角度来看，结果是多因素决定的，也就是说，导致它停止的方式有很多种，你唯一能说的就是它一定会停止，因为它可能因为宏观事件而停止，比如外部资本提供者要求的回报率门槛变了。

I think it's actually overdetermined in a statistical sense, meaning that there are so many different ways it can stop that the only thing you can say is that it's going to stop because it could stop because of a macro event that changes the hurdle rate that external capital providers are looking for.

如果突然我要求的回报率变成高个位数，而不是6.5%，那么数据中心项目，尤其是那些底层代币价值不断缩水的项目，竞争力就会大大下降。

If I'm suddenly looking for, you know, high single digits and not six and a half anymore, well, then all of a sudden data center projects with their deflating underlying token comprising looks much less competitive.

这会带来巨大的变化，因为现在超过一半的数据中心项目资金，或者说一半的资本，都来自外部融资。

So that changes things dramatically given that more than half of data center projects now or half of the capital for data center projects now are external financing.

所以这会让情况发生剧变。

So that changes things dramatically.

资本提供者撤回资金，这显然是一个原因。

The the so so the the the providers of capital pulling back is a is a an obvious source.

然后，显然，还有IPO后的现象，这些公司必须在代币价值不断缩水的情况下产生有竞争力的回报，然后向高端市场进军，却发现随着他们向高端市场进军，回报并不存在，而他们还在继续大举投资资本开支。

And then, obviously, the the post IPO phenomenon of having these these companies having to generate competitive returns on the back of a deflating commodity and then moving up market and discovering the returns aren't there as they move up market and they continue to spend aggressively on CapEx.

投资者很快就会对此感到不满，正如我们长期关注这个领域所知道的那样。

Investors become, you know, unhappy about it very, very quickly as we know from hanging around this stuff for a long time.

所以，用不着太多因素，人们就会觉得这个投资机会远没有我之前想的那么有吸引力，因为他们不得不立刻放弃我原本以为他们在卖的东西，现在还得往高端市场走，去追应用。

So it wouldn't take very much to have people feel like this is this is a much less compelling investment opportunity than I felt like because they're having to immediately abandon the thing that I thought they were selling, and now they're having to move up market and chase applications.

我对这个已经没那么兴奋了。

I'm not that excited about that anymore.

所以，你看，这里面有各种各样的因素。

So, like, you know, there's a host of these of these different pieces.

另一个很明显的就是——而且我们已经听到一些风声了——随着这越来越变成国与国之间的生死存亡之战，你可能会看到出口管制的实施。

Another one obviously is and we're seeing rumblings of this already is that as this becomes increasingly the state versus state existential battle that you could see export controls instituted.

所以，我们不能用中国的模型。

So, we can't use Chinese models.

中国的公司也不能用美国的模型。

Chinese model Chinese companies can't use US models.

市场开始碎片化了。

We start balkanizing the market.

碎片化的市场看起来会慢得多，也小得多。

The balkanized market looks much slower and smaller.

嗯，我不知道我愿不愿意为此买单。

Well, I don't know what I'm willing to pay for that.

那情况就不同了。

That changes things.

政府介入。

Government involvement.

我们已经在讨论Anthropic或者OpenAI可能让美国持有5%的股份。

So we're talking already about Anthropic or, I guess, it was OpenAI having potentially a 5% US share in it.

我愿不愿意？

What am I willing?

作为投资者，我对此怎么看？

How do I feel about that as an investor?

我想让美国政府成为我公司的共同投资者吗？

Do I want The US as a co investor in my company?

这会带来什么改变？

How does that what does that change?

那我应该愿意付多少倍呢？

What what multiple should I be willing to pay?

你知道，你可以沿着这些思路一直想下去。

You know, you can go down all of these paths.

如果这些东西真的是资本密集型，就像铁路或者公用事业那样，正如我一直说的，而且我们在微软和其他一些超大规模云服务商那里已经看到了这种情况，那么就需要重新估值。

And if these things are truly CapEx intensive, much like the, you know, the railroads or much like utilities as I've argued, and we see this already in in Microsoft and some of the other hyperscalers, then there's a rerating required.

我不愿意为一家本质上具有公用事业级资本支出、并且回报率逐渐下降到类似公用事业水平的公司支付30倍的市盈率。

I'm not willing to pay a 30 times price earnings multiple in a company that essentially has u u a utility class CapEx usage and, you know, sort of an asymptotic decline back towards a a more utility like hurdle rate.

所以有很多种可能的结果，而唯一不会出现的情况就是它真的成为AGI的看涨期权。

So there's so many ways this can break, and the way it doesn't break is if it's actually a call option on AGI.

对。

Right.

这似乎是唯一的路。

That seems like the only way.

因为，我在看，我列了一个清单，列出了你可以为这些大型AI实验室销售的任何东西价格最终都会下降这一事实提出的各种论点。

Because, like, I'm looking at the made a list of like the different arguments that you can make for like the fact that anything these big AI labs are gonna sell will the price will inevitably come down.

这撑不起那个投资。

It won't support the investment.

我们已经聊过其中一些了，但比如开源模型就能让专有模型的成本降下来。

We've covered a number of them, but like the open source models out there can make the proprietary model costs come down.

你以前也说过这个。

You've talked about this in the past.

还有小模型。

There would be small models.

对。

Okay.

所以现在有些小模型在某些领域做得和大模型一样好，它们也能把成本拉下来。

So you have small models out there that are doing as good of a job in some areas of as the big models and they can bring the cost down.

我还记了一条，是关于扎克伯格的。

Another thing I wrote down to Zuck.

对吧？

Right?

马克·扎克伯格可能觉得，不让OpenAI和Anthropic主导下一个范式对他有利，而且他已经开始行动了。

Mark Zuckerberg probably sees it to his advantage to like not have OpenAI and Anthropic dominate this next paradigm and he's already trying.

首先，从开源开始。

First, started with open source.

现在他推出了自己的新专有模型，但成本只有大约25%。

Now he's starting with his new proprietary model, but the cost is like 25%.

再加上所有这些超级应用的出现，这有点像这些公司在产品方面的希望，我们俩刚才都认为这会更重要。

Then you add the fact that all these super apps that are coming out which is sort of like the prayer for these companies on the product side which we just both I think agreed is gonna be more important.

好吧，OpenAI会有一个超级应用，Anthropic也会有一个超级应用，还会有其他公司的，谁知道呢。

Well you're gonna have a super app from OpenAI, you'll have a super app from Anthropic, you're gonna have a super app from you know, who knows what.

突然间，你就会想，如果我能用更便宜的其他超级应用，为什么还要花这么多钱用这个呢？

And all of a sudden, you're just like, why am I, you know, paying all this money to use the super app if I could just use a different one for cheaper and

嗯，而且甚至更不。

Well, and even and even more no.

不。

No.

当然。

For sure.

而且更根本的是，这些超级应用或者随便你怎么叫，我觉得它们就像缰绳。

And even more fundamentally, these super apps or whatever you wanna call it, like I I think of them as harnesses.

对吧？

Right?

它们坐在上面，协调模型在做什么。

They sit on top and kind of orchestrate what models are doing.

所以像缰绳，比如，你知道，quad code，像codex，我觉得它正在改名。

So harnesses like, you know, quad code, like codex, which I think is being renamed.

但不管怎样，codex，像OpenCode之类的，所有这些越来越像，我用的比喻是，就像你有一群调皮的孩子，模型就像调皮的孩子，而缰绳就像非常非常能干的保姆，所以它们带着调皮的孩子，让它们真正做有用的事情。

But anyways, codex, like OpenCode or whatever, All of these increasingly are like the analogy I use is is it's kind of you've got a bunch of bratty kids and the models are kind of bratty kids and the harnesses are kind of like really really high functioning nannies and so they take the bratty kid and they make them actually do useful stuff.

对吧？

Right?

所以，是的。

So Yeah.

过去十八个月我们看到的进步，很大程度上其实是靠给那些调皮的孩子套上缰绳，也就是请来高效的保姆，而不是模型本身结构上的改进。

Much of the improvement we've seen in the last eighteen months has really been about the imposition of harnesses, effective nannies sitting on top of bratty kids, and not about the actual structural improvements in the models themselves.

这其实是个很大的误解，但它反映了未来的趋势：投资者会越来越多地审视这些，然后问，为什么我还要继续花十亿美元去搞一次大规模的训练，而模型可能十八个月后就过时了？因为别自欺欺人了，GPT-5.6并不是一次大规模的训练。

And that's a a sort of a huge misunderstanding, misunderstanding, but but it's it's reflective of where we're going in the future that investors increasingly are going to look at this stuff and say, well, why am I continue why are you continuing to spend a billion dollars on a huge training run for a model that may be out in eighteen months because let's not kid ourselves, like, GPT 5.6 was not a massive training run.

它只是在一个两年前就预训练好的基础模型上做了相对较小的增强，我们看到的多数进步都来自缰绳和后训练，比如所谓的人类反馈强化学习（RLHF），以及其他事后加入的工具。

This was a a relatively modest enhancement on an existing foundational model that was pre to pre trained pre trained probably two years ago now, and most of the gains we're seeing are harnesses and post training, was things like what are called reinforcement learning with human feedback, RLHF, all of these other tools that are now coming in after the fact.

一旦投资者深入了解，看到越来越多这样的情况，他们就会质疑，为什么我们要在预训练上花这么多钱？

Once investors look under the hood and see more and more of this, they'll be questioning, why are we spending so much on pretraining?

如果模型的大部分进步都来自后训练、RLHF和其他工具，甚至量化之类的，那为什么还要进行十亿美元级别的训练？

Why are you doing billion dollar training runs anymore if most of the gains and models are coming from post training in RLHF and some of these other tools or even, like, quantization and whatever else?

这些公司将面临巨大的压力来削减开支，这会对超大规模云服务商和整个行业产生巨大的连锁反应，因为那是他们赖以生存的生态系统。

There's gonna be immense pressure for these for the companies to cut back on that spending, which will have huge knock on consequences across the hyperscalers and across the board because they're that's the food the ecosystem they live in.

那是他们依赖的食物链。所以，另一个潜在的失败方式是，当人们意识到，随着对缰绳的依赖增加，这些训练开支中很大一部分或越来越多的部分可以被砍掉，而对未来模型效用几乎没有任何影响时，就会出问题。

That's the food system they rely So it's another way that this can potentially fail is when the realization strikes that a lot of this or increasing fraction of this training expenditure could be could be wiped out with almost zero consequence for the future utility of what we of what we will get from these models, given the increasing reliance on harnesses.

好的。

Okay.

那么接下来，你懂的，很自然的问题就是，音乐停了会怎么样？

And so then, you know, the obvious follow-up here is, well, what happens when the music stops playing?

你明白吧？

You know?

如果真的停了的话。

If it does.

对。

Yeah.

那么然后

So then

假设你是对的。

Assuming you're right.

因为我们说的，又是那种压缩的巨额投资周期，很多公司可以说把未来都押在上面了。

Because we're talking about, like, again, a a compressed massive investment cycle with, like, many companies, you know, betting, say, you could say their future on it.

所有这些表外融资，我们今天其实没怎么深入聊，但，嗯。

All this, like, off balance sheet financing, which we really didn't get into today, but, Mhmm.

这不是用传统方式融资的。

It's not being financed in traditional ways.

你知道，我们又看到信用违约互换重新出现或成为焦点。

You know, we're seeing the return or the highlight of credit default swaps again.

嗯。

Mhmm.

那谈谈如果这个垮了会发生什么。

So talk a little bit about what happens if this goes under.

嗯，这在很多方面类似于全球金融危机期间发生的情况，我们会发现它如何在整个经济中扩散，因为机构投资者中越来越多的部分，数学上，由于这是过去六到十二个月高收益和投资级债券发行中如此大的比例，而且未来这个比例还在增长，他们必须是这些东西的持有者。

Well, so it's it's in many ways analogous to what happened during the global financial crisis is that we find out how it has metastasized across the economy because there's an increasing fraction of the institutional investor population that mathematically because this is such a large fraction as a percent percentage terms of issuance in both high yield and prime, It's for high both high yield and investment debt over the last six to twelve months, and it's a growing fraction of it going forward.

数学上，他们必须是这些东西的持有者。

Mathematically, they must be holders of this stuff.

所以不管你喜不喜欢，你都已经身在其中了。

So you're you're you're in that you're in it whether you like it or not.

而且人们意识到，即使只是持有标普500指数基金，他们也身在其中，因为超大规模企业和AI相关股票的集中度现在大约超过40%。

And so people are realizing that they're in it even by holding an S and P 500 index fund because of the the concentration of hyperscaler and and AI related names, which is something like 40 odd percent now.

所以即使你试图分散投资，你仍然深陷其中，但情况比这更隐蔽，因为它藏在那些被委婉地称为高等级投资债券里，而这些债券越来越多地被与超大规模企业相关的债务所占据。

So you even by trying to diversify, you're still you're still in it, but it's even more insidious than that because it's sitting inside of what euphemistically might think of as high higher grade investment bonds that are increasingly being taken up by hyperscaler related debt.

然后这又会迅速在像PIMCO这样的地方首先显现，接着他们转手，最终它会进入某家保险公司。

And then that in turn is rapidly it shows up at a place like PIMCO first, and then they'll flip it and it ends up inside of, you know, some insurance company.

它还会蔓延到欧洲的银行。

It'll be it'll be, you know, spreading across European banks.

它会出现在所有我们曾经惊讶地发现美国房地产和CMBS债务的地方，随后还有2007、2008、2009年期间的信用违约互换和信用违约互换指数。

It'll be in all the same places that we were, like, startled to find real US real estate and CMBS debt, and then subsequently, credit default swaps and credit CES squares back in the back in the 2007, 2008, 2009 period.

所以这完全是照搬同一套剧本，只是换了不同的资产类别。

So we're gonna it's literally following the same playbook just with a different asset class.

这太可怕了。

That is scary.

那你打算怎么应对？

So how are you playing it?

我的意思是，你是个投资者。

I mean, you are an investor.

你是在做空某些东西，还是你有什么计划？

Are you, like, shorting certain things, or what is your plan here?

其实我的日常工作大部分是风险投资。

So I very much so my day job in large part is in venture capital.

所以大多数情况下，我们就是不投这个领域。

And so for the most part, we just don't invest in it.

我们就是……其实怎么围绕AI投资并不明显，因为作为风险投资人，最糟糕的事情之一就是进入一个有上千人参赛的马拉松。

We just it's it's not obvious it's not obvious how to invest around AI because one of the worst things you can do as a venture capitalist is get into a marathon where there's a thousand participants.

他们都在起跑线上。

They're all at the start line.

他们都资金充足，训练有素，然后你就会想，天哪。

They're all well funded and well trained, and it's like, oh my god.

我得熬过所有这些人才行。

I'm gonna have to outlast all these people to get to the finish.

所以你真的要选好位置，尽量避开那些资本集中、而且容易导致回报不佳的领域。

And so you you really have to pick your spots and try to stay away from these sectors where people are concentrating capital and doing it in a way that leads to much poor returns.

所以大多数情况下，我们在很多不同领域都非常活跃，但就是不碰AI，这有点奇怪，但并不是因为不相信AI。

So for the most part, you know, we've been we're very active in a host of different areas, but not not AI, which is perverse because it's not because we don't believe in AI.

而是因为我们认为从结构上讲，作为投资者，AI是个很糟糕的领域。

It's because we believe it's structurally a terrible place to be as an investor.

然后从个人资产配置的角度来说，我非常不愿意投入新的资金到任何广泛的指数基金或被动投资类别，已经两年多没这么做了。

And then in on a on a more personal level in terms of, you know, assets, I just very loathe to to commit any haven't committed new capital to any sort of broad index class passive categories in over two years for that reason.

因为不管我愿不愿意，之前的投资已经让我在这个资产上的配置比我想象的要高很多了。

Because you just you whether I like it or not, prior commitments now amount to a much larger commitment to this asset than I would like already.

所以只要我还在市场上有点资产，就已经算是过度投资了。

So I'm already overinvested in this stuff just by the fact of having a pulse and having some assets in the market.

所以必须非常小心，不能让它在不知不觉中增长得太快。

And so you have to be very careful about not having it grow in a really, you know, in a way that you wouldn't otherwise have noticed.

所以从个人和职业角度，我采取了不同的方法，但两者其实是相互映照的。

So this sort of personally and professionally, I take a different approach, but they're all kind of mirrors of the each of the other.

对。

Right.

现在，这不是一个投资建议播客。

Now, is not an investment advice podcast.

我必须得这么说。

I have to say that.

但你不是那种，对于这么确信这一切都会崩盘的人，你却没有采取任何个人做空的大仓位来从中获利。

But you're you're not like for someone with such conviction that this is, you know, all gonna come down, you're not taking any, like, personal big short position where you could benefit.

我的意思是，你甚至有一个你认为这会在一年内下跌的时间点，而且

I mean, you even have a time when you think this goes down in, like, a year and a

半年。

half.

当然。

Sure.

当然。

Sure.

当然。

Sure.

对。

Yeah.

对。

Yeah.

对。

Yeah.

所以这不是我的工作，但你知道，说实话，我更喜欢晚上睡个好觉。

So it's not my job, but I mean, you know, I and I get more pleasure from sleeping well at night is the honest answer.

但我要说的是，我正在和两家大型对冲基金合作，他们做了一些相当复杂的头寸，我已经和他们合作了将近一年。

But I will say so I'm I'm in a I'm working with two very large hedge funds who put on some fairly complicated positions that I've been working with for close to a year now.

所以我间接地，所有偏见都摆在桌面上，与几笔非常大的交易密切相关。

So I am indirectly all biases on the table, very closely associated with a couple of very large trades related to stuff.

而且，你知道，如果成功了，对我对他们都有好处。

And, you know, if it works out, it works out for me and them.

我明白了。

I see.

在我们走之前，我能再问你两个问题吗？

Can I ask you two more before we go?

好的。

Yeah.

当然可以。

Sure.

还是说你得走了？

Or do you have to run?

好的。

Okay.

我只是想问问你关于中国的事。

I just wanna ask you about about China.

中国在这方面的方法很不一样。

So China has a very different approach here.

对吧？

Right?

实际上，你可以告诉我我是否说错了。

It is effectively it is the you could tell me if I'm wrong here.

据我了解，政府基本上在背后支持很多这样的事情。

From my understanding, the government is basically backing a lot of this.

所以如果一切都上涨，那就像政府拨款浪费了，你知道，这种事在全球各地每周三都会发生。

So if everything goes up, it's just like government allocation, you know, wasted which like is that's, you know, happens around the globe on a Wednesday.

对吧？

Right?

这不像他们的整个系统会崩溃。

It's not like, you know, their whole system collapses.

那你觉得他们是否比美国系统更能免受影响？美国系统更依赖私人投资于数据中心和AI训练。

So do you think that they're insulated much more than The US system which relies much more on the private investment in data centers and AI training?

你知道，假如这一切开始崩溃，中国仍然会拥有技术，你知道，最终这是对的。

You know, given like, let's say this all starts to collapse China will still have the technology and you know, that's ultimately right.

我想我们都同意，你知道，最终重要的是什么，其实，我不知道。

I think we agree what, you know, I mean what matters in the end will be like actually, I don't know.

我不想替你把话说出来，但说到底，我们剩下的东西确实很重要。

I don't want to put words in your mouth, but like effectively what we're left with is pretty important.

所以中国可能不会出现经济崩溃，而且，你知道，他们还能继续往这个技术里投资，并且感觉良好。

So China might have no economic collapse and, you know, a technology inside of this thing that they can keep investing in it and and feel good about.

在一定程度上，中国面临的问题——我最近做了很多关于中国的研究——就像电池制造业和太阳能行业曾经发生的那样，全国各地都有巨大的动力去建设这些设施，以给中央政府留下深刻印象。

To a degree, the problem that China has, and I've been doing a lot of work on China lately, is that much like what happened with battery manufacturing, much like what happened with solar, that there is this huge incentive across the country to impress the central government by building these things locally.

所以中国总理最近一直在告诫各省省长，别再建那么多数据中心了，因为这现在是新热点。

And so the Chinese premier has been recently cautioning the provincial governors, stop building so many data centers because this is now the new thing.

之前是电池工厂。

It was battery plants.

再之前是太阳能电池制造，而四十年前，是水电站大坝。

It was solar cell manufacturing, and for a while and forty years ago, was hydroelectric dams.

所以中国有消费者消费不足、地方和中央政府过度支出的历史，这导致了大规模房地产投资和鬼城。

And so China has a history of under of consumers underspending and regional and central governments overspending and which led to, you know, massive investment in real estate and ghost cities.

我预计你会看到同样的现象，尽管社会后果会小一些，一旦这一切都变成——就像过去十年他们在公寓楼、住宅和工业空间上的过度建设一样。

I expect you'll see the same phenomenon, albeit with a little less social consequences once all of this turns out to be it'll be much like what happened with their overbuilding in apartment buildings and and residential and industrial space over the last decade.

他们确实有点动摇，但并没有因此崩溃，这

Which they I mean, they definitely shook a little bit, but they didn't crumble from it, which

很有启发意义。

is instructive.

这很有启发意义。

Which is instructive.

我觉得情况会有点像那样。

And I think it'll look a little like that.

这在很大程度上是因为这个经济不像西方经济体，尤其是美国那样依赖消费支出。

And that's in large part because this is an economy not as reliant on consumer spending as Western economies in The US in particular.

对。

Yep.

所以保罗，我能总结一下你的观点吗？基本上就是，我们这里有一种正在商品化的技术，实际上，除了销售纯粹的智能之外，任何进一步的举动都不会容易。

So Paul, can I can I summarize what your position is, which is basically, we have a technology here that is commoditizing that is effectively, you know, any move beyond just selling sort of like pure intelligence is is not gonna be easy?

与此同时，我们正在大规模建设数据中心，基于它会带来回报的承诺，投入了大量资金，但最终，成本会比人们预期的要高得多。

And alongside that, we have a data center build out that is getting a ton of money based on this prompt on the promise that it will pay off, but ultimately, we will be much more expensive than people anticipate.

这两个因素加在一起，必然会导致崩溃。

And that is gonna lead those two factors combined will lead to an inevitable collapse.

对。

Yeah.

我只想补充一点，从结构上看，数据中心之所以会变成更棘手的生意，即使其他所有环节都顺利，是因为这项技术比现代史上任何技术都更快地进入市场，并且比任何技术都更快地达到十亿用户。

And the only the only piece I would add to that is that structurally one of the reasons why the data centers will become an even more fraught business even with all of the other pieces working out is that this technology came to market faster than any technology in modern history and reached a billion users faster than any tech.

所以表面之下，存在巨大的低效。

So under the hood, there are vast inefficiencies.

科技行业非常擅长消除巨大的低效，无论是通过推出新的芯片，还是改进软件编译器，或者其他任何方式。

The technology industry is very good at wiping away vast inefficiencies, whether it's through the launching of new silicon or improving in software compilers or anything else.

所以总的来说，我们应该预期代币通缩会持续，甚至在未来加速，因为这东西进入市场的速度太快，而且提高效率的机会太多了。

So that's all a long way of saying that we should expect token deflation to continue and even accelerate in future because of how quickly this stuff came to market and how many opportunities there are to drive efficiencies.

所以这给底层经济带来了巨大的压力。

So that creates just incredible pressure on the underlying economics.

好的。

Okay.

最后一个问题。

Last one for you.

经历了这一切之后，AI会是什么样子？

What does AI look like after all this?

就是说，虽然在你看来前面会有问题，但总会有赢家。

Like, is there, you know, even though that there's a prob there there will be problems ahead in your view, there will be a winner.

技术会继续进步。

The technology will continue to advance.

你觉得，我是说，我知道基于AGI的看涨期权来投资可能不明智，但你认为这技术有可能走向那个方向吗？

Do you think I mean, you know, I I know that investing based off of a call option on AGI might be unwise, but do you think that there's a chance that that is where this technology goes?

在你看来，未来会是什么样？

What does the future look like in your perspective?

不是这一代技术。

Not this generation of technology.

大型语言模型有一个深层的结构性问题，就是它们无法轻易地实时更新模型权重。

There's this deep structural problem with large language models that they can't, they can't easily update the model weights in real time.

所以从这个角度看，它们不是动态系统。

So in that sense, these are not dynamic systems.

而且，你知道，Yan Lakun和其他人，之前在Meta的研究员，现在自己去做世界模型了。

And, you know, Yan Lakun and others, the former researcher at at at Meta is now off doing his own world model thing.

有很多人都会这么说。

Sort of there's lots of people who appoint who who will say the same thing.

所以我怀疑这不是正确的路径，但我确实认为这是一项极其有价值的技术，从某种意义上说，它会消失。

And so I doubt that this is the this is the path, but I do think it's incredibly valuable technology that in a sense will disappear.

然后，它会变得像电力一样。

And in that, it will become like electricity.

它是一种基础设施。

It's a utility.

它会成为许多其他持续运行的事情的基础，而且我可能不会比知道现在给我的MacBook供电的水电站大坝更清楚我的token是从哪里来的。

It will become underlying a host of other things that go on all the time, and I will no more know who provides my tokens than I do from which hydroelectric dam the power came from that's powering my MacBook right now.

所以，OpenAI和Anthropic这些公司，它们的未来是？

So the OpenAI and Anthropics of the world, their future is?

就像水电站一样。

Like a power dam.

我不知道他们在哪里，也不知道他们是谁，但我想他们存在，而且他们会获得类似公用事业的回报率。

I don't know who I don't know where they are or who they are, but I guess they exist and they'll earn utility like rates of return.

好的。

Okay.

保罗，非常感谢你。

Paul, thank you so much.

真的很感谢你今天抽出时间。

Really appreciate your time today.

是的。

Yeah.

当然。

Sure.

没问题。

No problem.

好的。

Alright.

太好了。

Great.

好了，各位，记得订阅保罗的通讯。

Well, folks, do sign up for Paul's newsletter.

网址是 paul kedrosky dot com。

It's at paul kedrosky dot com.

今天真的很棒。

This has been great.

希望我们还能再合作。

I hope we can do this again.

再次感谢保罗，我们下期大科技播客再见。

Thank you again to Paul, and we'll see you next time on big technology podcast.