OpenAI 如何引发美国经济衰退？——与史蒂夫·艾斯曼的对谈

我拿到数字了。

I got the numbers.

它涨了50%。

It jumped 50%.

三个月内涨了43%。

43% in three months.

本季度达到168亿。

To 16.8 billion for the quarter.

所以三个月内涨了43%，但相比第二季度140%的增长率有所下降。

So it jumped 43% in three months, but that's down from a growth rate in the second quarter of 140%.

当然，那不可能持续。

Well, that's not sustainable, of course.

对，但关键是，我觉得tokenmaxxing的结束开始产生影响。

Well, yeah, but the point is, I think the end of tokenmaxxing is starting to have an impact.

所以他们通过tokenmaxxing赚的钱比他们持续经营能合理赚到的要多——

So they were making more money as a result of tokenmaxxing than they reasonably could have made on an ongoing basis-

是的

Yes

……一旦公司看了预算，就会说：‘我们显然不能以这个速度在AI上花钱，尽管这项技术很棒。’

... once companies looked at their budgets and said, "We obviously can't spend at this rate on AI, as great as the technology is."

没错。

Correct.

好的。

Okay.

所以一个有趣的连锁反应是，在当前的利率水平下，新数据中心融资的供应显然会变得更难获得。

So one interesting ripple is that with rates where they are, the supply of financing for new data centers is obviously going to be harder to come by.

绝对。

Absolutely.

更难证明其合理性。

Harder to justify.

不过，需求方面，尽管随着开放权重模型的出现，情况可能略有变化，但供应限制仍然到处都是。

The demand side of it, though, even though it might be changing a little bit with the open-weight models, there are still supply constraints all over the place.

所以如果供应受限而需求继续增长，那市场格局可能会变得很有意思。

So if the supply gets limited and the demand continues to increase, that could make for an interesting dynamic in the marketplace.

确实有可能。

It could.

现在的一个问题是，全国各地的数据中心都遭到了很多反对。

Part of the problem right now is that there's a lot of pushback on data centers all over the country.

这个很难量化，但是——

It's very hard to quantify it but-

什么不太受欢迎？

What's less popular?

呃……

Uh...

我刚想说——

I was about to make a-

我不知道。

I don't know

……一个不太好的玩笑，但是有——

... bad tasteful joke, but there's-

也许是道奇队。

Maybe the Dodgers.

好吧。

Fine.

但是——现在反对声音太大了，数据中心的大规模建设似乎会大幅放缓。

But- There's so much pushback that all this, the build-out of data centers seems to be it's going to slow by a lot

这对卖算力的人来说是利好。

Which would be bullish for people that sell compute.

数据中心少了。

Less data centers.

我不确定这是不是利好，因为这意味着你会买更少的算力，因为——

I don't know if it would be bullish because it would mean you're going to buy less compute because-

因为价格

Because the price

……你根本建不了数据中心。

... you can't build a data center.

如果建不了数据中心，你的算力就没地方放。

If you can't build a data center, you got no place to put your compute.

但现有的算力会变得更加抢手。

But what compute exists becomes more in demand.

很可能是这样。

That's probably the case.

好的。

Okay.

所以其实我想问你这个问题，因为我觉得这是故事的一部分。

So I actually wanted to ask you about that because I think that's part of the story.

我觉得对AI的需求不会变。

I don't think the demand changes for AI.

我觉得就像我们面前这群人，他们刚花了几天时间——

I think like this group in front of us, they've just spent a couple of days-

我同意。

I agree.

关键在于价格。

It's the price.

关键就是-

It's the price-

……AI的价格才是最重要的。

... of the AI that matters.

所以眼前这帮人刚花了两天时间跑遍了华尔街的所有投行。

So this group in front of us has just spent two days going into all the investment banks on Wall Street.

每一家投行在27年给AI和AI相关项目以及招聘的预算，都会比26年和25年大幅增加。

Every one of them is going to have a substantially larger budget for AI and AI-related projects, and AI-related hiring in '27 than they had in '26 than they had in '25.

企业层面的这部分需求是不会变的。

That part of it at the enterprise level is not going to change.

另一件大概率不会变的事是消费者需求，我觉得这主要归功于ChatGPT，而且——

The other thing that probably won't change is consumer demand, which I think is mostly ChatGPT, and-

还有Muse。

And Muse

……

...

现在Muse在智能体方面。

now Muse, on the agentic side.

还有Gemini，现在每次谷歌搜索实际上都已经是AI了。

And Gemini, every Google search is now an AI de facto.

所以这个也不会消失。

So that's not going away either.

任何在使用这项技术的人都不会说：“你知道吗？

Nobody that's using this technology is going to say, "You know what?

我真的不需要这个，”或者“我还没把它融入我的生活，我要回到2022年的做事方式。”

I don't really need this," or, "I haven't made this a part of my life, and I'm just going to go back to the way I was doing things in 2022."

我觉得这些都不太可能发生。

I don't think any of that's really going to happen.

所以债券的价格，发债的难度，其实不会真正改变需求端。

So the price of the bonds, the difficulty in selling debt is not going to really change the demand side.

我觉得变的是谁有能力去建设产能来满足那个需求。

I think what changes is who gets to build the capacity to serve that demand.

而且可以说，如果像CoreWeave和Nebula这样的公司没法发债，也没法按他们想要的程度供应算力，那球就又回到了超大规模云厂商那边，亚马逊、Alphabet、微软，他们的地位比一年前更强了，那时候可能债务更便宜，或者债券的胃口更大。

And arguably, if the CoreWeaves and the Nebulases of the world are not able to sell debt and supply as much compute as they want to, the ball just goes back into the court of the hyperscalers and Amazon, Alphabet, Microsoft, they're in an even stronger position than they were in a year ago when maybe debt was cheaper or there was a bigger appetite for bonds.

从投资的角度，你怎么看这个想法？

What do you think about that idea from an investing standpoint?

超大规模云厂商的地位只会更强。

The hyperscaler's position just strengthens.

我同意超大规模云厂商的地位确实增强了。

I agree that the hyperscaler position has strengthened.

我最关注的是这个业务太集中在Anthropic和OpenAI身上了，如果打价格战，那些都不重要。

What I'm most focused on is this business is so concentrated from Anthropic and OpenAI, that if there's a price war, none of that matters.

好的。

Okay.

整个市场都会回调。

The whole thing will correct.

如果三四年后发生这种情况，可能就无所谓了，因为到时候会有很多其他公司来消化这些算力。

Now, if three, four years from now that happens, it probably won't matter because there'll be a lot of other companies to take the compute.

但现在，市场太集中了，你最好祈祷那两家公司表现好。

But right now, it's so concentrated that you better hope that those two companies do well.

看起来现在的定价更多是基于使用量，而不是像过去15年那样，企业软件是按SaaS模式销售的。

It seems that the pricing is based on usage now more so than the last 15 years, the way enterprise software was sold was SaaS.

你按人头付费，比如某家公司有1000名员工。

You pay us per head, so such and such company has 1,000 employees.

Salesforce会按每个员工收80美元，这就是SaaS合同。

Salesforce is going to take $80 per employee, and that's the SaaS contract.

现在，它已经转向了按使用量付费的模式。

Now, it's migrated more toward a pay what you use model.

对我来说，这比我们之前以为他们会怎么卖这些服务要更可持续。

That feels more sustainable to me than the way we thought they were going to sell these services.

而且价格也要贵不少。

Well, it's also a lot more expensive.

对，不过我觉得企业对这些费用的掌控力会更强一些，当然也可能并非如此。

Yes, but I think that companies have more control over those, or maybe it's not that way.

我是觉得成本确实更高了，所以企业会更注重控制开支，也会更多地让员工去使用那些更便宜的模型，这就会给 Anthropic 和 OpenAI 带来降价的压力。

Well, I think it's more expensive, and so companies are going to be more cost-conscious, and they're going to get their employees to use the cheaper models much more, which is going to put pressure on Anthropic and OpenAI to cut price.

明白了。

Okay.

这是你的看空观点。

That's your bear case.

好的。

Okay.

你觉得我们正在见证这一趋势的开端吗？

And you think that's something that we're witnessing the beginning of or?

我觉得我们可能确实看到了它的开端。

I think it's potentially that we're seeing the beginning of it.

好的。

Okay.

那传导机制是怎样的？OpenAI出了问题，事情发生，然后我们就陷入经济衰退，全面经济衰退？

What's the transmission mechanism by which OpenAI has a problem, things happen, and then we're in an economic recession, economy-wide recession?

那我们假设有一场全面的价格战。

So let's imagine that there's a full-blown price war.

好的。

Okay.

对吧？

Okay?

我们就假设它发生了，好吗？

Let's assume it, okay?

我不知道会不会发生，但我们就假设它发生了。

I don't know if it's going to happen, but let's just say that it does.

Anthropic的收入增长转为负值。

Anthropic's revenue growth turns negative.

没人会再给他们投钱了。

Nobody's going to fund them.

他们得削减自己的承诺了。

They got to cut back on their commitments.

甲骨文直接完蛋了，因为甲骨文一半的积压订单都来自OpenAI。

Oracle goes right down the toilet because half of Oracle's backlog is just from OpenAI.

顺便说一句，它的股价现在交易得像——

By the way, its stock price is trading like-

像已经完蛋了一样。

Like it's in the toilet.

对，没错。

Yes, exactly.

所以，给那些不知道的人说一下，去年第三季度左右，甲骨文发布财报之前，股价是230美元。

So just for those of you who don't know, last year, around the third quarter before Oracle reported, the stock was $230.

然后他们发布了财报，显示积压订单达到了3000亿或4000亿，我记得是4000亿。

And then they reported, and they showed that their backlog had gone to 300 or 400, I think 400 billion.

三个月内就涨了三千亿。

It was up 300 billion in three months.

然后所有人都疯了，股价从230美元涨到330美元。

And everybody went nuts, and the stock went from $230 to $330.

你还记得吧，就两天之内。

You remember this, like in two days.

我从没见过一只股票这样涨100点。

I never saw a stock go up 100 points like that.

这可是市场上最大的股票之一。

One of the biggest stocks in the market.

超级大。

Huge.

就两天。

In two days.

然后，不知道是谁，可能是戴维森的吉尔·卢里亚，发了一份报告说，这三千亿全是OpenAI带来的。

And then, I don't know who came out, it might've been Gil Luria from Davidson, put out a report to say that the entire 300 billion was just OpenAI.

然后所有人都吓坏了。

And everybody just freaked out.

那就是最高点了。

That was the top.

然后股价从330美元跌到230美元，现在只有130美元了。

And then the stock went from $330 to $230, and now it's at $130.

因为人们不信任——

Because people don't trust-

他们不信任，你不能依赖一个亏损的公司。

They don't trust it, that you can't rely on a company that loses money.

你把未来押在了一个疯狂烧钱的公司上。

You've hinged your future on a company that bleeds cash like a lunatic.

这几天股市真是很有意思。

This has been a very interesting couple of days in the stock market.

我同意。

I agree.

所以，就像我昨天说的，通胀数据很高，大家当时真的觉得‘完了’，心想‘这情况不会好转了……’

So, as I said yesterday, a hot inflation print, and it really felt like people were like, "Oh, shit," like, "It's not getting-"

确实很高。

It was hot.

情况很糟。

It was bad.

根本不可能……

It's not-

今天早上情况更严峻了

And this morning was hotter

……形势并没有好转。

... it's not getting better.

今天我们看到了生产者物价指数。

We saw producer prices today.

昨天是消费者数据，今天是生产者物价指数。

Yesterday was consumer, today was the producer prices.

昨天也不太好，你之前那些存储股，本来是市场的宠儿，结果被狠狠打击了。

Also not good Yesterday you had the memory stocks, which were the belle of the ball, killed.

有趣的是，西部数据，它已经回撤了54%。

Fun fact, Western Digital, it's in a 54% drawdown.

54%？

54%?

对，54%。

54.

哇。

Wow.

而且Steve，它今年至今还是涨了100%。

And Steve, it's still up 100% year to date.

这就说明之前的涨势有多疯狂。

Just to show you how crazy the run was.

我从来没听说过，一只股票被腰斩了，还能涨这么多——

I've never heard of that, a stock getting cut in half and still being up double-

100%。

100%

……年初至今。

... year to date.

真是难以置信。

It's unbelievable.

所以昨天收益率立刻飙升了。

So you had yields spike immediately yesterday.

顺便说一句，那些在最高点买入的人，这句话对他们可没什么帮助。

By the way, the person who bought it at the peak, that sentence doesn't help them.

就算它还是跌了100%，也帮不上忙。

Doesn't help that it's still down 100.

确实帮不上忙。

Doesn't help.

对。

No.

收益率飙升了，而Anthropic显然不是上市公司，但有一家基于区块链的公司叫Hyperliquid，交易量很大。

You had yields spike, and Anthropic, obviously it's not a publicly listed stock, but there is this blockchain-based company called Hyperliquid that does a lot of volume.

过去24小时内，Anthropic的股票交易额达到了1760万美元。

Over the last 24 hours, Anthropic has traded $17.6 million worth of stock.

这个东西，它是一个市场，而且运作得相当不错。

And this thing, it's a market, and it works pretty well.

它在SpaceX的IPO上表现得非常好。

It worked very well for the SpaceX IPO.

所以这就是Anthropic的交易。

So this is Anthropic trading.

这就是Anthropic在Hyperliquid上的交易。

So this is Anthropic on Hyperliquid.

明白了。

I see.

所以昨天Anthropic表现不佳，显然。

So Anthropic was doing bad yesterday, obviously.

然后收盘后，他们的营收数据出来了。

And then the numbers came out after the bell of what their revenue was.

结果暴跌了。

It puked.

半导体股在盘后也崩了。

Semis puked in the after-hours.

大家都慌了。

People are freaking out.

昨晚推特上全是坏消息。

Last night, Twitter was black.

一片漆黑。

Jet black.

大家都吓坏了。

People are freaking out.

然后今天早上——

And this morning-

我并没有慌。

I wasn't freaking out

……你非常冷静，我看到了。

... you were very calm, I saw.

今天早上，很有意思，我觉得Anthropic涨了5.5%。

This morning, very interestingly, I think Anthropic is up 5.5%.

收益率大幅下降，而今天飙升的股票都是市场上最垃圾的股票。

Yields are coming way down, and the stocks that are ripping today are the biggest pieces of shit in the market.

比如Lenoir，我觉得是今天标普500里表现最好的股票。

Like Lenoir is, I think, the best performing stock in the S&P today.

因为房地产。

Because of housing.

所以任何跟利率沾点边的股票都涨。

So anything even remotely rates-related.

市场叙事这样转变，真是太好笑了。

And it's so funny how the narrative of the market changes like that.

上周还说，谁在乎利率啊。

Last week it was, "Who cares about the interest rates?

我觉得它碰不到超大规模企业。

I guess it can't touch hyper scalers."

我觉得它碰不到这个。

"I guess it can't touch this."

然后市场突然开始恐慌，讽刺的是，那就是顶部。

And then all of a sudden the market starts to freak out, and ironically, that's the top.

股市里就是这么有意思。

And it's so funny how that works in the stock market.

利率的顶部，我想在底部我们确实需要坏消息，因为就一天。

And the top in interest rates, and I guess we needed bad news for the stock market at the bottom, because it's just one day.

Anthropic从低点大幅反弹。

Anthropic's ripping off the lows.

标普和等权重指数也从低点大幅反弹。

S&P and the equal weight are ripping off the lows.

这是地球上最棒的游戏。

It is the best game on the planet.

你对利率下降的速度感到惊讶吗？

Are you surprised by how quickly rates came down?

我其实挺惊讶的。

I'm actually quite surprised.

通胀数据非常糟糕。

The inflation numbers were very bad.

我觉得大家都很恐慌。

I guess there's such panic.

人们都在抢购国债——

People were just buying treasuries-

但这就是我们学到的

But here's what we learned

……作为一种避险方式。

... as a form of safety.

我们学到的是这样的。

Here's what we learned.

过去几个月里，大家一直在争论一个问题，我很喜欢你上次说的，关于赤字担忧很多是装腔作势，我同意。

The question, or the debate over the last couple of months, and I love what you said last time about the deficit worries being a lot of virtue signaling, and I agree.

那是个问题——

It was a question-

我说过吗？

I said that?

我什么时候说的？

When did I say that?

你说过。

You did.

在我们这儿说的。

With us.

哦，上次说的。

Oh, last time.

这句话说得真好。

That's a good line.

你当时可是发了一通牢骚。

You went on a rant.

那段话特别受欢迎。

That was very popular.

这就像一团乱麻，很难跟市场理清楚。

It's a hard ball of yarn to untangle with the market.

就像在问，为什么？

It's like, why?

是因为这个还是那个？

Is it because of this or that?

我们并不总是知道，但有些日子我们会发现是什么在推动交易，昨天我们就发现了。

We don't always know, but some days we find out what's driving the trade, and yesterday we found out.

是不是因为世界各国政府的财政不负责任？

Is it the fiscal irresponsibility of our governments around the world?

不，是人工智能。

No, it's AI.

是人工智能在推动交易，因为我们一看到交易出现疲软或走弱的迹象，收益率就从5.4一下子降到5.1。

It was AI driving the trade because the moment we saw a hint of weakness or softness in that trade, the tenure went from 5.4 down to 5.1 like that.

所以全都是人工智能。

So it's all AI.

所以我们并不是突然对财政什么的更放心了。

So we didn't all of a sudden get more comfortable in fiscal whatever.

我觉得不是。

I don't think so.

不是。实际发生的是，人们说：“好吧，我猜为了推动人工智能交易而进行的债券抛售会放缓，因为投资者的兴趣不在了。”

No. What happened was people said, "Okay, I guess all the bond selling to fuel the AI trade is going to slow down because the investor appetite isn't there."

所以买国债。

So buy treasuries.

对。

So right.

所以国债有了喘息的空间，价格就回升了。

So treasuries got some breathing room and came in.

好的。

All right.

这挺有意思的。

This is interesting.

在过去一年里，CoreWeave的CEO，请展示图表5A，丹尼尔。

So over the last year, the CEO of CoreWeave, chart 5A please, Daniel.

CoreWeave的CEO一直在大量抛售股票。

The CEO of CoreWeave has been selling a ton of stock.

为他辩护一下，我们有一张图表显示价格、销售情况以及他卖出的位置。

In his defense, so we've got a chart showing the price, the sales, and where he sold.

总共卖了6.82亿美元的股票。

It's $682 million worth of stock.

为他辩护一下，他不是某天醒来突然说：“哎呀，我不喜欢这只股票了。”

In his defense, he's not waking up one day and saying, "Ah, I don't like this stock anymore."

他就是一直在卖。

He just sells all the time.

他就是一直在卖。

He just sells all the time.

这些都是提前计划好的。

This is all planned in advance.

而且，我敢肯定其中很多是为了支付他获得的一些股票的税款。

And again, I'm sure a lot of this is to pay the taxes for some of the stocks that he's granted.

尽管如此，你会觉得，在一家受到如此关注的公司里，他至少应该对股东说：“好吧。

Nevertheless, you would think at some point with a company that is under such scrutiny, that he would say to his shareholders, "All right.

我不卖了。

I'm not selling.

这已经不合理了。

This doesn't make sense anymore.

我相信这家公司。”

I believe in the company."

他并没有这么做。

He has not done that.

而且昨天，我不知道用‘雪上加霜’这个词是否合适，但史蒂夫，我很想听听你的看法。

And yesterday, I don't know if adding insult to injury is the right word, but Steve, I'd be curious to hear your take on this.

《金融时报》写了一篇文章，标题是‘CoreWeave瞄准无限保险资金来资助AI建设’，因为这是故事的重要部分。

The Financial Times wrote a story, CoreWeave targets, quote, "Unlimited insurance money to fund the AI build-out," because that is a big part of the story.

你能想象像那样摆姿势说：‘现在是我该利用我的保险合同来接管的时候了’吗？

Could you imagine posing like that and say, "This is the time that I think I should make my pitch from my insurance contracts to take over?"

所以杰德，我觉得马克·沃尔特应该把洛杉矶道奇队卖回给他自己的保险公司。

So Jade, I think Mark Walter should sell the LA Dodgers back to his own insurance company.

那是他买来的。

Which he bought it.

然后用那些收益投资CoreWeave。

And use those proceeds to invest in CoreWeave.

哇。

Wow.

那就会把两个潜在的丑闻撞到一起，变成一个超级大新闻。

That would be a collision of two potential scandals into one big story.

哦，他的墙够大的。

Oh, he got a big enough wall.

咱们把它搞得更复杂一点。

Let's compound it.

史蒂夫，这是市场里人们想称之为下一个大空头的领域之一，很多人在这里插旗。

Steve, this is part of the areas of the market that people, to the extent that people want to call the next Big Short, this is an area where people are planting their flag.

我完全同意。

I totally agree.

人寿保险很危险。

Life insurance is dangerous.

咱们回到马克·沃尔特，因为我刚做了个播客聊这个。

Just let's come back to Mark Walter, because I just did a podcast about it.

他买了洛杉矶道奇队，买了洛杉矶湖人队，还买了好多其他体育队。

He bought the LA Dodgers, he bought the Los Angeles Lakers, he bought a bunch of other sports teams.

而且他没有披露这件事。

And he didn't disclose it.

他就是因为想买才买的。

He just bought them because he wanted to.

听着，我也想拥有洛杉矶道奇队，但我没那么多钱。

Listen, I'd like to own the LA Dodgers too, but I don't have the money.

所以他让保险公司借钱给另一个实体——

So he had the insurance company lend money to another entity-

那个实体也是他拥有的。

Which he owns

……也是他拥有的。

... which he also owns.

但这掩盖了一个事实，那就是他打算用这笔钱亲自掌控——

But it obscures the fact that he is going to use that money to personally take control of the-

然后他就买了洛杉矶道奇队。

And he bought the LA Dodgers.

而且最后还真成了。

And it worked out.

对，确实成了。

Yeah, it did work out.

价格涨上去了。

The price went up.

他买了洛杉矶道奇队，应该能经营得不错。

He bought the LA Dodgers, it'll be able to do well.

还有湖人队。

And the Lakers.

湖人队，花了大概20亿美元投进去？

The Lakers, what, $2 billion, put in money?

他确实投了。

He did.

所以我不知道是他说的还是他们说的。

So I don't know if he said this or they said it.

我看到了那些话。

I saw the words.

我觉得马特·莱文写过这个。

I think Matt Levine wrote about this.

有人把这称为，如果说有犯罪的话，那也是没有受害者的。

Somebody referred to this as, to the extent that there is a crime, it's victimless

目前来说确实是这样。

At this point, it is

因为价值——

Because the value of the-

因为价值本来就会上涨。

Because the value would've well went up

反方观点会说，等等。

The counterpoint would be, wait a minute.

你不应该这么做。

You're not supposed to do this.

当然，这是违法的。

Well, of course, it's illegal.

但如果奶奶知道她投资的那个合同——

But if Grandma knew that she was investing in a contract that-

她就不会买那个合同了。

She would not have bought that contract

没错。

Exactly.

而且你也不可能拿到3%的利息，你得付12%。

And you wouldn't have gotten the money at 3%, you would've had to pay 12%

在那之后。

After that.

所以——

So-

顺便说一句，这里有个有趣的事实。

By the way, here's fun fact.

所以马克·沃尔特斯的保险公司总部在特拉华州，那里有个特拉华州的保险监管机构。

So Mark Walters' insurance companies are based in Delaware, and there is a Delaware insurance regulator.

那么问题来了，你们觉得现在特拉华州的保险监管员在当监管员之前是做什么的？

So question, what do you think the current Delaware insurance regulator did before he was an insurance regulator?

猜猜看。

Take a guess.

他之前是干什么的？

What was his job?

对，他之前是干什么的？

Yeah, what was his job?

他靠什么谋生？

What did he do for a living?

猜猜看。

Take a guess.

是金融相关的吗？

Was it something in finance?

不是。

No.

哦。

Oh.

猜猜看。

Take a guess.

他以前是播客主播吗？

Was he a podcaster?

他以前是警长。

He was a sheriff.

哦，那完全说得通。

Oh, that makes perfect sense.

而且他还是——

And he's the-

而且他还是保险监管员。

And he's the insurance regulator.

好的。

Okay.

这完全说得通。

That makes perfect sense.

听着，我对警长没什么意见。

Look, I have nothing against sheriffs.

上帝保佑。

God bless.

他们保护——

They protect-

没有警长，就没有副手——

No sheriffs, no deputies of the-

他们保护我们——

They protect us-

当然。

Of course.

……但他对保险一窍不通。

... but he doesn't know anything about insurance.

那好，问题来了。

So all right, so here's the question.

很多在Substack上写文章的人都在把保险公司投资AI数据中心这件事联系起来。

A lot of people writing on Substack are drawing these connections between insurance companies that are now investing in AI data centers.

我想问你的问题是，这对持有保险单的人来说风险有多大？

I guess the question to you is, how risky is that for the people that hold insurance policies?

那些写这些文章的人会不会被证明是对的，也就是说，对普通老百姓来说，系统性风险更大，他们不是投机者，他们就是实实在在拿着保单的人？

And are these people that are writing about this going to be proven correct that there's more systemic risk for regular people who are not speculators, they're literally holding policies?

这是不是人们关注的领域……

Is this an area that people-

我真不知道怎么回答这个问题。

I just don't know how to answer that.

好的。

Okay.

我真的不知道。

I really don't.

你并不是特别担心这个，还是说——

You're not particularly worried about it, or-

不是我不担心，只是我真的没有答案。

It's not that I'm not worried about it, I just don't have an answer

好吧。

All right.

史蒂夫，在结束之前，我们还有两件事想跟你聊聊。

Steve, there's two more things that we want to do with you before we wrap up.

我想问问你之前提到的那个政治反弹，不只是关于数据中心建设，而是更广泛地说。

I want to ask you about something that you mentioned, the political backlash, not just over the building of data centers, but just broadly speaking.

我听了你最近一期节目，你采访了一位在联合国谈AI风险的人。我觉得现在事情已经发展到白宫提出了一个四点计划来监管AI的程度。

I listened to a recent episode you did with someone who actually spoke at the UN about the risks of AI, and I feel like this has now gotten to a point where we have sort of a four-point plan from the White House about how AI will be regulated.

听起来好像是各人自扫门前雪，但会成立一个委员会。

And what it sort of sounded like was it's every man for himself, but there'll be a commission.

这个委员会将由人工智能行业的人组成，基本上，指导方针就是如果你发明的东西失控并造成损害，你要承担责任。

The commission will be made up of people that work in the AI industry, and basically, the guidance is you have the liability if something you invent goes crazy and causes damage.

我的看法是，是的，我喜欢这样，但责任由谁来承担？

And my perspective is, yeah, I like that, but liability for who?

是这些公司的投资者要写支票承担责任，还是有人创造了真正有害的东西，而他们是高管，个人必须面对后果？

Liability for the investors in one of these companies to write a check, or liability that somebody created something really harmful and they're an executive, and they personally will have to face the consequences.

我想听听你对目前这种监管会如何运作的看法。

I guess I'd love to know what your opinion is of the current state of that sort of regulation is going to work.

如果你有替代建议，应该是什么？

What should it be if you have an alternate suggestion?

所以我认为，如果你经营一家人工智能公司，而你的公司去攻击别人或做了坏事，你的公司要负责。

So I think that if you run an AI company and your AI company goes out and hacks somebody or does something bad, your company's liable.

就是负责。

It's just liable.

现在，更有趣的问题是，我是那家公司的CEO，我知道这会发生，但我不在乎，结果还是发生了。

Now, the more interesting question is, I'm the CEO of that company and I knew this was going to happen, and I didn't care, and it happened anyway.

真的会有人来找我麻烦吗？

Will anybody actually go after me?

2008年发生了什么？

What happened in 2008?

这正是我接下来要说的。

That was my next statement.

嗯，没有任何人个人受到追究。

Well, nothing happened to anyone personally.

2008年，尽管那些高管犯下了人类历史上最严重的欺诈，但没有任何人受到追究。

In 2008, nothing happened to any executive, even though they committed the worst fraud in human history.

所以我有个播客，采访了一位同行播客主，她做了很多关于企业欺诈的节目。

So I had a podcast where I was interviewing a fellow podcaster, and she does a lot of stuff on corporate fraud.

我们聊到欺诈，说CEO应该被追究责任，她几乎是恳求地说。

And so we were talking about fraud and that CEOs should be held accountable, and she's like pleading.

她说：“史蒂夫，我们能做些什么？”

She says, "Steve, what can be done?"

然后我说：“什么也做不了。

And I said, "Nothing.

因为什么都不会发生——

Because nothing will be done-

……很遗憾。”

... unfortunately."

所以——

So-

嗯，到现在肯定也不会了。

Well, by now, certainly not

……如果2008年没人进监狱，他们也不会因为黑客行为进监狱。

... if nobody went to jail for 2008, they ain't going to jail for hacking.

好的。

Okay.

那现在的监管真的能阻止任何事情吗？

So is the current state of regulation really going to stop anything?

或者我挺悲观的，我觉得非得出点大事才行，等风波过后，可能才会出台些规矩。

Or I'm very cynical, I think we have to have a crisis, and then in the aftermath, maybe there'll be some rules.

我完全这么想。

That's 100% the way I think about it.

事情向来都是这样发展的，对吧？

That's how it always goes, right?

一直都是这样。

That's how it always is.

好的。

Okay.

你觉得政府持股、挑选赢家这种做法怎么样？

What do you think about the government taking stakes, picking winners?

你肯定很喜欢吧。

You must love that.

我觉得那样会很糟糕。

I think that would be awful.

太糟糕了。

Just awful.

你不喜欢政府入股任何AI实验室或项目的想法吗？

You don't like the idea of the government taking stakes in any of the AI labs or projects?

不喜欢，因为那样的话，大家都会觉得它们大到不能倒，然后它们就会得到救助。

No, because then everybody would assume that they'd be too big to fail, and they'd get bailed out.

我不想让政府插手。

I don't want the government involved.

好的。

Okay.

你觉得现在这些超大规模云服务商的架构，它们……

Do you think that the way the hyperscalers are currently set up, that they...

我们看看现在的股票，看看亚马逊、Alphabet，除了苹果和特斯拉，在排名前七的公司里，我们在华尔街称它们为“七巨头”。

We look at the stocks now, we look at Amazon, Alphabet, with the exception of Apple and Tesla, of the top seven companies, we call them the Mag Seven on Wall Street.

在这前七家公司里，几乎都涉足这个行业，现在连Meta也加入了，它建了自己的数据中心供自己使用——

Of the top seven companies, they're all pretty much in this business, including now Meta, who has built data centers for their own use-

但他们也可以把数据中心对外开放，让别人租用算力，如果他们想的话。

... but they could turn them outward and have other people rent compute from them if they wanted to.

你觉得未来五到十年，这些公司会变成十万亿美元市值吗，还是说我们差不多已经接近它们规模的自然上限了？

Do you think that over the next five to 10 years, we will see these companies become $10 trillion, or are we pretty much butting up against what the natural limit on their size is?

我觉得它们的规模没有自然上限。

I don't think there's a natural limit on their size.

好的。

Okay

但对我来说，关键的问题是，这会不会像互联网那样发展？

But to me, the intellectual question is, does this sort of play out like the internet?

我在奥本海默工作的时候，亨利·布洛杰特就在我隔壁，因为他也在奥本海默。

When I was at Oppenheimer, Henry Blodget was literally across the hall from me because he was at Oppenheimer.

他走到销售大厅，说即将创造巨额的财富，互联网将征服世界。

And he got out on the sales floor and he said That dynastic levels of wealth are about to be created, and the internet's going to conquer the world.

他说得完全正确。

And he was 100% correct.

他只是对第一代互联网公司的判断错了。

He just wasn't correct about the first generation of internet companies.

是第二代互联网公司，利用第一代打下的基础，才走向了辉煌。

It was the second generation of internet companies that used the rails that were created by the first generation to go on to glory.

也许这里也会发生类似的事情。

Maybe something like that happens here.

我不——

I don't-

所以，就像美国在线、网景、雅虎——

So, like AOL, Netscape, Yahoo-

它们都是——

They're all-

……Excite。

... Excite.

没错。

Exactly.

那些公司最后都没成……马特，我能想到的第一代公司里，只有亚马逊——

None of those companies ended up... Matt, the only first gen I can think of is Amazon-

就是亚马逊。

Is Amazon

……最后发展起来了。

... that went on.

对。

Okay.

我不是说现在比过去好或差，只是情况不同。那时候有太多公司根本不该上市却上市了。

I'm not saying that this is better or worse, it's just it's different in the sense that back then there were so many companies coming public that had no business doing so.

说得对。

That's fair.

而现在，就两家公司包揽了一切。

And now we have two companies that are everything.

是的。

Yes.

这倒是挺公平的。

That is fair.

你觉得Anthropic今年会上市吗？

Do you think Anthropic will come public this year?

他们最好上市。

They better.

好的。

Okay.

对，我也这么觉得。

Yeah, I agree.

因为我觉得他们现在真的需要上市，因为整个tokenmaxxing会让他们的数据开始变得很难看。

Because I think they actually need to go public right now because the whole tokenmaxxing is going to make their numbers start to look really bad.

就是说他们得趁热打铁。

Like they have to strike while the iron is still hot.

他们确实需要。

They need to.

我的阴谋论是，Anthropic必须在公布第三季度财报之前上市。

My conspiracy theory is that Anthropic needs to go public before they report third quarter.

好的。

Okay.

你觉得为什么OpenAI看起来没那么着急？

Why do you think OpenAI doesn't seem to be in as much of a rush?

我猜他们的情况也差不多——

I would guess that they're in similar-

我不知道。

I don't know.

我本来以为他们会着急的。

I would think they would be.

我不太理解。

I don't understand it.

好的。

Okay.

你觉得这次IPO——

Do you think the IPO-

我觉得其他方面就是他们的表现没那么好。

Other than they're just not doing nearly as well.

你觉得这次IPO会受欢迎吗？

You think the IPO will be well-received?

Anthropic的？

Anthropic?

我换个角度来解释。

Let me frame it this way.

SpaceX是我们见过规模最大的IPO。

SpaceX is the largest IPO we've ever seen.

推特上有很多悲观的言论，毕竟那里本来就是负面情绪聚集的地方。

There was a lot of pessimism on Twitter, where pessimism usually lives.

大家都预测它会搞乱指数基金，把资金从其他股票抽走，用来支撑它庞大的体量。

All of the predictions about it would screw up the index funds, it would take money away from all these other stocks so that they could support the size of it.

所有那些我们...

All of the things that we...

内部人士抛售会压垮股东。

The insider selling would crush the shareholders.

它会以35美元开盘，然后一旦内部人士解禁可以卖出，就会跌到50美元。

It would open at 35 and collapse to 50 once insiders were unlocked and able to sell.

人们说了那么多理由，说这要么对股票本身，要么对所有人来说都是一场灾难。

There were so many reasons why people said this is going to be a disaster, either for the stock itself or for everyone.

但那些都没有发生。

And none of that happened.

它开盘了，跌了，然后又涨了，然后又跌回去。

It opened up, it fell, then it went up again, then it fell back.

但它表现得像一只正常的股票。

But it's acting like a normal stock.

一只正常的技术股。

A normal technology stock.

你觉得像Anthropic这样的公司，规模差不多大，会不会也受到类似的欢迎，然后我们很快就习惯它的存在了？

Do you think something like Anthropic, which will be roughly the same size, will be similarly received and just we'll quickly get used to it being in existence?

我不知道。

I have no idea.

好的。

Okay.

好，我问你一个问题。

All right, let me ask you a question.

它们上市的时候你会投资吗？

Do you invest there when they go public?

我会买吗？

Would I buy it?

不会。

No.

你会买吗？

Would you be a buyer?

绝对不可能。

No way.

绝对不可能。

No way.

那我问你一个你可能会回答的问题。

Let me ask you a question that you might answer.

好。

Okay.

我试试看。

I'm taking a shot.

请讲。

Go ahead.

好的。

All right.

那么，今年盈利增长了30%。

So, earnings have grown 30% this year.

投资者们都在说，"别以为这能持续下去。"

Investors are saying, "Don't think it's sustainable."

前瞻市盈率——

The forward PE-

我们愿意支付的价格，已经大幅下降了。

... we're willing to pay, has come down dramatically.

这让你看涨还是看跌？

Does that make you bullish, bearish?

你不在乎吗？

Do you not care?

你对这种动态怎么看？

What do you think about that dynamic?

我对估值不是那么敏感。

I am not valuation sensitive in that way.

好吧，但市场在告诉你什么？

All right, but what's the market telling you?

市场纯粹是叙事驱动的。

The market is pure narrative.

纯粹是叙事，而现在唯一重要的叙事就是人工智能。

It is pure narrative, and there is only one narrative that matters right now, and that's AI.

所以，你告诉我人工智能的叙事会怎么发展，我就能告诉你股市会怎么走。

So you tell me what happens to the AI narrative and I'll tell you what happens to the stock market.

就是这样。

Period.

好。

Good.

对。

Yep.

这才是关键。

All that matters.

对。

Yep.

所以，我觉得这话没错，很多人说股市不代表经济。

So, I think that's true, and a lot of people say the stock market is not the economy.

我觉得现在某种程度上确实是这样。

I think right now it kind of is.

我觉得我们现在说的是那些还没上市的公司，竟然有能力引发经济衰退。

I think we're talking about companies that aren't even public yet having the ability to cause a recession.

所以那些已经上市的公司，我真的觉得市场所有的表现都来自某种AI交易的衍生品。

So the companies that are public, I really think all of the performance in the market is coming from some derivative of the AI trade.

哦，我完全-

Oh, I completely-

而且现在债券市场，我们昨天听迈克尔说的，我们学到了发行债务的能力对当前收益率有多重要。

And now the bond market, we learned, as Michael said, we learned yesterday how important to prevailing yields, the ability to sell debt.

我真的觉得这就是那种经济变成了股市的情况，可能不会永远持续，但就在此刻，我觉得比以往任何时候都更真实。

I really think this is one of those situations where the economy did become the stock market, and it may not last forever, but right in this moment, I think it's more true than ever.

如果是这样的话，这个市场保持在接近高位或者不发生某种灾难性事件有多重要？

If that's the case, how important is it that this market remain near highs or not have some sort of calamitous event?

这感觉就像是个反馈循环——

It almost feels like the feedback-

我不这么看。

I don't think of it that way.

好吧。

Okay.

我还是回到我最初的观点。

Again, I go back to my original point.

你看标普500，40%是科技股。

If you look at the S&P, 40% is tech.

再加上谷歌、亚马逊之类的，就到50%了。

Add Google, Amazon, blah, blah, blah, you get to 50%.

是50%。

It's 50.

再加上投行、GE Vernova、Quanta、Eaton和Rockwell，就占到标普的65%了。

Add the investment banks and GE Vernova and Quanta and Eaton and Rockwell, and you're at 65% of the S&P.

是的，工业股也是……

Yeah, the industrials are-

这都是一笔交易。

It's all one trade.

好的。

Okay.

就是人工智能，没有别的，讨论结束。

It's AI, period, end of discussion.

好的。

Okay.

所以，只要需求保持，而且有人愿意为这些建设提供资金……

So, so long as the demand holds up and somebody's willing to finance the build-

那就没问题。

It's fine.

好的。

All right.

没事。

It's fine.

另一个你熟悉的领域，房地产市场。

Another area that you know something about, the housing market.

好的。

Okay.

好的。

Okay.

这是自金融危机以来最糟糕的住房环境。

This is the worst housing environment since the GFC.

毫无疑问。

No question.

虽然情况远没有那么严重，但确实很糟糕，而且我觉得难以置信，经济的三分之一处于萧条水平，然而——

It's obviously not nearly the same thing, but it is bad, and I find it unbelievable that a third of the economy can be depression level and yet-

而且没人关心。

And nobody cares

……没人关心。

... nobody cares.

因为AI，一切看起来都很好，对吧？

Things are good because of AI, right?

AI。

AI.

谢天谢地。

Thank God.

谢天谢地，有AI在，股市和经济目前还能撑住。

Thank God for AI, for the stock market and the economy's purpose for right now.

2022年对房屋建筑商来说是非常奇怪的一年，因为他们当时火得不行，而在房市冻结的情况下，建筑股本不该这么火。

2022 was a very weird year for the home builders because they were on fire, and home building stocks should not be on fire in a housing freeze.

实际情况是，所有有房的人都被套住了。

The thing that happened was everybody that owns a house is stuck.

没人能卖房，因为利率太高了。

Nobody could sell it because rates are too high.

没有买家，也没有卖家。

There's no buyers, there's no sellers.

那你有什么？

What do you have?

太棒了。

Amazing.

新建住房。

New housing.

Lennar和其他公司，像Pulte和KB，他们赚翻了。

And Lennar and all of the others, Pulte and KB, they feasted.

他们拿到了优惠。

They were given concessions.

那是一场狂潮，一场混战。

It was a frenzy, a free-for-all.

而那种局面已经结束了。

And that dynamic is over.