本集简介
双语字幕
仅展示文本字幕，不包含中文音频；想边听边看，请使用 Bayt 播客 App。
一般来说，IPO 规模每占市场市值的 1%，市场表现就会比正常情况差 4%。
In general, the market does about 4% worse per percent of market cap that comes in in an IPO in a given year.
所以如果我们看到 6% 的 IPO 规模，那就比正常情况差 24%。
So if we are seeing 6% coming in, that's 24% worse than normal.
如果大家都缺算力，那我当然愿意买一块五年前、能耗更高的 GPU。
If we're all short of compute, then sure, I'm going to pay for a five year old GPU that is less energy efficient.
但如果现在算力突然过剩，那信贷方面就会出现大规模违约。
But man, if now the world is suddenly awash in compute, we will have massive defaults on the credit side.
股权方面，我们也会看到公司倒闭。
We will see companies on the equity side go bust.
超大规模计算公司也会遇到问题，他们的投资回报会很差。
We will see a problem for the hyperscalers where their return on investment turned out to be lousy.
那这是不是意味着市场会下跌 18%？
And is that a minus 18% in the market?
不是。
No.
那市场会跌35%到50%。
That is a minus 35 to 50% in the market.
本期节目由Sarmaya Partners赞助播出，他们管理着一只回归实物的ETF，代码是LENS。
This episode is brought to you by Sarmaya Partners, who manages a return to tangible ETF LENS, ticker l e n s.
Sarmaya主题ETF，专为回归实物而设计。
The Sarmaya thematic ETF built for a return to tangibles.
更多信息请访问sarmayaetf.com，详情请见节目说明。
Learn more at sarmayaetf.com and see the show notes for disclosures.
我们开始吧。
Let's get into it.
大家好，欢迎收听《别人的钱》。
Hello, and welcome to Other People's Money.
今天请到的是GMO的资产配置联席主管Ben Inker，GMO目前管理着超过800亿美元的资金。
I am joined today by Ben Inker, co head of asset allocation at GMO, which manages over $80,000,000,000 today.
Ben，感谢你今天来做客。
Ben, thank you for joining me today.
非常感谢你邀请我，Max。
Thanks so much for having me, Max.
我想从你最近在第三季度信函中提出的一个重要观点开始，具体来说，你认为股票供应是导致AI泡沫破裂的催化剂，并且你预计这个催化剂将在未来十二到十八个月内显现。
I wanna start out with the big call that you recently made in your Q3 quarterly letter, specifically that you think that equity supply is the catalyst that will bring down the AI Bubble and that you expect that catalyst to play out in the next twelve to eighteen months.
你看到了什么，第一，让你认为AI是泡沫？
What is it that you're seeing, one, that has you saying AI is a bubble?
第二，跟我谈谈我们已经看到的股票供应，以及为什么你预计会有更多供应对市场造成压力？
And two, talk to me about equity supply that we've already seen and why you're expecting more of it to weigh on the market?
嗯，这里有很多内容。
Well, there's a lot there.
我们先从比较简短的版本开始，然后我们可以深入讨论任何你想了解的细节。
We'll start with the relatively quick version and we can get into as much detail as you want.
为什么我们认为AI是泡沫？
Why do we think AI is a bubble?
我想从最高层面来说，因为它是一项变革性的技术变革。
I'd say kind of at the highest level, it is because it is a transformational technological change.
变革性的技术变革通常会催生泡沫。
Transformational technological changes normally create bubbles.
我们一次又一次地看到这种情况，无论是运河、铁路、汽车、电气化还是互联网，这些事件，每个人都能说，哦，天哪，这会改变世界。
We've seen it time and time again, whether it was canals or railroads or autos or electrification or the internet, these events, everybody can tell, oh my God, this is gonna change the world.
所以他们说，嗯，我真的很想参与其中。
And so they say, well, I really wanna be a part of that.
这些变革性技术的难点在于，很难确切知道钱会在哪里产生，谁会受益。
And the difficulty with these transformational technologies is it's hard to know exactly where the money is going to be made, who is going to be the beneficiary.
事实证明，大多数时候，投入的大部分资金并没有很好的投资回报。
And it turns out most of the time, most of the money that gets invested doesn't have a great ROI.
而且在早期，投资回报看起来不错，对吧？
And since in the early days, one, the returns on investment look good, right?
因为你看到需求真的在飙升。
Because you are seeing demand really ramp up.
而且由于所有的兴奋情绪，往往会有大量资本涌入。
And because of all of the excitement, you tend to get huge amounts of capital coming in.
现在，只要有大量资本涌入，往往就会带来问题，对吧？
Now, anytime you have huge amounts of capital coming in, that tends to create a problem, right?
通常情况下，稀缺才能带来高回报。
It is normally the case that scarcity creates high returns.
所以，如果大量资本涌入，就会引发问题。
And so, if you have a lot of capital coming in, you create issues.
随着资本不断涌入，即使在一段时间内需求增长快于供应，最终，当所有参与供应建设的人似乎都赚得盆满钵满，越来越多的人想加入时，你最终会达到一个点，供应超过了需求。
And as more and more capital comes in, even if it was the case that demand was growing faster than supply for some period of time, Eventually, as everybody who participated in the building of that supply seems to be getting rich and more and more people want to be involved, you eventually hit a point where now supply is bigger than demand.
而这种情况的问题在于，不仅仅是那些新增的投资最终回报不佳。
And the problem with a situation like that is it's not just the incremental investment that then winds up having a lousy return on investment.
它还会波及整个领域的投资，甚至可能摧毁这些投资。
It goes and takes the entire corpus of investment in the space and can destroy that.
所以，当第一条铁路在伦敦和曼彻斯特之间建成时，利润非常可观。
So when the first railroad was built between London and Manchester, it was fantastically profitable.
到英国铁路繁荣结束时，已经有六条铁路了。
By the end of The UK railway boom, there were six.
其实根本不需要六条。
And there was no need for six.
但问题是，一旦有了六条，哪条铁路都赚不到钱了。
But the thing is when there's six, no railroad is gonna be making any money.
所以从第一层来看，这很像历史上那些泡沫出现时的情形。
So at the first level, man, this looks a lot like historical situations where we have seen bubbles.
第二，当我们看这个行业的基本特征时，发现很多都涉及算力的提供或者AI模型的提供。
Second, when we look at the basic characteristics of this business, What we see is a lot of this is about either the provision of compute, or the provision of AI models.
而且在这两种情况下，我们看到这些更像是竞争性业务，而不是赢家通吃的业务。
And what we see in both cases is those look more likely to be competitive businesses, than winner take all businesses.
要想获得真正高的资本回报，需要的是赢家通吃的情况，比如因为有网络效应、规模效应等等。
The way you have a really good return on capital is to have a situation where the winner takes all because of network effects, because of returns to scale, what have you.
你不想看到的是切换成本低、选择之间差异不大的情况。
What you don't want to have is a situation where the switching costs are low, where the differentiation is not that profound between the choices.
很难找到一种情况，切换成本低、很多东西相对可比，但资本回报却很高。
It is very difficult to find a situation where switching costs are low, lots of things are relatively comparable, and the return on capital is really high.
很高。
High.
对，这正是我们看到的。
Well, that is exactly what we're seeing.
不管是这周Anthropic的模型最好，还是OpenAI的模型最好，还有来自其他实验室和其他国家的开源模型，很明显人们都在切换。
Whether it's, you know, this week, Anthropic has the best model, OpenAI has the best model, and then also the open source models that have come out from other labs from other countries, It's clear that people are switching.
而且在很多情况下，人们甚至会在不同查询、不同用例之间切换，有些公司专门帮你优化这种任务级别的切换。
And in many ways, people are switching even between queries, between different use cases, their entire businesses where their job is to help you optimize this switching at a task by task level.
没错。
Exactly.
而且这是必须发生的，对吧？
And that needs to happen, right?
如果我们要让人工智能得到广泛使用，它就必须便宜。
If we are going to get really broad use of artificial intelligence, it needs to be cheap.
但问题在于，便宜的东西很难同时做到利润丰厚。
And the problem with it being cheap is it is hard for something to be cheap and be wonderfully profitable.
所以，我们认为这很可能最终会被证明是一场泡沫。
So, we think this is likely to prove to have been a bubble.
那么，问题是我们为什么认为供应会成为那个驱动因素？
Now, the question of why do we think that supply is going to be the driver of that?
回顾过去几十年的市场，我们看到越来越多的证据表明，市场在供应方面变得越来越缺乏价格弹性。
As we look at the market in the last few decades, what we see is more and more evidence that the market is becoming more price inelastic when it comes to supply.
所以你想一想，如果你持有某只股票的股份，别人想从你这里买一些，他们需要推高价格才能让你卖出。
So if you think about it, if you own shares of a stock, somebody else wants to come in and buy some from you, they need to push up the price to cause you to sell.
但如今的市场中，有越来越大的比例的人永远不会因为那个原因而卖出，对吧？
But the market these days consists of increasingly large percentages of people who will never sell for that reason, right?
如果你想从英伟达那里获得一些股份，你需要推高价格，但你永远不可能让一个标普500指数基金因为价格被推高而向你卖出。
If you wanna get some shares from Nvidia and you need to push up the price, you are never going to get an S and P 500 index fund to sell to you because you pushed up the price.
你也永远不可能让一个英伟达2倍杠杆ETF因为价格被推高而向你卖出英伟达的股份。
You are never going to get an Nvidia 2X ETF to sell Nvidia to you by pushing up the price.
而且，越来越多的主动型基金经理会说，天哪，我对英伟达的低配程度只能承受这么多。
And increasingly, if you've active managers who are saying, Oh my God, I can only afford to be underweight Nvidia to a certain extent.
如果你推高英伟达的股价，我就更低于标准配置了，所以我需要买一些。
And if you push up the price of Nvidia, I am more underweight, so I need to buy some.
当你推高价格时，你不会让那个投资者卖给你。
You are not going to get that investor to sell to you when you push up the price.
我们看到的是市场中的足迹，这个市场越来越受供应变化驱动。
What we see is the footprints in the market of a market that is increasingly driven by what happens to supply.
而我们多年来第一次突然获得了大量供应。
And we are suddenly for the first time in years getting significant supply.
特别是，我们即将看到SpaceX、然后是Anthropic，再然后是OpenAI上市，成为历史上最大的IPO。
In particular, what we are getting is SpaceX soon to be Anthropic and then soon to be OpenAI coming public as the largest IPOs in history.
坦白说，SpaceX是历史上最大的IPO，这是一件大事，但750亿美元虽然很多，其实也不算多。
And frankly, you know, SpaceX largest IPO in history, a big deal except $75,000,000,000 is a lot of money, but it isn't.
750亿美元在一个75万亿美元的市场里，并不算多。
$75,000,000,000 in a $75,000,000,000,000 market is not much.
在接下来的十二个月里，SpaceX另外的2万亿美元将可以出售。
Now over the next twelve months, the other $2,000,000,000,000 worth of SpaceX becomes available to sell.
其中相当一部分会被卖掉。
And a decent chunk of that will be sold.
而另一边需要有买家接盘。
And it needs to get bought on the other side.
这些股票需要被投资者以整手买入，而他们为了买这些股票，会卖掉其他股票。
It needs to get bought in round numbers by investors who will be selling other shares to do so.
现在我想插一句，因为IPO时只放出少量流通股这个想法并不新鲜。
Now I wanna step in because this idea that at the IPO, a small amount of the float is what's gonna come out is not new.
SpaceX IPO的新颖之处在于解锁期的分批安排。
What was new with the SpaceX IPO was the staggering of the unlocks.
传统上，通常就是180天。
Traditionally, it's just a one hundred and eighty day.
所有人被锁定180天，之后就可以卖出了。
Everybody is locked up until one hundred and eighty days, and then you get to sell.
这种疯狂的解锁期分批安排，是不是也说明他们自己也意识到了这个问题？
I mean, did the crazy tranching of the unlock schedule kind of tell you that they understand this problem too.
就连这些股票的发行方也清楚地知道这种供应量带来的影响。
Even the issuers of this equity acutely are aware of the effect of this level of supply.
对。
Yes.
虽然他们可能不担心我所担心的，因为当SpaceX上市时，SpaceX关心的是SpaceX的股价。
Although they may not be worried about the same thing I'm worried about because when SpaceX goes public, what SpaceX cares about is the price of SpaceX shares.
他们并不真正关心市场上其他股票的价格会怎样。
They do not really care what happens to the prices of the rest of the shares in the market.
但没错，这些绝对是历史上最大的IPO。
But yes, these are far and away, again, the largest IPOs in history.
如果有一个180天后所有人都能卖出的悬崖事件，那对市场来说会非常难以应对。
And if you had this cliff, you know, one hundred and eighty day everybody can sell event, you know, that would have been incredibly difficult for the market to handle.
所以我觉得，SpaceX是意识到了这一点的。
So I think, SpaceX was aware.
我觉得，虽然我不能说他们具体做法是最有效的方式，但让所有这些股票不一次性涌入市场是有道理的。
I think, while I can't claim that what they did in detail is the most efficient way of doing it, It makes sense to not have all of these shares come into the market exactly at once.
分批上市是有道理的，因为这可是很大一笔钱。
It makes sense to have some staggering over the time because this is a lot of money.
而且现在并不是每个持有SpaceX股票的人都会在解禁第一天就卖掉，但我仍然觉得分批上市很有意义。
And now not everybody who owns SpaceX will sell it on the first day they are able to, but still, I think it makes a lot of sense to stagger it.
在接下来的大约十二个月里，几乎所有持股的人都会有机会卖出。
It is going to be the case that over the twelve months or so, almost everybody who owns it is going to have the opportunity to sell.
Anthropic的情况也会一样。
And the same thing will be true for Anthropic.
OpenAI的情况也会一样。
The same thing will be true for OpenAI.
总的来说，假设OpenAI和Anthropic真的上市，它们加起来可能占美国总市值的6%、7%甚至7.5%，这比我们以往在任何十二个月里通过IPO增加的市值都要多。
And in aggregate, are talking about assuming OpenAI and Anthropic do come public, maybe six, seven, 7.5% of the aggregate market cap of The US, which is a greater increment of market cap via IPO than we have ever seen in a twelve month period.
你能拿1999年、2000年那会儿来比较一下吗？
Can you compare it to ninety nine, two thousand?
我觉得那之前是纪录保持者。
I think that was the record before.
那时候市场市值占比是多少？
What percentage of market cap did we see come there?
那次事件中，我们看到大约2.54%的市场市值进入。
In that event, we were seeing somewhere between 2.54% of the market coming in.
到目前为止，我们从未见过一年内超过4%的市场市值进入。
We have never actually seen a year where greater than 4% came in up until now.
所以，这将是第一次。
So, this would be the first time.
而需要担心的是，历史上我们看到最大IPO数量的三年是1999年、2000年和2021年。
And the reason to be concerned is the three years in history that we have seen the largest amounts were 1999, 2000 and 2021.
所有这些年份之后的年份，市场都经历了困难时期。
And all of them had subsequent years where the market had a difficult time.
一般来说，市场市值每通过IPO进入1%，市场表现就会比正常情况差约4%。
In general, the market does about 4% worse per percent of market cap that comes in an IPO in a given year.
所以，如果我们看到6%的市值进入，那就比正常情况差24%。
So, if we are seeing 6% coming in, that's 24% worse than normal.
现在，这是增长，对吧？
Now, is a gain, right?
市场通常是上涨的。
The market normally goes up.
所以，这并不是说市场一定会崩溃，但确实会带来重大挑战。
So, this is not saying there is unquestionably going to be a market route, but it does create a significant challenge.
所以，当你说在6%的基础上，加上外汇乘数，比正常情况差24%，我的意思是，如果只取6%的平均值，听起来我们可能预期大约下降20%的一年。
And so when you say 24% worse than normal on that 6% with the forex multiplier, mean, I you just take the 6% average, that sounds like approximately a down 20% year is what we might be able to expect.
但那是没有调整估值的情况。
But that's not adjusting for valuation.
显然，在那些时期，2021年的典型例子就是那些所谓的“arc”名字。
Obviously, in those periods of time, you know, the poster child of 2021 was very much the arc names.
而那个篮子，显然，纳斯达克、标准普尔500，它们下跌的幅度更合理，并且恢复了。
And that basket, obviously, the Nasdaq, the S and P five hundred, they fell at much more sort of reasonable levels and recovered.
那个篮子确实挣扎了很长时间。
That basket really struggled for a long time.
你说震中的时候，你指的是什么？
What do you look at when you say the epicenter?
因为我们谈到的很多股票要么是新上市的，要么还没上市，但有很大一部分已经上市的股票被视为人工智能股票，无论是半导体，还是一些电力及工业领域的公司。
Because a lot of the stocks that we're talking about are either fresh to the market or haven't even come out yet, but there is a huge component of stocks that are already listed that are considered to be AI stocks, whether it's semiconductors, some of the the power and industrial names.
你认为当前市场的真正震中是什么？它可能比市场整体预期的18%到20%的跌幅还要大。
What do you consider to really be the epicenter of the current market that might see a greater drawdown than, say, you know, 18, 20% that you might expect for the market overall?
当然，如果人工智能泡沫破裂，那么受益于人工智能上涨的股票就会首当其冲。
Well, certainly if AI is a bubble that bursts, it is the stocks that benefited from AI going up.
这确实包括超大规模计算公司、半导体公司、一些硬件技术公司，还有软件公司，对吧？
And that does include the hyperscalers, the semi names, some of the tech hardware names, pieces of software, right?
软件今年已经遭受重创，从ChatGPT事件后到一年前为止的最大赢家，变成了今年的重大输家。
Software already got hit pretty hard this year, having gone from kind of the largest winner in the post ChatGPT event up until a year ago to being a significant loser this year.
但那些从人工智能中受益的公司，如果形势逆转，它们面临的风险也最大，对吧？
But the companies that have benefited from AI are the companies that have the most at risk should things turn the other way?
那么究竟谁的风险最大呢？
And exactly who is most at risk?
关于超大规模计算公司，有一点是，天哪，这些公司在ChatGPT出现之前就已经非常优秀了。
One of the things about the hyperscalers is, man, these were awesome companies before ChatGPT.
他们的业务很棒，对吧？
They have awesome business, right?
Meta可能从模型中赚不到太多钱，或者他们卖计算机什么的。
Meta may or may not make a lot of money off of the models or they're selling computer, what have you.
但天哪，那是个好业务。
But man, that's a great business.
而且即使扎克伯格和其他人对AI赚钱的期望没有实现，这个业务也不会消失。
And that business does not disappear even if hopes by Zuckerberg and everybody else for how much money they're going to make off of AI don't come true.
这对甲骨文来说可能更难。
That may be tougher for an Oracle.
甲骨文业务不错，但他们现在想做的事情相对于他们当前业务的规模来说很大。
Oracle has a good business, but the scale of what they are trying to do now relative to the size of their current business is big.
对Core Weave或EBS这样的公司来说，问题更大，因为他们要么完全依赖AI业务，要么他们试图建立的业务比现有业务大得多。
It's a bigger issue for a Core Weave or an EBS or somebody who is investing a ton on the back of either a business which is entirely about AI, or is much smaller than the business they are trying to build.
所以影响当然取决于你的估值有多高，对吧？
So the impact should certainly depend on how expensive you are, Right?
如果你像SpaceX那样，以销售额的100倍来交易，即使业务本身没问题，也有很大的下跌空间。
If you are trading like SpaceX at whatever, a 100 times sales, there's a long way down you could go even if the business is fine.
然后还有一些业务可能最终并不顺利，它们可能会下跌100%。
Then you have some businesses that may not turn out to be fine, and they can go down a 100%.
大家好。
Hey, everyone.
希望你们觉得我的问题有价值。
Hope you're finding value in my questions.
我有个快速的问题想问你。
I've got a quick question for you.
你的投资组合中有多少实际上依赖于物理世界？
How much of your portfolio actually depends on the physical world?
如果你拥有标普500指数，超过三分之一是科技公司。
If you own the S and P five hundred, more than a third is tech.
能源、材料和贵金属，大约占5%。
Energy, materials, and precious metals, about 5%.
Sarmaya Partners认为，下一个周期属于实物资产，也就是那些十年来投资不足、不受重视的能源和金属行业。
Sarmaya Partners believes the next cycle belongs to tangibles, the energy and metal sectors that have been underinvested and underloved for a decade.
他们称之为“回归实物资产”，而LENS，股票代码l e n s，即Sarmaya ETF，是一只主动管理型ETF，用来表达这一观点。
They call it a return to tangibles and LENS, ticker l e n s, the Sarmaya ETF, is an actively managed ETF to express this view.
主动管理正是关键所在。
An active is the point.
团队会随着周期演变，跨行业、跨市值、跨国界地追踪这些主题，在发现机会的地方增加权重，在看起来过热的主题上削减权重。
The team follows the themes across sectors, market caps, and borders as the cycle evolves, adding where they see opportunity and trimming themes that look overextended.
矿业、能源、贵金属、关键金属，全部集中在一个主动管理的股票代码LENS中。
Miners, energy, precious metals, critical metals, all in one actively managed ticker, LENS.
当通胀来袭时，股票和债券可能同时下跌。
When inflation runs, stocks and bonds can fall together.
LENS增加了实物资产的敞口，有可能在股票和债券两方面都提供针对通胀和地缘政治风险的多元化效果。
LENS adds real asset exposure with the potential to diversify both the stock side and the bond side against inflation and geopolitical risk.
想了解更多关于LENS ETF的信息，包括完整的对比数据和表现，请访问sarmayaetf.com。
Learn more about the LENS ETF, their full data in comparison, including performance at sarmayaetf.com.
投资涉及风险，包括可能损失本金。
Investing involves risk, including the possible loss of principal.
投资前必须仔细考虑基金的投资目标、风险、费用和开支。
The fund's investment objectives, risks, charges, and expenses must be considered carefully before investing.
这些及其他重要信息都包含在招股说明书中，可在sarmayaetf.com获取。
This and other important information is contained in the prospectus, which can be obtained at sarmayaetf.com.
投资前请仔细阅读招股说明书。
Please read the prospectus carefully before investing.
现在我想再做一个类比，回到1999年和2000年。
Now I wanna do another analogy back to ninety nine two thousand.
我觉得你在信中提到的一点很有意思，就是事后看来，我们很容易说，哦，我们当时光纤建设过度了。
I thought one of the interesting things you put in your letter is it's really easy in hindsight to look back and say, oh, we were overbuilding fiber.
但在股价刚开始下跌的时候，我们其实还没到那个阶段。
But at the time that share prices started to come down, we hadn't really reached that stage.
现在回头看很容易说，我们都意识到我们在过度建设，这就是泡沫破裂的原因。
And it's really easy to look back now and go, well, we all realize we're overbuilding, and that's what undid the bubble.
但事实并非如此。
But that wasn't really the case.
那你觉得我们会在这种情况发生之前得到AI资本回报的答案吗，还是会像99年2000年那样，在供应开始造成压力的时候，我们还不清楚AI的需求到底有多大？
So do you think that we're gonna have an answer on AI return on capital before this happens, or is it gonna look like ninety nine two thousand where we're not really gonna have the answer on how much AI demand there's gonna be by the time this supply starts to weigh?
嗯，时间会告诉我们。
Well, time will tell.
但你说得完全对。
But you're absolutely right.
在互联网泡沫时期，泡沫在问题规模明朗之前就已经破裂了。
In the Internet bubble, the bubble rolled over before it was clear, the scale of the malinvestment.
这种情况很可能在这里重演。
That could very easily happen here.
而且经常如此。
And it often does.
但这并不是唯一的情况，市场在资本支出达到顶峰之前就达到了顶峰，而且在我们了解过度投资的规模之前就已经达到了顶峰。
But that wasn't the only situation where kind of the market peaked, one, before we saw the peak in CapEx and before we understood the scale of the over investment.
在人工智能和计算方面，与铁路的类比非常贴切，因为建造一个大型数据中心需要数年时间。
And in the case of AI and compute, the analogy with the railroads is a really apt one because it takes several years to build a giant data center.
你需要提前数年预测需求，而这些东西在几年前并不存在。
You need to be forecasting years ahead of time what demand is going to be for something that didn't really exist a few years ago.
所以，真的很难做到准确。
And so, it is really hard to get that right.
而且风险极高。
And the stakes are incredibly high.
所以，我的意思是，我认为最终我们会在这些数据中心上投资过多，然后会有巨大的减损。
So, by this, you know, I am saying, I think in the end, we will wind up investing too much in these data centers, and there will be giant write offs.
我想你也可以说，另一方面，我们不会建得足够多。
I suppose you could say on the other side, we will not build enough.
关于我们不会建得足够多这一点，我想说的是，这种情况迟早会改变，因为投资回报越好，整个经济就越会被吸引去做这件事。
The thing I would say about we will not build enough is sooner or later that will cease to be true because the better the return is on investment for doing this, the more it is going to suck the entire economy into doing this.
我们已经看到这种情况了，对吧？
We are already seeing that, right?
利率这么高，房地产市场基本就没什么动静了。
With interest rates where they are, the housing market is kind of dead.
大多数非住宅建筑项目也基本停滞了。
Most non residential construction is sort of dead.
唯一没受影响的就只有数据中心了。
The only thing that has been unaffected is the data centers.
如果你看看AI以外的经济投资，其实一直在下降，因为资本成本太高了。
And if you look at investment in the economy outside of AI, it has been falling because the cost of capital is high.
但在AI领域，投资并没有下降。
It is not falling in the AI space.
所以，如果你在考虑，好吧，那我该投资哪里呢？
And so, if you were looking and saying, okay, well, where do I wanna invest?
在我看来，投资那些投资不多的地方通常是个好主意，因为你需要稀缺性。
To my mind, investing where there isn't a lot of investment going on is generally a pretty good idea because you want scarcity.
你得去做别人没在做的事。
You want to be doing something other people are not doing.
现在投资数据中心，我觉得有点吓人。
To be investing in data centers today, I think it's a slightly scary thing to do.
时间会证明一切，但我们觉得现在有很多比AI领域更好的投资方向。
And time will tell, but we think there are lots of better places to be investing today than in the AI space.
你怎么把AI从那些行业里剥离出来？
How do you pull AI out of some of those sectors?
比如，工业领域传统上你会听到很多人强调缺乏投资。
Like, industrials traditionally are an area that you'll hear people pounding the table on lack of investment going into it.
人们说所有钱都涌向科技了。
People saying all this money is pouring into tech.
我们没有投资工业经济，但那些股票其实表现不错。
We're not investing in our industrial economy, but those stocks have actually been doing well.
同样，很多人多年来一直说新兴市场会好转。
Likewise, you know, a lot of people who've been saying for years, emerging markets are gonna turn around.
确实如此，但结果发现又是AI的故事。
They did, but it turns out that it was the AI story again.
似乎很多传统上表现与科技行业不同的领域，现在都出现了AI的影子，而且它……
It seems to be at the scene of the crime of so many of these areas that traditionally have been diversified in their performance from the technology sector, and it it...
它真的渗透进来了。
It's really creeped in.
它已经渗透到GDP里了。
It's creeped into GDP.
如果把AI相关的支出和由此带来的增长剔除掉，GDP看起来会弱得多。
If you disaggregate out the spending from AI and the growth that's coming from that, GDP looks much, much weaker.
所以我想知道，你觉得……
And so I wonder where you...
你觉得哪些领域还没有被AI触及？
Where are the areas that you think AI hasn't touched?
一个难点是，那些尚未被AI触及的领域，未来可能会被AI淹没。
One difficulty is the areas where AI hasn't touched are potentially areas that will get swamped by AI in the future.
但我们今年看到的一个有趣现象是，那些原本不被视为AI受益者的公司，突然大家发现它们其实是受益者。
But one of the interesting things we've seen, and it's been very much a story of this year, is that companies that were not viewed as AI beneficiaries, suddenly people realized they were.
比如我们的日本股票团队持有Toto这只股票已经有一段时间了。
Like our Japanese equity team had owned the stock Toto for quite a while.
他们喜欢Toto是因为他们生产的马桶质量非常好，而做好马桶是一门相当不错的生意。
They liked Toto because they make really good toilets, and making really good toilets is a pretty good business.
结果发现，Toto因为对陶瓷的了解比地球上任何人都多，所以还有一个很小的副业，就是制造这些陶瓷非反应性柱体，用来在半导体制造过程中搬运芯片。
It turns out Toto, by virtue of knowing more about porcelain than anybody else on the planet, also had this tiny side business, which was creating these porcelain nonreactive sort of columns to move semiconductors from place to place.
所以突然间它就成了大受益者，因为如果你要建更多这样的工厂，你就需要更多这些东西。
So suddenly, you know, it becomes a big beneficiary because if you're going build a lot more of these factories, you need more of these things.
危险在于，如果Toto的定价开始只考虑这些半导体相关的东西，那就令人担忧了。
Danger becomes if total starts to be priced as if the only thing that matters is these semiconductor related things, then it's worrying.
但现实是，如果AI要消耗大量资源，很多人都会受益。
But the reality is if AI is going to be a huge consumer of everything, lots of people are going to benefit.
我们在寻找的是，如何构建一个不仅仅是押注AI的投资组合？
What we are looking for is how do we build a portfolio that isn't simply a bet on AI?
它有一个AI的加成。
It has an AI kicker.
对，它们要么有AI的加成，要么本身就是好公司，处于有利位置，好事自然会降临到它们头上。
Yeah, that have an AI kicker or they are good companies, they are in a good place, good things can happen to them.
但当我们和客户交流时，发现很多人从别人那里听到，你的投资组合必须完全押注AI。
But what we see when we talk to our clients is a bunch of them are hearing from other people, your portfolio needs to be simply a bet on AI.
AI是唯一重要的事情。
AI is the only thing going on.
所有增长都在那里。
It is the place where all the growth is.
而我们会说，嘿，即使你相信AI真的会带来巨大的投资回报，你也不希望你的投资组合押注在单一事物上。
And we would say, hey, even if you believe AI is going to actually have a great return on investment, You do not want your portfolio to be a bet on a single thing.
那我们怎么扩大范围呢？
So how do we broaden this out?
我们怎么把范围扩大到那些价格合理、资本回报率不错的企业，这样作为股票投资者，我们能期待获得股票回报。
And how do we broaden this out into businesses that are reasonably priced, have a decent return on capital, and so as equity investors, we can expect an equity return from.
我想回到你提到的供应影响，以及市场结构变化如何随时间影响供应。就像主动投资者被市场排挤一样，这使得资金流动和供应的影响在决定市场走势时变得更加重要，而且在很多方面成为主导因素。
I want to go back to the effect of the supply that you see coming and the way that market structure changes have the impact of supply over time, just the way that the active investor has been crowded out of the market, it's just making the impact of flows and of supply that much more important in determining things and in many ways, the dominant factor.
当然，如果我们看过去三四十年，被动投资的崛起是非常巨大的。
Certainly, if we look over the past thirty or forty years, the rise of passive has been huge.
它从80年代占市场不到10%，到现在可能占了一半。
It went from under 10% of the market in the 80s to maybe half now.
除此之外，主动管理者也不像以前那么主动了，他们比过去更担心跟踪误差。
In addition to that, active managers are less active than they used to be, need to worry more about tracking error than they used to.
所以，他们需要担心自己与所对比的指数之间的差异。
So, they need to worry about how they differ from the index that they're measured against.
在一个顶部越来越集中的市场里，这让他们更不愿意调整投资组合，因为害怕跟踪误差。
In a market that has become increasingly concentrated at the top, that makes them less willing to move their portfolio around for fear of the tracking error.
虽然散户是市场的重要组成部分，但散户没有这种代理问题，不会那么在意跟踪误差。
While retail is a significant piece of the market and retail doesn't have these sort of agency issues where it is very concerned with tracking error.
说实话，大部分散户并不关注他们支付的价格，也不关注资产的估值。
Honestly, a big portion of retail is not focused on the price that they are paying, on the valuation of the assets.
市场要很好地吸收供应变化，就需要有投资者愿意根据价格的微小变动来调整自己的投资组合，而现在愿意这样做的投资者越来越少了。
What you need in order for the market to do a good job absorbing a change in supply is to have investors who are prepared to change their portfolio in response to a relatively small change in prices and fewer and fewer investors are prepared to do that.
被动投资者从定义上讲就不能这么做，他们不被允许。
The passive investors definitionally can't, they're not allowed to.
即使是成长型投资者，也比价值型投资者更不可能这样做。
Even growth investors are less likely to do it than value investors.
如果你因为认为英伟达是世界上最好的公司而持有它，那么当英伟达上涨10%时，你可能会说，这更证明了它是世界上最好的公司。
If you own Nvidia because you think Nvidia is the best company in the world, then Nvidia goes up by 10%, you might say, well, it's more evidence that it is the best company in the world.
如果你因为通用汽车便宜而持有它，当它上涨10%时，你可能会说，哦，好吧，我在通用汽车上赚了不少钱。
If you own GM because it was cheap and it goes up by 10%, you might say, oh, okay, well, I've made some good money on GM.
它不是世界上最好的公司，现在也许我能找到更便宜的了。
It wasn't the best company in the world and now maybe I can find something cheaper.
所以，就市场吸收供应而不引起价格大幅变化的能力而言，有价值导向的投资者是有帮助的。
So, in terms of the ability of the market to absorb supply without large changes in price, it is helpful to have value oriented investors.
而这类投资者在市场中的占比比过去小多了。
And they are a much smaller piece of the market than they used to be.
而且我觉得这种情况短期内不会改变。
And I don't think that's growing anytime soon.
我们确实看到，处理供应问题并不完全需要长期投资者。
We do see it is certainly the case that you don't 100% need long term investors to deal with supply.
所以，想想Citadel、James Street、Hudson River Trading这些公司在做什么，他们就是在寻找市场上的这些小扭曲，试图把它们熨平，这些扭曲可能在五秒、五分钟或者五小时后就会消失。
So if you think about what Citadel and James Street and Hudson River Trading and all of those guys are doing, they're looking for these little distortions in the market that they can try to iron out that will go away in the next five seconds or five minutes or maybe five hours.
所以，其中一部分可以由高频交易公司来处理。
So, some of this can be dealt with by the high frequency trading firms.
但还有相当一部分需要投资者来完成。
But a fair bit needs to be done by investors.
而现在愿意在价格变动时调整投资组合的投资者确实越来越少了。
And there's simply fewer investors who are prepared to change their portfolio when prices change.
这意味着，而且市场里所有的证据都表明，你需要更大的价格变动才能引发必要的交易。
And that means, and again, the evidence is all there in the markets, that you need a bigger change in price to cause the trading that will need to get done.
那么，你觉得高频交易者明白这一点吗？他们知道这个情况，他们是不是只能继续持有……
And so would you say that the high frequency traders, they understand this, they know this, and they're just gonna have to hold on to...
如果他们想减轻影响，他们就得自己承担更多风险，对吧？
If they want to dampen the impact, they're just gonna have to take more risk themselves?
我得说，那根本不是他们的商业模式，但Jane Street在七月份确实因为做了一些高频交易员通常不做的事情而损失了不少钱。
I would say that is simply not their business model, but Jane Street managed to lose a bunch of money in July doing things that high frequency traders don't normally do.
所以也许他们对高频交易的执着程度并没有人们认为的那么高。
So maybe they are less religious about the high frequency thing than they are given credit for.
但综合来看。
But among the things together.
而那些公司正是高频交易员所使用的，对吧？
And it is those companies that the high frequency traders use, right?
好的，有人想卖掉一大笔微软的股票，我来吸收这些。
Okay, somebody wants to sell a big block of Microsoft and I'm going to absorb that.
我想做的是在把微软股票处理掉之前先对冲一下。
What I want to do is hedge that before I can get rid of the Microsoft on my own.
所以，我需要一些和微软非常相似的股票。
So, I need stocks that look a lot like Microsoft.
我们在那些频率稍低一点的多策略对冲基金里也看到了同样的情况。
We see the same things, you know, in the slightly less high frequency in the multi strategy pod shops.
他们非常擅长找出某个特定领域里的机会，但他们交易的股票或资产其实都是相似的。
They do a very good job of figuring out where in a particular space there is the opportunity, but they're really not trading in stocks or assets that are unalike.
而这对他们来说风险太大了。
And that is just too risky an activity for them.
而且我不认为这种情况短期内会改变。
And I don't imagine that is going to change anytime soon.
当你使用大量杠杆时，你根本无法承担某些类型的风险。
You know, when you are running with lots and lots of leverage, you simply cannot take certain kinds of risk.
现在我想再多谈谈风险。
Now I I wanna talk about risk a little more.
我们一直关注股票方面，但这里还有很大的债务和信贷部分。
We've been focused on the equity side, but there's a big debt and credit component here.
资本成本上升的部分原因，正是有大量资本涌入这个领域来资助建设。
It's part of what's pushing the cost of capital up is just how much capital is coming into the space to fund the build out.
我们开始看到数据中心有点占据了私人信贷市场，占据了贷款市场，而且这些是私人信贷公司，是银行。
We're starting to see data centers sort of taking over the private credit markets, taking over the loan markets, and it's private credit firms, it's banks.
人们担心这些风险敞口的系统性影响。
And people have concern as to the systemic nature of these exposures.
我们谈论的金额相当惊人，无论是银行对特定项目的直接贷款，还是向非银行金融机构贷款，然后他们再用这些资金去发放更多贷款。
I mean, it is pretty staggering the amount we're talking about here, whether it's direct lending on specific projects from banks or it's lending to non bank financial institutions who are then using that to go give more loans themselves.
如果股权估值出现回落，无论是否导致AI需求放缓，都会引起动物精神的改变。
If we do get a pullback in equity valuations and it causes whether we see the demand for AI slowdown or not, causes a change in the animal spirits.
你是否担心信贷市场以及所有这些贷款的系统性风险？
Are you worried at all about credit markets and the systemic nature of all this lending?
嗯，首先我要说的是，我对信贷市场非常担忧。
Well, I mean, the first thing I would say is I am worried about credit markets full stop.
从我们的角度来看，信贷利差最近有点异常地紧。
Credit spreads are sort of weirdly tight these days from our perspective.
它们在九月份有所扩大，但尽管如此，我们仍然处于历史百分位的第十三位左右，无论是高收益债券还是投资级债券。
They widened out in September, but still, think we're something like the thirteenth percentile versus history for high yield and the thirteenth percentile versus history for investment grade.
关于AI，有一点可以说，对吧，就算我们假设AI会发展得非常庞大，以至于你不用担心AI领域的信用风险。
And one of the things you can say about AI, right, even let us assume that AI is going to be huge enough that you don't have to worry about credit risk within the AI space.
但几乎不可能完全不担心，因为作为信贷提供方，如果事情发展得特别好，你能赚一些钱，但不会太多。
It is almost inconceivable that you shouldn't worry at least a little bit because one of the features of being a credit provider is if things turn out fantastically well, you make some money, but not a ton.
而如果事情发展得不好，你会损失很多。
And if things turn out badly, you lose a lot.
所以，信贷的风险回报是不同的。
So, the risk reward is different in credit.
这也是为什么信贷市场自然不愿意把所有的钱都投到一个特定领域。
And that's one of the reasons why credit naturally is resistant to wanting to throw all of its money at a particular area.
如果你认为AI的投资回报率会比其他地方高得多，而且你是股票投资者，你可能会说，是的，我想在那里全押。
If you think the return on investment in AI is going to be so much higher than everywhere else, and you are an equity investor, maybe you say, yes, I wanna do an all in bet there.
但在信贷领域全押是没有意义的，因为正面你只赢一点，反面你会输很多。
But it doesn't make any sense to do an all in bet on the credit space because, you know, heads you win a little bit, tails you lose a lot.
你需要分散投资。
You need to be diversified.
所以，我们现在开始遇到这样的情况：在AI相关的债务中，分散投资的需求开始显现出来。
So, we are starting to hit the situation where the need for the diversification is biting in AI related debt.
我们也在看到一个，我认为，与AI无关的问题正在冲击信贷市场。
We're also seeing, I would argue, an unrelated problem that is hitting the credit markets.
我之前说过，利差，也就是你在高收益债券中获得的回报，相对于历史水平来说是偏低的。
I said that the spreads, the amount that you're getting paid in high yield is low relative to history.
对于高质量的垃圾债来说尤其如此。
And that is particularly true for high quality junk debt.
BB级和B级的债券，它们的利差真的很紧。
The BBs and the Bs, their spreads are really tight.
但CCC级，也就是评级更低的债券，情况就不再是这样了。
The CCCs, the lower rated debt, that is no longer the case.
我认为截至九月底，CCC级债券的交易利差显著高于平均水平，大约在第七十四个百分位，而第一百个百分位是历史最宽的水平。
I think the CCCs as of the end of September were trading significantly wider than average, something like the seventy fourth percentile where one hundredth is the widest ever.
我们在GMO的信贷团队开始试探性地进入这个领域，但还没有真正兴奋起来，因为我们看到那些高风险结构中有很多是在2020年、2021年利率为零的时候建立的。
Our credit team within GMO is starting to dip its toes in, but is not really excited yet because what we see in those really risky structures is a lot of them were built in 2020, 2021 when interest rates were zero.
这些资本结构在利率为零的时候是可行的。
And they are capital structures that worked when interest rates were zero.
现在利率到了5%，这种资本结构就行不通了，必须进行重组。
And now with interest rates of 5%, the capital structure simply does not work and you need to have a restructuring.
所以现在有些债权人或借款人已经陷入困境了。
So there are creditors or there are borrowers who are in trouble now.
而且正如你所说，AI领域对资本的需求是无穷无尽的。
There is also this insatiable demand, as you said, for capital in the AI space.
他们越来越需要支付溢价才能吸引 lenders 来借钱给他们，因为多元化已经成了问题。
And they are increasingly having to pay a premium to get lenders to come in to lend to them because the diversification is getting to be a problem.
如果所有计划中的投资都到位的话，这会变成一个严重的问题，对吧？
And if we get all of the investment that is slated to come, this will become a profound problem, right?
因为我们会看到AI领域的公司债务集中度超过我们现代信贷市场历史上所见过的任何情况——当然，铁路时代除外，那时候基本上所有债务都是铁路债务。
Because we will see a concentration in the AI space within corporate debt that is greater than anything we have, well, I mean, not anything we have ever seen in the railroad times, basically all debt was railroad debt, but in the modern history of the credit markets.
而且即使AI债务没有出现大规模违约，这也会带来麻烦。
And that will create trouble even if we don't get kind of a massive wave of defaults of the AI debt.
确实很讽刺，就在我们因为人们过度集中在软件行业而得到提醒之后，他们却选择把所有这些钱投到那个可能是唯一在自我颠覆的行业。
It is certainly ironic that right after we get the reminder because people got over concentrated in software, they've chosen to throw all of this money at the sector that is perhaps the only thing that is disrupting itself.
比软件被颠覆得更快的只有AI本身，而人们正是在信贷市场上转向这个领域。
The only thing getting disrupted faster than software is AI itself, and that's where people are turning in the credit markets.
但我也想谈谈这对其他行业的影响，因为资本成本正在上升。
But I also want to talk about the interplay to those other sectors, because the cost of capital is rising.
这在很多方面正在挤掉其他借款人，他们只能看着自己将要支付的利率。
It is in many ways choking out other borrowers who are just looking at the rates that they're going to have to pay.
一是因为基准利率在上升。
One, because of the rising base.
无风险利率已经涨了很多，但也是因为市场内部的竞争。
The risk free rate has come up so much, but also because of this competition within the market.
你之前提到过房地产，说房地产现在完全处于低谷。
You talked about housing earlier, how housing is just absolutely on the floor.
如果我们继续看到这种情况，你觉得还有哪些行业无法承受更高的借贷成本？
I mean, if we continue to see this, what are the other sectors that you think are not going to be able to withstand the higher cost of borrowing?
这对任何依赖大量借贷的行业来说都是个问题。
I mean, it is a problem for anything that relies on lot a of borrowed money.
所以，显然房地产就是这样。
So, obviously housing does that.
汽车行业也是这样，各种耐用消费品都是如此。
The auto sector does that kind of durable stuff of all kinds.
所以，那些不是直接支撑数据中心的长线投资，全都得不到资金。
So, long dated investments that are not directly powering data centers is all being starved.
但这也不是什么天大的坏事，对吧？
And that's not the worst thing in the world, right?
说到底，作为投资者，你是在提供资本。
At the end of the day, as investors, you are providing capital.
把钱投到资本稀缺的地方，通常是一门不错的生意。
And providing capital into places where there isn't a lot of capital tends to be a pretty good business.
所以，虽然这确实会带来一些社会问题，对吧？
So, whereas there are societal problems, right?
如果我们没有建造足够的房子，那确实不好。
If we are not building enough houses, that's bad.
但如果我们没有建造足够的房子，向建造房子的公司提供资金，实际上结果会相当不错。
But if we are not building enough houses, supplying capital to the companies building houses actually turns out pretty well.
所以我们看到，越来越明显的是，所有非AI的东西都开始受到挤压。
So, we are seeing it is increasingly the case that everything non AI is starting to get squeezed.
而且，如果AI的建设继续按照人们假设的速度增长，情况还得变得更糟。
And again, this is going to have to get a lot worse if the AI buildout continues growing at the rate that people are assuming it will.
所以不幸的是，你可能会陷入整个经济都押注在AI上的局面，这可不是你想要的。
So unfortunately, you can wind up in a situation where kind of the entire economy turns into a bet on AI, which is not really where you would want to be.
但如果我们想按照我们试图投资的速度来投资，这很难避免。
But it is sort of hard to avoid if we are going to invest at the rate that we are trying to invest.
那政府投入呢？
What about the government put?
我们看到所有这些AI领袖都在向现任政府靠拢，甚至在某些方面呼吁自我监管。
I mean, we're seeing all of these AI leaders cozying up to the current administration, even in some ways calling for self regulation.
你可以说他们也看到了这一点。
You could argue that they're seeing this too.
AI已经渗透到各个角落，规模之大意味着即使事情放缓，也不一定会突然停滞。
The scale with which AI has kind of gotten its tentacles into every corner means that that if things do slow down, it doesn't even have to come to a crashing halt.
只是放缓。
Slow down.
可能会出现整个经济都受影响的情况。
You could have a scenario where it is it is bad all over the economy.
如果政府不提供支持，可能会造成很大的问题。
And if you don't have the government there to backstop it, it could create a very big problem.
你觉得这是真的吗？
Do you think that that is real?
你觉得政府会提供支持吗？
Do you think the government backstop is there?
不幸的是，我认为政府已经负担不起作为后盾的角色了。
Unfortunately, I don't think the government can afford to be the backstop anymore.
我们有很多债务。
We have a lot of debt.
我们的赤字非常大。
We are running a very large deficit.
我不知道钱从哪里来，才能不破坏政府债券市场的稳定。
I don't know where the money comes from to do this without destabilizing the government bond market.
如果你破坏了政府债券市场的稳定，造成的损害可能比你能帮助的还要大。
And if you destabilize the government bond market, you kind of wind up doing more damage than you can help.
好消息是，如果这有点像互联网周期和光纤过度建设的重演，那个周期里损失了很多钱。
The good news is if this is sort of a rerun of the internet cycle and the overbuild out of fiber, there was a lot of money lost in that cycle.
那并没有破坏太多东西，因为那些钱不是由高杠杆实体借出的。
It did not destabilize much of anything because it was not lent by massively leveraged entities.
不是银行借出的。
It was not lent by the banks.
也不是那些承诺自己无风险的实体借出的。
It was not lent by entities that had promised that they themselves were not risky.
展开剩余字幕（还有 270 条）
所以，只要损失发生在合适的地方，即使损失很大，也不会引发灾难。
And so, you can have significant losses without a disaster as long as the losses live in the right places.
目前来看，这些损失大多都发生在合适的地方。
And at the moment, they largely seem to live in the right places.
不过，如果人工智能的建设突然停滞，造成大量损失，那确实会很糟糕。
Now, it will stink if the AI buildout kind of comes to a screeching halt and there's a bunch of losses.
我们在2001年、2002年经历过一次衰退。
We got a recession in kind of 2001, 2002.
那次衰退不算太严重，但确实发生了，而且利润大幅下降。
It wasn't a horrible recession, but we got one and profits came down a lot.
利润并没有持续低迷太久，衰退总是会发生的。
They didn't stay down that long and recessions happen.
在衰退期间，政府会降低利率，而这次政府有足够的空间来降息。
And in recessions, governments lower interest rates, and there is plenty of scope for the government to lower interest rates this time.
所以，我并不是说特朗普能救场。
So, it's not that I think Trump can come to the rescue.
我只是不知道这笔钱从哪儿来。
I just don't know where that money comes from.
但美联储，从现在的利率水平来看，或者至少从2027年的利率水平来看，如果经济突然停滞，它有很大的空间来降低资本成本。
But the federal reserve from where rates are now, or at least where rates are going to be in 2027, had plenty of room to lower the cost of capital if the economy comes to a screeching halt.
现在的问题是，经济运转得相当热，通胀比我们想要的要高，而且投资需求很大。
Right now, the danger is the economy is running pretty hot, inflation is higher than we would like, and there's a lot of demand for investment.
可能都是AI驱动的，但投资需求确实很大。
It may all be by AI, but it's a lot of demand for investment.
所以，这不断推高商品和服务的价格，以及资金的价格。
So, that keeps pushing up both prices of goods and services and the price of money.
但在经济衰退时，这种情况很快就会逆转。
But in a recession, that turns around pretty quick.
我认为美联储可以介入并降低利率。
And I think the Fed can step in and lower rates.
而且，很多进行这些投资的实体其实负担得起这些投资。
And again, a lot of the entities that are doing this investment can afford this investment.
微软可能无法从它的数据中心获得理想的投资回报，但我不认为微软会面临债务违约的风险。
Microsoft may not have the return on investment it would like from their data centers, but I am not expecting Microsoft is going to have to default on its debt.
对。
No.
而且有些人在关注甲骨文的情况，说这是不是第一块倒下的多米诺骨牌。
And some people are looking at what's happening with Oracle and saying, is the first domino to fall.
并不是说他们负担不起数据中心，而是他们没有获得监管批准。
It's not that they can't afford the data centers that they didn't have the regulatory approval.
所以很容易被这些头条新闻带偏，看到裂缝在形成。
So it is really easy to get caught up in some of these headlines and see the cracks are forming.
但如果不仔细研究，就容易被误导。
But if you don't look into it closely, you can get led astray.
你有没有看到哪些领域让你觉得担心，比如人们无法履行义务，还是说现在一切还在正常运转？
I mean, are you seeing any areas that say this is concerning to me that people are not being able to meet their obligations, or are the wheels continuing to turn right now?
目前来看，我们在公开信贷市场上并没有看到太多困境的迹象。
So right now, we don't see a lot of evidence of distress in the public credit markets.
在私募市场里，透明度低得多，很难说清楚。
In the private markets where there's a lot less visibility, it is hard to say.
我记得前几天读了一篇文章，说私募信贷的违约率，看问谁，有的说1%，有的说6%，还有的说19%。
I think I was reading an article the other day that said the default rate on private credit is 1%, 6%, or 19% depending on who you ask.
1和19之间差距可大了。
And there is a big gap between one and nineteen.
我不敢说谁是对的。
And I do not claim to know who's right there.
但总的来说，我们没看到太多困境。
But in general, we aren't seeing a lot of distress.
而且说实话，全球经济表现比我预期的要好。
And in general, frankly, the global economy is doing better than I would have expected.
虽然我们在大力建设AI，但想想欧洲的情况，想想世界上很多地方，尽管经历了严重的能源冲击，而且它们并不是AI建设的核心，世界却表现得惊人地好。每当世界表现得出乎意料地好，困境就会比原本要少。
We've got the AI buildout, but if you think about how Europe is doing, how much of the world is doing despite a very significant energy shock, and the fact that they are not central to the AI buildout, The world is doing surprisingly well, and whenever the world is doing surprisingly well, there is less distress than there would otherwise be.
现在，我们来谈谈那个能源冲击。
Now, let's talk about that energy shock.
正如你所说，欧洲尽管面临这些挑战，表现却出人意料地好。
As you said, Europe is doing surprisingly well despite it.
你可以说，欧洲以及亚洲的一些国家，当然还有全球南方，也就是世界上最贫穷的国家，它们对高能源价格的感受最为深切。
You could argue that them and maybe some countries in Asia and obviously the global South, the poorest countries in the world that feel high energy prices the most acutely.
但在发达国家中，欧洲和亚洲那些不生产能源的国家，可以说是受影响最大的。
But in a developed world, the non energy producing countries in Europe and as well Asia are arguably the most exposed.
如果这场危机继续像现在这样持续下去，你觉得我们会不会开始感受到那种痛苦？
If we continue to have the crisis stretch on as long as it has, Do you think that we're gonna start to feel that pain?
嗯，我们肯定已经在某种程度上感受到那种痛苦了。
Well, we must already be feeling that pain somehow.
柴油价格到处都很高。
The price of diesel is really high everywhere.
这确实增加了货物运输的成本。
That really increases the cost of getting goods from place to place.
这也增加了农业成本。
It increases the cost of agricultural.
这会推高很多其他东西的成本。
It increases the cost of a bunch of things.
问题就来了，什么时候这不再只是成本增加的问题，而是我们根本没法做到的问题？
And the question comes in, when does this stop being an issue of increased cost and starts being an issue of we simply cannot do this?
我不太清楚这两者之间的转折点在哪里。
I don't have a really good feel for when those two things change.
全世界一直在消耗它的储备，不可能永远这样下去。
You know, the world has been shooing through its reserves and it can't do that forever.
我觉得没人确切知道储备降到多低才会开始出问题。
I don't think anybody knows exactly how low it can take those reserves before problems start.
而且现在还有个有意思的地方，就是到底有多少石油被扣在市场上没放出来，其实不太清楚，对吧？
And at this point, it's also interesting, it's not entirely clear how much oil is being held off the market, right?
石油还是能通过霍尔木兹海峡。
Oil is getting through the Strait Of Hormuz.
它还是能通过红海，无论是往北还是往南。
It is getting through the Red Sea, both the northward way and the southern way.
这更麻烦一些。
It's more of a pain.
虽然更贵了，但还是在发生。
It's more expensive, but it is happening.
而且我认为高价格肯定在影响物价，这导致央行不得不加息，而它们原本并没有预料到需要这样做。
And I think the high prices are definitely affecting prices, and that is causing central banks to raise rates in a way that they were not expecting to need to do.
这种情况要持续多久才会演变成粮食安全危机？
How long does this go on before it turns into kind of a food security crisis?
我不知道。
I don't know.
我真希望我知道。
I wish I did.
嗯，我想换个话题，谈谈分配的问题。
Well, I wanna switch it maybe to more of an allocation question.
传统上，债券被视为用来对冲股票的工具。
Traditionally, bonds have been looked at to offset stocks.
能源实际上一直是与债券和股票呈负相关的东西。
Energy has really been the thing that has been inversely correlated with both bonds and stocks.
你说过把整个投资组合押在一件事上是个坏主意。
Now you said, you know, making your entire portfolio a bet on one thing is a bad idea.
我认为只增加到两件事可能也是个坏主意。
I'd argue only upping it to two is probably also a bad idea.
但如果你一直把50%投资在科技股，50%投资在能源上，过去十一年里有十年你都在赚钱。
But if you had been 50% invested in tech stocks and 50% invested in energy, you have made money 10 of the last eleven years.
同时长期投资科技和能源，这真是一个绝佳的杠铃策略。
It has been an incredible barbell to be long tech and long energy.
而且能源冲击这个问题似乎不仅仅是眼前的问题。
And this idea of energy shocks doesn't really seem like it's a just a right now problem.
在一个去全球化的世界里，国家之间冲突更多，这似乎可能会比过去几十年更频繁地发生。
In a deglobalized world with with more conflict between nations, it does seem like this might be something that we can expect to be more regular than it was maybe in the past few decades.
那么作为分配者，你是否增加了能源的份额？
So as an allocator, have you been increasing the share of energy?
那你觉得，在这个相关性不断变化的世界里，能源是不是应该作为一个战略性的配置？
And do you think about energy as a strategic allocation in this world where correlations are shifting?
我们确实一直很喜欢能源类股票，更广泛地说，是资源类股票，因为它们具有分散风险的特点。
So we have definitely liked energy stocks and let's say more broadly resource stocks for quite a while because of their diversifying characteristic.
大多数股票之间的相关性都很高。
Most stocks are highly correlated to each other.
不管你是按天、按周、按月还是按年来看，都是如此。唯一一组随着时间跨度拉长，相关性反而持续下降的股票，就是资源公司。
This is true whether you look on a daily basis, a weekly basis, a monthly basis, a yearly basis, the one group of stocks where that correlation keeps dropping as the timeframe gets longer is the resource companies.
所以我们喜欢资源类，是因为对其他人来说很糟糕的情况，对它们来说可能并不糟糕。
So we like resources because there is a bad case for everybody else that isn't a bad case for them.
它们并不能抵御所有类型的通胀冲击，但其中有一部分，它们确实能很好地应对。
They are not guaranteed against all kinds of inflationary shocks, but there are a chunk of them that they are very well positioned for.
所以，是的，我认为从长期来看，如果我对其他一无所知，我会超配的股票类别中，资源类会是其中之一。
And so, yes, I think on a secular basis, groups of stocks that I would be over allocated to if I knew nothing else, One would be the resource space.
另一个就是高质量股票，因为它们在经济上风险更低，而且你似乎不需要用更低的长期回报来换取这种风险的降低。
The other would be high quality stocks because they are less risky economically, and you don't seem to have to pay for that reduced risk with a lower long term return.
所以，这两类就是我说的，如果你把我扔进市场里，我什么都不知道，我会不会过度持有某些类别的股票？
So, those are the two groups that I would say, if you just dropped me in the market and I knew nothing about what was going on, are there any groups I'd over own?
那就是优质股和资源股。
It'd be quality and resources.
现在，当我们真正构建投资组合时，关键问题是，好吧，但它们的定价如何？
Now, when we're actually building portfolios, the crucial thing is, okay, but how are they priced?
当人们都说‘哦，是的，我们需要尽可能多地持有能源’时，你不想持有能源。
You don't want to own energy when people are saying, oh yes, we need to own as much energy as possible.
当优质股成为潮流时，你不想持有优质股。
You don't want to own quality stocks when quality is all the rage.
而今天，我们说我们宁愿持有非美国股票而不是美国股票，因为价格更低。
And today, we say we'd rather own non US stocks than US stocks because the prices are lower.
我们宁愿持有价值股而不是成长股，因为那个差距真的很大。
We'd rather own value stocks than growth stocks because that spread is really wide.
这个差距可能会保持很大，对吧？
It's possible it's gonna stay wide, right?
我们之前聊过，股票供应方面的弹性越来越小。
We talked earlier on about this increasing inelasticity of supply with regard to stocks.
这可能意味着，成长股的持有者往往在价格上涨时不卖，而价值股的持有者在价格上涨时确实会卖，这两者估值之间的差距可能会一直保持。
And maybe that means that the spread between the valuations of growth stocks where the holders tend not to sell if price goes up, and value stocks where people really do sell if price goes up, maybe that stays wide.
但关键是，即使这个差距一直存在，对价值股来说也是个很大的优势，因为价值股和成长股都不是一成不变的公司群体。
But the thing is, even if that does stay wide, that is a lovely advantage for value stocks because neither value nor growth are a static group of companies.
如果你在年初买了一批价值股，到年底再看，有些已经不再像价值股了。
And if you buy a group of value stocks at the beginning of the year and you come back at the end of the year, some of them no longer look like value stocks.
这是好事，因为它们看起来像成长股了。
And that's a good thing because they look like growth stocks.
当价值股和成长股之间的折价越大，你投资组合中有10%的股票晋升到成长股行列时，在差距大的时候，这简直太棒了。
And the bigger the discount is between the value stocks and the growth stocks, when you get 10% of your portfolio that graduates into the growth universe, that is wonderful when the spread is wide.
成长股那边也会发生同样的情况。
And the same thing happens on the growth side.
你买了一批成长型公司，到年底再看，有10%已经不像成长型公司了。
You buy a bunch of growth companies, you come back at the end of the year and 10% of them don't look like growth companies.
嗯，如果你付了很高的溢价，那就不妙了。
Well, if you paid a really big premium, that's bad.
所以自2018年以来，价值和成长之间的差距普遍存在，但近年来，这对价值来说其实非常有利。
And so we have seen since 2018 generally widespread between value and growth, but in recent years, that's actually worked out really well for value.
有意思。
Interesting.
你说除美国以外的股票，美国以外的一些地方和这个AI主题有关联。
Now when you say ex US stocks, some of the places outside of The US are tied to this AI theme.
你是不是在低配像韩国和台湾这样的高飞市场，它们确实从AI主题中获益不少？
Are you underweighting the high flyers like Korea and Taiwan that have have certainly benefited from this AI theme?
我想简短的回答是没有，因为我们超配新兴市场，而它们是新兴市场中最大的两个市场。
I guess the short answer is no because we overweight emerging markets, and they're the biggest two markets in emerging.
我们先撇开中国不谈。
Let's hold aside China for the moment.
所以如果我们超配新兴市场，我们就会有一堆这样的股票。
So if we're overweight emerging, we have a bunch of these stocks.
但我想说的另一点是，确实有很多公司在韩国和台湾受益很大。
But the other thing I would say is yes, there are a bunch of companies who have been really big beneficiaries in Korea and Taiwan.
总的来说，它们的估值并没有涨太多。
In general, their valuations haven't gone up that much.
真正发生的是它们的利润暴涨了。
What has happened is their profits have exploded.
相比价格低很多的时候，我们现在对这些公司的热情没那么高了。
And we're less excited about those companies than we were when the prices were much lower.
但台湾的情况并不是到处都是帕兰提尔那样的公司，那些公司确实表现很好，但定价却好像它们会成为史上最了不起的公司一样。
But this isn't a situation where Taiwan is full of Palantirs, you know, companies that have legitimately done very well, but are priced as if they are going to be the most amazing companies in history.
这些公司之所以表现好，是因为利润涨了十倍，而股价只涨了七倍。
They are companies that have done very well because their profits just went up tenfold, and the price of the stock just went up sevenfold.
所以我们仍然持有，因为利润增长的速度比股价快。
So, we are still holders because the profits have been going up faster than the prices.
我们持有的少了一点，因为我们不认为这些利润能完全持续下去。
We hold a little bit less because we do not believe that those profits are fully sustainable.
嗯，这正是我接下来想谈的，就是利润增长的可持续性。
Well, that is exactly where I wanted to go next, which is the sustainability of the profit growth.
这在新兴市场确实如此。
It's true in emerging markets.
在美国也是如此。
It's true in The US.
标普500的盈利增长非常惊人。
The S and P five hundred earnings growth has been spectacular.
我们即将迎来又一个财报季。
We're about to see another earnings season.
人们预测这种惊人的盈利增长会持续下去。
People are predicting that spectacular earnings growth to continue.
你觉得这能持续多久？
How sustainable do you think that is?
然后我还想谈谈这些历史上具有周期性的行业和业务，大家最关心的问题是，我们是不是正处于盈利的顶峰？
And then also, I wanna talk about these historically cyclical industries, historically cyclical businesses where the big question on everyone's mind is, are we at peak earnings?
比如说，你看看那些存储类股票，它们看起来便宜，但周期性股票往往在最不该买的时候显得便宜。
If you look at the memory stocks, for instance, they screen cheap, but cyclical stocks tend to screen cheap at the worst times to buy them.
所以作为价值投资者，我很想听听你对这些历史上周期性行业盈利过高问题的看法。
And so as value investors, I would love to get your perspective on the over earning question of these historically cyclical industries.
所以，特别是当你观察美国的盈利能力时，发生了一些极其引人注目的事情，正如你所说，标普的利润创历史新高，标普的资本回报率也看起来处于历史最高水平。
So, particularly when you look at profitability in The US, there is something utterly fascinating, has gone on, which as you said, profits for the S and P all time highs, return on capital in the S and P looks to be at all time highs.
如果你细分来看，排名前10或前50的股票，它们的盈利水平在历史上从未如此之高。
If you break that down, you know, the top 10 stocks or the top 50 stocks have never made money like this before in history.
接下来的50只股票，它们的日子也过得很好。
The next 50 stocks life is really good.
虽然不是历史上最好的，但历史上最好的盈利爆发期是2022、2023年从疫情中走出来的那波巨大盈利。
It's not quite the best in history, but the best in history was kind of that twenty twenty two, twenty twenty three huge burst of profitability is coming out of COVID.
现在，再往下看，比如标普中排名250到500的股票，感觉是，对，还不错。
Now, then you get to, let's say, you know, stocks two fifty to 500 of the S and P, and it's like, yeah, think it's pretty good.
可能处于历史百分位的第七十位左右。
Maybe it's seventieth percentile versus history.
再看中型公司，可能只有历史水平的四成，小型公司可能只有两成。
You go to the mid caps, and maybe it's fortieth percentile versus history and you go to the small caps and maybe it's twentieth percentile versus history.
所以在美国，我们看到一个很有意思的现象：从整体企业界来看，你会觉得这肯定是周期性的高峰，但深入表面之下，在周期高峰时，小型公司本该赚得盆满钵满。
So in The US, we have this fascinating thing where if you look at the overall corporate sector, you would say this must be a cyclical peak, but you dig in under the surface and at a cyclical peak small cap should be minting it.
它们赚的钱会比大公司少，因为历来如此，但相对于它们自己的正常水平，它们应该表现得很好，但实际上并没有。
I mean, they'll be making less money than large caps because they always do, but relative to their normal, they should look wonderful and they don't.
所以关于美国的利润，我们想说的是，利润似乎越来越由市场权力驱动。
So one thing we would say about profits in The US is they seem increasingly to be driven by market power.
美国在大多数行业已经变得竞争性大大降低。
The US has become a much less competitive place in most industries.
竞争少的地方，利润就很好。
And the places where there's less competition, it's been great for profits.
我看不到这种模式会明显逆转。
I don't see that pattern obviously reversing.
政府在反垄断方面并不是特别积极。
The government is not being really aggressive on antitrust.
但有一个现在变得极其重要的前提，就是在科技领域，那些科技垄断巨头们目前正在投资一个比他们通常所处的领域竞争激烈得多的领域。
The caveat that is now an incredibly important caveat is on the tech side, those tech monopolists are currently investing in a space, which is much more competitive than the space they normally live in.
当你建设数据中心时，有一点非常重要，那就是数据中心需要具备互操作性，因为你需要确保如果有人想迁移到你这里，他们可以做到。
When you are building data centers, one really important thing about data centers is you really want interoperability because you need it to be the case that if somebody wants to move to you, they can.
否则，你就没有能力去抢占市场份额。
Otherwise, you don't have the ability to take share.
所以数据中心本质上是一个没有差异化的业务。
So data centers is inherently an undifferentiated business.
我的意思是，并非完全没有差异化，但这是一个很难做到深度差异化的业务。
I mean, not entirely undifferentiated, but a difficult to profoundly differentiate business.
现在有五家公司正试图将算力商品化，并把它变成期货。
There are five companies right now trying to commoditize compute into futures.
期货的核心意义就在于它们之间是可以互换的。
The whole point of futures is that they're interchangeable with each other.
所以我们可以达成共识，算力是一种商品，而且它也正是以这种身份被出售的。
So we can agree that compute is a commodity and that's what it's being sold as.
对。
Exactly.
而且你知道，现在AI模型的交付竞争挺激烈的。
And you know, the AI model delivery right now it is pretty competitive.
你之前提到，每个新模型出来都会超越前一个。确实，我自己用AI的时候，有个模型路由器在帮我判断，怎么最高效地完成任务。
You were saying, as each model comes out and leapfrogs the one before, and certainly, you know, when I am using AI, I have a model router that is trying to figure out, okay, well, what is the most efficient way to get this thing done?
它能在十几个模型里选。
And it gets to pick among over a dozen models.
完全不存在什么锁定效应。
The lock in is completely nonexistent.
不只是我有个下拉菜单可以随便选。
It's not just that I have a drop down menu and I get to pick any of them.
而是我的软件会根据每个问题，自动切换到不同的模型。
It is that I have software that each question, it is moving to a different one.
所以这看起来也是个竞争很激烈的行业。
So that also feels like a pretty competitive industry.
而这并不是美国在盈利方面成为世界羡慕对象的原因。
And that is not why The US became the envy of the world from a profitability perspective.
所以我认为美国市场结构的某些部分不太可能改变，而且规模大仍然是有好处的。
So I think certain parts of The US market structure are not likely to change, and it's still gonna be good to be big.
但最奇怪的是，在市场的最顶端，这些公司向来是资本轻的公司，不需要大量投资，投资回报率却很高。
But the very strange thing is at the very top of the market, man, these were always capital light companies that didn't need to invest in ton and had huge returns on investment.
而现在它们把自己变得资本重了很多，并且在一个竞争激烈的领域投资，很难理解资本怎么能获得超常回报。
And now they have made themselves much more capital heavy, and they're investing in a competitive space where it's hard to understand how there's going be an extraordinary return on capital.
即使有回报，也很难理解谁能够负担得起所有这些费用。
And if there is, it's really hard to understand who's going to be able to afford to pay all of this money.
你谈到了我们对变化有多敏感。
You talked about, you know, just how sensitive we are to change.
听起来资本管理的方式似乎并不太适应变化，但如果美国前十的股票价格连续三年持平或下跌，会发生什么？
It doesn't sound like much of the way capital is managed is set up to deal with change, but what would happen if we had three years of sideways to negative price action in the top 10 US stocks?
比如，什么……
Like, what...
市场参与者会怎么反应，考虑到它对全球回报的持续重要性？
How would market participants react to that given how important it has been for sustaining returns for for the entire world?
世界其他地方很多资金都押注在美国股市上涨上。
So much of the rest of the world is invested in US stocks going up.
金融体系有多少是建立在那些股票会继续主导行业、保持定价权并增长盈利这个假设上的？
I mean, how much of the financial system is predicated on the idea that those 10 stocks are going to continue to dominate their industries and have that pricing power and the ability to grow their earnings?
一方面，我认为那些公司非常有能力继续主导它们目前所在的行业。
On the one hand, I think those companies are very well positioned to continue dominating the industries that they currently dominate.
只是随着它们的投资方向变化，这部分业务可能变得不那么重要。
It's just that may become a less important part of their business, given where their investment is going.
股市能承受最大股票横盘一段时间的情况吗？
Can the stock market, you know, absorb a situation where the very largest stocks go sideways for a while?
是的，实际上这是常态，对吧？
Well, yeah, because actually that's the normal pattern, right?
如果你看过去一百五十年，通常最大股票的表现是落后的。
If you look over the last one hundred and fifty years, in general, the largest stocks underperformed.
那为什么历史上一直是这样呢？
Now, why has that been true historically?
嗯，首先，我们是用市值来定义大小的。
Well, one, we are defining size in terms of market cap.
所以，要成为全球最大的公司，一种方式就是被高估。
So one way you get to be the largest company in the world is by being overvalued.
另一种方式，历史上，规模是有限制的，因为要增长10%所需的投资，如果你像AT&T那么大，那投资就太大了。
The other is historically there have been limits to scale where the given the investment you need to grow by 10%, if you are AT and T is so big.
你知道，如果我们是在1950年代或1960年代，天哪，除非经济增速那么快，否则你根本做不到。
If you know, we're in 1950s or 1960s, one, man, you can't do it unless the economy is growing that fast.
而且，要做到这一点，你需要投入巨资，这样一来，一旦出错，后果就相当严重了。
And two, you would need to invest so much in order to do that, that the stakes of being wrong become kind of catastrophic.
现在，世界变化的一个方面是，如今的巨头们比过去的巨头们垂直整合程度要低得多。
Now, one way that the world has changed is the giants today are much less vertically integrated than the giants of the past.
如果英伟达需要自己建工厂，并且要提前四年预测，因为建一座工厂需要四年时间，来猜测全世界需要多少GPU加速器、AI加速器，那将是一场高风险的赌注，而且他们可能会输，因为AI的发展速度可能不如他们预期的那样快。
If NVIDIA needed to build its own fabs and needed to do that guessing four years ahead of time, if it takes four years to build a fab of just how many GPU accelerator, AI accelerators the world is going to need, that would be a very high stakes bet And they could lose because AI grows less fast than they want.
他们也可能会输，因为人们想要的是TPU，而不是他们的GPU，或者AMD占了上风，或者有别的公司冒出来抢了他们的首发。
They could also lose because either people want TPUs, not their GPUs, or AMD gets an advantage over them, or somebody else comes in and steals their launch.
台积电做这个投资的风险比英伟达做这个投资的风险要小。
It is less risky for TSMC to do that investment than it would be for NVIDIA to do that investment.
不过台积电做这个投资还是有一定风险的，因为他们现在投资了很多。
It's still kind of risky for TSMC to be doing that investment because they're making a lot of investments today.
而且所有这些公司都在大量投资。
And all of these companies are investing a ton.
而且这种情况有可能变得相当糟糕。
And it is possible for that to turn out badly enough.
不一定要糟糕到灾难性的程度。
It doesn't have to be catastrophically badly.
只要结果不是那么好，股价就不会上涨。
It just if that turns out not that great, the stocks aren't gonna go up.
是的，世界会适应，但世界也会改变。
And yeah, the world will adapt, but the world will change.
而世界可能变化的一个方式就是，这些股票不再称霸全球。
And one of the ways the world might change is if these stocks stop being world beaters.
人们会说，也许我不想让我的投资组合里有那么多美国资产了。
People will say, maybe I don't wanna have as much of my portfolio tied up in The US as I do.
也许他们会说，也许我不需要把那么多投资组合放在股票上了。
Maybe they will say, maybe I don't need to have as much of my portfolio tied up in stocks as I do.
这不会是世界翻天覆地的变化，但在边缘上，这些变化很重要，而且它们会悄悄影响你。
It won't kind of be a profound change in the world, but at the margin, these changes matter and they kind of sneak up on you.
在2012年，也就是新兴市场最火热的时候，人们说当然你需要超配新兴市场，对吧？
You know, in 2012, at sort of the height of the bricks, people said, you know, of course you need to be overweight emerging, right?
那里才是增长的地方。
That's where the growth is.
不要投资发达国家。
Do not invest in the developed world.
他们这么做的时候，正好赶上新兴市场彻底走下坡路，接下来十年新兴市场都没有盈利增长。
They did that just in time for emerging to fundamentally roll over and there was no earnings growth in EM for the next ten years.
现在人们已经记不起还有什么时候不是这样想的：当然要大幅超配美国。
And now people do not remember a time when it wasn't the case that of course you want to be massively overweight The US.
现在，已经两年了，美国的表现落后于世界其他地区。
Now, it's been two years where The US has lost to the rest of the world.
等到人们意识到这一点，并且这成为普遍共识——哦，美国已经过了巅峰，去别处投资吧——到那时，这就会开始变成坏主意。
By the time one wakes up to that and it becomes the conventional wisdom, oh yeah, The US has passed its peak, invest elsewhere, That will start being a bad idea by then.
我上一个问题问得太多了。
I gave you too many parts of the last question.
所以我想回到其中一个部分，就是周期性盈利的故事。
So I wanna turn back to to one part, is the cyclical earnings story.
这些周期性行业一直受到这种长期趋势的影响，以及关于盈利峰值的问题。
These cyclical industries that have been exposed to this this secular trend and the question of peak earnings.
你怎么看待那些按远期盈利看起来便宜，但我们对是否已经达到周期性峰值存在巨大不确定性的股票？
How do you think about stocks that screen cheap on their forward earnings, but there's tremendous uncertainty as to whether we have hit the cyclical peak?
从历史上看，现在不是买入像内存制造商这类公司的时候。
Historically, this would be a bad time to buy a company like the memory makers.
他们是商品生产者，对吧？
They are commodity producers, right?
他们完全是价格的接受者。
They are complete price takers.
正如你所说，他们在周期的高峰期，市盈率看起来最低。
And as you say, they look cheapest on a PE basis at a peak time in their cycle.
所以不仅增长会消失，实际上还会变成负增长，对吧？
And so not only does the growth go away, it actually turns negative, right?
如果美光今天的毛利率是88%，我觉得就算不对，也差不了多少。
And if Micron today has a gross margin of 88%, I think if I'm not right, I'm really close.
他们可以卖出更多的内存芯片，但如果利润率回落到历史水平，收入仍然会下降。
They could sell a lot more memory chips and still have their revenue drop if their profit margins were to fall to anything like historic levels.
所以，并不是说这些人不会看到收入下降。
So, it is not the case that these guys can't see their revenue fall.
不仅仅是他们的利润率面临风险。
It's not just that their margins are at risk.
如果他们的利润率下降，由于利润率太高，他们的收入就会面临风险。
If their margins come down, their margins are so high that their revenue will be at risk.
另一方面，我觉得难以判断的是，大多数大宗商品的需求增长速度相当缓慢。
I would say on the other side, the thing that makes this difficult to handicap is the rate of demand growth for most commodities is pretty slow.
这主要受经济增长的制约，而经济增长每年实际只有几个百分点。
It's kind of governed by economic growth, which is only a few percent real per year.
这就像一个终极限制，就像利润不可能超过GDP的100%一样，因此，在某个时候，他们必须停止比GDP增长得更快。
It is an ultimate limit in the same way that profits cannot be more than 100% of GDP, and therefore at some point they need to stop growing faster than GDP.
但这样说没什么实际意义，因为好吧，从GDP的15%到100%之间还有很大的空间。
But that's not a useful thing to say because all right, you've got a lot of room between whatever 15% of GDP and 100.
我不知道这次内存短缺会持续多久。
I don't know how long this memory shortage is going to last.
我猜我们很快就会看到另一面，这些公司不会是一个很好的投资，但与其他一些AI受益者不同，至少它们的交易水平使得你当前的现金流在你总投资成本中占很大比重。
My guess is we're going to see the other side soon enough that these companies won't be a great investment, but unlike some of the other AI beneficiaries, at least they are trading at levels where your current cash flow is a really meaningful part of your total investment cost.
所以，如果这种情况持续三年，那就太好了。
So, if this goes on for three years, that's great.
我不知道这会不会持续三年，但这确实是个棘手的问题。
I don't know whether it's gonna go on for three years, but it is a tricky problem.
我真希望这些家伙的估值很高，这样我就能说，对，别碰他们，因为那太蠢了。
I wish these guys were trading at high valuation, so I could say, yeah, just don't own them because that's stupid.
但市场也不是完全没脑子。
But the market isn't being entirely stupid.
现在我最喜欢问那些注重价值的投资者的问题是：美光便宜吗？
It's my favorite question to ask value minded investors right now is is is Micron cheap.
这确实是个让人一时答不上来的问题。
It is a true stumper at this moment in time.
但我想回到泡沫破灭的话题上，因为你特别关注了股权供应这个因素，以及它可能带来的影响，并试图把它单独拿出来分析。
But I wanna just go back to the sort of bubble popping question because, you know, you focused so much on the equity supply component, the impact that that can have and tried to isolate that.
我觉得人们想到泡沫破灭时，想到的不是一年内下跌18%。
And I think when people think of a bubble popping, they don't think of an 18% decline in a year.
他们想到的是2000年纳斯达克下跌90%那种情况。
They think about 90% drawdown in the Nasdaq in 2000.
而且，是多种因素结合在一起才会导致那样的事情发生。
And there's a combination of factors that combine to make an event like that happen.
你觉得除了股权供应是戳破泡沫的那根针之外，还有没有其他因素，比如AI需求放缓，或者信贷事件导致借贷和建设停滞？
Do you see the recipe for that beyond just the equity supply being the thing that pricks the bubble, but a slowdown in demand for AI, a credit event that makes it so that the lending and the borrowing stops for the build out.
我们这里有哪些因素，会让情况变得比平均水平差20%、差24%更极端？
I mean, what factors do we have here to get something that is more extreme than just 20% worse, 24% worse than average?
所以现在，很多支出都是内部循环的。
So right now, so much of this is incestuous spend.
比如一半的计算能力，也许我现在说的数字不对，但我们就假设我说的数字是对的，因为肯定有一个数字是正确的。
Like half of the compute, maybe I'm wrong at this point, but let's pretend I am not wrong with the number I am stating because there has got to be a number that is right.
比如，如果现在一半的支出都花在训练新AI模型上，结果发现，天哪，我们在训练AI模型上花了太多钱。
Like if half of the spend right now is on training of new AI models, And it turns out, oh my God, we spent way too much money on training AI models.
所以，不应该有一半的计算能力都用在训练AI模型上。
And so, it shouldn't be the case that half of compute is going to be training AI models.
应该只有10%的计算能力用在训练AI模型上。
10% of compute is going to be training AI models.
所以突然之间，我们有了更多的算力可用，因为用于那个当前用途的部分减少了。
And so, suddenly we have a lot more compute available because less is dedicated to this thing that it is currently dedicated to.
这之所以发生，是因为我们发现，当没人真正在AI上花大钱时，他们的支出一年翻五倍也没问题。
And that is occurring because what we're finding out is, hey, when nobody was spending anything meaningful on AI, it was fine to have their spend quintuple in a year.
但现在他们真的需要看到这笔支出的投资回报，而回报并没有真正到来。
But now they really need to see the return on investment of that spend, and it's not really coming in.
所以推理的支出就下降了。
So the spend for inference comes down.
然后突然之间，我们从算力严重短缺的世界，变成了一个算力过剩的世界，这有很多好的连锁反应，因为它减缓了旧数据中心折旧的速度，对吧？
And then suddenly we go from a world where we are hugely short compute, which has lots of lovely knock on effects because it slows down the depreciation rate of the old data centers, right?
如果大家都缺算力，那我当然愿意为五年前能效较低的GPU付钱。
If we're all short of compute, then sure, I'm going to pay for a five year old GPU that is less energy efficient.
但如果现在世界突然算力泛滥，那五年的数据中心还不如直接关掉，因为那样成本更低。
But man, if now the world is suddenly awash in compute, well, that five year old data center might as well just turn off because it costs me less.
我可以在这个最先进的数据中心用更少的电力生成一个token，而且那里有充足的容量。
I can use fewer watts to generate a token in this state of the art data center and I got plenty of capacity there.
如果发生这种情况，那信贷方面就会出现大规模违约。
If that happens, then we will have massive defaults on the credit side.
我们会看到股权方面的公司倒闭。
We will see companies on the equity side go bust.
我们会看到超大规模数据中心运营商面临问题，因为他们的投资回报率很差。
We will see a problem for the hyperscalers where their return on investment turned out to be lousy.
那是不是市场下跌18%？
And is that a minus 18% in the market?
不是。
No.
那是市场下跌35%到50%。
That is a minus 35% to 50% in the market.
我不确定这一定会发生，但很容易发生。
I'm not sure it is going to happen, but it would be easy for it to happen.
再说，如果你更专注于AI领域，尤其是单一业务的AI公司，因为如果微软在数据中心上的投资回报率很差，微软的股价也会下跌，但微软还是会没事的。
And again, if you were more specifically in the AI space, particularly the more single line AI space, because again, Microsoft will go down if their return on investment on the data centers was lousy, but Microsoft will be okay.
Core Weave 就不行了。
Core Weave will not.
Super Micro 可能也撑不住。
Super Micro may not be.
而且这也不是说一定会发生，但说真的，如果真的发生了，我们也不该太惊讶，因为每一次真正改变世界的技术革命，都是这个模式。
And it's not that it is guaranteed to happen, but man, we really shouldn't be that surprised if it does, because that is the pattern we have seen for every truly revolutionary technological change that really has changed the world.
铁路深刻地改变了世界。
Railroads profoundly changed the world.
天哪，它们甚至改变了人类的基因，因为人类突然能以以前不可能的方式四处移动。
Hell, they changed the DNA of humans because humans could suddenly move around in a way that they couldn't before.
汽车真的改变了世界。
The automobile really changed the world.
电气化深刻地改变了世界。
Electrification profoundly changed the world.
互联网深刻地改变了世界，纳斯达克跌了90……
The Internet changed the world in a profound way, and the NASDAQ went down 90...
好吧，也许不是90%，但至少是80%。
Well, maybe not 90%, but at least 80.
所以当它发生的时候，如果它真的发生，你不会回头说，哇，天哪，那真是个震惊。
So when it happens, if it happens, you wouldn't turn around and say, wow, my God, that is a shock.
从现在来看，对大多数人来说，这感觉就像互联网泡沫时期一样震惊，就像铁路繁荣之后一样震惊。
From here for most people, it feels like a shock just as it did in the internet bubble, just as it did in the aftermath of the railway boom.
但这确实可能发生。
But it can happen.
我们在季度报告中提出的观点是，它可能在我们真正意识到过度投资是个问题之前就开始发生了。
And the point we were raising in the Quarterly is it might start happening before we actually understand that the over investment was a problem.
因为在互联网泡沫时期，泡沫已经彻底破裂了，我们才意识到，哦，天哪，互联网流量实际上并没有每三个月翻四倍。
Because in the internet bubble, the bubble had well and truly burst before we figured out, oh my god, internet traffic was not actually, you know, quadrupling every three months.
在那成为证据之前，其他事情拖累了它。
Other things dragged it before that became evidence.
好的。
Alright.
好消息是，说到解锁，我们有个日期。
Well, the nice thing is when it comes to the unlocks, we have a date.
我们有时间表。
We have a timeline.
我们知道那些是什么。
We know what those are.
当然，一旦我们获得那些S-1文件的正式开放，它可能是一个搁置文件，他们可以把它放在搁置架上。
Certainly, once we get the official opening of those s one filings, it could be a shelf filing and they can just put it up on the shelf.
但我们会有一个时间表，所有这些事情都会发生。
But we will have a timeline for all of these things happening.
好的，Ben，非常感谢。
Well, Ben, thank you so much.
人们可以去GMO网站阅读那份季度信，也可以关注你和同事们发布的研究。
People can go read that quarterly letter on the GMO website as well as follow the research that you and your colleagues put out.
非常感谢你来参加节目。
Thank you so much for coming on the show.
很高兴能来。
Happy to do it.
希望你喜欢这期节目。
Hope you enjoyed that.
我想特别感谢今天的赞助商LENS，股票代码是L-E-N-S，也就是Sarmaya主题ETF，它提供实物资产敞口，并且有可能让60/40的投资组合更加多元化。
I wanna say a big thank you to today's sponsor, LENS, ticker l e n s, the Sarmaya thematic ETF, which gives real asset exposure and the potential to diversify a sixty forty portfolio.
请访问sarmayaetf.com阅读招股说明书。
Read the prospectus at sarmayaetf.com.
披露信息在节目说明里。
Disclosures in the show notes.
感谢收看。
Thanks for watching.
关于 Bayt 播客
Bayt 提供中文+原文双语音频和字幕，帮助你打破语言障碍，轻松听懂全球优质播客。