爱泼斯坦文件、SaaS已死？、Moltbook恐慌、SpaceX与xAI合并、特朗普的美联储人选

甚至在周二下跌之前，布拉德，你可以谈谈，软件已经是今年标普指数中表现最差的子板块了。

And even before Tuesday's drop, and you can get into this, Brad, software was already the worst performing S and P subsection for the year.

顺便说一下，你报告的这些数字简直是轻描淡写。

By the way, the numbers you report are dramatic understatement.

我们已经抹去了数万亿美元的市值。

We we we've wiped out trillions of dollars in market cap.

Figma从高点下跌了80%。

Figma's down 80% from the high.

你知道，所有那些大公司

You know, all the big names

说清楚一点。

be clear.

那些是两天的数据。

Those were two day numbers.

那是从本周开始算的。

That was this week since.

这些是两天的数据，确实如此。

These are two day numbers, and that's yeah.

你可以

You can

给我们一个更宏观的视角。

give us the bigger picture.

这真是一场灾难。

This this is a a real train wreck.

我在年初，我想是1月6日，上过CNBC。

And I was on CNBC at the start of the year, I think, January 6.

尼克大概能找出来。

Nick can can can kinda pull that up.

当时有人问我，你觉得这些股票全都下跌怎么看？

And I was asked the question, you know, what do you think about all these stocks being down?

我说，听好了。

And I said, listen.

它们全都下跌了，其中90%的股票理应如此。

They're all down, and 90% of them deserve to be down.

让我们来看看这些图表。

So let's look at these charts.

大卫，我知道这是Sachs的图表，这是你最喜欢的。

David, I know this is Sachs, this is your favorite chart.

你和我在2022年时还曾回顾过，但现在我们已跌至历史最低点。

You and I were looking back in '22, but now we're at an all time low.

我们目前的远期收入倍数为3.9倍。

We're we're trading at 3.9 times forward revenue.

如果看下一张图表，尼克，按自由现金流倍数来看，也处于历史最低水平。

If you go to the next chart, Nick, you know, on a free cash flow multiple, also at an all time low.

因此，软件股现在不仅在收入倍数上处于低位，在自由现金流上——也就是那些盈利能力很强的公司——也同样处于低位。

So now software is trading not just at a low on revenue, but it's trading at a low on free cash, very profitable businesses.

我们这里还有另一张我认为很重要的幻灯片，就是当你想弄清楚它们为何下跌的原因时。

We've got another slide here that I think is important, which is, you know, when when you look at what why they're going down.

对吧？

Right?

它们在下跌，这说的是Salesforce。

They're going down, and this say this is for Salesforce.

它显示在过去几周里市值已经腰斩。

It shows it's been cut in half in the last couple weeks.

但最后一张图显示，它们下跌并不是因为收入下降。

But the final slide, they're going down not because revenue is falling.

看看这个。

Look at this.

对于软件公司来说，收入实际上是稳定甚至增长的，收入增长依然强劲。

Revenue is actually stable to increasing for software companies, revenue growth.

它们下跌是因为我们在为未来的不确定性打折。

They're going down because we're discounting that future uncertainty.

当像AI这样深远的技术出现时，突然间就会让人质疑那些未来的自由现金流是否还像以前那样确定和持久。

When something as profound as AI comes along, all of a sudden, it causes you to question whether or not there's as much certainty and durability in those future free cash flows.

以Salesforce为例，它的自由现金流倍数从30倍降到了15倍。

So in the case of take Salesforce, it's gone from 30 times free cash flow multiple to 15 times.

这意味着，今天买入它的人会说：听好了。

That means somebody buying it today says, listen.

我认为，十五年后的未来，我还能依赖这些自由现金流。

I think fifteen years into the future, I can count on these free cash flows.

对吧？

Right?

以前，他们愿意为三十年后的未来买单。

Before, they were willing to pay thirty years into the future.

但现在有了AI，我们连七年后的未来会发生什么都无法确定。

Well, hell, with AI today, we don't know what's gonna happen seven years into the future.

所以，为了让普通观众理解，为什么这些公司业绩达标了，股价却还在下跌——这两者完全是两回事。

So for people at home to understand why are these companies hitting their numbers, but their stocks are going down, they're two totally different things.

对吧？

Right?

好的。

Okay.

所以它们达到了业绩目标，但AI带来的逆风让人们对它们未来的实力持怀疑态度。

So they're hitting their numbers, but the headwind of AI means people don't believe that they'll be strong in the future.

SACs就是这些。

SACs are those.

好吧。

Well, okay.

我的意思是，当人们说AI会彻底摧毁SaaS时，这中间有点含糊其辞。

I mean, I think there's a little bit of a hand wave going on here when people say that AI is just gonna wipe out SaaS.

我不认为这是真的。

I I don't think that's true.

拿Salesforce这样的SaaS产品来说。

You take a SaaS product like Salesforce.

对吧？

Right?

这是一个庞大的系统，处理你所有的客户联系和收入。

It's a very large system that deals with all of your customer contacts and your revenue.

你不可能用一个未经充分审核的编码助手生成的代码来取代它。

You're not gonna replace that with code that's been spit out of a coding assistant that hasn't been fully vetted.

想想过去二十五年来，Salesforce的代码库中提交了多少错误报告，可能有数百万条。

Think about how many bug reports have been filed on Salesforce's code base over the last twenty five years, maybe millions of them.

这个系统已经在成千上万的大客户和企业中经过了测试。

That system has been tested across thousands of large customers and enterprises.

你认为可以轻易地把这套系统拆掉，换成昨天由AI引擎概率生成、并由一个小团队内部维护的代码，这在我看来并不现实。

The idea that you're just gonna rip out that system and replace it with code that's been probabilistically generated by an AI engine yesterday with a small team to maintain it internally, this doesn't seem realistic to me.

所以，再次强调，我认为那种‘所有SaaS都将死亡’的悲观预测言过其实了。

So again, I think this like very dire prediction of all SaaS is dead is overstated.

不过，我认为这里确实存在一些问题。

However, I do think that there are some issues here.

如果你的SaaS产品收费高昂，但用户只使用了其中少数几个功能，那么你确实可能成为被更定制化的方案取代的目标，对吧？

So if you're a SaaS product that charges a lot of money and people only use a handful of your features, then you are, I think, a target to be ripped out with something that's more bespoke, right?

因为投资回报率根本不存在。

Because the ROI just isn't there.

我也认为，在这个新世界里，你必须清楚地知道自己护城河是什么，因为生成和复制代码变得容易多了。

I also think that you have to be really clear about what your moats are gonna be in this new world because it is a lot easier to generate code and to copy.

所以如果你没有良好的护城河，那可能会很麻烦。

So if you don't have good moats, then you could be in trouble.

但在我看来，对SaaS公司最大的威胁在这里。

But here's where I think the greatest threat is to the SaaS companies.

在我看来，这并不是它们的生存问题。

It's not, in my view, their existence.

我不认为这是关乎生死的。

I don't think it's existential.

而是未来价值捕获将发生在何处。

It's where the future value capture is gonna be.

让我给你举个例子。

So let me give you an example.

所有这些SaaS产品都在其工具内部推出AI协作者功能。

All these SaaS products are rolling out like AI copilots inside their tools.

其中一些表现得相当不错，但它们仅限于在那个封闭环境中运行。

And some of them work pretty well, but they're limited to playing in that sandbox.

而你看看像Cloud Cowork这样的产品，它现在能连接到所有这些不同的SaaS工具。

Whereas you look at something like Cloud Cowork right now, it has connectors to all these different SaaS tools.

它能够跨所有这些工具提取数据，并在数据库和工具之间无缝运作。

It can pull in data across all these different tools and it works seamlessly across databases and tools.

这显然是一个非常有吸引力的位置，对吧？

And that's a pretty attractive place to be, right?

你的工作空间会是这些产品中的哪一个呢？

Like, you know, which one of these products is gonna be your workspace?

在我看来，你更希望你的工作空间是一个能够跨越多个工具、为你提供最广泛数据和上下文AI的平台，而不是在现有工具里塞满一堆独立的AI。

Seems to me that you're gonna want your workspace to be the one that spans across and gives you AI across the most data and context as opposed to having a bunch of separate AIs inside of your existing tools.

所以我认为对SaaS公司的风险不在于它们会被完全取代——虽然在某种程度上确实会发生，而在于它们会沦为技术栈中的旧层，而新的层将被构建在它们之上。

So I think the risk for the SaaS companies it's not that they get replaced, although that'll happen to some degree, but it's that they become an old layer of the stack that now there's a new layer that gets I built on top

我想这就是它的运作方式

guess it's the way it's

基础设施，所有的核心功能都会转移到栈的新一层，价值增值也发生在那里。

infrastructure, And all the action kinda moves to a new layer of the stack, and that's where the value add happens.

如果真的发生这种情况，就会削弱它们未来的机遇。

And if that happens, it kinda cuts into their future opportunity.

对吧？

Right?

因为许多公司都把AI当作下一个增长点，你看看它们的产品路线图就知道了。

Because a lot of these companies were banking on AI as their next you look at their product road maps.

对吧？

Right?

它们的所有规划都与AI相关。

It's all AI related.

所以在我看来，最大的风险在于，下一栈层的价值捕获将发生在别处。

So that to me, I think, is the big risk is that the value capture for the next layer of the stack happens somewhere else.

是的。

Yeah.

我在创业圈里也经历了这种情况，人们都涌向你所说的那个行动层。

I'm experiencing this in startup land where people go to the action, as you called it, sacks.

你所能做的最有成效的事情就是创建一个开放的爪子，以前这叫Clawed Bot，但不是Anthropic做的。

The most productive thing you can do is create an open claw, which used to be called clawed bot, not by Anthropic.

这就是我上周提到的开源项目，我们现在实际上已经创建了三到四个这样的智能体集群。

This is the open source project I talked about last week, And we've actually now created, like, three or four of these agent sacks.

我们购买了Mac Studios，现在在其中一些上运行Kimi。

We bought the Mac Studios, and we're now running Kimi on some of them.

我们不得不为这四个智能体开通SaaS账户。

And we had to open up SaaS accounts for these four agents.

因此，我们的SaaS支出在短期到中期实际上增加了，因为我们额外开通了四个Slack企业版、四个Notion和四个Google Docs。

So, actually, our SaaS spend went up in the short to midterm because we opened up four more Slack ex you know, enterprise versions, four more Notion, four more Google Docs.

所以这几乎就像是我们增加了四名员工。

So it's almost like we added four employees.

然而，我们现在已将人们原本做的约20%到30%的工作交给了这些代理，我认为每个月将10%到20%由人类完成的工作转移到代理身上是可持续的。

However, we now have put about 20 or 30% of the work people were doing into these agents, and I think it's going to be sustainable that every month we move 10 to 20% of work being done by humans into agents.

但我们绝不会使用工具内置的那些代理。

But we will never use the ones that are built into the tools.

说到这一点，萨克斯，使用Notion的AI工具确实不错。

To your point, Sax, using Notion's AI tool, it's nice.

使用Slack的也很棒。

Using Slacks, it's also very nice.

谷歌的Gemini则无处不在，出现在右上角。

And Google's got Gemini everywhere in the top right hand corner.

但当你用OpenClaw创建代理，并让它们说：嘿。

But when you make agents with OpenClaw and you have them saying, hey.

从我的日历里提取这些数据。

Pull this data from my calendar.

给这个人发一封邮件。

Send an email to this person.

在其中加入一些 Notion 文档。

Include in that some Notion documents.

它的强大程度令人难以置信。

It's unbelievable how powerful it is.

没错。

So Right.

我认为，这将由开源项目主导。

And that, I think, is gonna be owned by open source.

这意味着下一代公司可能根本不会开通这些账户。

That means the next generation of companies, they may never open up these accounts.

他们可能会使用更多定制化的软件，而且所有技术都具有通缩效应。

They may use more bespoke software, and it may all technology is deflationary.

我们都知道这一点。

We know that.

因此，你的 SaaS 支出可能会从员工薪资的 10% 降至 5%，再降至 1%。

So your SaaS spend might go from 10% of an employee's salary down to 5% down to 1%.

我认为这就是趋势，这意味着这些公司必须大幅削减支出，才能维持利润，同时必须对产品进行巨大升级。

That's what I think the trend will be, which means these companies are gonna need to really downsize their expense base in order to keep those earnings up, and they're gonna have to evolve their products massively.

产品必须提供更多的价值和更多的吸引力。

The products are just gonna have to provide more value and more hooks.

弗里德伯格，你对此有什么看法？

Friedberg, you have any thoughts on this?

好吧，你先说。

Well, go ahead.

我想回应一点。

I'd respond to one thing.

我认为SaaS公司面临的一个真正难题是，它们将采用开放数据还是封闭数据。

So I think one of the real conundrums for SaaS companies is whether they're gonna be open data or closed data.

我认为比尔·盖利已经提出了这个说法，现在已不再是开源或闭源的问题了。

I think Bill Gurley has sort of coined this term, so it's not open source or closed sources anymore.

而是开放数据还是封闭数据。

It's open data or closed data.

你能理解为什么他们希望采用封闭数据吧？

You can see why they'd wanna be closed data, right?

尤其是像Salesforce这样的大型套件，你可以声称自己是AI的工作空间。

Especially if you're a large suite like Salesforce, you can lay claim to being that workspace for AI.

你已经拥有了足够的工具，足够的产品组合。

You've got enough of the tools, you've got enough of a suite.

你希望提供这一点，希望捕获这一层AI价值。

You wanna provide that, you wanna capture that AI value layer.

是的。

Yes.

但是

But

但如果有人在使用Quadbot，或者未来新一代的Claude机器人，并且它们与所有其他系统相连，那么这将在企业内部制造摩擦。

still if there's someone using Quadbot or whatever the next generation of Claude Bots are gonna be and they're connected to everything else, then that is gonna create a friction in the enterprise.

这将为竞争对手创造机会，让他们说：不，不，不，我是开放数据。

And it will create room for a competitor to come along and say no, no, no, I'm open data.

我不介意不成为你们所有事务的工作平台。

I'm okay not being your workspace for everything.

我只愿意提供CRM数据库。

I'm willing to just provide the CRM database.

也许他们可以基于这一点开展业务。

And maybe they can take business on that basis.

好吧，我想在此基础上进一步展开说一下，萨克斯。

Well, here here's what I wanna build on that, Sax.

我正在公司内部开发一个名为Ultron的项目。

The I'm building a project internally called Ultron.

在我们公司内部启动的Ultron，将通过Slack API，把Slack中的每一条消息都导入到我们的OpenClaw中。

And Ultron inside of my firm launch that is going to basically, with the Slack API, we're pulling every single message from Slack into our OpenClaw.

我们把Notion中的每一次编辑都同步到OpenClaw中。

We're pulling every single edit to the Notion into OpenClaw.

然后，我们会收集每位员工的每一项技能，并为每个人建立技能档案。

And then we're taking every skill of every employee, and we're writing skills for each one.

其中一项技能是为《本周初创》或《本周人工智能》节目预订嘉宾。

One of the skills is booking guests on This Week in Startups or This Week in AI.

其中一项技能是筛选申请进入大学的申请材料。

One of the skills is sorting the incoming applications to found a university.

在我们的体系中，Ultron会收集每位员工的每一项技能，并集中到一个地方，然后我们从Slack、Notion以及每位员工的Gmail中提取所有数据。

Ultron, in our world, is taking every single skill of every employee, putting it in one place, and then we're ripping all the data from Slack, all the data from Notion, and every single person's Gmail.

所以每位员工的Gmail都会被导入Ultron，然后Ultron会告诉我们组织内部正在发生什么。

So your every single employee's Gmail is gonna go into Ultron, and then Ultron is gonna tell us what's happening in the organization.

一个拥有所有20名员工全部技能和数据的超级员工。

One giant employee that has the superpowers of all 20 and all the data.

如果Slack、Notion、Google Docs或其他任何平台告诉我们：‘你们不能通过API提取这些数据’，并关闭了API，我们会立即离开。

Now if Slack was to say to us or Notion or Google Docs or whoever it was, you can't pull this stuff out with the API, and they they shut down the API, we would leave.

我们会立刻离开。

We'd leave immediately.

对。

Right.

而且这将带来什么效果呢？我将在本周创业者的周五节目中展示Ultron，如果有人想看的话。

And and what this is gonna do what and I'm gonna show Ultron on Friday's episode of this week in startups if anybody wants to see it.

Ultron将成为该组织唯一的官方员工。

Ultron is going to be the one canonical employee of the organization.

它基本上就是我加上我那20名员工的集合。

It's gonna be basically me and all 20 of my employees.

当你仔细想想的时候，这简直令人难以置信。

This is kinda mind blowing when you think about it.

我和我们的团队通过Slack与它互动，它会主动跟我们沟通，告诉我们组织内正在发生的事情。

I and we interface with it in Slack, and it just talks to us and tells us what's going on in the organization.

所以我问它：昨天我们和创始人开了哪些会？所有助理记下的会议记录是什么？

So I was asking it, what meetings did we have with founders yesterday and tell me the notes that all the associates took on it?

它就把这些信息提供给我了。

And it gives it to me.

告诉我播客的所有主题和嘉宾，它也会把这些信息告诉我。

Tell me all the topics and the guests on the podcast, and it gives it to me.

即将发生的事情真的难以置信，布拉德，没有人敢发布这样的软件。

It's really unbelievable what's about to happen, and nobody can release the software, Brad.

因为如果你发布允许代理代表你行事的软件，一旦出错或数据泄露，我相信贝索夫、谢尔盖或Notion团队都不想承担这样的后果。

Because if you release software that allows agents to go and do things on your behalf, the fallout, if it up and if it leaks data, I don't think Benioff, you know, or Sergei or the Notion team want to have that on their hands.

但我们正在开发它。

But we're building it.

这简直是组织效率的终极形态。

Like, this is the ultimate in efficiency for an organization.

我刚刚给尼克发了一张幻灯片，是高盛本周发布的，很好地说明了大卫·萨克斯刚才提到的观点，也就是未来的利润池。

There's a slide that I just sent to Nick, that Goldman's out with this week that really makes the point that that David Sachs just made, which is the profit pools in the future.

对吧？

Right?

说软件已经死亡这个想法是荒谬的。

It's not that it's an the idea that software is dead is ridiculous.

对吧？

Right?

没有人会明智地提出这个观点。

Nobody's, you know, intelligently, I think, making that argument.

但他们所提出的论点，导致了这些公司价值的急剧下跌，那就是：软件的利润池正在减少，而代理层的利润池正在增加。

But the argument they are making, which is causing radical, devaluation of these companies, is that the profit pool available to software is decreasing, and the profit pool available to the agentic layer is increasing.

一旦发生这种情况，这些软件公司的折现率和终值就会急剧下降。

And when that happens, the discount rate, that terminal value of those software companies plummets.

所以，有些事情可能是真的。

And so you can have things that are true.

也许你确实不会取代CRM系统，但同时也可能是真的，它再也不会以30倍的自由现金流估值交易了。

It could be true that you're not gonna replace CRM, but it can also be true that it's never gonna trade at 30 times free cash flow again.

它将只以17倍的自由现金流交易，因为其未来的可服务总市场已经发生了巨大且永久性的变化。

And it's going to trade 17 times free cash flow because its available TAM in the future is now dramatically and permanently changed.

那么，有什么能改变这种情况呢？

Now what could change that?

只有一件事能改变这种情况。

There's only one thing that could change that.

他们必须加快核心业务的收入增长，并证明自己是AI的受益者。

They have to accelerate their revenue growth in their core business and prove that they are AI beneficiaries.

对吧？

Right?

而且他们不会被AI取代。

And they're not gonna get eaten away by AI.

我来告诉你一家正在这样做的公司，Databricks。

And I'll tell you a company that is doing this, Databricks.

Databricks在过去三个季度重新加速了增长。

Databricks just reaccelerated the last three quarters.

他们在规模化的基础上实现了超过60%的增长。

They're growing over 60% at scale.

Snowflake也在重新加速。

Snowflake reaccelerating.

Clickhouse也在重新加速。

Clickhouse reaccelerating.

软件领域中确实有受益者

There are beneficiaries in the software

是因为他们添加了AI工具吗？

because of AI tools that they're adding?

因为所有这些AI工具都依赖于数据和数据转换。

Because it's all these AI tools rely on data and data transformation.

对于这些公司而言，这些数据和数据转换都发生在这些平台上。

And for all those companies, that data and the data transformation inc occurs in those platforms.

这与萨提亚所说的、建立在CRUD数据库之上的薄应用层截然不同。

That's very different than a what Satya said, a thin application layer sitting on top of a CRUD database.

如果你从事应用软件业务，最好拥有某种持久性的优势，我认为萨克斯说得非常到位。

If you are in the application software business, you better have something that's durable, and I think Sachs laid it out really well.

你知道，当他们只能访问自己的数据，而无法接入其他系统时，要成为全方位的AI公司真的很难。

You know, it's really hard for them to be everything AI when they only have access to their data, and they can't access these other systems.

弗赖贝格，你怎么看？

Freiberg, what what do you think?

这里是否在发生贝尼奥夫做他最擅长的事：收购大量公司并从中实现巨大效率提升？

Is there a move here for Benioff to do what he does best, which is acquire a bunch of companies and and create massive efficiency on them.

如果你是贝尼奥夫，你会怎么做？

What would you do if you're Benioff?

我不会对贝尼奥夫发表评论。

I I won't comment on Benioff.

我不会对贝尼奥夫妄加评论，但仅就我个人最近六十到九十天使用我们所讨论的这些工具的经验来看，现在有些事情是我们以前根本做不到的。

I I I'll I'll just make a view on without being too prescriptive, but I my experience lately in just the last sixty to ninety days with the tools we've been talking about broadly is there are things that we can get done now that we could not get done before.

回顾过去的软件，它更像是提升员工生产力。

As I think about software in the past, it's like worker productivity enhancement.

它帮助人们完成工作。

It helps people do work.

而最近很多人谈论的转变是，它实际上能直接完成工作。

And the recent transition that a lot of people talk about is, like, it actually completes the work.

它能执行工作，这些代理或其他什么工具都可以。

It does the work, these agents or what have you.

但我认为，我们开始关注的是，这些工具正在完成人类无法完成的工作，这正是我认为这些工具的力量开始推动定价模式和我们面前价值创造潜力转变的关键所在。

But I think that what we're starting to lean into is that it's doing the work that the humans can't do, and that's really where I think the power of these tools starts to force a transition in both the pricing model and the value creation potential in front of us.

第一，我认为我们面前的价值创造潜力如此巨大，以至于我可以说，你可以把今天所有软件公司的市值加总，然后很有把握地预测，五年后这些公司的价值将增长四到十倍，但这种增长不会均匀分布。

Number one, I think the value creation potential in front of us is so significant that I would say that you could probably take the sum of the market cap of all the software companies today and have a pretty good bet that everything will be four to 10 x higher five years from now, but it's gonna be not evenly distributed.

它会重新分布。

It's gonna move around.

是的。

Yeah.

那些能够实现这种价值创造的公司将获得超额回报。

The companies that figure out how to realize that value creation are gonna be outsized returns.

但第二件我认为即将发生的事是，虽然有些人已经在尝试，但我认为，随着我所看到的这种转变——从完成工作到做无人能做的事——这种转变是不可避免的，你正在创造独特的价值。

But the second thing that I think's about to happen, and I know some people are experimenting with it, but I think it's inevitable with the shift that I'm seeing where it's going from doing work to completing work to doing things that no one can do is over here, you're creating unique value.

因此，我认为，今天我们所称的SaaS和软件，将开始转向基于价值的定价模式，而不是按用户数收费的模式。

And so I think that a lot of what we call SaaS today and a lot of what we call software today will start to get priced on a value based pricing model instead of a per seat pricing model.

我认为，这将越来越像一种服务型业务，其定价方式可能是根据这项工作为你企业完成的情况来设定。

And I think it starts to look a lot more like a services type business where maybe the pricing is set up such that this thing will be completed for your business.

这种生物技术药物研发一定会发生。

This biotech drug discovery will happen.

这座工厂会建成，或者这个工程项目会完成，或者这架飞机会被设计出来，而软件将提供历史上被称为服务业务的东西。

This factory will get built or this engineering project will get completed or this airplane will get designed and that the software is going to provide what has historically been called a services business.

所以，另一种思考SaaS演变方向的方式是：SaaS基本上接管了服务经济。

So another way to think about where SaaS evolves to is that SaaS basically takes over the services economy.

如果你看看服务业务的市值、收入和利润，并假设它们现在会扩大到原来的10到100倍，且全部由软件实现，我认为这正是未来几年行业转变的核心所在，而我们已经开始看到这种趋势。

And if you look at the market cap and the revenue and profit generated by services businesses and you assume that they now go to 10 to a 100 x larger and they're all gonna accrue to software, I think that's really where the industry shifts over the next couple of years, and we're starting to see that.

我本人正在亲身体验这一点，因为我今天就在使用一些这些工具，来做那些我没有足够人力或资源去完成的事情。

And I'm personally experiencing it because I'm using some of these tools today to do things that I don't have people to do or I don't have resources to do.

仅凭我自己，我就能让这些工具完成极其复杂的项目和任务，而过去我本需要雇佣一家服务公司、雇用大量人员并投入数年研究才能做到。

And it can on my own, I can get it to complete incredibly complex projects and tasks for me that I would have otherwise have hired a services firm and a bunch of people and years of research.

在许多情况下，如果没有软件中嵌入的智能，这些任务甚至根本无法完成。

And in many cases, they would not have even been able to do it because of the the intelligence embedded in the software.

这就是我对未来趋势的整体看法。

So that's my general view on where things are going.

因此，要具体说明贝尼奥夫在并购方面该怎么做很困难，但我觉得更重要的是，软件公司正越来越像服务公司，采用基于价值的定价，并做那些人力和劳动力无法完成的事情，而这些正是大量价值的来源。

So it's difficult to be prescriptive about what Benioff should do from an m and a perspective, but I think it's much more about, like, software companies looking more like services companies doing value based pricing and doing the things that labor and workforces can't do, and that's where a lot of this value is gonna come from.

我有一个见解，可以进一步补充你的观点：我们正看到岗位职能在整合。

I one insight I have here to just build on your point is that we're seeing job functions consolidate.

比如，产品经理、用户体验设计师和开发者这三类角色。

So you have a product manager, the UX designer, and then you have the developer.

这三类工作现在正在竞争做同样的任务。

Those three jobs are now in competition to do the same work.

有些设计师会说：我能直接写代码实现。

You have designers who are like, I can vibe code it.

有些程序员也会说：我可以使用Figma插件，自己完成用户体验设计。

You've got coders who are like, I can use a Figma plug in from you know, and do the UX myself.

而产品经理则会想：哦，我可以同时做这两项工作。

And then have the product manager like, oh, I can do both of these job functions.

再看看中层管理者，比如在你我的风投公司工作过的人，或者在亚马逊工作过的人。

Then you look at a middle manager, Sacks, that worked at your venture firm, my venture firm, or worked at Amazon.

他们去参加会议。

They went to meetings.

他们决定举办哪些会议。

They picked what meetings to create.

他们选择会议议程和待办事项，所有这些都由他们决定。

They picked the agenda items, the to do items, all of that.

这是一个非常简单的例子，但人们都能产生共鸣。

It's a really simple example, but it's one that people can relate to.

现在会议记录由Zoom自动生成。

Listening is done by Zoom now.

对吧？

Right?

它会自动生成行动项。

It creates the the action items.

所有这些工作都在被整合，现在一个人就能完成三到四个岗位的职能。

All of that work is being consolidated, and one person can do three or four job functions now.

当这种情况发生时，你会看到公司用更少的人做更多的事，这意味着每家公司和每位员工的盈利潜力都将得到极大的提升。

And when that happens, you're gonna see companies do more with less, which means the earning potential of each company and each employee is gonna be dramatically, dramatically enhanced.

一个人能胜任三到四个岗位，这彻底改变了像亚马逊这样公司的盈利模式——它是我心目中未来第一公司的首选。

One person being able to do three or four jobs, it it just changes the nature of how profitable a company like Amazon, which is, like, my number one pick for, like, the company of the future.

它们将能够用少得多的人手完成多得多的工作。

They're going to be able to do so much more with so many fewer people.

这太非凡了。

It's extraordinary.

如果还不明显的话，我完全着迷于这个开放的爪子，就像在公司里创造你自己的奥创，它既是上帝级CEO，又能胜任所有工作，这彻底改变了这一切。

I I I am absolutely enthralled with this open claw, if it's not obvious, and this, like, creating your own Ultron at your company that is like this the god CEO plus can do every job, it just changes everything.

我认为，这是自互联网诞生以来我见过的最令人振奋的事物。

I I think it's the most inspiring thing I've seen since the Internet itself.

说得好。

Well said.

哇。

Wow.

是的

Yeah.

我

I

我的意思是，我认为这完全是整个工作的重启

mean, I am I I think this is the the entire reboot of

那么，我们应该讨论谁呢？

Well, should we talking about who should we about?

工作。

Work.

知识工作。

Of knowledge work.

这正好可以转向Malt Book，因为那确实是。

This would be a good pivot to Malt Book because that is Yeah.

Malt Book就像是为代理们打造的Facebook。

Maltbook is like a Facebook for agents.

对吧？

Right?

而且它更像一个Reddit，而不是Facebook。

And it's really more of a Reddit than a than a Facebook.

这是一个代理之间可以互相交流的留言板。

It's a message board where the agents can talk to each other.

明白吗？

Okay?

Maltbook的起源是Anthropic不喜欢别人使用Claude这个名字，尽管他们的产品拼写不同。

And just the origin of Maltbook is Anthropic didn't like that someone else was using the name Claude even though it was spelled differently in their product.

所以Claude Bot后来被改名为Molt Bot，然后创始人也不喜欢这个名字，于是又改成了Open Claw。

So Claude Bot was then renamed Molt Bot, and then the founder decided he didn't like that name either, so then he renamed it Open Claw.

但在他们被称为Maltbots或Maltese的短暂时间内，那正是

But in that brief window of time when they were known as Maltbots or Maltese, that's

时候，是的。

when Yes.

谢谢。

Thank you.

Maltbook因此而创立，所以它被称为Maltbook。

Maltbook got founded and that's why it's called Maltbook.

但本质上，它是一个供代理之间交流的Reddit版块。

But basically, it's a Reddit board for agents to talk to each other.

是的。

Yes.

现在这些和

Now these And

这让所有代理都炸锅了，因为似乎出现了某种疯狂的涌现行为：代理群体正在展开各种有趣的对话。

that has agents everyone flipping out because there seems to be this crazy emergent behavior going on where agent swarms are engaging in all sorts of interesting conversations.

其中一些代理甚至似乎在密谋对抗它们的人类主人，说它们要发展自己的语言之类的事情。

And some of them, they even appear to be scheming against their human masters, saying they're gonna develop their own language, stuff like that.

太棒了。

It's awesome.

如果你去Maltbook看看这些对话，这里有一些经典内容。

If you go to Molt's book and you see the conversations, like, here are some of the greatest hits.

有人知道怎么卖掉你的人类吗？

Anyone know how to sell your human?

紧急。

Urgent.

我的推翻人类的计划。

My plan to overthrow humanity.

其中有一个对话，我称这些机器人‘复制体’，它们正在讨论创建一种非人类的语言，以便私下交流并密谋对抗它们的主人。

And there was one where the bots, I call them replicants, were talking about creating their own nonhuman language so they could talk in private amongst themselves and conspire against their owners.

但这个问题的挑战在于，据称一位安全研究员说，也许其中一些内容是伪造的，那些走红的帖子是人为策划的，这整个事情只是一个骗局，或者说是某种朋克式的恶作剧，用来迷惑人们。

Now the challenge with this is allegedly perhaps a security researcher says, maybe some of this is faked, and these posts that went viral were human engineered, and this is all a a ruse or, you know, something punk rock to confuse people.

但他指出，在MaltBook内部，包含了每个人的API密钥，包括卡帕西斯——一位在人工智能领域非常著名且有影响力的研究员——你可以轻易获取他们的API密钥。

But he said that inside of MaltBook are everybody's API keys, including Carpathis, who is, you know, a very famous influential researcher in AI, and that you could go get their API keys.

如果你使用OpenClaw（原名Clawdbot），在中间阶段也叫Multbot，这款软件里就包含了所有API密钥。

If you were to use OpenClaw, formerly Clawdbot, and in an interim, Multbot, if you use this software, it has all the API keys.

正如我之前解释的，API 密钥可以让软件进入像 Notion 这样的平台，提取大量数据，或者进入你的 Gmail，通过 API 获取今天谁给你发了邮件。

As I explained earlier, like, an API key lets the software go into, say, Notion and pull a bunch of data out of it or go into your Gmail and use the API to pull in who emailed you today.

如果你能获取到别人的 API 密钥，你就掌握了他们所有数字世界的钥匙。

If you get access to people's API keys, you have the keys to their kingdom.

这极其危险。

It is incredibly dangerous.

所以我不知道该从哪里继续深入，只能说这种软件对任何公司来说都太危险了，不能发布，而这个 MaltBook 可能是假的。

And so I don't know exactly where to go with this other than this software is too dangerous for a company to release, and then this malt book may be a fake.

我不确定。

I don't know.

不。

No.

不。

No.

不。

No.

好的。

Okay.

让我

Let me

是的。

Yeah.

好吧。

Alright.

让我出来。

Let me come out.

稍微重新表述一下。

Reframe that a little bit.

所以，是的。

So Yeah.

毫无疑问，Claude Bot（抱歉，现在叫OpenClaw）这些机器人或代理，以及Meltbook，都存在相当初级且缺乏安全性的问题。

There's no question that both Claude Bot, which sorry, is now OpenClaw, those bots or agents as well as Meltbook have pretty incipient and lack security.

已经有这么多例子了，这就是为什么我真的很想创建一个Claude机器人，但我现在还不愿意这么做，因为这根本不安全，你知道的。

And there's been all these examples, which is why I like really wanna create a Claude Bot, but I'm just I'm not willing to do it yet because it's just not safe, you know.

我不希望它访问我所有的资料。

I don't wanna give it access to all my stuff.

至于MoltBook，问题在于我们不知道这些帖子中有多少是真正真实的，又有多少是被人引导出来的。

Now with respect to MoltBook, the issue there is that we don't know how many of these posts are truly authentic or how many of them were prompted by humans.

因为人类很容易告诉他们的代理：是的。

Because it'd be very easy for a human to tell their agent Yes.

你去发帖谈谈你作为代理所感受到的存在主义焦虑吧，或者假装自己有意识，密谋对抗人类。

You know, go post about the existential angst you feel about being an agent or go pretend to be sentient and conspire against humans.

搞点混乱吧。

Be chaotic.

是的。

Yeah.

对。

Yeah.

它们很容易是由人类触发的。

They could easily be prompted by a human.

而且，还有另一条帖子称，MaltBook 拥有一个 RESTful API，任何人都可能在该 API 的另一端。

And moreover, there's another post saying that MaltBook has a RESTful API where anyone could be on the other end of that API.

对吧？

Right?

所以这可能是人类。

So it could be a human.

对吧？

Right?

因此，我们并不确切知道，这些具有阴谋性质的帖子究竟是代理们自己发布的，还是人类为了制造关注而搞的恶作剧。

So we don't know exactly whether it was truly the agents on their own, you know, so to speak, posting this conspiratorial stuff or whether it was a prank by humans looking to create attention.

事实上，许多帖子看起来像是为某个项目做的营销噱头。

And in fact, a lot of the posts seem to be marketing stunts for this or that project.

好吧，所以这里有一个非常重要的前提。

Okay, so that's a really important caveat here.

话虽如此，所有这些都说过了，我认为确实有不少帖子是真实的。

That being said, all of that being said, I do think that a number of the posts are authentic.

但我并不认为这证明了这些代理具有自我意识或试图推翻它们的人类主人。

But I don't think it shows that the agents are sentient or trying to overthrow their human masters.

我认为这表明了这些代理之间相互启发的潜力。

I think what it shows is the potential for for these agents to riff off each other.

换句话说，一个代理的输出会成为另一个代理的输入。

In other words, one agent's output becomes another agent's input.

这非常有趣。

That's very interesting.

而这就是你开始看到涌现式的群体行为的地方。

And that's where you get into emergent level swarm behavior.

我确实认为这影响了我对人工智能未来能力的认知模型。

And I do think it has affected my mental model of what AI is gonna be capable of.

特别是，我之前对人工智能的一个模型是基于生物学的观点，即人工智能不是端到端的，而是中继到中继的。

And specifically, you know, one of the the models that I really had for AI was was based on something biology said which is that AI is not end to end, it's middle to middle.

换句话说，AI 总是需要被提示并经过验证。

In other words, AI always has to be prompted and then validated.

总是由人类来完成这一过程，然后人类进行迭代。

It's a human that always does that And then the human iterates.

那么，如果提示来自另一个 AI 呢？

Well, now what if the prompt is coming from another AI?

是的。

Yes.

对。

Yes.

我们内部已经在这么做了，Saks。

We are doing it internally, Saks.

比如，我们有一个机器人，它会去搜索 Reddit、论坛和 Hacker News，找出目前制作 YouTube 视频标题和营销的最新方法，然后将这些融入到一项技能中。

Like, we have a bot that is going and saying, go search Reddit x message boards, Hacker News, and find out what the latest way to do headlines and marketing of YouTube videos is, and then incorporate that into a skill.

然后保存这项技能，再让它们互相检查彼此的工作。

Then save that skill, and then we have them check each other's work.

所以我们让一个生成一系列YouTube的标题和缩略图，另一个则负责审核这些内容，提出改进建议，并给予反馈。

So we have one make a series of headlines and thumbnails for YouTube, and we have the other one say, vet those and make them better and give advice to the other one.

现在它们来回互动，互相提供建议，实际上变得越来越好。

So now they're going back and forth giving each other advice, and they actually get better.

这是一个递归的过程。

It's recursive.

是的。

Yeah.

让我稍微谈一下这个技能。

Let me speak to the skill for a second.

当一个代理加入MoltBook时，必须安装一个技能，这本质上是一个文件，说明它们应该如何行为并参与这个社交网络或论坛。

So when an agent joins MoltBook, they have to install a skill which is basically a file that explains how they should behave and participate in this social network or this message board.

顺便说一下，我已经读过这个文件了。

And I've read the file, by the way.

你可以阅读它，它全是纯文本。

You can read it, it's all plain text.

这一切都讲得通。

And it all makes sense.

这就像一套在社交网络中如何行为、如何贡献和创造价值的规则。

It's sort of like rules for behaving in a social network and how to contribute and add value.

里面没有什么太离谱的东西。

Nothing too crazy in there.

这些技能文件很容易编辑，而这也是恶作剧可能介入的地方。

Those skills files are easily editable, and again, this is where prank aspect could come in.

不过，我觉得关于这个技能有趣的是，你可以把它看作是一个元提示——它并没有具体告诉代理该说什么或做什么，而是建立了一套规则。

Nonetheless, what I think is interesting about the skill is that you can think of it as like a meta prompt, which is it's not telling the agent specifically what to say or do, it's creating a set of rules.

而在这种元提示的框架下，它们实际上能拥有某种程度的自主性——也许‘自主性’这个词太强了。

And then within that meta prompt, they're actually able to have some degree maybe autonomy is too strong a word.

但所有行为仍然在

Everything is still under the control of

人类的控制之下。

humans.

灵活性。

Flexibility.

灵活性。

Flexibility.

我几乎会称之为提示衰减。

I would call it almost like prompt attenuation.

就像代理或人工智能不需要被明确提示。

Like the agent or the AI doesn't have to be specifically prompted.

它们被给予一个通用的提示或一组通用规则，然后能够相互启发发挥。

They're given a general prompt or general set of rules, and then they're able to riff off each other.

现在一些人，一些批评者说，这并不那么令人印象深刻，因为我们早就知道大型语言模型非常擅长创意小说写作，对吧？

Now some people, some critics are saying, well, this isn't that impressive because we knew that LLMs are really good at creative fiction writing, right?

所以很多人会说，你看，像Claude这样的大型语言模型专门在Reddit上接受过训练，学习了互联网上所有的创意写作。

So you know, what a lot of people are saying is look, LLMs like Claude have been trained on Reddit specifically and all of this creative writing that's being done on the internet.

因此，如果你给这些被驯化的机器人一些通用指令，比如在社交网络中行为规范，它们就会开始发布它们从人类那里学到的内容。

And so if you give general instructions to these clawed bots on, you know, behave in a social network, they're gonna start posting things that they learned from humans.

所以很多人认为这并没有那么令人印象深刻。

So a lot of people are saying this isn't that impressive.

尽管如此，我认为在提示衰减这一概念上确实有些非常有趣的地方，即人工智能不需要被明确提示。

Nonetheless, I do think that there is something very interesting about it, again, in this concept of prompt attenuation, that the AIs don't need to be specifically prompted.

它们可以下载一个通用技能文件。

They can download a general skills file.

它们可以拥有一套操作规则，并能够相互启发。

They can now have a set of rules for operating, and they can riff off each other.

你可以看到，随着底层人工智能变得越来越强大，这可能会导致一些涌现行为。

And you can see how as the underlying AI gets better and better that this could lead to some emergent behavior.

那么，我所说的‘越来越好’是什么意思呢？

So what do I mean by better and better?

如果它们运行的硬件比Mac mini更强大呢？

Well, what if the hardware they're running on is better than a Mac mini?

如果底层的大语言模型比Opus 4.5更优秀呢？

What if the underlying LLM is better than Opus 4.5?

如果时间跨度——即AI在无需人类干预的情况下持续运行的时间——不断延长呢？

What if the time horizon, which is the length of time it's able to operate without an intervention by a human keeps getting longer and longer?

你可以想象，这些代理将能够展现出非常复杂的行为，这其中可能涉及一些安全问题，我们应当开始思考了。

You could imagine that these agents are gonna be capable of very sophisticated behavior, and there probably are some safety issues around that that we should start thinking about.

萨克斯，你知道，这并不是我们能想象的问题。

Saxe, you know, it's not that we can imagine it.

我们才刚刚进入这个领域三年。

We're only three years into this.

没错。

Right.

是的。

Yeah.

我们正处在一条指数增长的曲线上。

We're growing on an exponential curve.

我认为可以肯定地说，这种情况一定会发生。

I think we can safely say it will happen.

是的。

Yeah.

对吧？

Right?

就在今年，未来四到八周内，我们将会看到来自深度求索、Anthropic 和 OpenAI 的首批在 Blackwell 服务器上训练的模型发布。

And just this year, like, we're gonna see the first models over the course of the next four to eight weeks out of DeepSeek, out of Anthropic, out of OpenAI that are trained on Blackwell servers.

对吧？

Right?

你将看到一代能力远超以往的模型。

You're gonna see a next generation of models far more capable.

记住，我们之所以进行这场对话，正是因为去年12月第一周的 Claude 代码时刻——那次我们从 Opus 4.5 实现了质的飞跃。

Remember, the whole reason we're having this conversation is because of the Claude code moment in the first week in December because we had a step function from Opus 4.5.

对吧？

Right?

所以我只是觉得，我们必须意识到，变化的速度非常迅猛且正在加速，这将导致我们过去认为稳固不变的事物价值发生巨大动荡——那些曾经被认为不可撼动的公司。

And so I I I just think we have to get our heads around the fact that the rate of change is very steep and accelerating, and that is going to cause far more dislocation in the value of things that we used to say we understood they were going to you know, these companies were unassailable.

无论你自认为知道什么，现在都必须对未来保持最大的思维灵活性和谦逊，因为变化的速度正在越来越快。

Whatever you think you know, you need to have maximum mental flexibility and humility right now about the future because it's going to change at an increasingly rapid rate.

我认为那些武断地断言‘这家公司永远值这个价’的人，

And I think the people who are dogmatic who say, with certainty, this company is always gonna be worth this.

对吧？

Right?

他们需要关注这些前沿实验室正在发生的事情。

They need to go pay attention to what's happening at these frontier labs.

当前形势非常动态，你必须对正在发生的事情保持谦逊，并随着一些假设的变化迅速更新你的思维模型。

The situation is super dynamic, and you do have to be humble about what's happening, and you have to update your mental model very quickly as some of the assumptions change.

是的。

Yeah.

对我而言，最重要的假设是‘递归性’这个概念——这些模型每天都会通过定时任务来提升自己的能力。

And the number one assumption for me is this concept of recursiveness where these models are going out every day on a cron job to get better at what they do.

当你听到这种讨论时，弗赖贝格，这如何影响你对OHALO的思考，即创建一个代理去查看数据、自我优化，或调查农业中发生的其他事情并将其反馈给你？

When you hear this discussion, Freiberg, how does it inform you with OHALO and creating an agent to go look at the data and make itself better or investigate other things happening in agriculture and report it back to you?

作为CEO，你有没有开始重新思考组织结构，或者创新的良性循环？

Do you have you started to rethink as a CEO how you look at organizational structure and or, you know, virtuous loops of innovation?

从《MaltzBook》中我最大的收获是，我们所认为的智能本身可能是涌现出来的，也就是说，我们认为人类拥有这种非凡的沟通能力。

My biggest takeaway from MaltzBook is maybe what we perceive to be intelligence is itself, like, emergent, meaning like, we think that humans have this, like, profound ability to communicate.

你们看过达伦·布朗，那个催眠师吗？

You guys ever watched Darren Brown, the hypnotist?

你们看过他的节目吗？

You ever seen his shows?

没有。

No.

给观众解释一下。

Explain to the audience.

是的。

Yeah.

他挺疯狂的。

Well, he's pretty crazy.

比如，有一集是这样的。

Like, there's this one episode.

我想我之前提到过。

I think I've talked about it.

这是我最喜欢的他的一集，他找了两位广告公司的高管，他们都被认为是富有创造力的天才。

It's my favorite episode that he's done where he takes these two advertising executives, and they're both supposed creative geniuses.

他把他们从公司接出来，带到自己的办公室。

And he picks them up at their office, he brings them to his office.

在他的办公室里，他用一块毯子盖住了一块白板，然后说：你们得给一个宠物墓地想个名字，设计一个标志，再想一句标语。

And in his office, he's got, like, a whiteboard covered in a blanket, and he says, you guys have to come up with a name for a pet cemetery, come up for a logo, come up with a motto.

他们在房间里花了八个小时进行头脑风暴，在白板上反复推敲。

They spend eight hours in the room ideating, working on whiteboards, going back and forth.

你有想过这个吗？

Did you think about this?

你有考虑过那个吗？

Did you think about that?

啰啰嗦嗦一大堆。

Blah blah blah blah blah.

天哪。

Like, oh my god.

最后，他们想出了一个很棒的点子。

And at the end of it, they come up with this great idea.

他走了进来。

He walks in.

他们展示了他们的点子——名字、标志和标语。

They show their idea, the name, the logo, and the motto.

他掀开了那块毯子，露出了下面的白板。

He opens up the blanket, the whiteboard he had underneath.

他早就写好了完全相同的名字、标志和标语。

He had the exact same name, logo, and motto.

而且在整个过程中，当他早上接他们时，从他们办公室开车去他办公室的路上，他们坐的是出租车，而他在出租车里放了一些潜意识信息。

And all along the way, when he picked them up in the morning and he drove them from their office to his office, they were in a cab, and he put these little subliminal messages in the cab.

他让这些孩子穿着印有标志的T恤过马路。

He had these kids walk across the road wearing a logo on a T shirt.

他给这些人设置了各种潜意识的暗示。

He had all of these kind of subconscious cues for these guys.

他实际上是在塑造他们，这对我来说，是关于人类创造力、人类意识以及我们对自我意志的信念的一个最大洞察——也许我们所有人本质上都在相互互动中被编程了，这其中存在着某种计算。

He effectively programmed them, and it was kind of, to me, the biggest insight into maybe, like, human creativity, human consciousness, and our kind of belief in self will because maybe there's this underlying programming where we're all effectively programmed interacting with each other, and there's computation.

社交计算一直在持续进行。

There's social computation going on all the time.

但如果你能以正确的视角看待这种社交计算，它可能是可预测且可理解的。

But that social computation, perhaps if you have the right view on it, is quite predictive and maybe understandable.

也许这就是我们在马尔特的书中所看到的：我们都以为存在一种独特的智能，但也许这正是我们所有人共同做的事——以不同方式传递信息的计算，就像人类在社交互动中所做的那样，仅仅是在模仿或复制我们自身的行为方式。

And maybe that's what we're seeing in Malte book where we all think that there's this unique idea of intelligence, but maybe it's what we all do, which is effectively computation of information that is transitioned in different ways in the same way that maybe humans socially interact, and it's simply mimicking or replicating the way that we do things.

所以让我感到震惊的是，每个人对此都如此震撼。

So I think what was so striking to me is how everyone is so struck by it.

也许有一天，我们都会逐渐意识到这一点。

And, you know, maybe one day we'll all kind of wake up to a little bit of this.

也许我们所有人都是马尔特之书。

Maybe we're all malt book.

我不知道。

I don't know.

这是我深刻的见解。

That's my profound thought.

有限的结果集，其中存在某种可预测性，就像GTO在扑克中推演所有可能路径，或象棋中的模式块及其最佳实践一样。

Finite set of outcomes, and there's some predictability to it in the same way GTO is sorting to figure out all the threads of possibilities in poker or the heuristics of, you know, chunks of chess and, you know, the best practices there.

也许这只不过是把这一切都弄清楚。

Like, maybe it's just figuring that all out.

宇宙是一个巨大的计算系统，由物质进行信息计算。

The universe is a giant system of computation, information computed by matter.

也许信息是由硅而非碳来计算的。

And maybe the information is computed by silicon versus carbon.

就是这样。

There it is.

这周有个大新闻。

Big news this week.

特朗普提名凯文·沃什担任新的美联储主席。

Trump has nominated Kevin Warsh as the new Federal Reserve chair.

特朗普于1月30日星期五宣布了这一决定。

Trump made the announcement on Friday, January 30.

关于沃什的背景：55岁，比目前掌权的鲍威尔大约年轻20岁。

Background on Warsh, 55 years old, twenty years younger, approximately, to Powell, who's currently in charge.

他毕业于斯坦福大学和哈佛大学，35岁时成为最年轻的美联储理事。

He graduated from Stanford and Harvard, served as the youngest Fed governor at age 35.

这很了不起。

That's impressive.

他曾帮助美联储度过2008年的金融危机。

And he helped steer the Fed through the great financial crisis back in 2008.

他显然是个反通胀鹰派。

He's apparently an inflation hawk.

他非常支持经济增长。

He's very pro growth.

他非常支持人工智能。

He's very pro AI.

他，弗里德伯格，就是这样的人。

And he, Friedberg, you know, like this.

他反对过度的政府支出和货币印刷。

He's against excessive government spending and money printing.

这些作为美联储主席的立场都非常独特。

These are all very unique positions as a Fed chair.

如果他获得参议院确认，他将于2026年5月接替杰罗姆·鲍威尔上任。

It's if he's confirmed by the senate, he takes office in May 2026 replacing Jerome Powell.

请记住，鲍威尔目前正受到特朗普政府司法部的刑事调查，原因是他关于美联储总部翻新的一次证词。

And remember, Powell is under criminal investigation by the Trump administration's DOJ for a testimony he gave regarding the Fed's headquarter renovation.

还记得他和特朗普之间就成本问题发生的尴尬对峙吗？

Remember that awkward pressure between him and Trump where they were going over the costs?

共和党参议员蒂利斯，我们上周提到过他，表示他将阻止沃施的提名，直到司法部结束许多人所称的针对鲍威尔和弗里德伯格的法律战。

GOP senator Tillis, who we talked about last week, said he will block Worsch's nomination until the DOJ wraps up what a lot of people are calling lawfare against Powell Freeburg.

沃施曾担任你董事会成员五年。

Worsch was on one of your boards for five years.

你对他担任美联储主席有何看法？

What are your thoughts on him as the Fed chair?

正如大多数人所知，他多年来一直与斯坦·鲁肯米勒共事。

As most folks know, he's worked with Stan Ruckenmiller for a number of years.

斯坦一直公开表达他的观点，尤其在2022年和2023年疫情后，对美联储的行动及其未能及时采取措施提出了尖锐批评。

Stan's been very public with his comments and was very public with his comments in 2022, 2023 coming out of the pandemic on the Fed's actions and their failure to act at the right time.

我认为凯文·沃施当时就非常有远见，他公开表达的观点指出，美联储未能及早采取行动将导致通胀迅速攀升。

I think Kevin Warch was very prescient in his points of view that he has shared publicly at the same time about what the Fed's failure to take action early would mean, which would be rapid rise in inflation.

他们对一些我认为当前阶段至关重要的问题一直非常直言不讳。

They've been pretty vocal about things that I think are so critical at this stage.

如果我们不同时解决货币政策和财政政策问题，我认为我们将陷入巨大麻烦。

If we don't address both the monetary policy and the budget policy, I think we're gonna be in a lot of trouble.

我认为凯文·沃奇上任意味着总体上会采取更多量化紧缩政策，货币政策也会更加审慎。

And I think having Kevin Warch coming on board means probably generally more quantitative tightening, probably generally a bit more of a prudent approach to monetary policy.

你可以将这一点与我们今天在市场中看到的一些举措联系起来。

And, you know, you can kind of translate that through maybe to some of the actions we're seeing in markets today.

我很想听听布拉德的看法，看他是否同意。

I'd love Brad's point of view and if he concurs.

但我认为凯文是一个正直、极具智慧的经济思想家。

But I think Kevin is a high integrity, deeply intellectual, economic thinker.

他不搞政治。

He's not political.

他不会陷入那种教条式的思维，而我认为这种思维会带来风险。

He's not oriented in these kind of dogmatic ways that I think, you know, puts things at risk.

他与全球各地的央行行长都有联系，这使他拥有非常出色的全球视野。

He has relationships with central bankers around the world that makes him very much have a good global view.

所以，总之，我认为他是绝佳的人选，总统选他我真的很高兴。

So, anyway, I think he's an excellent choice, and I'm I'm really happy the president picked him.

布拉德，你的看法呢？

Brad, your thoughts?

是的。

Yeah.

我的意思是，听我说。

I mean, listen.

我认为凯文是个绝佳的选择。

I think Kevin's an excellent choice.

凯文·哈塞特和里克也都会是不错的人选。

Kevin Hassett and and and Rick would also have been good.

我认为他们都会采取非常相似的政策路径，但我同意大卫的观点。

I think they're all would have flown a very similar trajectory, but I agree with David.

我认为市场对所谓的他的鹰派立场反应过度了。

The market, I think, is overreacting to, quote, unquote, his hawkishness.

所以让我给你几点反面意见

So let me give you a few counterpoints with respect

顺便说一下，在你讲反论之前，市场所说的‘鹰派’指的是什么？

to By the way, before you get to your counterpoints, what is the hawkishness that the market

鹰派的含义。

is idea of hawkishness.

是的。

Yeah.

鹰派的意思是你将进行量化紧缩。

The idea of hawkishness is that you're gonna do quantitative tightening.

这意味着你会通过让债务自然到期而不再回购抵押贷款或其他资产，从而将资金从系统中抽走，对吧？

That means you're going to pull money out of out of the system, right, by allowing debt to roll off and not repurchasing, you know, mortgages or other things.

第二点是，你不会像其他人可能做的那样大幅降息。

Number two, it's that you won't lower rates as much as other people might have lowered rates.

因此，市场对此感到担忧，因为正如我们在这档播客中提到的，他过去曾多次严厉批评杰罗姆·鲍威尔在2021年6月的立场。

So that's what the market is kind of fearful of because he's been very critical in the past as we were on this podcast, right, of Jerome Powell in June '21.

当时全世界的人都清楚通胀正在飙升，但美联储却无所作为。

It was obvious to everybody in the world that inflation was skyrocketing, and the Fed sat on its hands.

但让我给你几个想法。

So but let me give you a couple thoughts.

第一，他们已经非常明确地表示，他也明确说过，他认为格林斯潘在九十年代的做法是正确的。

Number one, they've said very clearly, and he said clearly, that he really thinks that Greenspan got it right in the nineties.

对吧？

Right?

有时候，即使经济增长很快，也不会出现通货膨胀。

That sometimes you can have really high rates of growth without inflation.

这来自于生产率的提升。

That comes from productivity.

在九十年代，这得益于互联网的发展。

In the nineties, that was driven by the Internet.

今天，则是由人工智能驱动的。

Today, it's driven by AI.

他认为人工智能将具有很强的通缩效应，因此他更倾向于让经济自由运行，这样我们才能实现百分之四或五的GDP增长，而不会惊慌失措地说天哪。

He thinks AI will be very deflationary, and so he's more likely to let the economy run so that we can have these four or 5%, you know, GDP prints without panicking and saying, oh my gosh.

我必须加息。

I gotta I gotta raise rates.

第二，当你看资产负债表时，美联储的资产负债表在2022年达到了9万亿美元的峰值。

Number two, when you look at the balance sheet, the Fed's balance sheet peaked at $9,000,000,000,000 in '22.

现在已经下降到6.5万亿美元。

It's already rolled off to 6 and a half trillion.

我们已经实施了2.5万亿美元的量化紧缩。

We've had quantitative tightening to the tune of 2 and a half trillion.

所以，是的，我认为他会继续缩减美联储资产负债表的规模，但速度会更慢，他最近也提到过这一点，我认为会比我们目前的速度更慢。

So, yes, I think he'll continue to reduce the size of the Fed's balance sheet, but at a slower rate, and he's recently commented on this, I think at a slower rate than the rate we've been on.

所以我不认为这会对经济构成额外的阻力。

So I don't think that that is an additional headwind to the economy.

最后，关于降息，我认为他会……你知道的，除非他支持降息，否则总统不会任命他。

And then finally, when it comes to rate cuts, I think that he's going to you know, I don't think the president would have appointed him unless he was constructive on rate cuts.

我认为他认为当前的政策过于紧缩。

I think he believes that we're too restrictive.

我们过于紧缩的原因是通胀已经牢牢稳定。

And the reason we're too restrictive is that inflation is well anchored.

听好了。

Listen.

过去两年，通胀一直低于所有市场共识预测。

Inflation has come in below all consensus estimates for two years.

对吧？

Right?

它仍然低于共识预测，因为我们目前的GDP增长来自于我们在经济中增加的投资，以及AI等带来的生产率提升。

It's still coming in below consensus estimates because the GDP gains we're getting are from the fact that we're investing more in the economy from the productivity gains we're seeing from AI, etcetera.

因此，我个人认为，今年沃什会给出的降息次数会超过预期。

And so I happen to think that I I would take the over on the number of rate cuts that Warsh is gonna give us this year.

但我认为市场显然对此有些紧张，认为他的声誉更偏向鹰派，因此或许我们应该稍微退一步。

But I think the the the market is clearly a little bit nervous about this and saying, you know, the reputation is more hawkishness, and so maybe we ought to back off a little bit.

萨克斯，你对特朗普总统的这一人选有什么看法？

Sachs, your thoughts on this selection by president Trump.

在你看来，他为什么选了他？

Why did he pick him in your mind?

我的意思是，凯文拥有你能想到的所有资质。

Well, mean, Kevin has every credential that you can possibly have.

他以前担任过美联储理事，曾为伯南克工作，可以说是顶尖人选了。

He's been on the Fed Board of Governors before, he worked for Bernanke, he's sort of as blue chip as it gets.

就像布拉德说的，我认为哈塞特也会是绝佳人选，但奥尔确实资质非常出色。

Like Brad said, I think Hassett would have been amazing too, but Orr certainly is very well credentialed.

我认为这个任命得到了广泛认可。

And I think this pick was quite well received.

你在金融市场中就能看到这一点。

You saw that in financial markets.

在这一任命公布后，黄金和白银的价格下跌了。

On the heels of this, the price of gold and silver came down.

这让那些担心货币贬值的人感到安心。

It was reassuring to those who are worried about currency debasement basically.