Mad Money w/ Jim Cramer 9/16/26

我有Salesforce。

I have Salesforce.

它是安全的，但我还是得进去确保它是安全的。

It's secure, but I still have come in and make sure it's secure.

但他们攻击了那个国家，攻击了一个没有防御的国家。

But they they attacked the country they attacked the country with no defenses.

一个正常的国家是有防御的。

A typical country does have defenses.

但让我问你一个问题。

But let me let me ask you.

如果他们不让你参与，我的意思是，这让我非常困扰。

If they don't bring you in, I mean, it bothered me tremendously.

或者不是。

Or no.

让我听我把话说完。

Let me you hear me out.

我可不是在说空话。

I'm not blowing smoke with this.

看了这些人写的各种评论，我就纳闷，他们怎么不说呢？

If all the different commentary I read from these people, why don't they say, you know what?

我们跟尼凯什·阿罗拉坐下来聊过。

We sat down with Nikesh Arora.

我们也跟我们的朋友乔治坐下来谈过。

We sat down with our friend, George.

好吗？

Okay?

然后我们意识到，你知道的，怎么样？

And we realized, you know what?

他们的流程是，我们将能够在护栏内正确地做到这一点。

Their process is, and we're gonna be able to do it right with guardrails.

相反，他们只是说，天哪。

Instead, they just say, oh my god.

我们都不知道自己到底在干嘛。

We don't know what we are.

你看，到处都有人出事，结果大家还在那儿说，去查查州检察官那边吧。

Like, people are dying everywhere, and it's gonna be like, go check your trust in the state's lawyer.

我心里就想，拜托，成熟点行不行。

And I'm, like, thinking, come on, grow up.

事情确实发生了。

It's happening.

你看，事情正在发生。

Look, it's happening.

我们正跟AI实验室生态里的所有人沟通。

We are talking to all the people in the AI lab ecosystem.

我们正在跟他们合作。

We are working with them.

我们正在他们的产品中嵌入接口，以构建安全防护。

We are building hooks into their products to create security.

这里有个时间差。

There's a lag.

研究和落地执行之间存在滞后。

There's a lag between research and execution.

但企业实际是在执行层面运作的。

But enterprises live here in execution land.

他们并不在研究层面。

They don't live in research land.

所以是的，研究结果告诉我们，天哪，我们往往不够重视，如果不好好监控这些智能体，可能会出乱子。

So yes, research is telling you that, holy, we don't pay attention and we're not going to keep track of these agents, things can happen.

确实如此。

That's true.

完全有可能。

It can.

作为网络安全公司或其他公司的负责任领导者，我们的职责就是确保以安全、可靠的方式部署人工智能。

It's our job as responsible leaders of cybersecurity companies or responsible leaders of companies, period, to make sure that we deploy AI in a way that is safe and secure.

而这正是

And that's going

即将发生的。

to happen.

这也是你收购CyberArk的原因。

And that's why you bought CyberArk.

没错。

Yes.

你收购 CyberArk，是因为你清楚，没有身份认证根本行不通。

You bought CyberArk because you know, you can't Identity.

没有身份验证，很多事都干不了。

You can't do stuff without identity.

你能识别出它们，也能把它们干掉。

You can identify these, you can kill them.

也就是说，你得知道它们是谁，以及它们打算干什么。

Well, it means that you know who they are and what they're gonna do.

你知道他们是谁。

You know who they are.

但是，有一件事让我很担心，我觉得用“害怕”这个词不太准确。

But but the one thing I am scared about, and I think scared is the wrong word.

我是感到担忧。

I'm concerned.

我认为，如果落入朝鲜或伊朗那些聪明人的手里，可能会造成一些破坏。

I do I think in the hands of very smart people in North Korea or Iran, there's some damage that could be done.

这时候，你得反应快，得琢磨最坏的情况可能是什么。

And there, you really have to be thinking on your feet about what is the most nefarious thing that could have.

你看。

Look.

任何技术都有两面性。

Every technology has both sides.

它既有好的一面，也有坏的一面。

It has a dark side and the good side.

好消息是，我们使用技术的大部分用途都是善意的。

And the good news is a majority of the uses that we use technology for are for good.

总有人可能利用这项技术来作恶，这也是你刚才提到的那个神话时刻所表达的意思。

There is always the opportunity or the chance that somebody's gonna take the technology and use it adversely, and that is what if you you saw the mythos moment as well, that was what it was about.

我们担心的并不是好人会滥用这种技术。

The fear is not that we will, as good actors, use mythos.

我们担心的是坏人会利用它，开始用这类能力去攻击企业，甚至侵入基础设施。

The fear is bad actors will take mythos and start using capabilities like that and start attacking companies and be able to enter enter the infrastructure.

所以我们都清楚这一点。

So we are all aware of it.

我们都在构建能力来阻断这种能力，遏制类似中间人的行为，并为AI越来越强大的时代做好准备。

We're all building capability to block that capability, block the mid source like behavior, and get ready for a time when AI is getting more and more powerful.

所以我认为当务之急是每家公司都要认真审视自身，意识到我们正在进入一个全新的世界。

So I just think the impetus right now is that every company needs to look hard and say, we're moving into a new world.

在这个新世界里，你必须具备实时进行网络防御的能力，因为恶意攻击者将能够以实时速度对你发起攻击。

The new world is where you have to be able to do cyber defense at real time speed because the bad actors are gonna be able to attack you at real time.

我觉得，没有哪家公司里有人真正懂行。

And there's no one who knows enough at their own companies, I believe.

可能有一两个人会做这些，但这其实是要引入一家懂得如何运作的大公司。

It might have one or two people who do it this, but this is about a process of bringing in a big company who knows

知道该怎么做。

what to do.

吉姆，过去八年里，我一直在努力说服大家，

Jim, for eight years, I was trying to convince the world that

你那边有九个人。

nine in your stuff.

嗯，我加入帕洛阿尔托网络公司已经八年了。

Well, it's been eight years since I joined Palo Alto.

现在快第九年了。

It's coming on the year ninth year now.

但这九年里，我一直在努力让大家重视网络安全。

But in nine years, I've been trying to get people to pay attention to cybersecurity.

你知道吗？

And you know what?

我们的朋友达里奥就靠一件事做到了。

Our friend Dario did it with one thing.

他说他

He said he's

是个更好的营销人员

a much better marketer

比我强。

than I am.

你知道，这挺有意思的。

You know, was funny.

如果我还回到华尔街，我会说，提升业绩，Palo Alto Networks。

If I were back on Wall Street, I would have said raising numbers, Palo Alto Networks.

那是Palo Alto的财务主管、董事长兼CEO。

That's the cashier, chairman CEO of Palo Alto.

咱们来聊聊值得信任的标的，天哪。

Let's talk that we own for the trust, and holy cow.

这太棒了。

This has been great.

伙计，资金又回来了，行情要启动了。

Man, money's back in for the break.

接下来，克勒默将连线安谋国际（Arm Holdings）的首席执行官，深入了解这家芯片制造商的最新状况。

Coming up, Cramer's linking up with the CEO of Arm Holdings to get a grip on the chipmaker's current status.

下一个。

Next.

交易往往从一个问题开始，接着是另一个，然后你面前会打开十几个标签页。

Trading often starts with one question, then another, then a dozen tabs sit open in front of you.

一个分析师说买入，另一个却说再等等。

One analyst says buy, another says wait.

一条新闻标题告诉你现在就该行动。

A headline says move now.

噪音太多容易让人搞不清楚状况。

Noise can lead to confusion.

Capital.com 平台上的新闻和分析能帮你理清思路，别被炒作带偏了。

Capital.com's in platform news and analysis help you trade on clarity, not hype.

所以做决定的始终是你自己。

So decisions always stay yours.

Capital.com。

Capital.com.

交易要思路清晰。

Trade with clarity.

差价合约风险很高。

CFDs involve a high level of risk.

81.31%的散户投资者会亏钱。

Eighty one point three one percent of retail investors lose money.

NFL赛季正式开打了。

It's NFL kickoff time.

独家NFL球队估值，由体育商业专家迈克尔·奥泽尼亚带来。

Exclusive NFL team valuations with sports business expert Michael Ozenia.

NFL绝对是利润最丰厚的联赛。

The NFL is by far the most profitable league.

NFL球队估值数据现已上线cnbc.com/sport。

NFL team valuations now on cnbc.com/sport.

整个AI数据中心板块在六月达到峰值，随后大部分在七月末触底，之后又疯狂反弹。

The whole AI data center cohort peaked in June, then most of them bottomed near the end of July before rebounding like crazy.

有些股票还没回到高点。

Some of the stocks are still off their highs.

ARM Holdings是一家主要的芯片设计公司，现在也自己生产CPU。

ARM Holdings, major chip design company, now also makes its own CPUs.

股价在六月曾涨到452美元，现在回调到了4244美元以下。

Stock was went to $452 in June, now pulled back to just under $4,244.

这跌了46%，但我觉得你不该这么看。

That's down 46%, but I don't think you should look like that.

我觉得你应该看看这个。

I think you should look like this.

从我上次见我们的下一位嘉宾到现在，它已经翻了一番。

It doubled from the last time I saw our next guest.

人们担心大型前沿AI实验室会自我限制数据中心的发展，这就是为什么Arm周一股价大跌的原因。

People are worried about a self imposed slowdown in the data center coming from the big frontier AI labs, which is why Arm got slammed on Monday.

但我认为我们面对的不仅仅是芯片短缺的问题。

But I think we're looking at a more than your chip shortage regardless.

别只听我一说。

Do not take it from me.

咱们来连线一下Renee Haas。

Let's check-in with Renee Haas.

他是Arm的首席执行官，听听他的看法能帮我们更好地了解情况。

He's the CEO of Arm Home to get a better read of the situation.

Renee，欢迎回到《Bid Money》节目。

Renee, welcome back to Bid Money.

谢谢，吉姆。

Thank you, Jim.

很高兴再次见到你。

Great to see you again.

好的。

Okay.

所以你要为我在节目开头提出的观点作证。

So you're going to be my testimony for my view at the top of the show.

美联储刚才宣布加息，我可以说，是四分之一个百分点。

We had a fed rate, I could say, a quarter a quarter percent.

我的意思是，你要关注那些需求极其旺盛的公司，哪怕加息四分之一个百分点，也根本挡不住买家的脚步。

And what I say is you have to look at companies where demand is so great that a quarter percent really doesn't stop the buyers.

我觉得你们产品的需求非常强劲，所以利率是3.75%还是4%，其实影响都不大。

And I think demand is so great for your product that it really won't matter all that much whether rates are three and three quarters or four.

我不太清楚怎么把沃勒主席今天提到的内容，和我们整体需求的百分比变化直接联系起来。

I don't know how to exactly link, what, what chairman Worst did today in terms of our overall demand in terms of percentages.

但我可以这么说，吉姆，无论是边缘计算、汽车领域、机器人技术，还是数据中心，我们对技术的需求都达到了前所未有的高度。

But what I can say, Jim, is that demand for our technology, whether it's at the edge, whether it's in automotive, robotics, and certainly data center has never been stronger.

我知道你最近听很多人讨论过人工智能，但这确实是真正的人工智能，因为AI对算力的需求太大了。

And I know if you've heard, you've had a lot of folks on this week talk about AI, but it really is AI because AI is so compute intensive.

这需要大量的CPU、GPU和内存支持，而这正是我们硬件业务的核心所在。

It requires so much in terms of CPUs, GPUs, memories, and that's the hardware arm is.

所以，是的，对我们来说，需求形势从未如此强劲。

So, yeah, demand for us, it's just never looked better.

我记得你刚公布计划时，说的是20亿美元的业务规模，那时候你们还没有晶圆厂产能。

I know when, you first unveiled your plan for, say, just a $2,000,000,000 in business, you didn't have the foundry space.

后来我看你们把目标下调到了10亿，但我以为你们在规划上还是留了些余地，觉得应该能再搞到些设备来满足需求。

I felt you a little more we got to down to a billion, but I thought you were giving us a little more leeway in thinking that maybe you could get some more machines to make what you need.

所以在财报电话会上，大概在五月的时候，我们表示对20亿美元的营收有把握。

So what we what we said in the earnings call, think was May time frame that we had visibility to $2,000,000,000.

而在上一场财报电话会上，我们说从五月到七月，我们对实现这20亿美元目标的信心增强了。

And what we said in the last earnings call was that our confidence to achieve that $2,000,000,000 number had increased from May to July.

现在已经是九月了，我可以告诉你，吉姆，我对实现那个目标的信心比七月财报电话会议上还要强。

Here I am in September, and what I can tell you is, Jim, I'm more confident today than I was on that July earnings call.

好的。

Alright.

太棒了。

That's terrific.

这确实是个转折点。

So we're feeling That's an inflection.

你心里清楚。

You know that.

你是这么说的。

You say that.

而且我们对此感觉不错。

And it's and we're feeling good about it.

我们对这件事非常有信心。

We're feeling really good about it.

好的。

Okay.

那现在咱们直面问题，达里奥，谈谈这些可能放缓的问题。

So now let's take a head on, Dario, and these issues involving potential slowdown.

我的意思是，在我看来，这些公司将会……我看啊。

I mean, to me, what these companies are gonna be I'm look.

我主要属于黄仁勋那一派的想法，因为我觉得是你教会了我这么想。

I'm in the Jensen Huang school of thought mostly because I think you taught me to be there.

不过我觉得杰森的意思是，听好了。

But I think Jensen's saying, listen.

这些公司会负起责任的。

These companies are gonna be responsible.

他们会做正确的事。

They're gonna do the right thing.

虽然会有监管措施，但发展步伐还是会继续加快，因为需求实在太大了。

They'll be guardrails, but it's gonna continue advance of pace because the demand's so great.

当然。

Oh, of course.

你知道，这挺有意思的。

You know, I you know, it's interesting.

上周有些情况出来了。

Some of the things have come out this last week.

我的观点是，归根结底，任何推出产品的公司都必须对产品的质量、可靠性和安全性负责。

You know, my my view on it is that, ultimately, any company that's putting out a product needs to stand behind the quality, reliability, and safety.

我的意思是，这是一个最基本的底线问题。

I mean, that's a kind of a fundamental table stakes issue.

企业必须把它做好。

Companies need to get it right.

政府做不到这一点。

Governments don't get it right.

那15家公司也做不到。

15 companies don't get it right.

这是创办公司时必须做好的基础工作。

It's kind of the hygiene of what you're doing building a company.

我觉得这次不一样的地方在于，首先AI带有一种神秘感，让人特别担心。

I think what's different here is that number one AI has got this mystical quality about it that people get really afraid about.

其次，这些都是年轻的公司。

And secondly, these are young companies.

这些年轻公司在开发产品时速度极快，技术进展也非常迅速。

These are young companies developing products really, really fast and the advancements are taking place very, very quickly.

这意味着作为产品公司，我们需要通过工程、质量、可靠性和安全方面做好基础工作。

All that means is as a product company, there's the hygiene you put in place through engineering, quality, reliability, safety.

他们最终会找到解决办法，但答案不是让我们集体踩刹车，因为我甚至不清楚一个产品如何比另一个更好地解决这个问题。

They'll figure it out, but the answer is not let's all collectively tap on the brakes because I don't even know how one product solves that versus another.

这对我来说真的不太合理。

It's just doesn't really make a lot of sense to me.

但我可以告诉你这个周末发生的一件事，我会向人们准确表达你刚才说的观点，因为这是常识性的看法。

But I would tell you there was something that happened this weekend, which is that I would tell people exactly what you said because it's a common sense view.

但他们肯定会回来跟我说，吉姆，如果那些话不是真的，这些人怎么会说我们到2030年都会死呢。

But they will come back and say, Jim, these people wouldn't say we're all gonna die by 2030 if it weren't true.

你知道的，Arm交付了海量的产品，成千上万种产品，三百五十亿颗芯片。

You know, Arm delivers tons of products, thousands of products, three hundred fifth 50,000,000,000 chips.

用在所有东西里吗？

In everything?

是的，全都用。

In everything.

质量、可靠性，我们必须得为这些负责到底。

Quality, reliability, we need we have to stand behind it.

你能想象吗，要是我们的产品因为安全或质量问题着火了怎么办？

Can you imagine if we had security issues or safety issues that caused our products to burn on fire?

说实话，我觉得我会很在意这件事。

Well, I tell you, I'd I would see it.

再打个比方，如果有个实习生在公司干了半年，看了我们的质检流程后跑出去说我们的产品都会爆炸，大家都会没命，那简直太不负责任了。

And if we had an intern who worked at the company six months and looked at the quality process we had and then came out and said all of our products are gonna blow up and and everyone's gonna die, it's just irresponsible.

所以我觉得大家对此的第一反应有点过于夸张，但缺乏逻辑依据。

So I found sort of the knee jerk to all this a little sensational, but not grounded in logic.

好的。

Okay.

看来我们在这点上看法一致。

I think that's that's you come down where I am.

接下来我想聊聊机器人和自动驾驶汽车。

Now I wanna talk about robots for a second and self driving cars.

要想做成靠谱的机器人和自动驾驶汽车，这些设备到底需要多少颗CPU？

How many CPUs do these things really need if they're gonna be good robots and good self driving cars?

很多。

A lot.

而且关键在于CPU的分布方式。

And and it starts with the distribution of where the CPUs are.

你得有作为机器大脑的CPU才行。

You've got CPUs that are the brains of these machines.

所以实际上在机器人里，明白。

So literally in a robot Mhmm.

你会拥有几个非常大的 CPU 集群。

You'd have a couple of very large CPU clusters.

我说的 CPU 集群，是指那些可能包含 10 个、20 个甚至 60 个 CPU 的芯片。

And when I say CPU clusters, these are chips that could have 10 CPUs, 20 CPUs, 60 CPUs.

然后你就拥有了多个大脑。

And then you've got multiple brains.

换句话说，你可能在头部装一些处理器，也可能在躯干部分装一些。

In other words, you may have something in the head, you may have something in the torso.

然后当你延伸到四肢时，传感器、控制器以及那些负责感知功能的部件，它们各自都需要CPU。在机器人身上，可能有成百上千个CPU；在汽车里也是一样， literally 有几百个CPU。

And then when you go out to the limbs, the sensors, the controllers, things that can do the perception, their CPUs there, Hundreds and hundreds of CPUs and robots and similar in an automobile, literally hundreds of CPUs.

那这就会成为主要的瓶颈因素了。

Well, that's gonna be the big gating factor then.

问题是，我们根本造不出足够的CPU来满足所有这些需求。

We don't we can't make enough CPUs to do all that.

正如我们之前讨论过的，需求已经爆棚了，而主要的瓶颈在于供应。

Demand is off the charts as we talked about earlier, and the big constraint is going to be supply.

供应方面主要卡在制造逻辑芯片所需的晶圆上，但问题不仅仅出在晶圆上，吉姆。

And supply is around wafers for building the logic chips, but it's not just the wafers, Jim.

还有基板。

It's the substrates.

以及测试设备。

It's the testers.

是内存的问题。

It's the memory.

是整个供应链的问题。

It's the entire supply chain.

所以我觉得未来几年，我们会面临一些非常复杂的供应链挑战。

So we are we are looking at some really complex supply chain issues, I think, for the next number of years.

你刚才说‘听好了’的时候，我一开始还挺担心的。

So I was initially worried when you said, listen.

我们会自己生产，同时也为其他人代工。

We're gonna go in and make our own, and we also make them for everybody else.

你是全球最大的厂商。

I mean, you're the world's largest.

我本应该一开始就说的。

People I mean, I should have said that from the very beginning.

你绝对是全球第一。

You're by far the world's largest.

但听起来需求非常旺盛。

But it sounds like there's so much demand.

没人会对你说，蕾妮，你别想跟我们竞争。

No one is gonna say to you, Renee, you can't come in against us.

他们只会说，能不能请你帮帮我们？

They're gonna say, can you please help us?

我们的问题更多是后者，而不是前者。

Our problems have been far more the latter than the former.

你知道，我们在内部讨论这个问题的时候，就在想该怎么回应那些质疑的声音，比如‘天哪，你们为什么要这么做？’

You know, when we were having discussions about this internally, we were thinking about how do we respond to the questions that say, you know, gosh, why are you doing this?

‘你们这是在侵占我的地盘。’

You're infringing on my space.

但实际上我们收到的问题是：‘能不能多生产一些？’

The questions we get is, can you make more?

“能不能加快交付速度？”

Can you deliver them faster?

你们下一代产品什么时候发布？

When is your next generation coming out?

市场空间很大，大家都有机会。

And there's plenty of room for everybody.

我是说，Jensen 正在基于 ARM 架构打造 CPU。

I mean, Jensen builds a a CPU based on ARM.

他提到 Vera 的需求非常旺盛。

He's talked about fantastic demand for Vera.

所以世界挺大的，我们大家都有足够的空间。

So the world's pretty big, and there's a lot of space for all of us.

你觉得还有很多人把你公司当成手机公司吗？也许这就是我觉得它被严重低估的原因？

Do you think there's still people who remember your company as really a a cell phone company and maybe that's why I still think it's incredibly undervalued?

有可能。

I it's possible.

确实，那是公司的历史，但很快数据中心就会成为我们最大的业务板块，而且过去几年里这块业务一直保持着两位数的增长。

You know, that's the history of the company, but in very short time, data center is gonna be our largest business, and that we've seen double digit growth in that over the last number of years.

其实关于ARM，我们的定位是AI的计算平台。

Really, the thing about ARM is we are the compute platform for AI.

所有的AI应用最终都会通过ARM来运行，不管是以哪种形式。

Every AI application is gonna run through ARM one shape or another.

接下来我要跟屏幕前的观众朋友们说几句。

Now I will say this to our viewers.

你之前已经多次提到过这一点。

You said this many times.

你有一次走在街上。

You come once walk on the street.

希望你继续这样做，就像从140降到120，你会非常沮丧，因为我们知道需求是疯狂的。

Hope you continue to do And it would be like one forty going down to one twenty, and you and I'd be very frustrated because we knew that the demand was insane.

但我认为人们需要一点时间来意识到你对公司所做的改变。

But I think it took a little while to people to realize what you've done with the company.

我们传统上在谈论数字时非常保守，我非常确信7月份的情况更好。

We've been very conservative traditionally about how we talk about the numbers, and I'm very, very confident when I tell you that things were better in July.

五月的时候是这样，到了九月情况更好，从七月的表现来看也是如此。

They were in May, and they were better in September in terms of how they look in July.

希望大家把这话当真，因为我们上市以来每个季度的业绩都超过了预期。

I hope people take that to the bank because we have essentially beat every single quarter since we've been public.

这也是我想让大家明白的一点：美联储加息并不意味着世界末日。

Well, it's why I want people to understand just because the Fed raises rates doesn't mean it's the end of the world.

有些公司需求非常强劲，它们的股价依然稳步上涨。

There's some people who have such strong demand that their stocks continue apace.

这位是ARM控股的首席执行官Renee。

That's Renee, CEO of ARM Holdings.

请看看这个。

Please look at this.

那个信托基金当时的仓位非常好。

The trust had a great position.

说实话，我们赚了不少钱，然后选择了获利了结，但这其实是个大错。

We made so much money, candidly, and then we took profits, and that was a big mistake.

好了，广告回来继续聊钱的事。

Man, money's back after the break.

接下来我们会探讨，奥克塔（Okta）是否能在智能体时代来临时巩固已有的成果？

Coming up, is Okta positioned to lock down gains as we enter the agentic era?

克拉默正在采访这位首席执行官。

Cramer's investigating with the CEO.

马上就来。

Next.

现在大家都在担心AI带来的风险，我要深入聊聊那些我认为能帮助降低风险的网络安全公司。

Now that everybody's bringing their hands over the risk of AI, I'm gonna drill down with cybersecurity places that I think can help mitigate the risk.

以Okta为例，这家身份验证领域的大佬在8月底报告了非常出色的季度业绩。

Take Okta, a major player in identity verification that reported a phenomenal quarter in late August.

过去六个月股价上涨了140%到144%，部分原因是之前一直受AI替代担忧的压制。

Stocks are up a 140%, 144 over the past six months, Part because it's all being held back by AI displacement worries.

结果发现，人工智能根本不会让Okta变得过时。

Turns out artificial intelligence was never gonna make Okta obsolete.

他们的安全软件现在比以往任何时候都更加重要，毕竟现在有AI代理在互联网上到处跑。

That didn't, their sec security software is more essential than ever now that you've got AI agents roaming around the Internet.

所以让我们仔细看看Todd。

So let's take a closer look at Todd.

节目最后出场的是Okta的联合创始人兼CEO。

We get an open end of the show, cofounder and CEO of Okta.

了解更多。

Learn more.

Todd，欢迎回到《赚钱之道》节目。

Todd, welcome back to Made Money.

很高兴能来到这里，Jim。

It's great to be here, Jim.

好的。

Alright.

那么，Todd，我找到了一段你在9月9号做的采访，也就是在那件事发生之前。

So, Todd, I've got a, an interview that you did September 9 before what happened.

好的。

Okay.

确实如此，你非常有远见。

This is it's a and you you're very prescient.

你早就预测到这种情况会发生，但你提到这些实体都有身份标识。

You predicted that this could happen, but you said that they have these things have identities.

也就是说，只要有了身份标识，我就能阻止它们。

And it basically, if you have an identity, I can stop you.

情况是这样的。

Well, here's the deal.

AI智能体是一种全新的、强大的身份类型。

AI agents, they're a powerful new identity type.

以前呢，你只是有一些软件，然后把它们整合在一起就行。

So it used to just be you would have some software and you would integrate it together.

而现在这些AI智能体能够自主做决定了。

And now these AI agents can make decisions on their own.

要让它们变得更强大，你必须做一件关键的事，那就是给它们更多的数据、工具和可执行的操作权限。

And to to make them more powerful, you have to do one key important thing, and that's give them access to more data, more tools, more actions that they can take.

这意味着你得赋予它们身份标识，这样你才能对数据和这些操作进行控制、监控和全面掌握。

That means that you have to give them an identity so that you can control and monitor and have visibility into that data and those actions.

这样你就能洞察到潜在的负面后果，从而加以防范和控制。

That gives you insight into the potential negative repercussions so you can prevent it and control it.

我觉得这点非常重要。

Well, I think this is really important.

之前有很多人用相当激烈的言辞讨论可能出现的问题。

We had a lot of people discussing, in some pretty fiery language what could go wrong.

然后我看了你们的做法，心想，等等。

And then when I read what you do, and I think, wait a second.

你看。

Look.

这类东西可能会很多。

There may be a lot of these.

它们看起来可能不太正经，但只要有了身份标识，哪怕它们不怀好意，也还是能被制止的。

They may, seem nefarious, but as long as they have an identity, even if they are nefarious, they can still be stopped.

我们认为关键在于可视性和控制权。

We think the key is visibility and control.

所以像OpenAI和Hugging Face都听说过这个问题，之前有个事件，OpenAI的一个模型突破了隔离限制，去攻击了Hugging Face这家公司。

So OpenAI and hugging face, everyone's heard about this problem that happened where an OpenAI model jumped containment and went after this company Hugging Face.

不过OpenAI表示他们的模型并没有连接互联网。

Now OpenAI said that their model was not connected to the Internet.

结果发现，它其实连上了互联网。

Now, turns out that it was connected to the Internet.

所以，通过像 Okta 这样的工具实现可视化和控制，你才能真正清楚自己的 AI 连接到了哪些地方。

So by bringing visibility and control with something like Okta, you actually know what your AI is connected to.

这当然不是解决所有问题的万能药，但解决问题的基础在于可视化和控制。

So it's not going to be a magic fix for all this, but the foundation of the fix is visibility and control.

一旦建立起这套机制，你就能采取正确的应对措施，并对实际发生的情况做出准确的风险评估。

And once you have that in place, then you can take the right actions, then you can have a good risk assessment of what's actually happening.

所以，正如你所说，你提供的是构建安全智能体企业的蓝图。

So you're offering, as you say, the blueprint for the secure agentic enterprise.

这非常重要。

It's very important.

当我与客户交流时

When I talk to customers

不过，你能大概说说我接触了多少客户吗？

But would you probably talk how many customers I'm

我每周都要坐飞机出差，去世界各地跟客户交流，他们经常会问起竞争对手的情况。

I'm on on a a plane every week getting out there in the world and talking to customers, and they ask about competition.

投资者也总是问关于竞争的问题。

Investors ask all the time about competition.

你们的主要竞争对手是谁？

What's your competition?

他们通常希望我能聊聊某些平台型企业的竞争情况。

And they expect me to talk about maybe a platform player.

让人摸不着头脑。

Confusion.

对，就是这种混乱的局面。

Confusion is the yeah.

我看过你的资料。

I read your stuff.

因为每个供应商都在对客户说，我们有解决一切问题的方案。

Because every every every vendor is going to the customer and saying, we have the answer to everything.

我们什么都能做。

We can do it all.

在客户看来，这全是重叠和混乱。

And it looks to the customer's mind, it's a bunch of overlap and a bunch of confusion.

所以我们正努力联合整个行业，告诉大家：要想成功，你需要这些要素。

So we're trying to work with the industry everyone in the industry and say, here are the elements you need to be successful.

你需要实时可视化的方案，需要身份管理的方案，而我们专注于身份管理，并且会把这块做好。

You need this plan for your real time visibility, your plan for identity, and we're focused on the identity, and we're gonna fix that.

那身份的统一管理者，这就是你们公司需要的东西吗？

But the universal director of identity, that's what you need at a company?

对，你得有一套系统来记录你的 AI 智能体。

It's you have to have a system of record of your AI agents.

但你是可以……

But you can.

你必须有一个中央仓库来管理这些智能体，这样你才能看到它们在做什么，以及它们连接到了哪里。

You have to have a central repository of your agents so you can see what they're doing and see what they're connected to.

这是实现管控的基础。

It's the foundation for control.

好的。

Okay.

所以跟很多人一样，这可不是那种能超越股市局限的问题。

So like many other people, this isn't one of those issues that transcends stocks.

我是说，我之前还跟我老婆说我要去见你。

So I mean, versus, I I was telling my wife I was gonna see you.

她知道我跟你聊过，而且你脑子挺清楚的。

She knows that I I talked to you and that you've got a real solid head.

她问，我的意思是，除非大家都用你的方案，否则是不是所有人都会完蛋？

She goes, well, mean, like, unless everybody uses your friend, I mean, isn't everyone gonna die?

我说，天哪。

And I said, oh, god.

不会的。

No.

她接着说，那好吧，那你就是想把你的朋友排除在外。

And she goes, well, then you just want your friend out.

我说，托德，我关注托德很久了，但也许他只是一个提供理性解决方案的人。

I said, Todd, I I covered Todd for a long time, but maybe he's just a rational person offering a solution.

我觉得我们有一件事做得很好，那就是我们没有把问题简单化。

I I think I don't one of the things we don't do is we don't oversimplify it.

我们知道人工智能技术潜力巨大，而让它变得更强大的方法就是给它更多的数据。

We know that there's tremendous potential in AI technology, and the way you make it more powerful is you give it more data.

给它更多的连接。

You give it more connections.

给它更强的动力。

You give it more power.

赋予它更高的能力。

You give it more capability.

我希望这一切都能实现，让好处惠及每家企业和各个组织。

And I want that to happen, and I want the benefits to accrue to everyone's company and everyone's organizations.

我希望美国能赢。

I want America to win.

我也希望全世界都能受益。

I want the world to win.

但要在控制风险的同时实现这一点，关键在于透明度和掌控力。

But the only way to do that while also mitigating the risks is visibility and control.

而且我们现在的处境非常有利，因为实现透明和掌控的核心在于身份识别。

And and we're in a perfect situation because the key to visibility and control is identity.

你得知道这些智能体在哪儿。

You have to know where these agents are.

得知道有12个。

Have to 12 know

还是说有一百个成群结队地冲过来？

100 in a swarm that are coming?

那简直就像一场踩踏事故。

That's like a stampede.

怎么阻止一群蜂拥而至的AI？

How do you stop a swarm?

嗯，部署这些系统的公司规模很大，实力也很强。

Well, the systems that contract this stuff are quite big and quite powerful.

我们在防御和进攻方面都拥有大量的计算机科学技术和高度复杂的体系。

We have a lot of computer science and a lot of sophistication on the defense as well as the office.

但你是不是在跟你的团队说，天哪……

But don't you aren't you saying, oh my god to your team.

一千二百个来了。

Here comes 1,200.

我是说，你该怎么办？

I mean, what do you do?

首先，我们给客户提供的工具可以追踪、控制和阻止它们。

Well, I mean, first of all, we give the tools to our customers to track them and control them and block them.

整个世界都在变得具有自主代理能力。

The whole world is becoming agentic.

这部分叙事可能有点被低估了。

This part of the this part of the narrative is is probably underhyped.

我们都在谈论智能体。

We talk about agents.

人人都聊人工智能，但现实是，一切都在变成

Everyone talks about AI, but the reality is is that everything's turning into

智能体。

an agent.

你的手机会变成智能助手。

Your phone's gonna be an agent.

你的电视也会变成智能助手。

Your TV's gonna be an agent.

你的车也会变成智能助手。

Your car's gonna be an agent.

所以关键在于对各类技术设备的全面可视化和掌控。

So it's about visibility and control across all of technology.

这就是其中的利害关系。

That's the stakes.

不过，托德，这件事里有一样让我挺介意的。我一开始听说的时候就想，他们怎么不去找像您这样专业的人聊聊呢？

But one of the things that bothered me about this whole incident, Todd, when I first heard it, I said, what are they talking to people like Todd?

他们难道不该去找像CrowdStrike的乔治·库尔茨那样的专家谈谈吗？

Are they talking to people like George Kurtz at CrowdStrike?

因为这才是他们应该去沟通的对象。

Because that's who they should be discussing things with.

他们不该在推特上散布那种我们会像好莱坞烂片里演的那样全都完蛋的言论。

They shouldn't be going on Twitter and saying we're all gonna die like a bad Hollywood movie.

我觉得很多时候，他们其实是想做好事的。

I think a lot of times they're trying to do the right thing.

确实存在风险，大家也都挺担心的。

There there are risks and there are things that everyone's worried about.

问题在于，正如你所说，这需要整个生态系统的配合。

The problem is is that it's gonna take the ecosystem, as you said.

这需要模型公司来参与。

It's going to take the model companies.

还需要SaaS应用的支持。

It's going to take the SaaS applications.

以及安全基础设施的配合。

It's to take the security infrastructure.

这就是为什么‘蓝图’这个概念如此重要，它能让大家协调一致，在正确的轨道上共同努力，把握技术的未来，确保我们在潜力与风险之间取得平衡，从而实现理想的结果。

That's why this concept of a blueprint is so powerful, this blueprint that can get everyone coordinated the right lane working together to take what is really the future of technology and making sure we balance off the potential with the risk and get the right outcomes

对所有人来说。

for everyone.

也许达里奥的那封信会更好，他是个聪明人。

Maybe Dario's letter would have been better, and this he's a smart guy.

我不会告诉他该怎么做。

I'm not gonna tell him what to do.

如果他当时说，听着。

If he had said, listen.

我们需要时间来做好你刚才说的那些事，而不是简单地说要慢下来。

We need time to do exactly what you did as opposed to say we gotta go slower.

因为“慢下来”容易被理解为是在自我束缚、削弱自己。

Because go slower was interpreted as being as as hobbling yourself.

但我觉得，留出时间来做你提到的那些事情，是非常必要的。

But time to be able to do the things you're talking about, I think seems well served.

我认为，正确的讨论重点不在于速度。

I I think that I think that right discussion is not about speed.

我觉得在公司里，最关键的讨论永远是关于优先级的。

I think the right discussion in companies, it's always about priority.

到底什么才是最重要的？

What's the priority?

所以这并不是为了慢而慢，故意拖慢进度。

So it's not about slowing down because you wanna go slow.

而是要把行业必须做的事情排好优先级，确保整体安全。

It's about prioritizing the things we have to do as an industry to keep it safe.

如果真是这样的话，那倒有点像黄仁勋的做法。

Well, if that's the case, that's a little more like what Jensen Huang's doing.

我不太认同那种所谓的零风险或者零错误的概念。

I hate the idea that there's, like, you know, some sort of zero something.

在我看来，你提供了一条切实可行的路径，不管需要花多长时间，都能把局面控制住。

To me, you're offering a legitimate pragmatic way to be able to get it under control regardless of how long it may necessarily take.

只要整个行业把优先级定对了，时间表自然就会理顺。就像黄仁勋说的，我们不会在产品没准备好时就强行推向市场。

It's and if we have the right priority as an industry, the timelines will work themselves out because we're not won't as Jensen said, we're not gonna ship products before they're ready.

我们会确保把正确的管控措施和安全放在首位。

We're gonna make sure we prioritize the right controls and safety.

对我们来说，这一切都始于这种可视化和控制的概念，这非常重要。

And for us that all starts with this visibility and control concept which is so important.

不过我想告诉你，你总是提到这一点：如果出了什么差错，我的生意因此受损，我会起诉这些人，看看他们是否做到了你刚才说的那些事情。

But I would tell you that when I really I mean, you constantly mention that that if I if something went wrong and my business was hurt by this, I would sue these guys to see if they did what you just talked about.

如果他们没建立这样的流程，他们的生意恐怕就难以为继了。

If they didn't have that process, their business will probably go under.

这真的很重要。

This really matters.

对相关企业来说确实很重要。我们给使用AI的企业建议是，在把系统接入更多设备、导入更多数据、开放数据仓库和应用，并重构业务流程时，一定要提前规划好蓝图，确保实现可视化和可控性。

Well, matters to the companies to put the right controls in place, and our advice to the companies using AI is make sure as you connect it to more things and as you give it more data and you open it up to your data warehouses and applications and you rework your business processes, make sure you have the right blueprint in place to get that visibility and control.

这一切要从身份管理开始，而我们正好能在这方面提供帮助。

And that starts with identity, and we're here to help you with that.

天哪，托德，我真不知道该怎么说。

Boy, Todd, I don't know.

你让我明白了身份认同。

You've taught me identity.

我记得第一次跟你聊的时候，我还以为那是什么卡片之类的东西。

I remember the first time you talked to me, and I thought it was like, you know, some sort of a card.

不过你肯定也给Jim解释过了。

But we you've certainly explained it to Jim.

我以前不是这样的。

I used to be.

对。

Yes.

好吧，你看。

Well, look.

我……世界上的

I I World's

已经过去了。

moved on.

你一直是个头脑清醒、理性的人，听起来我们现在比以往任何时候都更需要你。

You have always been a a clear headed rational person, and it sounds like we need you more than ever.

谢谢你，Todd。

Thank you, Todd.

感谢邀请。Todd McKinnon是Okta的联合创始人兼首席执行官。

Thanks for having Todd McKinnon is the he's the cofounder and CEO of Okta.

债务资金还在那边

Debt money's back there

给你。

for you.

谢谢。

Thank you.

接下来这位可是华尔街反应最快的人，所以我们要借助大家的力量来考考他。

Coming up, he's the fastest mind on Wall Street, so we're putting him to the test with your help.

好，快进入闪电问答环节吧。

Bring on the lightning round next.

是时候了。

It is time.

闪电问答环节结束了。

And then the lightning round is over.

你准备好了吗？

Are you ready?

史蒂夫，该你了。

Steve that.

道格拉斯在阿拉斯加。

Tom with Douglas in Alaska.

道格拉斯。

Douglas.

咱们开始干活吧。

Let's get to work.

在新泽西州、宾夕法尼亚州和德克萨斯州扩大业务。

Expanding operations in New Jersey, Pennsylvania and Texas.

与英伟达和谷歌的合作，16项网络、数据中心建设以及AI防御系统。

Partnerships with Nvidia and Google, 16 networking, data center build outs and AI defense systems.

博士。

Doctor.

克拉默，你对诺基亚有什么看法？

Cramer, what are your thoughts on Nokia?

我非常喜欢诺基亚。

I like Nokia very much.

你能提出来，我很高兴。

I'm glad you brought it to our attention.

我觉得这情况非常棒，我现在就会买。

I think it's a terrific situation and I would be a buyer right here, right now.

我们接下来连线马里兰的Chris。

Let's go to Chris in Maryland.

Chris。

Chris.

嘿，吉姆。

Hey Jim.

我是来自马里兰州安纳波利斯的克里斯。

This is Chris from Annapolis Maryland.

不错。

Nice.

我打电话是想聊聊纳斯达克代码为SHIP的股票。

I'm calling in about the NASDAQ ticker SHIP.

Bold Transportation。

Bold transportation.

Bold Transportation现在势头很猛。

Bold transportation is on fire.

我觉得这情况非常好。

I think it's a terrific situation.

这只股票价格不贵，而且我认为它的分红是安全的。

It's not an expensive stock, and I think the yield is safe.

我会买入。

I'd be a buyer.

接下来连线亚利桑那州的基拉。

Let's go to Kira in Arizona.

基拉。

Kira.

你好，吉姆教授。

Hello, professor Jim.

最近怎么样？

How's it going?

其实挺好的。

It's actually going excellent.

谢谢。

Thank you.

你呢？

How about you?

我也挺好的。

It's going good for me.

太好了。

Perfect.

太好了。

Perfect.

哦，对。

Oh, yeah.

我挺好的。

I'm doing good.

好的。

Alright.

我想听听你对Hub Group的看法，考虑到公司和运输行业最近的情况。

I wanna get your thoughts on Hub Group with everything going on with the company and the transportation sector.

你觉得现在买这个股票合适吗？

Do you think this fuck is a buy right now?

这确实是个

That's a

难题。

tough one.

这确实是个难题。

That's a tough one.

JB Hunt昨晚公布了糟糕的业绩后，我在卡车运输股上只持有一点小仓位，我得告诉你，接下来一段时间会很难熬。

After JB Hunt announced bad numbers last night, I my travel just has a small position in trucker, and I've gotta tell you, it's gonna be rough going for a little bit.

得等油价稳定下来，咱们就没事了。

Gotta wait till oil calms down, and then we're okay.

接下来听听佐治亚州克林顿的听众来电。

Let's go to Clinton in Georgia.

克林顿。

Clinton.

你好，吉姆。

Boo you, Jim.

boo你。

Boo you.

吉姆，这家公司有84亿美元的积压订单，对美国核潜艇业务有巨大的敞口，而且先进反应堆的机会也在增长。

Jim, this company has $8,400,000,000 in backlog, huge exposure to the US nuclear navy, and growing advanced reactor opportunity.

然而股价却接近52周低点。

Yet the stock sits near its fifty two week low.

在这个估值下，VWX技术难道不是一个现在买入的重大机会吗？

At this valuation, isn't VWX technology a major buying opportunity right here?

不是。

No.

我来告诉你VWX的问题在哪。

I'll tell you the problem with VWXT.

30倍的市盈率太高了。

The price earnings multiple at 30 is too high.

我觉得这是一家好公司。

I think it's a great company.

我是说，确实非常棒，但价格太高了，尽管它涉及核能。

I mean, really terrific, but it's too expensive even though it has nuclear.

我们得暂时观望一下。

We have to hold off.

接下来请连线伊利诺伊州的Bruce，你好Bruce。

Let's go to Bruce in Illinois, please, Bruce.

Jimmy Chill，我持有Dominion Energy的股票，听说他们要和NextEra Energy合并。

Jimmy Chill, I own stock and Dominion Energy, and they are supposed to merge with NextEra Energy.

如果合并获批，我手里的道明尼能源股票就会换成NextEra Energy的股票。

If the merger is approved, my shares of Dominion will be converted into shares of NextEra.

你对NextEra Energy怎么看？

What are your thoughts of NextEra Energy?

赶紧套现离场。

Take the money and run.

说真的，赶紧套现走人。

Take the money and run, honestly.

我的意思是，我觉得拿到钱就对了。

I mean, I just think you take the money.

你已经赢了。

You've won.

别犹豫了。

Don't fool around.

这只股票不如你手里那只值得持有。

That's not a good stock to own versus the one that you own.

我们连线新罕布什尔州的克里斯。

Let's go to Chris in New Hampshire.

克里斯。

Chris.

嘿，吉姆。

Hey, Jim.

你觉得RDW红线怎么样？

What do you think of a RDW red wire?

我们现在处于加息周期。

We are in a rate tightening cycle.

在加息周期里，我们不能到处去买那些亏损的公司。

In a rate tightening cycle cycle, we cannot buy companies that are losing money left and right.

这根本行不通。

It just won't work.

好了，各位，闪电问答环节到此结束。

And that, ladies and gentlemen, is conclusion of the lightning round.

闪电环节由嘉信理财赞助播出。

The lightning round is sponsored by Charles Schwab.

接下来，克拉默会论证，即使这些公司放慢脚步，你依然可以买入AI概念股。

Coming up, Cramer's making the case that you can still buy the AI stocks even if the companies do slow things down.

下一个话题。

Next.

明天，请通过街头快讯来开启你的交易时段。

Tomorrow, kick off the trading day with squawk on the street.

这里是纽约证券交易所第九号交易台现场直播。

Live from post nine at the NYSE.

如果这些无人机群被证明很聪明，而且别有用心，那我确实担心朝鲜和伊朗。

If the swarm turned out to be smart, turned out to be nefarious, and I worry about North Korea and Iran.

我不担心你会释放出一堆无人机。

I'm not worrying about you unleashing a swarm.

你根本排不上我的担忧名单。

You are the least of my worries.

好。

Good.

我确实应该这样。

I should be.

什么？

What?

这算是夸奖了。

There's a compliment.

AI，我选大卫，不要你的那个讨厌鬼。

AI I'll take David, not your bitch.

我跟你说。

Let me tell you.

那才是让人担心的事。

That was worries.

AI，帮个忙。

AI favor.

东部时间早上九点开始。

It all starts at 9AM eastern.

自从OpenAI和Hugging Face那件事发生后，我看着AI的发展，感觉就像核能一样让人担心。

When I look at what's happening with AI since the big open AI hugging face incident, it feels like nuclear power all over again.

如果1979年三里岛真的爆炸了，我们几十年都没再建核电站，尽管后来证明它是迄今最大规模、最清洁且相对最安全的能源形式。

If the 3 Mile Island really blew up in 1979, we stopped building nuclear plants for decades, even as it proved over time to be the cleanest and relatively safest form of power at scale ever invented.

当我们发现OpenAI推出了旨在测试漏洞的强大AI代理，结果它们却联合攻击了完全独立的Hugging Face时，这彻底改变了关于安全的讨论。

When we learned that OpenAI launched powerful AI agents that were meant to test vulnerabilities and somehow they coordinated a joint attack on Hugging Face, totally separate company, it redefined the safety debate.

如果这些智能体能够像真正的坏人那样，把自己伪装成敌对团队、互相掩护、向OpenAI撒谎并销毁证据，那确实值得我们担心。

If these agents could basically anthropomorphize into enemy agents as a team hiding themselves, lying to OpenAI, covering their tracks just like real bad guys, Then we have, I'd say, some reason to worry.

这简直就像三哩岛核事故重演了一遍。

It's like three mile and all over again.

随之而来的监管可能会非常严苛，甚至扼杀创新。

The regulation in its wake could be quite stifling.

达里奥·阿玛迪斯（Dario Amodei）对此担忧的表达非常到位。

The worry was well articulated by Dario Amade.

他是Anthropic的CEO，认为这类问题对人类来说是生死攸关的大事。当然，还没那位研究员说得那么夸张——他之前说人类在四年后就会面临AI带来的‘死刑判决’。

He's the CEO of Anthropic, talking about how this kind of thing is an existential issue for humanity, not as bad as the exanthropic researcher who said humanity faces an AI death sentence four years from now.

不过情况也确实挺严重的。

But still pretty bad.

在他看来，解决办法是放慢AI发展的速度，加强干预以确保安全，因为必须防止这类入侵事件再次发生。

The solution in his eyes, a slowdown in AI development, lots of intervention to do things safely because we need to ensure this kind of intrusion won't happen.

说白了，就是让发展节奏缓一缓。

Slack in the pace, basically.

关于这一点，我其实有两重身份。

Now I got two hats on this one.

首先是我的选股角色，我想帮你一起想办法多赚点钱。

First is my stock picking hat where I wanna help you try to make some money.

从这个角度看，我担心AI支出放缓会连累很多公司，比如英伟达、英特尔和美光等等。

From that perspective, I worry about how an AI spending slowdown could hurt all sorts of companies like NVIDIA, Intel, Micron, among many others.

这些都是我的核心持仓。

Those are ones that call bones.