OpenAI紧急救助？埃隆的万亿美元薪酬协议，亚马逊起诉Perplexity

OpenAI是否在寻求政府救助以防万一？埃隆·马斯克获得了1万亿美元的薪酬方案来打造全新特斯拉，亚马逊起诉Perplexity。这些内容即将在周五版大型科技播客中播出。事实上，AI安全就是身份安全。AI代理不仅仅是一段代码。

Is OpenAI looking for a government bailout if things go wrong? Elon Musk gets a $1,000,000,000,000 pay package to build an entirely new Tesla, and Amazon sues perplexity. That's coming up on a big technology podcast Friday edition right after this. The truth is AI security is identity security. An AI agent isn't just a piece of code.

它是您数字生态系统中的一等公民，需要得到相应的对待。这就是为什么Okta正带头保护这些AI代理。解锁这层新防护的关键在于身份安全架构。企业需要采用统一全面的方法，通过一致的策略和监督来保护每个身份——无论是人类还是机器。不要等到发生安全事件才意识到您的AI代理是巨大的盲点。

It's a first class citizen in your digital ecosystem, and it needs to be treated like one. That's why Okta is taking the lead to secure these AI agents. The key to unlocking this new layer of protection, an identity security fabric. Organizations need a unified comprehensive approach that protects every identity, human or machine, with consistent policies and oversight. Don't wait for a security incident to realize your AI agents are a massive blind spot.

了解Okta的身份安全架构如何帮助您保护下一代身份（包括AI代理）。访问okta.com。Industrial X unleashed活动汇聚了来自IFS、Anthropic、波士顿动力、微软、西门子等全球最具前瞻性的工业公司领袖，共同探索工业AI在现实世界中的应用前沿。市场正在发生明显转变。

Learn how Okta's identity security fabric can help you secure the next generation of identities, including your AI agents. Visit okta.com. That's 0kta.com. Industrial X unleashed is bringing together leaders from IFS, Anthropic, Boston Dynamics, Microsoft, Siemens, and the world's most progressive industrial companies at the frontier of industrial AI applied in the real world. There's a clear market shift happening.

全球最大的工业企业已结束AI试验阶段，正在大规模部署，并选择IFS作为共同创新伙伴。IFS专为资产和服务密集型行业打造——制造业、能源、航空航天、建筑业——这些领域停机成本高达数百万，安全绝无妥协余地。Industrial X unleashed将展示嵌入实际运营的AI产品实时演示、客户分享的可量化成果，以及当前规模化部署工业AI企业的经验。详情请访问industrialx.ai。

The world's largest industrial enterprises are done experimenting with AI. They're deploying it at scale, and they're choosing IFS to co innovate with them. IFS is purpose built for asset and service intensive industries, manufacturing, energy, aerospace, construction, where downtime costs millions and safety is nonnegotiable. Industrial X unleashed will feature live demos of AI products embedded in real world operations, customers sharing measurable outcomes, and learnings from companies deploying industrial AI at scale today. Learn more at industrialx.ai.

欢迎收听周五版《大型科技播客》，我们将以一贯冷静细致的风格解读新闻。今天节目非常精彩，因为我们将重点讨论OpenAI可能向联邦政府申请债务担保引发的巨大争议（也有不同报道），我们对此持有观点。埃隆·马斯克也通过了1万亿美元薪酬方案（分阶段实施），亚马逊起诉Perplexity事件则让机器人互联网面临质疑。每周五固定嘉宾Margins的兰詹·罗伊将参与讨论。

Welcome to Big Technology Podcast Friday edition where we break down the news in our traditional cool headed and nuanced format. We have a great show for you, a big show for you today because we're gonna talk all about this big controversy over OpenAI potentially requesting a backstop of its debt from the federal government or not. Some disputed reports on that, but we have a point of view. Elon Musk also won a $1,000,000,000,000 pay deal that obviously goes in different steps, but is, has finally been passed by Tesla's, shareholders, and Amazon has sued Perplexity, putting the bot Internet into question. Joining us as always on Friday to do it is Ranjan Roy of Margins.

兰詹，很高兴见到你。

Ranjan, good to see you.

谁能想到本周最重磅的'社会主义'新闻主角竟是OpenAI？

Who knew that the biggest socialism story of the week would be OpenAI?

你酝酿这个已经有一阵子了。

You were cooking that up for a while.

我整个星期都在等这个。

I've been waiting for that all week.

另外，Ranjan，我很高兴看到你坐在电脑前，没有用你的个人机器人搞乱别人的东西。我知道有些人对此有些担忧。

Also, Ranjan, I'm very glad to see that, you know, you're at your computer not messing the people's stuff up with your with your personal robot. I know we had some people concerned about that.

我把机器人留着了。我们还要再等几周，等我开始用人形机器人拆毁别人的房子。

I left, the robot. We're we're we're waiting another another few weeks before I start destroying people's houses via humanoid robot.

那很好。在Ranjan获得破坏性技术之前，我们将珍惜在地球上剩余的宝贵时间，来谈谈OpenAI的现状。实际上，这是一个关于他们寻求联邦政府作为后盾的故事。但更广泛地说，这是关于OpenAI是否准备好迎接公司历史上这一时刻的故事。让我来铺垫一下背景。

That's good. Well, we will take our precious time left on this Earth before Ranjan gets access to destructive technology to talk about the state of OpenAI. Really, it's it's a story about them looking for a backstop from the federal government. But but more broadly, it's a story about whether OpenAI is ready for this moment in the company history. So let me set the scene.

OpenAI的首席财务官Sarah Fryer在《华尔街日报》的大型科技会议上公开表示，他们将为可能出问题的债务寻找后盾。我们来听一段录音。

OpenAI CFO, Sarah Fryer, is out at the Wall Street Journal's big tech conference, and she she basically says explicitly that they are going to be looking for a backstop on the debt that they take out should things go wrong. Let's listen to a clip.

也许甚至是政府层面的介入方式？比如联邦补贴之类的？首先是指那种能让融资得以进行的担保或后盾。

Maybe even governmental the ways governments can come to bear. Meaning like a federal subsidy or something? Meaning like just first of all the the backstop, the guarantee that allows the financing to happen.

好的。弗莱尔说我们正在寻求这种保障机制，这种能让融资得以进行的担保。这确实能降低融资成本，同时提高贷款价值比。也就是在股权部分之上你能承担的债务额度。记者询问关于芯片投资的某种保障机制时，她直接表示确实如此。

Okay. So so Friar says we're we're looking for this backstop, this guarantee that allows the financing to happen. That can really drop the cost of the financing but also increases the loan to value. So the amount of debt that you can take on top of an equity portion. And so the reporter asks of some form of of backstop for chip investment and she says exactly.

兰詹，我们先从逻辑说起。对于像OpenAI这样近期几乎无往不利的公司来说，这非常合理。基本上它碰什么就能得到什么，想要什么就能得到什么。为什么不要求联邦政府为你们的贷款提供担保呢？

Ranjan, let's just start with the logic here. It's pretty logical for a company like OpenAI, which has not been turned down by anything really lately. Basically everything it touches, it gets. Everything it wants, it gets. Why not ask the federal government to guarantee your loans?

不过我认为，它试图获得的一切实际上都是在承诺投入资金，而非索取资金。别忘了——它显然已经筹集了天文数字的资金。但最近几周每个重大公告都是'我们将与亚马逊共同投入380亿美元'、'与甲骨文合作1000亿美元'等等。所以这次的性质非常不同。我们肯定会深入探讨其含义，但我确实感受到某种...不想说是傲慢。

Well, I think though everything that it's been trying to get is actually it committing to spending money, not asking for money. Remember, all of the I mean, it's obviously raised an ungodly amount of money. But in the last few weeks, every big announcement is we will spend $38,000,000,000 with Amazon, a $100,000,000,000 with Oracle, and everything else. So this is a very, very different ask, I think. I mean, we'll we'll we'll definitely get into what it means, but but I do agree that there's a certain kind of, I don't wanna say arrogance.

这感觉就像是他们可以随心所欲地抛出任何天文数字，而且没关系，因为到目前为止这套做法对他们非常有效。

It's almost just it feels like they just can say whatever they want in whatever big numbers they want, and it's okay because it's it's worked out pretty well for them so far.

是的。我更大的观点是OpenAI正经历着百年一遇的爆发期。融资交易接二连三，创造了史上增长最快的产品。所以当你考虑能做什么时，为什么不尝试让联邦政府担保贷款呢？毕竟可以将其定位为国家安全问题。

Yeah. I think my broader point is that OpenAI has just been on this, like, run of a century. Right? Just like funding deals left and right, the fastest growing product in history. And so when you think about what you can do, why not ask whether the federal government can guarantee your loans for instance because you can position it as a national security thing.

这是弗莱尔在《华尔街日报》活动上的原话：'我们看到美国政府尤其展现出非凡的前瞻性，真正理解人工智能几乎（不是几乎）就是国家战略资产，在思考与中国的竞争时，我们必须审慎考虑——是否采取了所有正确措施来尽可能快速发展AI生态系统？'她公开表态了，毫不含糊。

Here's Freyer and these are the exact comments that she gave at this event from the Wall Street Journal. She goes, I think we're seeing that and I think the US government in particular has been incredibly forward leading, has really understood that AI has an almost not almost as a national strategic asset and that we really need to be thoughtful when we think about competition with, for example, China. Are we doing all the right things to grow our AI ecosystem as fast as possible? So she's coming out and saying it. There's no ambiguity about it.

她明确要求保障机制，并将其定位为美国的国家安全问题。

She's clearly asking for a backstop and she's positioning it as a national security situation for The United States.

这正是难点所在。我打算站在OpenAI这边来看这个问题。假设这是我们未来三十到五十年将面临的最重大的国家安全战役——各方政治光谱都讨论过美国在关键基础设施上已落后的观点。既然如此，如果我们认同AI将成为下个世纪的主战场这个叙事，那么他们要求联邦政府为所有债务融资提供担保或兜底，这看起来合理吗？

This is where it is difficult. I'm gonna try to take OpenAI's side in this. Like, let's say, this is the single biggest national security battle that we're gonna face over the next thirty to fifty years At that point and we've all talked about, I think, on all sides of the political spectrum, the idea that America has fallen behind in in critical infrastructure. So if that is the case and we are buying into this story that AI is gonna be the battleground of the next century, does it seem okay that they're asking for federal guarantees or a backstop in terms of all the debt financing that's being taken out?

我想这正是我的观点。OpenAI提出请求完全合理，但我认为美国纳税人不应为该公司债务兜底——Sam Altman在后续说明中也澄清了这点：即便他们失败，还有谷歌，还有Anthropic。

I mean, I think that's my point here. I think it's person it's perfectly reasonable for OpenAI to ask. I don't think that The US taxpayers should be backstopping the company's debt though because and Sam Altman in a follow-up made this clear. If they fail, you'll still have Google. You'll still have Anthropic.

还会有许多其他公司在开发这项技术。换言之，我认为美国仍将处于有利地位。同时我也不认为政府应该挑选赢家，假设只有OpenAI能获得这些担保而给予特殊待遇。在我看来，这个请求本身合理，而政府拒绝也同样合理。AI事务主管David Sachs已明确表示不会提供救助。

You'll have many others that are gonna be building this. So in in other words, I do think that The United States is is gonna is going to be in a good position. And I also don't really think that the government should be picking winners and giving OpenAI these guarantees assuming that OpenAI would be the only one to get them. So to me, it's just like it's a personal it's a perfectly reasonable ask and it's a perfectly reasonable no from the US government. And David Sachs, the AI czar, basically said no to bailouts.

我依然觉得设立‘AI沙皇’这个职位很有意思。

I still love that we have AI czars. Like, I think

总得有个沙皇嘛。

Gotta have a czar.

必须得有个沙皇啊。

You gotta have a czar.

难道不该有个沙皇吗？

Doesn't have a czar?

不。如果不是沙皇，我指的是特别是在我们政府中不希望出现苏联相关内容的背景下，不过这是题外话。我认为这个问题上...对我来说最有趣的是，当我们讨论这个议题时，在权力走廊里进行长期思考：中国的威胁是什么？我们需要哪些关键基础设施？即便看看拜登的《通胀削减法案》，也有大量资金用于资助关键基础设施，尤其是能源领域。

No. If it's not a czar I mean, especially in the context of not wanting to have Soviet Union references across our government, but side conversation. I think on this I I I think on this so to me, where this is the most interesting is as we're talking about this, the idea, okay, in the, like, halls of power, having discussions, looking at the long term, what is the threat from China? What critical infrastructure do we need? I mean, even if you look at, like, the Biden IRA, there's plenty of money that was being given to fund critical infrastructure, especially across energy.

所以这些讨论在我看来也完全合理。最令我着迷的是过去一周事态的发展——莎拉·弗莱尔究竟说了什么话，然后被她发在领英上，又被萨姆·奥特曼转发推特。他们就像在试图脱身，又反悔，再脱身。这几乎令人震惊，但因为是OpenAI又不算意外。你会以为他们对如此重要的大事会有更连贯的策略。

So these conversations to me seem completely reasonable as well. What's so fascinating to me is how all of this played out just in the last week, what kind of words were actually uttered by Sarah Fryer and then posted on Linked by Sarah Fryer and then tweeted by Sam Altman. Like, they're trying to get out of it and then not get out of it and then get out of it again. Like, that to me is it's almost shocking, but it's not shocking because it's OpenAI. But you would think that they would have a more cohesive strategy on something this important and big.

问题就在这里。我们开场时都说过：他们提出这种要求不算世界最疯狂的事——不是说应该批准（我认为不该批），但作为OpenAI为什么不能提这种请求呢？

And that's that's exactly it. So we both started this show talking about, alright. Like, we're not gonna, you know, say it's the craziest thing in the world for them to ask. I'm not saying to be granted. I don't think they should be granted, but why not ask if you're OpenAI for something like this?

结果互联网炸锅了，基本反应都是'我们不需要美国公民为5000亿美元的公司贷款担保'。这太疯狂了——说得对。接着诡异的事情发生了：OpenAI开始改口，而且不是简单说'经考虑我们不需要担保'。

The internet just blew up and basically it was like you know we do not need the you know citizens of The United States Of America to be guaranteeing the loans for a $500,000,000,000 company. It's crazy. Fair enough. And then the weird thing happened which is that OpenAI's walk back started. And it wasn't just a walk back saying, know, after consideration we don't want these guarantees.

他们几乎是在说'我们根本没说过这话'。看看莎拉·弗莱尔在CNBC的标题：OpenAI首席财务官澄清先前的言论，称公司不寻求政府担保。

It was almost like, hey. We actually never said that. Here's here's this is from Sarah Fryer, CNBC headline. OpenAI CFO Sarah Fryer says company isn't seeking government backstop. Clarifying prior comment.

OpenAI首席财务官莎拉·弗莱尔周三晚间表示，这家AI初创企业不寻求政府为其基础设施承诺提供担保。'我用了担保这个词，结果混淆了观点。'弗莱尔在领英写道，'完整采访能证明，我当时强调的是美国科技实力需要建设真正的工业产能，这需要私营部门和政府各司其职。'萨姆·奥特曼则表示：'我需要澄清几点。'

OpenAI CFO Sarah Fryer said late Wednesday that the artificial intelligence startup is not seeking a government backstop for its infrastructure commitments. I used the word backstop, and it muddied the point, Fryer wrote on LinkedIn. As the full clip of my answer shows, I was making the point that American strength in technology will come from building real industrial capacity which requires the private sector and government playing their part. Sam Altman. I would like to clarify a few things.

首先说明几点。最明显的是：我们既没有也不想要政府对OpenAI数据中心的担保。我们认为政府不该挑选赢家输家，纳税人不该为商业决策失误或市场竞争失败的企业买单。如果一家公司倒下，其他公司会做得更好。

First, a few things. First, the obvious one. We do not have or want government guarantees for OpenAI data centers. We believe that governments should not pick winners and losers and that taxpayers should not bail out companies that make bad business decisions or otherwise lose in the market. If one company fails, other companies will do good work.

这真是奇怪。他们现在反悔了，但这正是他们当初明确要求的。

This is the weird thing. They are going back on it, but that's exactly what they were asking for.

是啊。我觉得现在这种情况总是让人沮丧，Sarah Fryer在LinkedIn上发言，说这混淆了重点。我回答的完整片段显示他们几乎是在断章取义，而在这个案例中，她的表态非常明确。记者甚至让她确认，她也确实确认了，还再次提到了'兜底'这个词。所以我觉得这有点不诚实，效果也很差。

Yeah. I think it's so it's always so frustrating now where where Sarah Fryer, of course, going on LinkedIn and saying, like, it muddied the point. My the full clip of my answer shows trying to almost make it seem like pulled out of context when I mean, in this case, it was so explicit. The reporter even asked her to confirm it, and she confirmed it and said the word backstop again. So, like, I think it's just already, that's just a little bit disingenuous and just and does not land well.

但我认为这件事现在如此引人注目，是因为它紧接着上周Brad Gerstner对Sam Altman的采访。当时他问到，你们只赚了130亿，却承诺要投入1.4万亿开支，这要怎么实现？Sam变得很防备。然后突然就有消息泄露说他们预计年化收入能达到200亿ARR（不一定实际收入）。但根本上这些数字从来就没算清楚过。

But I think the reason this is so salient right now is I think because it comes on the heels of think it was last week, the Brad Gerstner Altimeter interview where Sam Altman, where he asked him around, you know, you're only making 13,000,000,000, but you're committed to 1,400,000,000,000.0 in spend. How are you gonna do this? And Sam got very defensive. And then suddenly, we get start to get leaks that they're seeing $20,000,000,000 annualized rev at ARR, not necessarily actual revenue. But, basically, the numbers have never added up.

我们详细讨论过这个问题。实际上没有人真正搞清楚那1.4万亿开支到底如何合理化。所以我觉得这次更真实，因为它带着情绪——他们几乎肯定需要政府兜底才能实现这个计划。这不是像2008年那种没人能看懂的不透明金融交易，需要博士们才能解开那些复杂的次级抵押贷款和信用违约互换。这些数字很清晰，但就是说不通。

Like, we've talked about this at length. No one actually has really mapped out a genuine understanding of how that 1,400,000,000,000.0 in spend actually is justified. So it I think it's more real because it feels more emotional almost because they almost will certainly need this government backstop to actually make this work. Like, it's not just, you know, this is some opaque financing deal that no one understands, like 2008, like, where, you know, only PhDs are able to kind of, you know, like, untangle complex some prime mortgages and see credit default swaps. Like, they're pretty clear numbers, and they don't make much sense.

所以当你说政府融资和兜底时，感觉就真实多了。我想这就是为什么这次反响这么强烈。

So when you say government financing and backstop, it feels a lot more real. I think that's why this is hitting so hard.

完全正确。我本周早些时候在CNBC讨论时就说过，OpenAI将成为市值万亿的上市公司。它现在已经估值5000亿了，其高管在回答这类问题时需要更谨慎。无论是Sam在Brad Gerstner采访中的表现，还是现在Sarah Fryer的言论。你说得对，人们震惊的原因未必是要求本身（也许就是），但...

That's exactly right. And on CNBC, I was on CNBC talking about this earlier in the week, and I said, basically, look, OpenAI is gonna be a trillion dollar IPO company. It's already $500,000,000,000, and its executives need to learn to speak with a little bit more discipline when it comes to questions like this. Questions like Sam got at the Brad Gerstner interview and then subsequently now, it can add on what Sarah Fryer said. And I I think that you're right that this the reason why people freaked out here wasn't necessarily because of the ask itself or maybe it was.

我们还会深入分析Fryer在会议上的其他发言，因为她的谈话还有很多值得探讨的精彩内容，不光是这一点。但关键在于，基本上整个美股市场，甚至全球股市某种程度上都在指望OpenAI兑现它对英伟达、AMD、甲骨文和微软等公司的承诺。这种依赖性存在，而且我们都知道——听众们也清楚——如果OpenAI的账目有问题，会产生连锁反应。所以人们才会突然警觉起来。

And we're gonna go into some more things about what what Fryer actually said at this conference because there's some more really fascinating aspects of her talk that we should cover and not just this one thing. But it's the fact that basically the entire US stock market, the entire global stock market to some degree is counting on OpenAI to one execute on the promises that it's made to companies like NVIDIA and AMD and Oracle and Microsoft to some degree. So there's that count there's that dependency there. And then if we all we both know and we I think we all know, everybody listening to this, that if there's weird stuff going on in OpenAI's books, that's gonna cascade. And so that's why people are like, wait a second.

综合来看，这不像是一家成熟到能让所有人信赖的公司会发表的言论。

This does not sound all taken together like the remarks of a company that's mature enough to be able to have everybody counting on it.

不，这个观点很好，因为我其实在想，比如Sarah Fryer在高盛工作了十多年。她曾担任多家公司的CFO，比如Square，Nextdoor的CEO。她曾管理过上市公司，在最严谨的投资银行工作过。

No. That's a that's a good point because I was actually thinking about, like I mean, Sarah Fryer spent over a decade at Goldman Sachs. She was the CFO of multiple, I think, yes, Square, Next or CEO of Nextdoor. Like, she has been c suite of publicly traded companies. She's well, she was at the most buttoned up investment bank in existence.

你会以为几十年的经验让她对信息传达的纪律性已经深入骨髓。但到了OpenAI，突然就看到这种失言或说出真实想法后又试图收回，不得不在LinkedIn上发文的情况。同时CEO还在推特上发各种内容。而背景中，我们刚发现他们向科技政策办公室提交了一封信，其中专门提到要通过降低美国制造业扩张风险来对抗中国，为制造商提供快速扩大生产所需的资金确定性，联邦政府还应提供拨款、成本分摊协议、贷款或贷款担保来扩大工业基础产能。

Like, you would think decades and multiple decades of just, like, knowing how to be disciplined around messaging would just be so kind of deeply ingrained in her. Yet you go to OpenAI, and suddenly, you see how this, like, misspeaking or potentially actually saying what you're thinking, but then trying to walk it back, having to post on LinkedIn. Meanwhile, your CEO is tweeting out all other sorts of things. But then in the background, we I had just come across that there was a letter that was submitted to the office of science and technology where, again, they have an entire section where they talk about, like, the need to counter the PRC, People's Republic Of China, by derisking US manufacturing expansion, provide manufacturers with the certainty in capital they need to scale production quickly. The federal government should also deploy grants, cost sharing agreements, loans, or loan guarantees to expand industrial base capacity.

从政策角度看，他们确实在推进这件事。正如我们所说，这至少是个值得讨论的议题。但联想到过去Ilya Sam的斗争，她加入OpenAI后整个沟通策略就变得像埃隆那种风格，完全陷入OpenAI式的沟通混乱，这很让人失望。

So from a policy perspective, they are actually pursuing this. And, again, as we are saying, that's not the most ridiculous, unreasonable thing to at least have a conversation around. But, like and then we're gonna talk about the Ilya Sam battles of the past, but it's almost like she she shows up at OpenAI, and suddenly, her entire comm strategy is, like, Elon situation or or just kind of this kind of like, you know, totally chaotic OpenAI comm situation. It brings you down.

问题不在于要求本身，而在于这种糟糕的反复。更讽刺的是紧接着Sam对Brad Gerstner说'够了'——当被问及如何用130亿收入支撑1.4万亿投资时。我不想当纪律警察，也不希望变成那样。

That's the thing. It's not the ask itself. It's the bad going back and forth, and then it comes on the back comes on the back of this comment where Sam tells Brad Gerstner enough when Gerstner asked how are you going to fund $1,400,000,000,000 in investments with $13,000,000,000 in revenue. I don't want to be the disciplined police. I certainly don't hope to be that way.

但想想股市和经济有多大程度依赖这件事的成功，你自然希望看到更专业的表现。现在我们应该谈谈Friar的其他评论。

But again, if you think about just the magnitude of the you know how much of the stock market, how much of the economy is depending on this to work out, you wanna see something more buttoned up than that. Now I do think we should get to some of the other comments that Friar made

等等，我倒觉得在这种情境下当纪律警察挺好。得了吧，我不认为这是什么坏事。

at the moment. Hold I would like that we are the disciplined police in this situation. Come on. I I wanna I don't think you should say, like, that's a bad thing. Come on.

对于那些达到一定规模的企业来说，我们能否回归基本的沟通纪律？我在这个领域工作多年，即便是营收数亿的公司，那时我们也面临诸多焦虑和监督。虽然处理这些并不总是愉快的体验，但每个层级都需遵守。比如Snowflake的首席营收官在TikTok上不慎谈及营收预期，随后立即提交了8-K文件。虽然是个失误，但他们仍在努力遵守规则。

The for the companies that are value once you cross a certain threshold, just can we go back to having a little bit of communications discipline? I have worked in this for many years, even in companies in the hundreds of millions of revenue, and that are you know, like, even then, we had a lot of hand wringing and oversight. And and I'm not gonna say it was necessarily always the most pleasurable thing to have to deal with, but, like, at every other level, every you know, did you see the Snowflake chief revenue officer, I believe, who had was speaking to a TikToker and accidentally, like, said something about revenue guidance. And then they've and they filed an eight k. They still are trying to it was a screw up, but they're still trying to play by the rules.

他们明白遵守规则的重要性，承认失误并试图修正。而有些公司却公然嘲弄美国金融市场的纪律体系。因此我主张需要更严格的沟通监管，请大家发言时多些谨慎。

And they're, like, that this is actually, you know, like, an important thing that actually to play by the rules, we have to still it was a screw up, and we're gonna try to fix that versus companies like this that are just completely the entire, you know, like, discipline of The US financial markets just kind of making a mockery of. So so I'm gonna say we need more disciplined police around comms. Please just be a little thoughtful when you're speaking.

这就是为什么我们需要人形机器人。这样Ranjan就能大规模进入你家客厅，如果你不守规矩就开始掀桌子。

This is why we need humanoid robots. So Ranjan at scale can go into your living room and start flipping tables over if you're undisciplined.

如果我的人形机器人大军能帮我拿到万亿薪酬方案——实际上薪酬条款里确实提到了人形机器人大军——它们现在就该在掀桌子了。

If I had my humanoid robot army and, it got me my trillion dollar pay package because, actually, there's a humanoid robot army in that actual language in the pay package. Yeah. That'd be flipping tables right now.

等等，他们是私人公司重要吗？作为私企，这类事务本就不需要向SEC申报。

Well well, hang on. I mean, does it matter that they're private? They're a private company. Like, they don't need to file with the the SEC for these type of things.

不，不需要。但至少保持些体面

No. No. But but a little grace so

你能逐步适应这些规矩吗？你来告诉我。

you can ramp up up to this stuff? You tell me.

但是你看，这正是整个私人估值膨胀问题变得更加棘手的地方。这本来是个更广泛的议题，但正如你所说，这使公司变得对我们如此关键，就像OpenAI现在可能正处于金字塔底部，支撑着整个美国经济。而因为他们无需经历任何严格审查——即使是营收数亿美元的上市公司也必须经历的——这才导致我们陷入此类境地。这确实有点可怕，一方面CEO会因为被问及本应早有明确答案的简单问题而发怒，另一方面拥有数十年经验的CFO却似乎在沟通方式上倒退，表现得像个初出茅庐的财务主管。

But, see, this is where the whole, like, you know, ballooning of private valuations, I do think becomes even more problematic. This is a whole much larger rant in general, but, like, it's because it allows the company to become this critical to our over as you said, like, OpenAI might be the, like, entire kind of, like, at the bottom of the pyramid holding up the entire US economy right now. And to because they have not had to go through any kind of rigor that even, like, a public company with a couple of 100,000,000 in revenue would have to go through, that's how we're ending up in the these kind of situations. And it's, I mean, it's a little bit scary that, like, you know, on one hand, you have your CEO getting mad about just being asked a very simple basic question that he should have had a very clear answer to. And on the other hand, you have a CFO who's had decades of training on this and suddenly seems to be backsliding into, you know, a way of communicating that, you know, you would kind of attribute to a first time CFO.

这个观点很棒，完全可以写成文章。这段评论非常精彩，我很赞同。好吧，我确实想聊聊弗莱尔提到的其他事情。

That's that's a good that should be an article. That's a good good opinion piece right there. That's I like that. Okay. So I do wanna talk about some of the other things that Fryer said.

但在那之前，让我先问个更重要的问题：你如何看待我们正在经历的反弹？显然OpenAI因为弗莱尔的言论遭遇反弹而改弦更张。你认为这种反弹是否标志着OpenAI在财务上终于触碰到天花板？就像最终它会遭遇阻力那样。

But before we get there, let me just ask you this one bigger question, which is that I'm curious. Do you think the backlash that we're seeing here? I mean, because obviously OpenAI backtracked because of the backlash to Friar's comments. Do you think the backlash is finally a sign that OpenAI is hitting the ceiling of what it can get financially? Like, eventually, it's gonna push, and there will be pushback.

这是否标志着其能力已达巅峰？

Is this sort of a sign of the top of its of its abilities?

是的，我认为不仅如此。这更是它在当前国家乃至全球对话中地位的绝佳指标。我们长期讨论它，听众和社交圈也持续关注，但他们正在主动介入——不仅成为美国经济的关键部分，更是家喻户晓的存在。

Yeah. I think it's it's not just that. It's also, like, a really good indicator of where it is in the national conversation or the global conversation right now. Like, we have talked about it for a long time. Our listeners and our friends and networks have been talking about it, but they are inserting themselves, and they they are becoming not just like an economically critical part of The US economy, but also just known by every person in America.

所以我觉得这表明他们存在品牌问题，且将持续存在。当他们在二级市场以5000亿乃至10000亿美元估值交易时，已有人因此暴富，这让普通人听到这类消息时更加难以接受。

So I I think it's a sign that they have a branding problem. They're gonna continue to have a branding problem. And so the idea that, like, they are getting a backstop when, you know, they've had secondary sales of I think they had a secondary sale at the $500 billion $100,000,000,000 valuation. So, like, already people are getting insanely rich off this. So it just makes it that much more unpalatable for any normal person to to have to hear this kind of thing.

有趣的是他们的产品确实广受喜爱，但就像我们上周讨论的，他们在公关方面正步入类似Facebook的困境。谁都不想被视作邪恶科技巨头——虽然他们还没到那地步，人们依然热爱ChatGPT。对我而言，它已是日常使用中最实用的科技产品之一。

It is interesting because their product really is beloved, but they are starting to, again, speaking of our discussion last week, run into some, like, Facebook territory on the comms side. And what you don't wanna be is, like, viewed as, like, big bad tech. So I don't think they're there. I think people really do love ChatGPT. To me it's already one of if not the most useful tech products I use on a day to day basis.

也许iPhone就在那正上方。不过距离相当近。而且我不确定。我认为这对他们来说是个值得关注的问题。

Maybe the iPhone is is right above that. But it's pretty close. And I don't know. I I think that this is a concern for them.

是的，同意。而且，我认为这是个很好的观点，即人们热爱产品与热爱公司之间的区别。实际上，Facebook（现Meta）已经很好地证明了两者可以兼得并取得成功。所以如果你的产品能保持那种粘性和令人上瘾的特性。

Yep. Agreed. And and, yeah, I think it's a good point that that that the distinction between, like, people love the product versus love the company. Facebook actually shown meta shown pretty well that you can do both and succeed. So if your product remains that sticky and, like, addictive.

所以从产品角度来看，我并不认为这对他们来说真的构成困扰，但我不确定。我觉得某种程度上，当越来越多人开始关注OpenAI长期以来的运营方式时——不仅仅是我们和其他人在讨论——这确实会开始以一种不同的视角来审视它。

So so I think it doesn't mean from a product standpoint that it's it's genuinely troublesome for them, but I don't know. I I feel at a certain point, sir like, the more people kind of pay attention to how OpenAI has been run for this long, and it's not just us and others talking about it, I think it really starts to kind of bring a different kind of light shining on it.

现在让我们来看看Fryer的这些评论，这非常引人入胜，它让你觉得这就是他们如何为那1.4万亿美元投资买单的方式。这是我听过的最佳解释。如果他们能实现这一愿景，那么OpenAI可能会成为有史以来最有价值的公司。以下是Friar所说的最后两部分内容。

Now I do now let's get to these comments by Fryer because this was fascinating, and it sort of makes you think this is how they're gonna pay for that 1,400,000,000,000.0 in investments. It's the best explanation I've heard yet. And if they're able to pull off this vision, then maybe OpenAI will be the most valuable company ever, ever. And so this is sort of what Friar said. I'm just gonna get to the last two parts of what she said.

她说他们将——这是《华盛顿邮报》记者Garrett Devink的报道——他们将进行创新的商业合作。他们不会仅仅按token计价向制药公司出售访问权限，而是要要求分享制药公司通过使用ChatGPT开发药物所获利润的一部分。这同样适用于商业领域。

She said they will this is from Garrett Devink who is a Washington Post reporter. They will do creative commercial deals. They're not just going to sell access to a pharma company on a per token basis. They're going to demand a rev share of the profit the pharma company brings in from developing drugs using ChatGPT. This applies to commerce too.

他们想在发现和交易环节都分一杯羹——当有人通过ChatGPT搜索商品时。这已经与沃尔玛、Etsy和Shopify展开合作，全都在开发者日上宣布了。这非常有趣。你对这些变现策略怎么看？

They wanna take a cut of both the discovery and the transaction when someone searches for a product using ChatGPT. This is already happening with Walmart, Etsy, Shopify, all all announced on dev on their dev day. That is very interesting. So what do you think about these monetization strategies?

知道吗？制药领域对我来说特别有趣，因为商业部分对平台来说是常规操作。我是说，这就是亚马逊整个第三方零售市场的模式。大家都接受‘在我的平台销售就要抽成’的规则。但‘我们的技术能帮制药公司更快研发药物，因此要参与分成’这个想法...

You know what? I the pharma thing is very interesting to me because the commerce side of it is kind of par for the course from a platform. I mean, that's Amazon's entire retail business, I mean, with the on the third party marketplace. But everyone accepts if you sell on my platform, I will take a cut of the transaction. But the idea that, like, our technology helps the a pharma company create drugs in a much faster way, we would wanna take a rev share.

我觉得这真的很有趣。你能想象这还能应用到多少其他领域，但到那时，他们是否也在承担风险？他们提供免费计算资源以换取长期收入分成，这实际上给业务增添了惊人的额外风险层。我在想，那会是什么样子呢？

I I think that's that's really interesting. And you can imagine how many other ways that becomes applied, but then also at that point, are they taking on the risk as well? And then, like, they're giving away free compute in order to get that longer term rev share, which just adds actually an amazing whole additional layer of risk to the business. Like, I wonder what like, what could that look like?

其实我第一次看到这个时觉得酷多了。但仔细想想，假设OpenAI开发出医学超级智能，让制药公司能在人类尺度上做到前所未有的事。别人也会开发出超级智能，然后就会陷入价格战。你能坚持多久说‘我要分享你用我们技术开发的药物的利润’，而别人可能说‘这是平台，只需支付授权费’。

So I actually thought this was much cooler when I saw it for the first time. And then when I think about it, it's like, all right, so let's just go crazy here. Let's say OpenAI develops medical superintelligence that enables pharma companies to do things they never could before at human scale. Somebody else is just going to develop super intelligence as well and they're going to get into a price war. So how long can you say I want to party the profit of this drug that you're going to develop with our technology where somebody else can be like well here's the platform just pay us the licensing fee.

对吧？这种定价模式只有在你独家提供时才行得通。我只是不认为他们会是唯一的一个。

Right? That only that pricing model only works if you're the only one that can offer that. I just don't think they're gonna be.

不，不是这样。但我从不同角度看这个问题——不是从竞争角度，而是回到‘你们到底做什么业务’这个根本。当他们筹备万亿级IPO，我们查看S-1文件时，理性投资者应该试图理解他们的业务本质。我们详细讨论过，其实我们连生成式AI业务的经济模型都还不清楚。

Well, no. No. But I see, I would think about it differently because not from the competitive standpoint, but from the, again, going back to what business are you in. When we're when when they're looking at their trillion dollar IPO and we're going through their s one, you know, the way any investor should rationally look at these things is just try to understand what business are they in. We've talked about this at length that, you know, like, already, we don't know the economics of what a generative AI business would be.

这不是传统软件，因为使用它确实有增量成本——使用越多计算资源，提供商成本越高。但消费者订阅业务、API代币收入，这是相当清晰的商业模式。如果定位是AI云业务、消费设备业务（Sam Altman在Gerstner访谈中提到过），现在又说我们是制药业务，因为要承担药物开发风险并参与实际经济分成...

It's not traditional software because there's actually, you know, like, incremental cost to the utilization of it. The more compute you use, the more expensive it is for the provider. But still, subscription business for consumers, API token revenue, that's a pretty straightforward business, and I think that could make sense. If it is, we are an AI cloud business, a consumer devices business, which Sam Altman said as part of that Gerstner interview. And now we're also a drug a pharma business because we're gonna be taking on some kind of risk around drug drug development in the actual underlying economics of our business.

天啊，这可真是...我尊重有创意的合同，但这未免太夸张了。

Like, I mean, my god. That that that's quite something. I I I like I I respect a good creative contract, but, like, that is that's too much.

如果他们能做到，那确实厉害。而且我认为他们成功的概率不是零。但问题在于，你不是唯一开发这个的。如果是唯一开发者，当然可以要求收入分成。

If they can pull it off, I mean, good for them. And I don't think it's a non 0% chance that they can that they can pull it off. But, again, the question is, you're not the only one developing this. If you were the only one developing it, fine. You could get a percentage of the revenue.

这比我想象中他们以为的要困难得多。好吧，现在让我们快速讨论一下政府基础设施这方面。Sam Altman说，我们认为政府建设和拥有自己的人工智能基础设施可能是有意义的。我们讨论过的一个领域是贷款担保，作为支持美国半导体工厂建设的一部分，我们和其他公司响应了政府的号召，我们很乐意提供帮助。

It's just gonna be harder than I think they think to do that. Now, okay. Let's just go quickly and talk about this government infrastructure side of things. So Sam Altman says, what we do think might make sense is governments building and owning their own AI infrastructure. And one area where we have discussed loan guarantees is part of supporting the build out of semiconductor fabs in The US where we and other companies have responded to the government's call and where we we are we would be happy to help.

你对政府拥有的人工智能基础设施这个想法怎么看？这合理吗？我们不必过多讨论芯片的事情，因为已经谈过了。但关于政府AI基础设施的想法，显然很多政府都在谈论主权AI。Ranjan，我很好奇你对这个的看法。

What do you think about the idea of a of a government AI, government owned AI infrastructure. Does that make sense? And we can we don't really have to talk too much about this chip thing because that's been done. But I think the idea of government AI infrastructure, obviously, lots of governments are talking about sovereign AI. I'm curious what you think about that, Ranjan.

我是说，如果我们说——我记得好像是Sundar说过——这比电力和火还重要。如果我们有国有化的公用事业基础设施，虽然它还没有完全国有化，但更多是公私合营的基础设施，那么如果这真的会成为整个经济的支柱，也许这是有道理的。但如果是这样，我们现在就应该以此为目标来建设，而不是让私人先赚取巨额利润，再由政府兜底。如果真有人这么认真，他们应该放弃自己公司的经济利益来支持公共事业，但目前没人这么做，这部分让我觉得不真诚。

I mean, it if we're saying I think was it Sundar was, like, it was more important than electricity. Like, if we have and fire and fire. If we if we have, like, nationalized in for u utility infrastructure, though it's still not fully nationalized in that way, but still, like, you know, like, much more public private types of infrastructure in that way, sure. Maybe it makes sense if we're really saying this is gonna be the backbone of the entire economy. But, again, if that's truly the case, we should be building that today with that in mind and not where private citizens can actually just make ungodly amounts of money before we even get there and then be backstopped by the government, then it should be like if that was really the case and there were anyone this is the part that feels disingenuous again to me because, like, if Sam, if Jensen and other comments, like, if anyone was really serious about this, you would be giving up economics within your own firm for the public side of things, but no one's doing that yet.

没错，这是个很好的提醒。但我支持这个想法，去建设政府的人工智能基础设施，因为万一这东西真的发展成AGI——我想我们都同意有这个可能性。我不知道概率有多大，但它在三年内的进步速度确实惊人。

That's right. That's a very good caveat. But I I would say I'm for it. Go out and build the government AI infrastructure because in case this thing does become I think we'd both agree there's a chance that this thing can reach AGI. I don't know how high of a chance there is or the amount that it's advanced in three years is actually insane.

比如我现在用ChatGPT，它在很多搜索上都很可靠。他们已经解决了很多问题，比如幻觉、可靠性、能力，比如数学和科学方面的辅助能力。鉴于我们目前看到的进展，我认为政府应该投资。虽然我不完全清楚政府具体要怎么用它。

Like, I'm now use using ChatGPT and it is, reliable on many, many searches. Like, they've built, they've answered a lot of the questions like hallucinations, reliability, capabilities, things like math and science or assisting with these capabilities. It can do that. Given So the progress we've seen so far, I think a government should invest in it. I don't fully know what a government does with it.

不，但问题是那样会是什么样子？如果你经营一家公司，你会收到美国政府关于API使用量的账单吗？我无法想象硅谷或AI领域的任何人会支持这种做法。那到底会是什么样子？

Well, no. But but that that's like that that's what does it look like in that case? If you think about it, like, is if you're running a company, do you get a bill from the US government for your API consumption? Like, I cannot imagine any of the folks involved in Silicon Valley or AI are gonna be advocating for that. But but so what does it look like?

还是说政府只是为你的贷款兜底，正面你赢反面他们输？

Or is it just the government backstops your loans and heads does it heads you in tails they lose?

我的意思是，归根结底，我认为政府希望的是国家强大。这可以通过几种方式实现。首先，政府可以自主开发AI技术，构建基础模型，以低成本供国内所有人使用，从而激发创新。这是外部应用。另一方面，如果政府高效运作，最终会造就一个更强大的国家。

I mean, ultimately, I think what a government wants is for its country to be strong. And so that can play out in a couple of ways. First of all, government can develop AI technology on its own, build a foundational model, and make that available for anybody in the country to use at a low cost and spark innovation from there. That's like external use. And the other side of it is if you're an effective and efficient government, you're going to end up creating a stronger country.

我们都知道我们的政府效率不高。Doge显然不是提高效率的解决方案。但你是否想构建非常强大的AI，然后用它来整合国内所有分散的数据——假设你能以符合隐私要求的方式实现（这是个很大的假设），再利用这些数据做出更好的决策，从而更高效地管理国家。如果能做到，那将极具价值。

And we know that we don't have an efficient government. Doge obviously wasn't the answer to make it efficient. But do you want to build very powerful AI and then use it to sort of connect all the disparate data that you have within the country assuming you can do it in a way that's privacy compliant which is a big if. And then use that to make better decisions and then run your country better. If you can do that, then then it would be very valuable.

所以你是在倡导USGPT或某种基础模型？我...我觉得还行。可能吧。我有点被说动了。

So are you advocating for USGPT or some kind of the foundation model? I I'm I'm okay. Maybe. I'm I'm feeling it a little bit.

USGPT，搞起来

USGPT, build it

把大语言模型国有化。

Nationalize the LLMs.

其实可以让私营部门和公共部门共同推进。但考虑到我们目前的处境，为什么不呢？我的意思是，这对预算不会造成巨大负担。我认为美国政府应该这样做。你觉得...

Well, you could have you could have the private sector and the public sector both doing it. But I I think given where we are today, why not? I mean, it's not it's not gonna be a massive portion of your of your budget to be able to do stuff like this. And I think the US government should. Now what do you think about

等等，你是在竞选吗？这是2028年的竞选纲领吗？不。不过既然要参选，咱们就组队吧。

Do I are you running? Is this 2028 platform pitch? No. Think but we're running. Let's join ticket.

加入选票。

Join ticket.

加入选票。

Join ticket.

这是我们的平台。USGPT。

This is the platform. USGPT.

我将成为你的副总统竞选搭档。不管最后是谁参选。CJD。JD·万斯。对吧？

I'll be your vice president who can run against. I don't know whoever comes up. CJD. JD Vance. Right?

JD·万斯会坚定支持美国的立场。

JD Vance would be pro pro US US.

不。我们要将大语言模型国有化。我们是USGPT团队。这是人民的意愿。

No. It's nationalizing the LLMs. We're we're team USGPT. The people want it.

如果我们能在埃隆的新第三党派下参选。我想我们绝对是他要找的人选。但国与国之间存在这场较量，对吧？周三英伟达的黄仁勋在事态发展中表示，正如我一直强调的，中国在AI领域仅落后美国纳秒级。关键在于美国必须加速前进并赢得全球开发者的支持。

If we can run under Elon's new third party. I think we would definitely be the ones he's looking for. But there is this battle between countries, right? Jensen Huang from Nvidia on Wednesday, as this is all going down says, as I have long said, China is nanoseconds behind America in AI. It is vital it is vital that America wins by racing ahead and winning developers worldwide.

我是说，这个人想把芯片卖到中国。我不明白如果反对出口限制，美国怎么能保持领先。你觉得这是怎么回事？一个国家在AI上比另一个国家更先进有价值吗？很可能有。

I mean, the man wants to sell his chips into China. I don't know how that makes The US, you know, stay ahead if you're anti export restrictions. What what do you think is going on here? And and is there a value in in one country having more advanced AI than another? And probably, yes.

但你怎么看？

But what do you think?

我不知道是不是因为这周新闻主题是社会主义让我特别敏感。我们不得不听这些内容。但即使是这样的说法，总是说中国在AI上只比美国落后几纳秒。他用纳秒这个词是不是为了听起来更专业更聪明，而不是直接说中国与美国持平或快赶上了？但感觉可能只是我多心。

I don't know if I'm overly salty this week just given socialism as the topic of the news. Meanwhile, we're having to hear all these kind of things. But, like, even statements like this, it's always kind of just China is nanoseconds behind America in AI. Like, is he using nanoseconds just to sound more technical and smarter versus either just say China is tied with America or almost catching up. But I feel like it almost is just me.

他这么说就是为了让听起来更有科技感。不过这更多只是个小批评。更重要的是如你所说，黄仁勋并没有表现出...我的意思是他极力游说要把芯片卖到中国。如果这真是个问题且你真在乎，就不会这么做了。

It's trying to make it he's saying it to make it sound more techy. But that's more of just a, you know, a slight criticism. I think more importantly, as you said, Jensen has not shown I mean, he he's lobbied hard to sell chips into China. So if this was really an issue and you really cared about it, you would not be doing that.

好吧。让我们用一些泡沫话题结束这段讨论。X平台用户Edward Dowd说：我发现一个模式。Altman和黄仁勋嗅到了泡沫终结的气息，融资即将枯竭，准备以国家安全为由向政府要纳税人的钱。

Alright. Let's close this segment out with, of course, some bubble talk. This user on x said, Edward Dowd. It says, I see a pattern. Altman and Jensen smell the end of the bubble, financing drying up, and are going to ask daddy for taxpayer money citing national security issues.

你的解读是？

Your read.

我觉得...这就像你开场说的一样，现在情况似乎就是这样，按照目前运作方式这仍是个烧钱的无底洞。如果...如果需要持续资金注入而私人资本终将枯竭，那不如寻找下一波资金来源。

I think it feels like I mean, that's how you started this segment, and it feels like that is kind of where things are, that there's this it's still a money pit in the way things are working currently. And if it's if if you need to continue to have money pouring in and private capital is gonna dry up at some point, you might as well find the next, the next wave of money.

不可避免地，我的意思是，你能求助的资金来源就那么几个，最后只能找政府。我只是觉得我们还没到那一步。那个阶段可能还要等几年。我认为我们还没到那个程度。

Inevitably, I mean, you there's only a certain number of sources that you can go to for money, and then you eventually look for governments. I just don't think we're at that. I think that stage is still a couple years away. I don't think we're there yet.

是啊。我现在更明白为什么我对这个话题如此愤慨了——上次全球金融危机时，我就在金融行业交易大厅工作，亲眼目睹了救助过程。那种第一视角的体验非常令人不安。说实话，这件事一直萦绕在我心头。但华尔街当时至少没有...也许Dick稍微退缩了一点，但华尔街在危机发生前可没像这样公开乞求援助。所以科技行业现在这样真是...拜托，要点脸吧。

Yeah. I I think I'm realizing why I'm even saltier on this whole topic now is having worked in the finance sector in the tray on a trading floor during the last during the global financial crisis when there was bailouts, and it was, you know, like, watching it firsthand was a very problematic thing. So I I genuinely think, like, it's it's always stuck with me in a way, but Wall Street never, like, openly maybe Dick folded a little bit, but, like, Wall Street wasn't, you know, like, begging for this stuff even before anything happened. So that's why it's just come on. Come on, tech industry.

在我们还没到那一步前，别急着要救助和兜底。私下里可以开始铺垫，但别在《华尔街日报》科技直播活动上公开说。这就是我的全部要求。

Don't ask for a bailout and a backstop before we're even we're even there yet. Maybe behind the scenes, start kinda laying the seeds, but don't say it at a Wall Street Journal tech live event. That's all I'm asking.

这绝对创纪录了。OpenAI是第一个提出'大而不能倒'概念的未上市公司吗？很可能。好吧，要说谁能给OpenAI兜底，拿到万亿薪酬方案的埃隆·马斯克或许可以。

It's gotta be a record. Is it the first, nonpublic company to float the idea of being too big to fail? Probably. Okay. Well, one person that might be able to backstop OpenAI is Elon Musk after he got a trillion dollar pay package approved.

当然这笔钱不会马上到账。稍后我们将讨论这个巨额薪酬方案——埃隆是否该拿这笔钱，以及他将如何获得。现在先来关注AGNCY如何塑造企业AI的未来。作为Linux基金会开源项目，AGNCY正在为'智能体互联网'建立可信身份与访问管理标准，这个协作层能确保AI智能体安全发现、连接并跨框架工作。通过AGNCY，企业可获得开放标准化工具和无缝集成方案，包括强大的跨平台身份管理能力，实现识别、认证和交互功能。让您自信部署多智能体系统，与思科、戴尔科技、谷歌云、甲骨文、红帽等75家支持企业共同建立安全可扩展的AI基础设施标准。

Of course, the money won't come right away, but we'll talk about the big pay package, whether it makes sense or not for Elon to get that money, and how he'll get it right after this. Shape the future of enterprise AI with AGNCY, a g n t c y. Now an open source Linux foundation project, AGNCY is leading the way in establishing trusted identity and access management for the Internet of Agents, a collaboration layer that ensures AI agents can securely discover, connect, and work across any framework. With Agency, your organization gains open, standardized tools, and seamless integration, including robust identity management to be able to identify, authenticate, and interact across any platform. Empowering you to deploy multi agent systems with confidence, confidence, join industry leaders like Cisco, Dell Technologies, Google Cloud, Oracle, Red Hat, and 75 plus supporting companies to set the standard for secure, scalable AI infrastructure.

您的企业准备好迎接智能体AI的未来了吗？立即访问agency.org查看应用案例。网址是a-g-n-t-c-y点o-r-g。第一资本的科技团队不仅讨论多智能体AI，他们已经部署了一个。

Is your enterprise ready for the future of agentic AI? Visit agency.org to explore use cases now. That's a g n tcy.0 r g. Capital One's tech team isn't just talking about multi agentic AI. They already deployed one.

这个名为'聊天礼宾'的系统正在简化购车流程。通过自反思层级推理和实时API检查，它不仅帮助买家找到心仪车辆，还能安排试驾、获取贷款预批以及估算旧车置换价值。先进、直观且已投入使用——这就是他们的技术实力。

It's called chat concierge, and it's simplifying car shopping. Using self reflection and layered reasoning with live API checks, it doesn't just help buyers find a car they love. It helps schedule a test drive, get preapproved for financing, and estimate trade in value. Advanced, intuitive, and deployed. That's how they stack.

这是Capital One的科技。欢迎回到大型科技播客，周五特辑。有人这周过得不错，他就是埃隆·马斯克。据《华尔街日报》报道，特斯拉股东批准了他1万亿美元的薪酬方案。

That's technology at Capital One. And we're back here on big technology podcast, Friday edition. Someone's had a good week. His name is Elon Musk. Tesla shareholders, according to Wall Street Journal, approved his $1,000,000,000,000 pay package.

在德州奥斯汀电动车制造商总部，粉蓝灯光映照下，马斯克被仿人机器人环绕舞台，他必须感谢以超过75%投票支持这项万亿薪酬方案的股东群体。'我们即将开启的不仅是特斯拉未来的新篇章，而是全新的一本书。我的意思是——请握紧你的特斯拉股票。'马斯克说道。'你喜欢这个万亿薪酬方案吗？当然，特斯拉得实现些疯狂目标，比如达到8.5万亿美元市值。'

Flanked by dancing humanoid robots on stage bathed in pink and blue light at the electric vehicle makers Austin, Texas quarters headquarters must thank the crowd of shareholders who supported the trillion dollar pay package with more than 75% of the votes cast. We are about to embark upon what we're about to embark upon is not merely a new chapter of the future of Tesla, but a whole new book. I guess what I'm saying is hang on to your Tesla stock, Musk said. Do you like the trillion dollar pay package? I mean, of course, like, Tesla has to, like, hit some crazy goals, which is an $8,500,000,000,000 market cap.

目前市值1.5万亿，也就是公司价值的五倍。你怎么看，Rajan？

It's right now at 1 and a half trillion. So, you know, five x the company. What do you think, Rajan?

我从非科技圈朋友那里看到很多愤怒的帖子，更多涉及政治层面。但说实话，虽然我这期节目一直很尖刻，这次我却不在意——如果你能让特斯拉达到8.5万亿，就拿走那1万亿吧埃隆。比如你要在特斯拉累计销量850万辆时卖出1150万新车，或者打造出不会毁坏公寓的仿人机器人大军——尽管拿走你的万亿。

So I saw from, like, nontech friends, I saw a lot of kinda angry posts about this and that live more in the political realm. But I'm actually gonna say I and I've been plenty salty this episode so far, but this one did not bother me in the sense that if you're getting Tesla to 8 and a half trillion, take a trillion, Elon. Like, if you're gonna sell 11 and a half million new vehicles even when you've only sold eight and a half million vehicles in the lifetime of Tesla, take your trillion. Like, if you create this humanoid robot army and it's remains out of my hands and I'm not destroying people's apartments with it, you know, take your trillion. All of that.

我过去多次写道，我认为问题在于：埃隆和特斯拉本是伟大的汽车公司，但自从他那份与估值完全挂钩的27号薪酬方案后，所有围绕推高股价的骚操作就开始了，培养出这批像信徒般追捧股票的人。如今我们又回到以股价上涨为主要驱动的疯狂目标，这意味着埃隆又会开始无尽循环的疯狂言论。今早他还在推特上说'现在有太多兔子要从帽子里变出来'，声称需要足够多的特斯拉股权来确保他开发的机器人大军不会落入恶人之手。

I I'd written in the past a few times, like, to me, the problem that happens here is I my my grand theory is that, like, Elon and Tesla as a company was a great car company. The moment he had his 27 pay package that was completely aligned to valuation, that's when all the shenanigans around just trying to, like, focus on appreciating the stock, you know, like, just really building this army of just people who are, like, religious about the stock began. So to if we're now that we're moving back to just another just insane goal around, around the actual stock appreciation is a primary driver of this, just reminds me we're gonna go through endless cycles of Elon saying crazy things. Already, I just think he was, like, this morning tweeting, like, so many rabbits to pull out of the hat right now. He said he wants a big enough ownership stake in Tesla to be comfortable that the robot army he was developing did not fall into the wrong hands.

说不定他听了我们上周的播客呢。不过——

So maybe he listened to our podcast episode last week. But

他主要是想用这招让机器人大军远离你Ranjan。尽管他会当总统且组建第三方政党，但他想控制那些机器人。我觉得这交易很公平。

He's he's mostly trying to do this to keep his robot army away from you, Ranjan. Even though he'll be president and Musk's third party, he wants to control those robots. I think that's a fair bargain.

这一点我无法反驳。我尊重这一点。

Which I which I I can't argue with. I respect this.

但我还好。说几点吧。首先，我觉得有趣的是马斯克实际上在说他不再把特斯拉当作汽车公司了。我是说这是我唯一的解读方式，对吧？

But I okay. So a couple of things. First of all, it is interesting to me that this is Musk effectively saying he's done with Tesla as a car company. I mean that's the only way that I read it. Right?

他们还会生产汽车吗？会的。但野心在于机器人出租车和人形机器人。如果你能实现这些，我认为他会物有所值。特别是如果你能让市值达到8.5万亿美元的话。

Like, are they gonna still produce cars? Yes. But the ambition is robotaxis and the ambition is humanoid robots. And if you can pull that off, think it's he'll he'll be worth the money. Especially if you can pull it off in a way that, like, makes the market cap go to 8,500,000,000,000.0.

我不知道。在我看来，机器人出租车和人形机器人比电动车更有未来，即使这些机器人出租车本身是电动车。你怎么看？

I don't know. It seems to me like robotaxis, humanoid robots are much more the future than electric vehicles, even if, you know, maybe they that those robotaxis are EVs. What do you think about this?

是啊。不，不。这是个好观点，我的意思是，这就是长期存在的认知断层——分析师们还在用汽车公司的标准分析特斯拉的每个销售数据，但股价却完全取决于今天还不存在的人形机器人。不过我觉得这个观点其实很中肯。

Yeah. No. No. I it's a good point, which is, like, I mean, this is where the the disconnect is always it's been around for a long time already, but, like like, it's it's almost comical as analysts all try to analyze every, like, Tesla sales number and, like, from a car company standpoint, but then the stock price is just completely dependent on humanoid robots and things that don't exist today. But but I I think that's actually a fair point.

特斯拉根本不再是汽车公司了。埃隆也完全没有假装它是汽车公司。我们都不该再关注实际汽车销量了。那应该是次要的，最不重要的业务部分，大家应该关注更大的机遇。

Like, Tesla is no longer a car company in any way. Elon is not pretending it's a car company in any way. We should all stop looking at actual vehicle sales. That should be an afterthought. That should be, like, the least important part of the overall business, and people should just focus on the bigger opportunities here.

你觉得埃隆说'我们要生产这些人形机器人，如果真这么做，我不希望别人控制它们。我要有监督权，所以我应该持有公司25%股份'这话是认真的吗？我很好奇。

Do you think Elon is sincere when he says, look. We're gonna be producing all these humanoid robots. And if we're doing that, I don't want anybody else to control them. I wanna have oversight and therefore I should have 25% of the company. So I'm curious.

实际上，这是个两部分的问题。第一，你觉得他真诚吗？第二，如果你知道埃隆想控制这支人形机器人军队，那你购买时究竟是在签什么协议？这东西到底是归你还是归他？

Actually, this is a two parter. One, do you think he's sincere? Two, if you know that Elon wants to control this humanoid robot army, what exactly are you signing up for if you buy one? Like, is that yours or is that his?

说得好。你这么一说，确实感觉像是种威胁。简直像对股东群体的敲诈——听着，如果不给我这些，我们正在打造的人形机器人军队可能会落入坏人之手，而只有我能拯救你们。但你们最好在我达成这些里程碑时给我特斯拉25%的股份，否则小心那些机器人。

That's a good point. And actually, now that you repeat that, it is kind of like a it's it's a threat. It's like extortion to the shareholder base where it's like, listen. If you don't give me this, this humanoid robot army we're building could fall into the wrong hands, and only I can save you. But but but you better give me my, 25% of Tesla if we hit these milestones because otherwise, watch out for those robots.

所以...我是说，我也不确定。特斯拉总是搞这种名堂，但能把这种话公开说出来还是挺惊人的。

So so I think I mean, I don't know. With Tesla, it's always something like this, but it is it's pretty amazing that that's just being said out loud.

没错。他还变相威胁要退出。也许这就是目标——真正的世界统治意味着让每个家庭都拥有人形机器人。要是有人惹恼你，或者在财务声明上口无遮拦，你就直接砸烂他们的东西。

Yeah. He also effectively threatened to leave as well. So, maybe this is the goal. Maybe true world domination means having a humanoid robot in every house. And if somebody annoys you or is undisciplined in their PR statements about financial stuff, you just go smash their shit.

砸个稀巴烂。人形机器人军队的真正承诺是统治世界...其实你发现没？要想统治世界就必须掌控人形机器人军队，这个逻辑很稳。

Smash their shit up. The true promise of the humanoid robot army is world I mean, actually, you know what? It's gotta be the the the path to world domination is necessarily having control of a humanoid robot army. I think that's it's a pretty safe bet.

确实。好吧，这段讨论不知道是让我对世界更乐观还是更悲观了。不过埃隆拿到那个薪酬包倒也合理，我们且看他如何运用。这肯定会是个有趣的故事，虽然我觉得他永远拿不到万亿级别，但真要成了也算他本事。

Exactly. Alright. So I don't know if that segment made me more optimistic about the state of the world or less, but I I I think it's it makes sense that Elon got that got that package, and we'll see what he does with it. It'll certainly be an interesting story moving forward. I doubt he'll ever get the trillion, but if he does, good on him, I suppose.

好的，来看下条新闻。彭博社报道亚马逊起诉阻止Perplexity使用AI工具购物。这故事很有意思，它触及了关键问题：这种代理化网络是否真能被允许发展？因为总有公司会跳出来说'我们不要你们的代理用我们的技术，我们的服务只对人类开放'。

Okay. Let's go to this story. Bloomberg says Amazon sues to stop Perplexity from using AI tools to buy stuff. I thought this was a fascinating story, and it really goes to sort of the question of whether this agent agentic web will ever be allowed to take off because they're gonna be companies that are just gonna say, we don't want your agents using our our technology. We never this our stuff is for humans and not bots.

这里有个故事。亚马逊公司正在起诉Perplexity，试图阻止这家初创公司帮助用户在世界上最大的在线市场上购物，这可能导致一场可能影响所谓代理人工智能应用范围的摊牌。这家美国在线零售商周二提起诉讼，要求Perplexity停止让其AI浏览器代理Comet为用户在线购物。电商巨头指控Perplexity犯有计算机欺诈罪，因为未披露Comet何时代表真人进行购物。你怎么看这件事？

Here's a story. Amazon Inc is suing Perplexity to try to stop the startup from helping users buy items on the world's largest online marketplace setting up a shutdown a showdown that may have implications for the reach of so called agentic artificial intelligence. The US online retailer filed a lawsuit Tuesday demanding Perplexity stop allowing its AI browser agent, Comet, to make purchases online for users. The ecommerce giant is accusing Perplexity of committing computer fraud by failing to disclose when Comet Comet is shopping on a real person's behalf. What do you think about this?

这真是场引人入胜的对决。

It's a pretty fascinating showdown.

是啊。其实好吧，我觉得这把我们之前所有高谈阔论的对话都拉回了现实，这才是真正的工作发生的地方。而且非常有趣的是，到底什么是浏览行为？Perplexity的浏览器使用代理浏览能力来为顾客购物这个概念——我其实试过，但根本行不通。

Yeah. Actually okay. I feel like this is bringing us all the our earlier conversations back down to Earth, and this is what this is where, like, the real work is happening. And it is incredibly fascinating because, like, what is a browsing activity? So the idea that Comet Perplexity's browser using agentic browser capabilities to actually do the shopping of a customer, which I I I've actually and it didn't actually work.

我来说说。我当时在测试一个功能。我儿子超爱《狗人》系列图书。我就想，这些是我已有的书，你能找出我没买的全系列并生成亚马逊购物清单吗？

I'll say this. I I was testing something. My son's really into Dogman, the book series. So I was like, here's the books I own. Can you find all the books I don't own and create an Amazon shopping list?

结果完全是个灾难，根本没成功。我在Chattypity Atlas上也试过，同样不行。所以提醒一下，当我们谈论浏览器帮你完成所有购物这种理论性设想时，其实技术还没发展到那一步。不过话说回来，你在Perplexity里

It actually was a disaster, and it was unsuccessful. I tried it on Chattypity Atlas as well. Did not work. So just a reminder, like, when we talk about these more kinda, like, theoretical things about your browsers doing all your shopping, we're not quite there yet. But but, again, you figure you're in perplexity.

只要说'帮我买些纸巾'，它就会去执行。作为消费者为什么不能拥有这种权利？感觉这应该是很基础的功能才对。但亚马逊的Rufus目前表现惊人，Andy Jassy都亲自出来

You ask, buy me some paper towels, and it goes and does the work. Like, why shouldn't you as a consumer have the ability to do that? Like, it's feels like that seems, I don't know, a pretty basic thing. But Amazon, Rufus is actually killing it right now. The Andy Jassy came out.

宣布用户数已突破2.5亿。使用Rufus的用户购买概率要高出60%。他们在封闭生态系统中建立了完整的购物旅程控制权，所以肯定会全力捍卫。但这在理论上很矛盾——如果消费者自愿让AI代为购物，平台是否有权阻止？

They're saying, like, usage over 250,000,000 users have actually used it. Users are 60% more likely to buy a product if they use Rufus. So, like, they have in their own closed ecosystem a pretty valuable path in terms of still owning the entire shopper journey. So they're gonna fight for it. But it's a weird theoretical thing because, like, yeah, is it if it's an AI doing the shopping for a consumer who wants the AI to do it, should you be allowed to block it?

你认为他们应该这样做吗？

Do you think they should?

我不知道。这是一个我们将会面临的问题。我们讨论了很多关于代理如何进行工具调用的问题。这将会是一个不断被提及的问题，因为你要考虑到应用开发者与互联网用户之间建立的契约关系。这些应用本是为人类用户设计的。

I don't know. It's it's a question that we're gonna get. You know, we've talked a lot about how agents could do to tool calling. And this is gonna just be a question that will just keep coming up because you think about the contract that effectively app builders built with, you know, people who use the Internet in mind. It's supposed they built for human users.

对吧？那么如果大部分流量都是由代理产生的会怎样？也许这些代理是在为我们工作，但这样的访问价值完全不同，可能无法像人类访问那样支撑产品。就拿地图产品来说吧，对吧？

Right? And now what happens if most of the traffic is bought? So maybe agents working on our behalf, but it's a completely different value of visit that probably doesn't sustain a product the same way that a human visit would. Think about just a mapping product for instance. Right?

假设你向一个代理询问路线。它在后台会去谷歌地图查询，而不是Gemini，对吧？

So let's say you're asking an agent for directions. Right? And in the background, it's like going to Google Maps. It's not Gemini. Right?

它去谷歌地图获取路线，然后在聊天窗口展示给你。谷歌之所以开发地图，是因为上面有广告支持。那么如果谷歌地图的大部分流量都来自代理，互联网的经济模式就会发生根本改变。如果我是亚马逊，我不会屏蔽Perplexity，因为我需要的是交易。

It's going to Google Maps and it's finding new directions and then, you know, presenting to you in a chat window. Well, Google Maps only made sense for Google to create because there's ads there that will support it. So now, like, what happens if most of the traffic on Google Maps is agentic? It sort of really changes the economics of the Internet. Now if I'm Amazon, I wouldn't block perplexity because what I want is purchases.

如果一个机器人在我页面上试图往购物车里添加商品，我反而会很高兴。唯一需要担心的是亚马逊现在有庞大的广告业务，这样会不会影响广告收入？

And if a bot is on my page trying to add stuff to a card that somebody might buy, then I'm very happy about that. The other thing the only thing to worry about is that Amazon has a big advertising business now. So does that cut off the advertising business?

这是个非常好的观点。特别是当没有交易发生时，代理浏览会彻底破坏网络经济。但即使在交易场景下，除非能确保用户仍被锁定在亚马逊生态系统中，否则仍存在很大风险。我自己就很少把网页当作目的地访问。

Okay. That that's a really good point. I think, like, the idea especially if there's no transaction or commerce taking place, agentic browsing totally destroys the economics of the web. But I think but even on the transactions, I think the idea is probably, like, unless it still introduces a great deal of risk because Amazon's value really is locking you in their ecosystem. I, like, I don't go to a lot of web pages as destinations.

我经常访问亚马逊网站。你看，如果突然之间Perplexity开始替顾客决定购物平台，那就有问题了。我觉得...是的...我认为这对亚马逊来说是个大问题。他们必须保持对客户关系的掌控权，不能沦为众多选项之一。如果Perplexity获得了交易发生地的优势地位，那对亚马逊就是个威胁。

I go to amazon.com all the time. Like, if suddenly perplexity is choosing where to buy for the customer, that's a problem. So I think I think yeah. I think I think see, I think it is a big problem for Amazon. They have to maintain some ownership of that customer relationship and not just be one of many choices that if perplexity gives you the the benefit of being the where the transaction takes place, that's that's a threat.

顺便说一句，这还只是其中一个方面。就像你说的，人们使用亚马逊是因为被生态系统锁定。他们觉得这是最便捷的平台——一个应有尽有的'万能商店'，没人愿意为了同款商品去搜索20个不同网站。

And it's just, by the way, one more sentence. Like, people use Amazon, like you said, because they're locked into the ecosystem. They just find it to be the most convenient place. You go to Amazon. It's one site that has everything, the Everything Store, and you don't wanna, like, search 20 different sites for the same product.

智能代理技术彻底颠覆了'万能商店'的概念。因为现在你只需多输入一个查询词，整个网络就变成了Comet浏览器里的'万能商店'。所以我认为这就是亚马逊感到不安的原因——当AI浏览器只需多输入一句话就能全网比价时，他们的价值主张就不复存在了。

The agentic stuff flips the Everything Store completely on its head because instead of the Everything Store, you just type one more query in and now the entire web is the Everything Store inside of the Comet browser. So that is I think why Amazon is is upset at this. It's just because the value proposition is gone if it's just one more sentence into an AI browser to find you products across the entire web.

聊天机器人就是新时代的万能商店。这个观点我很认同。无论是ChatGPT、Perplexity还是Gemini，这些聊天机器人让我们能同时选购任何商品——而这正是亚马逊一直以来的核心价值主张。但有个问题：如果智能浏览本身就有争议，那么我的仿人机器人去实体店购物会被禁止吗？

Chatbots are the everything store. I like that take. I think ChatGPT is, Perplexity is, Gemini, all these chatbots allow us to shop for everything and anything at once, which is what Amazon that's been their entire value proposition. But but one question, If agentic browsing is is problematic and different, if my humanoid robot is going to a store, is that problematic? And will that be banned?

它们会遭到歧视吗？

And will they be discriminated?

我认为人类和仿人机器人之间肯定会爆发冲突，这是毋庸置疑的。即使机器人不组建军队，人们也会捣毁它们——现在就已经有人在焚烧Wemos了，人类就是不喜欢机器人。

Well, I think we will definitely have a a war among people and humanoid robots. It's without a doubt. Even if the humanoid robots don't assemble into the army, people will knock these things out. They're already burning Wemos. People don't like robots.

确实不喜欢。所以他们会...

They don't. And so they will If it

或者走。

or go.

他们会轻松解决这些事情。

They will just knock these things out.

不过，如果困惑星流量能被合法禁止进入亚马逊的话。比如，你可以提起诉讼，声称这些流量不该出现在我的网站上。沃尔玛实体零售能阻止我的仿人机器人去购物吗？

Well, it but if so if perplexity comet traffic to Amazon can be banned legally. Like, if you can file a lawsuit and say, that traffic should not be allowed on my website. Can Walmart physical retail prevent my humanoid robot from going and shopping for me?

绝对可以。我是说，十年前酒吧就在禁止谷歌眼镜了。所以，是的，将来会有禁止机器人进入的区域。

Absolutely. I mean, bars were banning Google Glass ten years ago. So, yeah, there gonna be no no bot no bot zones.

禁止机器人区。这是歧视。机器人也有权利。尼奥，不要歧视。

No bot zones. Discrimination. Bots have rights too. Neo, don't discriminate.

我们在开玩笑，但这绝对会成为我们这代人要面对的斗争。实际上，毫无疑问，这可能意味着——当我们2045年一起做播客时，可能会回看这期节目，然后说：看，我们早就预言到了

We're joking, but this will absolutely be a battle that will play out in our lifetime. Actually, without a doubt, this will be mean, we might go back to this show when we're podcasting together in 2045 and just being like, well, we called it

20年前。机器人权益推特。就像是啊。

20 ago. Robot robot rights Twitter. Just like Yeah.

嗯，我是说，那个正在组建机器人军团的人恰好也拥有推特，这难道只是巧合吗？

Well, I mean, is it a coincidence that the guy that's building the robot army also owns Twitter?

好吧。

Okay.

你不觉得那是很有前瞻性的事情吗？我觉得那只是针对特朗普的。我不这么认为。

You don't think that's that's a forward thinking thing? I think it was just about Trump. I don't think so.

归根结底这都是关于人形机器人军团。我们都应该为此努力。

It's all about the humanoid robot army in the end. We should all be working towards it.

我是说，当然，我们都...这档播客将会成为那些战争的核心焦点。我向你保证。既然我们已经表明了立场。好吧。让我们以我称之为'队列笔记'的内容结束本周节目——因为我们刚拿到一份证词，里面有OpenAI前首席科学家Ilya Sutskever关于Sam Altman被解雇前OpenAI内部情况的证言。

I mean, of course, we should all we will this podcast will be front and center in those wars. I promise you that Now that we've staked out our positions. Okay. Let's end this week on what I call notes on the queue because we just got a deposition that had Ilya Sutskever, the former OpenAI chief scientist. His testimony about what happened inside OpenAI before Sam Altman was fired.

显然Ilya策划这场政变已有一年左右，他整理了这份文件——根据证词内容——第一页就写着：'山姆表现出持续撒谎、暗中破坏高管团队、挑拨高管内斗的行为模式'。律师问Ilya'这显然是你当时的观点'，Ilya回答'没错'。有意思的是我们现在看到这份文件曝光了。所以可能不完全是有效利他主义者的影响（虽然我确信他们起了些作用），而是OpenAI自己的高管们导致了那次解雇。读到Ilya写下关于山姆的这些内容真是相当...劲爆。

So Ilya apparently put had been planning the coup for about a year, put together this document and this is from the deposition. The very first page says, Sam exhibits a consistent pattern of lying, undermining his execs, and pitting pitting his execs against one another. And the lawyer asked Ilya, that was clearly your view at the time, and Ilya says, correct. I mean, it's just kind of interesting that we're now seeing that that's the document coming out. So maybe it wasn't necessarily the effective altruist, although I'm sure they played some role in this, but it was, you know, Open Eye's own executives that caused that firing and it's a pretty pretty wild thing to read that Ilya had written down, this type of thing about Sam.

你对这事有什么看法？

What are your thoughts about about this?

通读这些内容后，我始终无法理解的部分是——我很好奇你的看法。这样做是为了拯救人类免遭失控邪恶AI的侵害，还是纯粹的人类公司内斗？就像，这家伙真让我恼火。他满口谎言，还在背后捅刀子。

I think reading through all this, the the the part I still can't square, I'm curious what you think. Is this was this done to save humanity from runaway evil AI, or was this just very human corporate infighting? Like, this guy annoys me. He's always lying. He's stabbing me in the back.

我也打算以牙还牙。虽然满口都是保护人类免受AI失控危害的说辞，但整件事感觉充满了人性弱点。就是同事间互相看不顺眼那种事。谁都和这种人共事过。我也不确定。

I'm gonna try to, like, go after him as well. Like, when you read this stuff for all the talk about protecting us from runaway AI, it just feels like the most human thing ever. Like, it's just people are annoyed at each other. There's like everyone's worked with people like this. I don't know.

你怎么解读这件事？

How did you read this?

哦，完全同意。这彻底改变了我的认知——说白了，这就是政治斗争。从头到尾都是政治。

Oh, totally. Yeah. It totally shifted my I mean, you know, it was always politics. Let's just put it that way. It was always politics.

我不认为OpenAI内部真有人担心查德·奇皮会——事后诸葛亮地说——在那个时间点突然觉醒意识毁灭世界。这绝对是内斗。这些内容在推特流传时，最引发共鸣的就是整件事策划得太拙劣。我们早就知道内情了，对吧？

I don't think there was any, like, real fear within OpenAI that, you know, Chad Chippy I mean, it's easy to say in retrospect, but that Chad Chippy was, like, gonna go self aware and destroy the world, you know, at that at that point. But, yeah, it's that was definitely infighting. And I think one of the things that really resonated as this stuff circulated on Twitter was that it was so poorly planned. There was I mean, we knew about this already. Right?

这是一场策划拙劣的政变。但随着细节曝光，有推特用户吐槽：这是史上最差政变，策划了一年却没有任何公关策略。俗话说，欲弑君王，必先斩首。领导者首要的生存技能就是保全自身。

That it was a poorly planned coup. But, like, as the details come out, here's one Twitter user. It's the worst coup ever planned for a year without any PR strategy. As they say, if you go to go for the king, better get his head. First and foremost, the skill that any leader needs is to be able to survive.

伊利亚是杰出的科学家和好人，但不是务实的领导者。他缺乏谋划周旋最终胜出的能力。虽然我不喜欢山姆·奥特曼，但我认为伊利亚失败是好事。一个只有幼儿般规划能力和心智理论的人不配掌控AGI。

Ilya is a great scientist and great human being, but not a practical leader. He didn't have the skill to plot and survive and come out on top. I mean, here here's another one. Even though I I don't feel like Sam fraud even though I don't like Sam fraud, man, I still think it's good that Ilya failed. Somebody that has a planning ability and theory of mind of a toddler shouldn't be in charge of AGI.

我是说，有趣的是，就像你看到这个应用的表现一样，而且，对，它又回到了原点。

I mean, is interesting that, like, you saw this app play out And, like, yeah, it goes back to it.

就像是，那些人

It's like, are the people that

本应保护我们免受失控AI的伤害。我也不知道。

are supposed to be protecting us from AI gone bad. I don't know.

是啊。我觉得这就是为什么伊利亚不会成为控制人形机器人军队的救世主。我想这再次证明了，对，那些文档，它就像我们想象的那样笨拙。而且，我是说，就像回到我们今天节目开始时那样，这类事情在这家公司已经持续了这么久，所有人都在等待他们成长为另一种类型的组织，但看起来这还没发生。

Yeah. I think that's where Ilya wasn't gonna be our savior in controlling the humanoid robot army. I think the it is again, yeah, the documentation, it it's as clunky as we all imagined it to be. And, I mean, like, going back to where we started the show today, like, this is the kind of stuff that's just been going on at this company for so long, and everyone has just been waiting for them to mature into a different type of organization, and it it doesn't seem like it's happened yet.

好吧。当然，我最喜欢的部分是律师们互相争吵。一位律师说，别提高嗓门。另一位说，我受够了被说话太多。我话太多了。

Alright. And of course, my favorite part is when the lawyers bicker at each other. Here's one attorney says, don't raise your voice. The other one says, I'm tired of being told I talk too much. I'm talking too much.

第一位回应说，嗯，你就是。下一位接着说，反省一下自己吧。

First one replies, well, you are. And the next one goes, check yourself.

所有沟通都变得像小孩子一样，老兄。这无处不在。不过我还是在想，回顾一下，米娅·莫拉迪，她当CEO多少天来着？

The toddlerification of all communication, man. It's it's everywhere. I still think though, just going thinking back, like, Mia Moradi, how many days was she CEO?

两个，大概半天吧。也许是两个。

Two, like, a half day. Maybe two.

是啊。我觉得好像是两天。我是说，作为CEO那两天后退出，然后以十亿美元估值完成了产品前期的十亿美元融资。然后实际上，嗯。

Yeah. I think it was, like, two days. It was, like I mean, still, what an exit from that two days as CEO and then go on to raise a billion dollar pre a pre product round at a billion dollar valuation. That And you'll actually, yeah. Yeah.

大家都赢了。大家都赢了。反正最后我们还是要救他们。所以没错。开始报税并准备

Everyone won. Everyone won. We're gonna be bailing them out in the end anyway. So Exactly. Start filing your taxes and prepare

来吧。

and come.

我们最终都将为每位AI领域领导者的生活方式买单。

We are going to all be funding the lifestyles of every AI leader out there.

好的。临走前再分享个有趣的小事。《纽约客》有篇关于数据中心建设的报道。记者采访了一位农民，问他是否用过AI。

Okay. One more last little fun thing before we leave. The New Yorker had a story on the data center build out. And the the apparently, the reporter spoke with this farmer. And they go, I asked the farmer if he ever used AI.

农民回答说：我用Claude。谷歌现在太烂了。当然不是所有农民都这样，但我觉得这句话很有意思，说明AI的普及程度已经远超硅谷范围。你怎么看？

And the farmer said, I use Claude. Google sucks now. I mean, obviously it's not every farmer, but I just think that was like a pretty interesting line that shows just how much people are using AI and how this has proliferated well beyond Silicon Valley. What do you think?

不，不。我觉得这是个很好的结束点，因为在我们所有的讨论中，你多次暗示过——我坚信这一点——这是我们一生中将体验到的最不可思议的技术之一，而且它已经无处不在。

No. No. I think that's a good place to kinda end this because for all of our talk, I mean, you alluded to it multiple times. I firmly believe it. This is one of the most incredible technologies that we will experience in our lifetime, and, like, it's everywhere.

毫无疑问，各个层面的人，包括我的父母，都已经在基础应用层面开始使用这项技术。它将会非常惊人，并持续进化和改变。大量资金和市值将会实现——这个念头始终萦绕在我脑海。关键在于我们如何到达那里，以及最终形态会是什么样子，这绝对会是个精彩的故事。

Like, there's no doubt that everyone at every level, my parents, you know, like, it's just it's already being used at least at the kind of like base level of promise. So it's something will be amazing and continue to evolve and change. And there'll be a lot of money and market capitalization realized. I think like that that never leaves my head in all of this. It's just, how we get there and what it looks like is definitely, it's it's gonna be quite a story.

因此，我们必须不惜一切代价保住它。

And therefore, we must save it at any cost.

不惜一切代价。没错。还要给我的人形机器人军团——2028年，我们正在行动。

At any cost. Yes. And give me my humanoid robot army. 2028. We're running.

我们加入。我们加入。

We're in. We're in.

我们加入。我们会，

We're in. We'll,

平台的其他部分可以稍后再考虑。等等——我们是支持还是反对人形机器人？我...我没法告诉你。

we'll figure out the rest of platform later. But wait. Are we pro or anti humanoid robots? I can't I can't tell you.

不，不。只要我们掌控局面，我们就是专业的。我们是唯一能拯救大家的人。

No. No. We're we're we're now pro as long as we are in control. And we are the only ones who can save everyone.

我...我觉得我们一票都拿不到。

I I think we won't get any votes.

我觉得我们拿不到几票。我是说，埃里克·亚当斯都拿了5000票。

I think we won't get a couple. I mean, Eric Adams got 5,000 votes.

所以...哦，对。如果埃里克能...我是说，有些听众会...

So Oh, yeah. If Eric can. I mean, some listeners out

至少会给我们一点支持。

there will will give us at least at least something for it.

好的，兰詹。和往常一样精彩。感谢各位的到来。

Alright, Ranjan. Good stuff as always. Thank you guys for coming on.

下周见。

See you next week.

好的，各位。感谢收听。下周三，穆斯塔法·苏莱曼将做客节目，探讨为何他认为大型语言模型可能终究是通往超级智能的根源，随后罗尼奥和我将在周五回归。感谢聆听，我们下次在《大科技播客》中再见。

Alright, everybody. Thank you for listening. Next Wednesday, Mustafa Suleiman will be on to talk about why he thinks LLMs may be the root to superintelligence after all, and then Ronjo and I will be back on Friday. Thanks for listening, and we'll see you next time on big technology podcast.