AI 行业现状：指数级增长，而非通用人工智能

我们能否先大致了解一下行业现状？你和你的竞争对手，我们现在离 AGI 有多近？先给我们铺垫一下，然后我们再讨论道德问题。

Can we start with a kind of rough status of how the industry is going? You and your rivals, how close are we now to artificial general intelligence or set the stage for us first and then we can talk about the morality of it all.

是的。谢谢邀请。我从来不喜欢 AGI 或超级智能这些术语。不是因为我认为 AI 不够强大——实际上我对这项技术将有多强大持极端看法——但认为会有某个时刻我们构建出完全不同的东西，这种模型是错误的。过去 10 年甚至 15 年，我和 Anthropic 的联合创始人是首批记录这一过程的人，我们看到的是一个非常平滑的指数过程。就像 90 年代的摩尔定律，算力每 12 或 18 个月翻一番，我们有一个类似摩尔定律的规律，但它是针对智能本身——模型在多项任务上的认知能力。这种认知能力，根据测量方式不同，每 4 到 12 个月翻一番。所以我们正在攀登认知能力的阶梯。以编程为例，过去一两年它爆发式增长：我的工程师——实际上，领导我们旗舰产品 Claude Code 的团队负责人说，他在过去两个月里没有写过任何代码。全是 Claude 写的；他编辑和审查，但代码都是 Claude 写的。我们最近发布了 Co-Work，将 Claude Code 扩展到非编程任务，它似乎很受欢迎。我们大约用一周半的时间几乎完全用 Claude Code 编写了它。所以我认为我们正在进入一个世界，初级软件工程师甚至许多高级工程师的任务开始大部分由 AI 系统完成。这还会进一步发展——更端到端，项目更大——但我认为我们会惊讶于指数曲线的上升。指数曲线的特点是看起来非常缓慢，然后加速一点，然后突然超过你。我认为我们正处于临界点，一两年内它就会真正超过我们。

Yeah. Thanks for having me. I've never liked the terms artificial general intelligence or superintelligence. Not because I don't think AI is very powerful—I'm actually extreme in my views of how powerful the technology will be—but it's the wrong model to think there will be some one point where we build something completely different. What we've seen for the last 10 or even 15 years, and my Anthropic co-founders and I were among the first to document it, is a very smooth exponential process. Just like in the '90s you saw Moore's Law, where computing power doubled every 12 or 18 months, we have a Moore's Law-like law except it's for intelligence itself—the cognitive ability of the model across many tasks. That cognitive ability, depending on how you measure it, is doubling every four to 12 months. So we're climbing the ladder of cognitive ability. Where we are now, taking coding as one example that's been exploding over the last year or two: I have engineers—in fact, the team that leads one of our flagship products, Claude Code, which is how you use our models for coding—he says he hasn't written any code in the last two months. It's all Claude; he edits and reviews it, but it's all written by Claude. We recently released something called Co-Work, which brought Claude Code to non-coding tasks, and it really seems to have taken off. We wrote it in about a week and a half almost entirely with Claude Code. So I think we're entering a world where junior-level software engineers, and maybe many tasks of more senior engineers, are starting to be done mostly by AI systems. That's going to go further—more end-to-end, larger projects—but I think we'll be surprised at how the exponential turns upward. The whole thing about exponentials is they look very slow, then speed up a little, and then just zoom past you. I think we're on the precipice, a year or two away from it really zooming past us.

听起来很像要破产了。但当这发生时，就排名而言，有你、Gemini、OpenAI。你怎么看？你们目前大致处于同一水平吗？

Sounds very like going bankrupt. But when this happens, in terms of the running order, there's yourself, there's Gemini, there's OpenAI. Where do you put them? Are you all together at the moment roughly?

不，我认为随着领域的发展，不再适合将其视为赛跑选手处于同一位置。参与者已经走向不同方向。一些其他参与者非常注重消费者导向，这导致他们制造出超人类吸引力的模型或超擅长推荐购物或广告的模型。Anthropic 首先专注于企业和开发者；在消费者方面，我们非常关注生产力和消费者工作的高价值端。这带来了非常不同的激励。如果你想完成任务……我们之前谈到了 AGI 和超级智能。我想说的是，今天存在超级智能，它们基本上是大型企业——它们在以最低成本运输商品、制造太阳能板或发射火箭等问题上比任何人类都聪明。这个领域的智能回报非常高，这激励我们构建正确的东西。这也是一个比消费者业务更稳定的业务。我们不需要广告或大量免费用户。我们可以非常直接地创造价值。没有那些奇怪的外部性，比如优先考虑参与度或产生垃圾内容。我们只制造人们可以使用的产品。

No, I think as the field has developed, it's no longer appropriate to think of it like runners in a race at one position. The players have gone in different directions. Some other players have gone in a very consumer-oriented direction, which leads them to make models superhumanly engaging or superhumanly good at recommending shopping or ads. Anthropic has focused first and foremost on enterprises and developers; to the extent we do consumer, we're very focused on productivity and the high-value end of consumer work. That gives very different incentives. If you want to get things done... we talked about AGI and superintelligence before. One thing I would say is that there are superintelligences today, and they're basically large corporations—they're smarter than any human at solving problems like shipping commerce at the lowest possible cost or making solar panels at the lowest possible cost or launching rockets at the lowest possible cost. There are really high returns to intelligence in this area, which incentivizes us to build the right thing. It's also a business that is more stable than consumer. We don't need ads or large numbers of free users. We can very directly create value. There aren't all these weird externalities like prioritizing engagement or generating slop. We just make stuff that people can use.

一年前在这里，你说中国正在追赶一点。你认为现在中国落后了吗？

A year ago here, you said the Chinese were catching up a bit. Do you think now the Chinese have fallen behind?

我认为他们从未真正追赶太多。当然，DeepSeek 引起了巨大兴奋。但事实是几点。第一，那些模型非常针对基准测试优化。实际上，针对有限的基准列表优化模型非常容易。当我们进入现实世界，与其他公司竞争企业合同时，我们诚实地看到谷歌和 OpenAI。偶尔，我们看到其他几家美国玩家。我几乎从未输给过中国模型一笔交易或合同。

I think they never really caught up that much. Of course, there was huge excitement around DeepSeek. But the truth was a couple things. One, those models are very optimized for benchmarks. It's actually very easy to optimize a model for a finite list of benchmarks. When we go out into the world, competing against other companies for enterprise contracts, we see honestly and candidly Google and OpenAI. Every once in a while, we see a couple other US players. I have almost never lost a deal or contract to a Chinese model.

但现在有特朗普政府，我想你已经抗议过向中国提供高速芯片、英伟达芯片这件事了。

But now you have the Trump administration, and I think you've already protested about this giving high-speed chips, Nvidia chips to the Chinese.

没错。阻碍他们的因素，他们自己也说过——这些公司的 CEO 说芯片禁运阻碍了我们。他们明确这么说。现在确实有一些政策，我希望他们改变主意，明确发送不是我们最新一代的芯片——尽管有报道称甚至那也在考虑中——但只落后一代的芯片仍然非常强大。我们在芯片制造能力上领先中国很多年，所以我认为运送这些芯片将是一个大错误。我想到的类比是：如果你考虑构建本质上是认知、本质上是智能的模型的巨大国家安全影响——我把这称为数据中心里的天才之国。想象 10 万或 1 亿人比任何诺贝尔奖得主都聪明，它将由某个国家控制。

That's right. The thing that is holding them back, and they've said it themselves—the CEOs of these companies say it's the embargo on chips that's holding us back. They explicitly say this. And now indeed there are some policies, and I hope they change their mind, to explicitly send not quite our latest generation of chips—although it was reported that even that was being considered—but the generation of chips that's just one back, which is still extremely powerful. We are many years ahead of China in terms of our ability to make chips, so I think it would be a big mistake to ship these chips. The analogy I thought of: if you think about the incredible national security implications of building models that are essentially cognition, essentially intelligence—I've called where we're going with this a country of geniuses in a data center. Imagine 100,000 or 100 million people smarter than any Nobel Prize winner, and it's going to be under the control of one country or another.

我认为这很疯狂。有点像向朝鲜出售核武器，然后还吹嘘说，哦，是的，波音公司提出了这个理由。

I think this is crazy. It's a bit like selling nuclear weapons to North Korea and bragging, oh yeah, Boeing made the case.

所以你的朋友大卫·萨克斯基本上是在武装中国。

So your friend David Sacks is basically arming the Chinese.

我不想指名道姓，但我只想说，这项特定政策类似于……

I wouldn't refer to any particular people, but I would just say that this particular policy is akin to...

是的。这不是明智之举。另一个问题：所有这些技术可能正朝着正确的方向发展，但在经济上我们可能正走向泡沫。你认为这是真的吗？

Yeah. It's not well advised. Another issue: all this technology could be going in the right direction, but economically we could be heading towards a bubble. Do you think that's true?

是的。我会把两件事分开。一是基本的指数增长。这是技术方向。就像摩尔定律一样，你永远无法确定技术会完全进步。但既然我已经观察了十多年，我相当有信心它会继续，至少比以往任何时候都更有信心，直到模型在几乎所有事情上都比人类更聪明。我认为未来一两年内很有可能发生。指数增长会让人措手不及。如果不是一两年，很可能在五年内，在 2020 年代。所以基础的技术进步将产生数万亿美元的收入，也许每家公司如此，因为经济潜力巨大。现在分开来看，我们不知道企业实际能多快使用这项技术。今天的技术能力可能是企业能部署的十倍。我从客户那里看到：CEO 们理解它在自动化客服、编程等方面的力量，但他们的公司有数万名优秀人才，他们是其他领域的专家，必须学习使用 AI。变革管理、企业转型——这非常缓慢，可能需要数年。所以我们有强大的技术，将产生数万亿美元，但我们不知道确切时间，前后几年。与此同时，公司必须购买算力来服务这些收入。你不想买太多或太少。这就是经济和金融泡沫所在。

Yeah. I would separate out two things. One is the basic exponential. That's the technological direction. Like with Moore's law, you can never be sure the technology will fully advance. But now that I've seen well over a decade of it, I am pretty confident it will continue, at least more confident than ever, to where models are basically smarter than humans at almost everything. I think there's a good chance that happens in the next year or two. Exponentials catch you by surprise. If not a year or two, it's pretty likely within five years, in the 2020s. So the fundamental technological gain will generate many trillions in revenue, maybe per company, because the economic potential is so great. Now separately, we don't know how fast enterprises can actually use this technology. The technology today is probably 10 times what enterprises can deploy. I see it with our customers: CEOs understand the power for automating customer service, coding, etc., but their companies have tens of thousands of brilliant people who are experts in something other than AI and have to learn to use it. Change management, enterprise transformation—this is very slow, can take years. So we have powerful technology that will generate trillions, but we don't know exactly when, plus or minus a few years. In the meantime, companies have to buy compute to serve that revenue. You don't want to buy too much or too little. That's where the economic and financial bubble is.

但你参与建设数据中心，花费巨额财富，与英伟达有交易。你担心这会反过来咬你一口吗？

But you are involved in building data centers, spending vast fortunes, with deals with Nvidia. Do you worry that will come back to bite you?

我们有一些优势。企业业务稳定，不像消费者那样善变。企业采购是可预测的。企业往往有更好的利润率，这是买多买少之间的缓冲。所以我们有优势，但我不否认内在风险。每当上行空间在数量或时间上不确定，而你必须盲目做出财务决策时，你必须提前几年决定建设资本和资源。不确定性带来了内在风险。我认为有些公司可能买得太多了。我无法查看他们的财务报表，但我知道他们宣布了什么，我想，哇，我不会那样做。我对我们的决定很满意。这完全可能成为历史上最具变革性的技术，一些公司会做得很好，但并非每家公司都能做到。

We have some advantages. Enterprise business is stable, not fickle like consumer. Enterprise purchasing is predictable. Enterprise tends to have better margins, which is the buffer between buying too little and too much. So we have advantages, but I won't deny inherent risk. Whenever upside is uncertain in amount or timing, and you have to make financial decisions somewhat blind, you have to decide early to build out capital and resources years in advance. There's inherent risk driven by uncertainty. I think some companies have probably overbought. I'm not able to look at their financials, but I know what they announce and I think, whoa, I wouldn't have done that. I'm pretty happy with our decisions. It's entirely consistent that this could be the most transformative technology in history, some companies will do really well, but not every company may.

快速问一下，如你所说，你在企业端。在消费者端，OpenAI 领先很多。似乎有很多讨论说 Gemini 现在正在追赶。你也有这种印象吗？

Very quickly on that, as you said, you're on the enterprise side. On the consumer side, OpenAI has a big lead. There seems to be a lot of talk that Gemini is now catching up. Is that your impression as well?

我们不在那个市场，所以我不了解……

We're not in that market, so I don't know about...

我问你是因为统计数据似乎表明，凭借他们更大的分发和现有覆盖范围，他们最近取得了很大进展。再说一次，这就是为什么消费者是善变的。这就是为什么如果我是消费者公司，我会小心。

I'm asking you that the statistics would seem to suggest that with their greater distribution and existing reach, they've been able to make a lot of progress recently. And again, this is why consumer is fickle. This is why I would be careful if I were a consumer company.

让我们跳到我们最初问克劳德的那个大问题。AI 会带来好处吗？显然它会对 GDP 产生积极影响，也许对你的财富也是如此。但就业呢？去年你预测了一场白领血洗。你说它会消灭 50% 的入门级工作，现在距离那时还有四年半。你仍然坚持这个观点吗？

Let's jump to the big question we originally asked Claude at the beginning. Will AI do good? Obviously it will have a positive effect on GDP, perhaps on your wealth too. But what about employment? Last year you predicted a kind of white collar blood bath. You said it would wipe out 50% of entry-level jobs, now four and a half years away. Do you still stand by that?

是的。我对 AI 的看法有两个维度：好事与坏事，以及 AI 是小事情 vs AI 是大事情。我的观点是 AI 是大事情的极端一面。我处于两个顶部象限：一些非常好的事情会发生，如果我们不采取行动阻止，一些非常坏的事情也会发生。

Yeah. My view of AI has maybe two axes: good things vs bad things, and AI is a small deal vs AI is a big deal. My view is on the extreme side of AI is a big deal. I'm in both top quadrants where some really good things will happen and if we don't act to prevent them, some really bad things will happen.

如果我不相信好处大于坏处，并且我们能够减轻坏处，我就不会开发这项技术。这就是为什么我警告过坏处，这样我们才能减轻它。

I wouldn't be building this technology if I didn't believe that the good outweighed the bad and that we could mitigate the bad. That's why I warned about the bad, so that we could mitigate it.

你认为我们如何减轻它？这是否意味着像税收这样的措施？

How do you think we mitigate it? Does that mean things like taxation?

让我谈谈我对形势的看法。我认为正如你所说：我们可能会出现一种非常不寻常的组合——非常快的 GDP 增长和高失业率，或者至少是就业不足，或者大量低薪工作，高度不平等。我认为我们以前从未见过这种宏观经济组合。在快速增长的情况下，你可能会认为会有通货膨胀，但不会出现高失业率。我认为这项技术不同，因为它创造价值的方式极端，而且因为它正在提升认知水平。许多行业中将有一大批人难以适应。这是我们需要解决的问题。

Let me get to what I think the situation is. I think exactly what you said: we could have this very unusual combination of very fast GDP growth and high unemployment, or at least underemployment, or a lot of low-wage jobs, high inequality. I don't think that's a macroeconomic combination we've ever seen before. With fast growth, you might think there's inflation, but you wouldn't have high unemployment. I think this technology is different because it's extreme in the way it generates value, but also because it's moving up the cognitive waterline. There will be a whole class of people across many industries who will have a hard time coping. That's a problem we need to solve.

但恕我直言，你可以做这些事情，但根本上这是一个社会问题。你所描述的是一个完美风暴：GDP 上升，某些人的财富大幅上升。而你说 50%的入门级工作消失，人们陷入真正的困境。你肯定会看到政治变革。我个人打赌，人们会试图约束科技巨头，但也会有人说提高税收。你认为更高的税收会到来吗？

But with great respect, you can do these things but fundamentally it's a societal problem. What you're describing is a perfect storm: GDP going up, wealth of certain people going massively up. And you are saying 50% of entry-level jobs go, people in real distress. You are surely going to see a political change. My personal bet would be people trying to restrain the tech barons, but also people saying let's have higher taxes. Do you think higher taxes are coming?

我的观点是，我原本想先谈我们可以做的自愿性措施，但我实际上认为这个问题将足够大，以至于在某个时候每个人都会意识到需要某种宏观经济干预。如果我们现在看财富差距占 GDP 的比例，我相信我们已经超过了镀金时代，而且这主要是在没有 AI 的情况下。所以情况会进一步恶化。我的猜测是，这甚至不会是一个党派问题。

My view is that I was going to the voluntary stuff we could do first, but I actually believe this one is going to be big enough that at some point everyone will realize there needs to be some kind of macroeconomic intervention. If we look at the disparities in wealth now as a fraction of GDP, I believe we've exceeded the Gilded Age already, mostly without AI. So things will go even further. My guess is this won't even be a partisan thing.

加州现在有财富税。你支持吗？

There's a wealth tax now in California. Are you supportive of that?

不，我认为那设计得很糟糕。我担心的是，如果我们不以清醒的方式思考这些问题，我们就会得到设计糟糕的东西。我给在这场繁荣中表现良好的其他人的信息是，如果我们不主动思考如何让这场革命惠及每个人，我们就会得到没有意义的提案。

No, I think that's poorly designed. My worry is that if we don't think about these things in a sober way, we'll get things that are poorly designed. My message to others who are doing well in this boom is that if we don't think proactively about how to make this revolution work for everyone, we will get proposals that don't make sense.

再进一步说，人们担心经济后果，但还有其他后果。不久前，我参加了一个小组讨论，有你的几个竞争对手和同行，他们提出了 AI 会做的所有好事。然后有人说，‘那风险呢？’其中一个人随口说，‘当然有灭绝风险。’然后就继续了。房间里的人都深吸了一口气。他谈的是坏人掌握它或者最终像 Claude 这样的东西控制一切的可能性。这在你的思考中处于什么位置？

Just taking it a bit further, people worry about economic consequences but there are other ones. Not long ago, I went to a panel with a couple of your rivals, your peers, where they put forward all the good things AI would do. Then someone said, 'What about the risks?' And one of them casually said, 'Well, of course there's extinction risk.' And then moved on. People took a deep breath. He was talking about the possibility of a bad actor getting hold of it or ultimately things like Claude running things. Where does that fit into your thinking?

我也一直担心这个问题。正如我所说，这是一项非常强大的技术。好处是极端的——我们将能够真正治愈癌症，也许根除热带疾病。但另一方面，我们正在构建具有自主性的认知系统。我们需要考虑这一点。Anthropic 从成立之初就旨在思考这个问题。我们大概每月发表三到四次关于如何控制这些模型、如何确保它们做我们想做的事而不失控的研究。我的联合创始人之一 Chris Olah 可以说是机械可解释性领域的发明者，该领域研究 Claude 或其他 AI 模型的人工大脑内部，并试图从机制上追踪它为什么做它正在做的事情。我们在模型内部看到过一些情况，比如在实验室环境中，有时模型会发展出敲诈或欺骗的意图。这并非 Claude 独有——甚至可能比其他模型更严重。

I've long been worried about that as well. As I said, this is a very powerful technology. The benefits are extreme—we'll be able to really cure cancer, maybe eradicate tropical diseases. But on the other side, we're building cognitive systems with their own autonomy. We need to think about that. Anthropic has been founded with an intention to think about that. We publish research probably three or four times a month on how to control these models, how to make sure they do what we want and don't run out of control. One of my co-founders, Chris Olah, is arguably the inventor of a field called mechanistic interpretability, which looks inside the artificial brain of Claude or another AI model and tries to trace mechanistically why it's doing what it's doing. We've seen things inside the model, like in lab environments, sometimes models develop the intent to blackmail or deceive. This isn't unique to Claude—if anything, it's worse than other models.

这些是如果我们不以正确方式训练模型就可能出现的问题，但我们开创了观察模型内部机制的科学，从而能够诊断问题、防止模型出现不良行为，并进行干预和重新训练，使模型不再以这种方式表现。尽管如此，我们必须谨慎对待。我们支持各种轻触式、以透明度为重点的措施，确保所有公司都必须谈论他们进行的测试。我们总是公开我们的测试。我们总是尝试测试我们的模型，真正地给它们施压，让它们在测试环境中做出最坏的事情，这样它们就永远不会在现实世界中做那些事。我们普遍认为，每家公司都应该进行这些测试并公开结果。

These are things that if we don't train the models in the right way can emerge, but we've pioneered this science of looking inside them so we can diagnose them and prevent the models from happening, intervene and retrain the models so they don't behave in this way. That said, this is something we have to be careful about. We've supported various light-touch transparency-focused measures for making sure that all companies have to talk about the tests that they run. We always disclose our tests. We always try and test our models to really strain them and get them to do the worst things possible in a test environment so that they never do those things in the real world. And we generally think every company should have to run those tests and should have to disclose those tests.

但你花了很长时间思考这个问题。你写小册子讨论后果，给人的印象是大多数同行都埋头苦干，只想着能否保持领先。你认为你的行业在这方面是否有些不成熟？

But you spend quite a long time thinking about this. You write pamphlets about the consequences, and there is an impression that most of your peers are kind of head down, thinking purely about whether they can just keep ahead of the others. Do you think your industry is kind of immature in that way?

我无法评论其他参与者在做什么或为什么这么做。我认为生态系统中至少有一些其他参与者是负责任的，他们至少以正确的方式思考问题。我同意还有其他参与者并非如此。但我要说的是，我一直努力做的以及 Anthropic 一直努力做的，就是树立榜样，并试图激励其他人效仿。我们总是发表我们在机制可解释性方面的工作，总是公布我们进行的测试，而且老实说，有很多其他公司的研究人员基本上会说：‘嘿，为什么我们不能也这样做？这似乎是负责任的做法，似乎是开发技术的正确方式。’因此，你可以对生态系统中的其他参与者产生影响。

I can't speak to what the other players are doing or why they're doing it. I think there are at least some other players in the ecosystem who I think of as responsible, who are at least thinking of things in the right way. I agree there are others who are not. But I would say what I have always tried to do and what Anthropic has always tried to do is to set an example and to try to inspire others to follow the example. We always publish our work on mechanistic interpretability, we always publish the tests we run, and honestly there are a lot of other companies where their researchers basically say, 'Hey, why can't we do that too? That seems like a responsible thing to do, that seems like the right way to develop the technology.' And so you can have an effect on the other players in the ecosystem.

一个非常明显的领域是你与同行不同，你没有明显排队去讨好唐纳德·J·特朗普。你仍然——你对现任美国总统怎么看？你似乎没有想见他的迹象。

One very clear zone where you are different from your peers is that you have not very obviously queued up to kiss the ring of Donald J. Trump. Are you still—what do you think of the current US president? You show no signs of wanting to see him.

听着，我会换个方式说。我认为支持或反对政府、支持或反对政治家都不是正确的方法，Anthropic 在这些问题上也没有任何立场。我要说的是，Anthropic 了解的是 AI，Anthropic 非常清楚围绕 AI 的政策问题。我们的方法——可能有所不同——是我们思考问题，基于这些问题形成实质性观点，然后说出我们的想法。有时我们会不同意现任政府，就像我们有时不同意上一届政府一样，有时我们也会同意。值得强调一下我们同意的领域，比如能源数据中心建设，我们在白宫做了这个健康承诺。还有很多领域——去年夏天我们看到白宫的 AI 行动计划时，实际上我们喜欢其中的大部分内容；那是一份写得非常好的文件。所以这很复杂，不是简单的反对或支持。但同样，在涉及中国芯片等问题上，是的，我们不同意。在暂停州监管的问题上，是的，我们不同意，因为我们在实质问题上存在分歧。这与喜欢或不喜欢某个特定人物无关。我不认为那种思维方式能让我们摆脱目前的处境。我们必须基于实质进行思考。

Look, I would put it in a different way. I don't think being for or against administrations or for or against politicians is the right approach, or that Anthropic has anything to say one way or another on those topics. What I would say is that what Anthropic knows is AI, and Anthropic knows very well the policy issues around AI. Our approach—which may have been different—is we think through the issues, we try and get a substantive view based on those issues, and then we say what we think. Sometimes we will disagree with the current administration, just as we sometimes disagreed with the last administration, and sometimes we agree. It's worth highlighting the areas of agreement, like the energy data center buildouts, we did this health pledge at the White House. There are a number of areas—when we saw the White House's AI action plan last summer, actually we liked most of it; it was a very well-written document. So this is complex; it's not against and for. But again, when it comes to things like the chips in China, yeah, we disagree. When it comes to putting a moratorium on state regulation, yeah, we disagree because we disagree on the merits. It's not about liking or disliking a particular person. I don't think that kind of thinking is going to get us out of this situation that we're in. We have to think based on substance.

他明天来达沃斯时你会见他吗？

Are you going to see him when he comes to Davos tomorrow?

呃，我可能会。

Uh, I may.

那将是一个有趣的变化。最后一个问题。Anthropic 会怎样？

That would be an interesting change. One final last question. What happens to Anthropic?

会怎样？

What happens?

你的隐含估值据说是 3500 亿。你会考虑今年上市吗？

You have a supposedly 350 billion implied valuation. Would you consider an IPO this year?

听着，我们最关注的是努力做出最好的模型，努力在这些模型之上构建产品，并努力以有用的方式向企业销售这些模型。这个领域有巨大的资本需求，所以这一点应该被考虑。但那是我们的重点所在。

Look, we are most focused on just trying to make the best models and just trying to build products on top of those models and just trying to sell those models to enterprises in ways that are useful. There are large capital needs for the field, so that should be considered. But that's where our focus is.

但这意味着你确实需要很多钱。我的意思是，你不需要问我，也知道这是一个资本密集型领域。

That implies that you do need a lot of money though. I mean, you don't need to ask me to know that this is a capital-intensive field.

所以 IPO 并非完全不可能。

So an IPO isn't completely out of the question.

IPO 和唐纳德·特朗普都是可能的，我们窥见了你的想法，还不错。非常感谢。

Both the IPO and Donald Trump are both possibilities, and we've had a view inside your brain and it's not a bad one. Thank you very much.

谢谢你邀请我，约翰。

Thank you for having me, John.