蒂姆·库克的最后一年？科技巨头赛马会，Anthropic的盈利冲刺

丢掉那个。

Lose that.

看到了吗？

See?

非常简单。

Very easy.

不。

No.

这个字体看起来很棒。

It's it's a great looking font.

我原以为这种橙色会显得有点奇怪，但我很喜欢。

I may I I thought the the the color would be a little weird given how orange it is, but I love it.

我觉得很棒。

I think it's great.

不过我始终会选择Pro版，尤其是Pro Max，因为我最在意电池续航，而这显然是Air系列的一大短板。

And I I was always going to be, though, a pro and in in particular, a pro max guy because I care most about battery life, and and that's obviously a big ding against the air itself.

对吧？

Right?

就像，你需要额外的电池，那个外接电池组才能让设备达到完整的电池续航能力。

Like, that you need the the extra battery, the battery pack add on to be able to sort of get full battery life capacity out of the device.

我觉得，你知道，苹果现在有种尝试推出这些iPhone周边新产品的记录，甚至可以追溯到iPhone 5c，如果你还记得的话。

I think, you know, Apple now has a sort of a a track record of trying to launch these new sort of adjacent arms of the iPhone dating back to even the the iPhone five c, if you remember that.

对，那是第一款...哦对。

Like, right, that was the first Oh, yeah.

彩色iPhone。

Colorful iPhone.

它看起来非常漂亮，但销量似乎不太好。

And it and it looked very pretty, but it just didn't seem to sell very well.

所以他们只尝试了一次就放弃了。

And so they they only had sort of one one go of it.

然后，你知道，快进一段时间，他们显然在很多人呼吁后又重新推出了iPhone mini。

And then, you know, fast forward a bit, they obviously tried to reintroduce the iPhone mini after a lot of people seem to be clamoring for that.

但根据大多数人的说法，苹果最终停产了这款产品，这显然印证了这一点。

But again, by by most accounts that you hear and and obviously played out in the fact that Apple discontinued it.

这些产品就是卖得不好。

Like, these just weren't selling very well.

你可能会说，iPhone Plus系列也属于这一类。

And you might say, like, the the iPhone Plus lineup was also in that camp.

但基本上就是苹果一直在尝试拓展iPhone的这些周边产品线，但出于各种原因，它们就是没能站稳脚跟。

But it's basically like Apple keeps trying to do these adjacent new arms of the iPhone, and for whatever reason, they just don't take hold.

我认为iPhone Air尤其会显得很尴尬。

I think the iPhone Air in particular was always going to be weird.

我的意思是，早在它还只是传闻时我就写过，我觉得它会成为一款定位尴尬的中间产品，因为它既不是iPhone中的最高端机型。

I mean, I wrote about this well before when it was still just rumored that I felt like it would be sort of a weird, in a way, in between phone because it's not the highest of the high end of the iPhone.

对吧？

Right?

也不是苹果希望我们所说的'最经济实惠'的iPhone版本。

And it's not the cheapest, most affordable, as Apple would prefer, we say, version of the iPhone.

所以它有点处于中间位置。

And so it sort of sits in the middle.

是的。

Yes.

它看起来不错，拿在手里感觉也很好。

It it sort of looks nice and it feels nice in hand.

你知道，我自己没有，但在店里用过，手感确实很棒。

You know, I don't have one, but I used it in store, and and it's a great feeling phone.

但再次强调，它处于这种尴尬的定位。

But it's, again, sits in this weird spot.

它对于大众消费者来说太贵了，而且性能、续航和摄像头系统也不够顶级——这些现在应该是旗舰机的基本要求了。

It's like too too expensive to be sort of a mass consumer, you know, device, I think, and it's not fast enough and doesn't have the battery life and the camera system that sort of, you know, would be, I think, required now table stakes for the top of the line.

所以他们把它放在这种奇怪的位置上。

And so they just put it in this weird position.

我确实觉得有趣的是，他们把它和Pro系列更直接地挂钩了。

I do think it was interesting that they they sort of tied it more directly to, like, the pro lineup.

对吧？

Right?

它搭载了更快的芯片，他们试图从这个角度来定位它。

It has a faster chip and and sort of they're trying to angle it that way.

但再次强调，我认为它目前处于这种尴尬的中间设备状态。

But, again, I I view it as this, like, awkward in between device right now.

而且，不管它是否真的被推迟了，我认为毫无疑问的是，如果它大获成功，苹果会竭尽全力推出第二代产品。

And, you know, whether or not it was actually delayed, I think that there's no question that if it was a major hit, that Apple would be moving heaven and earth trying to get a second iteration out there.

对吧？

Right?

就像如果它能像MacBook Air那样受欢迎。

Like, if it was as popular as as the MacBook Air became.

对吧？

Right?

就像苹果公司一直保留这个品牌名称，因为那款设备太受欢迎了，他们能做到这一点。

Like, Apple has famously held on to that branding because that device is so popular, and they were able to do it.

但我要说的是，我同时也要指出，iPhone和MacBook Air最初也并非超级热门。

But I'll say that, and then I say it to say in the same breath that the iPhone the MacBook Air did not start out as super popular.

对吧？

Right?

它最初只是个过于昂贵的产品，对，就是那个很酷的演示——史蒂夫·乔布斯从信封里把它抽出来之类的。

It started out as a way too expensive sort of, yeah, cool demo with with Steve Jobs pulling it out of the envelope and and all of that.

所以它必须经历一个逐步被接受的过程。

And so, it had to work its way into that.

也许苹果认为他们可以凭借iPhone Air再次复制这种成功，但就目前而言，显然它还没有像预期那样腾飞。

And so maybe Apple thinks that they can do that again with the with the iPhone Air, but as of right now, it just hasn't taken off, obviously, is the way they would.

所以从你的话里我听出，你是喜欢17款的。

So you like what I hear you saying here is is you like the 17.

对那一款毫无留恋。

No looking back on that one.

噢，没错。

Oh, yeah.

是的。

Yeah.

所以17 Pro和Pro Max版本我用的这款，但即便是普通Pro版，小尺寸那款也是目前最快的机型。

So the 17 Pro and the Pro Max version that I have, but even the pro pro regular, the smaller version is just it's, you know, what it's the fastest one.

它拥有最好的摄像头。

It's got the best camera.

电池续航也最出色。

It's got the best battery life.

要是你还用着两代前的机型，换这个绝对值得——换我肯定升级。

Like and if you're two two two generations back, I would definitely upgrade if I'm if I'm you.

好吧。

Okay.

那听完我这就去下单。

Well, here I go after this.

这就挣扎着去苹果店再当一回剁手党

Make my crudge my way over to the app the Apple Store and say once again

我能问你一个相关但之前没深入讨论的问题吗？

Can I ask you one one tangential question that you hit upon, but we didn't really talk through?

因为我对你的观点很感兴趣。

Because I'm interested in your perspective.

你提到那篇关于蒂姆·库克退休的Feet报道有四名记者参与。

So you noted that the Feet report that talked about Tim Cook retiring had four reporters on it.

当我读到这篇300字的文章时，我就觉得，好吧。

When I read that and I saw that it's like a 300 word piece, and I'm just like, okay.

所以要么这四位记者都采访了同一个消息源，然后他们决定共同合作完成这个报道，要么他们各自有四个不同的消息源，或者有三种不同的消息源组合之类的。

So either these four reporters all talk to the same source, and they all have, like and they just all decided to work together on this thing, or they have four different sources, or they have three different you know, some mixture of different sources on this.

不过，你在新闻编辑部工作过应该知道。

But, like, you've worked in newsroom.

我早年曾在新闻编辑部工作过，TechCrunch有点不同，我会说它更独立，更像独立运营者而非传统新闻编辑部的风格。

I've worked in newsroom way back in the day, and TechCrunch was a little bit sort of different, much more independent, I would say, than a independent operator than than sort of working newsroom style.

所以我很好奇，当你看到这种情况时，这对你传递了什么信号？

And so I'm curious, like, when you see that, what what is what does that signal to you?

通常，这类大型署名报道的运作方式是：一位记者（可能是也可能不是主要负责该领域的记者）获得线索，然后其他三位在公司内部有不错人脉的同事会翻开他们的通讯录，给公司里认识的每个人打电话。即使他们实际贡献不大，只要投入了时间，就能获得署名。

Typically, way those big byline stories work is one reporter who may or may not be the main beat reporter gets the tip, and then three people who are you know, and and it becomes seen as such a big story by the masthead that three people who have decent sourcing within the company basically open up their Rolodexes and call every single person they know within the company, and they'll get a byline even if they didn't really contribute very much because they put the time in.

这是我的最佳猜测。

That's my best guess.

你觉得这到底是怎么回事？

What do you think it was all about?

不。

No.

我认为你——我觉得你很可能完全正确。

I think you're you're I think you're probably exactly right.

我是说，我一直在想，这些人是不是都有同一个消息来源？

I mean, you know, again, I was thinking through, like, do all these people have the same source?

但我觉得你说得完全正确。

But I think you're exactly right.

我认为是一个人获得了关于这起特定事件的线索，然后另外三人被拉进来是因为他们至少需要找到第二个消息来源来佐证。

I think it's one person has, you know, got a tip about this specific incident, and then the other three were sort of used because they need to get a sec at least a second source on this.

所以，他们动用四名记者的事实表明，他们获取的不止是第二消息源，至少还有几个关于讨论此事的内部人士的消息来源。

And so, you know, the fact that they, though, have four reporters on it suggests that they got more than a second source and that they maybe have a few sources at least about people who are talking about this notion.

但有趣的是，马克·古尔曼——我们之前讨论过，他最初报道了苹果接班人的变动以及库克可能最终卸任的消息——这次对Feet的报道进行了强烈反驳。

But it's interesting in the context that Mark Gurman, who obviously, as we talked about, was the person who originally reported on the turn of succession and and, you know, Cook potentially eventually stepping down, has pushed back pretty hard against the Feet report.

对吧？

Right?

他在通讯稿中直接点名质疑，表示不清楚对方从何处获得这个时间点的消息。

Like, he sort of called them out in his newsletter and saying, like, I I don't know where they're getting the timing of this.

就像我们刚才讨论的，考虑到库克的年龄等因素，这事迟早会发生。

Like, certainly, as we as you and I just talked about, like, you know, this is in the air that this is going to happen at some point just given Cook's age and everything else.

但目前没有任何迹象表明他听到了风声。

But there's no indication that he's heard.

而众所周知，作为目前消息最灵通的苹果记者，他理应能听到相关消息。

And obviously, he's, you know, by all accounts, the best sourced Apple reporter at this point that he's heard that there's anything.

但我也认为，如果这确实是个内部讨论的话题，知情者范围一定非常有限，因为这属于董事会决策层面。

But I also think, like, his sources within like, if this is a if this is really a discussion that's being had, like, amount of people who would be talking about this internally has gotta be pretty small because this is, like, a this is a board decision.

这首先是蒂姆·库克的决定，归根结底是他的选择。

This is a Tim Cook this is first and foremost a Tim Cook decision.

对吧？

Right?

这是由他来做决定的。

He's making this call.

现阶段没人能强迫他离职。

No one's gonna force him out at this point.

特别是在当前股价形势下更不可能，而且无论如何也没人能逼他下台。

Certainly not at this this point of the stock, and no one's gonna push him out regardless.

他一直以来都非常成功。

He's been so successful.

所以这确实取决于他的决定。

And so this is really his decision.

不过对于这些互相矛盾的报道之间的反复拉锯，我也不知道该如何解读。

And so I don't know what to make of the whole back and forth, though, between these conflicting stories.

有趣的是《Feet》杂志完全没有跟进报道。

It's interesting that the Feet hasn't sort of followed up at all.

他们确实没有发布更正，但据我所知，也没有坚持原来的说法。

They didn't certainly didn't issue a correction, but also hasn't, you know, hasn't doubled down as far as I know on it.

所以，是的，这件事就这么悬在那里。

And so, yeah, it's sort of just hanging out there as it is.

是啊。

Yeah.

这让我很困惑。

It's puzzling to me.

我是说，蒂姆·布拉德肖是那篇报道的第一署名记者，他确实是位报道苹果新闻的优秀记者。

I mean, Tim Bradshaw is the first byline on that, and he's he's a really good reporter who does cover Apple.

可能不是专职跑这条线的记者，但报道得足够多，也许有人向他透露了什么。

Maybe not as the beat reporter, but covers it enough that maybe somebody whispered something to him.

所以这类事件的发展总是耐人寻味。

So it is always fascinating how these things develop.

但是，同样地，就连古尔曼也写过一篇关于特努斯是明显接班人的报道。

But, again, like, even Gurman had a story talking about how Ternus was the heir apparent.

所以可能是Feet嗅到了风声，决定全力跟进并试图...是的。

So maybe the Feet kind of caught the scent there and decided to go full on in and try to Yeah.

深挖这个故事的内幕。

Digging what into the story was.

因为即使他们内部已经有了继任者，比如说，从一家没有接班人的公司变成有接班人的公司，你现在可以写个标题说他们加速了继任计划，也许会让事情看起来比实际更重要，但我仍然认为这其中有足够的真实性。

Because even if, like, they they have, like, a successor internally, that could be, like, you know, if they went from a company that didn't have a successor to one that does, now you could, you know, write the headline that they've ramped up succession planning and, you know, maybe make it seem more than it is, but I still think there's enough there that that this is very real.

是的。

Yep.

好的。

Okay.

好吧。

Alright.

我们休息一下吧。

Let's let's head to break.

在我们进入休息前，我想花点时间感谢Techmeme。

Before we go to break, I just wanted to quickly take a moment to to think Techmeme.

如果你访问Techmeme，可能会看到《大科技播客》被推荐在techmeme.com的播客栏目。

If you go to Techmeme, you probably have seen that Big Technology Podcast is one of the featured podcasts on techmeme.com.

我们非常感谢他们的推荐。

We definitely thank them for doing that.

Techmeme当然是一个追踪最新科技新闻的优秀网站。

Techmeme, of course, is a great website to keep up with the latest tech news.

最重要、最优质的新闻总是被置顶展示。

Always the best and most important stories are up at the top there.

当你不听《大科技播客》时，这是个保持信息同步的好渠道。

It's a great way to keep up with things when you're not listening to Big Tech podcast, of course.

好的。

Alright.

广告之后我们马上回来。

We'll be back right after this.

Capital One的技术团队不仅在讨论多智能体AI。

Capital One's tech team isn't just talking about multi agentic AI.

他们已经部署了一个。

They already deployed one.

它x称为聊天礼宾服务，它正在简化购车流程。

It's called chat concierge, and it's simplifying car shopping.

通过自我反思和带有实时API检查的分层推理，它不仅Isn't just help buyers find a car they love.

Using self reflection and layered reasoning with live API checks, it doesn't just help buyers find a car they love.

它能帮助预约试驾、获得贷款预审批以及评估置换价值。

It helps schedule a test drive, get preapproved for financing, and estimate trade in value.

先进、直观且已投入使用。

Advanced, intuitive, and deployed.

这就是它们的优势所在。

That's how they stack.

这就是Capital One的科技实力。

That's technology at Capital One.

如今，每一分钱似乎都应该更努力地工作，但决定把钱放在哪里却让人困惑。

These days, it feels like every dollar should be working a little harder, but figuring out where to put your cash can be confusing.

这正是Wealthfront的用武之地。

That's where Wealthfront comes in.

Wealthfront是一个以科技驱动的金融平台，旨在帮助你将储蓄增长为长期财富。

Wealthfront is a tech driven financial platform built to help you grow your savings into long term wealth.

截至2025年11月7日，他们通过合作银行提供的高收益现金账户能为未投资资金提供3.5%的年化收益率。

Their high yield cash account through program banks offers a 3.5 APY on your uninvested cash as of 11/07/2025.

而且没有月费，没有最低或最高余额要求即可享受该利率，你甚至可以在7天24小时内向符合条件的账户进行免费即时提款，几分钟内到账，让你的资金随时触手可及。

And there are no monthly fees, no minimum or maximum balance to earn that rate, and you can even make free instant withdrawals to eligible accounts in just minutes twenty four seven so your money can always be within reach.

目前，Wealthfront为新客户提供为期三个月、最高15万美元余额的额外0.65%年化收益率加成。

Right now, Wealthfront is offering new clients extra point 65% APY over the base rate for three months on up to a $150,000 balance.

这意味着当你首次开通现金账户时，总可变年化收益率可达4.15%。

That's a total of 4.15% variable APY when you open your first cash account.

立即访问wealthfront.com/bigtech进行注册。

Go to wealthfront.com/bigtech to sign up today.

这是为Wealthfront付费制作的客户见证。

This is a paid testimonial for Wealthfront.

可能无法反映其他用户的体验，且不保证未来的表现或成功。

It may not reflect the experience of others, and there's no guarantee of future performance or success.

Wealthfront经纪业务并非大额利率。

Wealthfront brokerage is not a big rate.

利率可能随时变动。

It's subject to change.

促销条款与条件适用。

Promo terms and conditions apply.

更多信息请参见节目描述。

For more information, see the episode description.

欢迎回到《大科技》播客，今天我们邀请到SpyGlass的MG Siegler。

And we're back here on Big Technology Podcast with MG Siegler of SpyGlass.

您可以在spyglass.org找到SpyGlass。

You can find SpyGlass at spyglass.org.

这是我最喜欢的出版物之一，强烈推荐你去看看，订阅他们精彩的通讯，并成为网站会员。

One of my favorite publications, definitely recommend recommend you check it out and sign up for the great newsletter and become a member of the site.

内容非常棒。

It's great stuff.

在Spyglass过去一个月的文章中，MG，你写到我们正处在这个大科技公司市值剧烈波动的时期。

In Spyglass, over the past month, MG, you wrote about how we are in this middle of this, like, very volatile moment for big tech market caps.

每周似乎都有新的大型科技公司成为地球上最有价值的公司。

Every week, it seems like there's a new big tech company that has the crown for the most valuable company on Earth.

英伟达长期占据这个位置。

NVIDIA has played that role for a long time.

微软长期占据这个位置。

Microsoft has played that role for a long time.

苹果长期占据这个位置，而谷歌，你知道的，已经超越了微软，看起来它可能正在向这个位置迈进，特别是当公司开始可能向Meta等其他公司出售其TPU芯片时，这可能会削弱英伟达的部分领先优势。

Apple has played that role for a long time, and Google, you know, has surpassed Microsoft and seems like it might be making its way there, especially as the company starts to potentially sell its TPUs, the TPU chips, to other companies like Meta, probably cutting into some of NVIDIA's lead.

让我问你一个关于这种波动性的问题。

Let me ask you a question about this volatility.

我是说，显然，如果你想想这些公司，我刚才提到的四家中有三家都在AI领域大展拳脚，而苹果嘛，苹果就是苹果。

I mean, obviously, if you think about all these companies, it's, you know, three out of the four that I just mentioned are big on the AI trade and Apple is, you know, Apple is Apple.

没错。

Yep.

你认为我们看到的这种波动，是否只是股市在试图判断谁将成为AI竞赛的赢家，然后把资金从一家公司转移到另一家？

Do you think the volatility that we've seen has just been, the stock market trying to figure out who the winner is gonna be in this AI race and sort of plowing money from, you know, one into to the other?

这就是这场赛马的本质吗？

Is that the nature of the horse race here?

我认为在最高层面上，是的。

I think at a at the highest level, yes.

我我觉得这个说法大致正确。

I I think that that's roughly correct.

有趣的是，就在几个月前，感觉英伟达几乎已经稳操胜券了。

It is fascinating how it felt like even just a few months ago that NVIDIA was basically running away with the game.

对吧？

Right?

他们达到了5万亿的估值。

They hit the 5,000,000,000,000 valuation.

当时只有微软，我记得是4万亿。

Only, I think, Microsoft at that point was at four.

就在不久前，我写过关于谷歌的文章，提到他们被严重低估了——当然股市已经证明我是对的。其实只要看看他们除了AI之外还有Waymo等项目在推进，任何人都能想到这点。

Google, not too long ago, I wrote about this, like, how I felt like they were significantly undervalued, which, of course, proved to be correct by the stock market, but it's sort of it's nothing, you know, that anyone couldn't have sort of thought about just looking at what they have on their plate beyond even AI with Waymo and everything else coming up.

但谷歌当时的估值只有英伟达的一半，可能还不到微软的水平，而现在他们强势反弹了。

But Google Google was, like, you know, valued at half of of it, certainly, NVIDIA and maybe even at Microsoft at one point just a few months ago, and now they've come roaring back.

而苹果，众所周知，过去十年大部分时间都处于领先地位。

And Apple, as you know, Apple's been in the front of the pack for most of the past decade.

但现在突然之间，正如你注意到的，我也在文章中提到过，这四家顶级公司之间的差距正在迅速缩小。

And now all of a sudden, as you're noting, and, you know, as I wrote a little bit about, they've they basically are collapsing up at that top echelon between those four companies.

现在关键就看当前关于AI的报道和舆论导向。

And it's just a matter of, like, what the current reports and and and narrative is around, in particular, AI.

看起来这些因素会把其中一家推上领先位置。

It seems like we'll push one of them to the lead.

确实感觉谷歌现在拥有这种势头，正是基于你提到的那些关于可能出售TPU的报道。

It does feel like that Google has that momentum right now based off of exactly what you're talking about with sort of the reports about, you know, potentially selling TPUs.

如果感觉他们可能不会失去这项业务的王冠，那显然会对英伟达产生直接影响。

And that obviously has a direct impact on NVIDIA if it feels like that they're going to maybe not lose the crown of that business.

但可以肯定的是，如果他们首次面临来自谷歌的业务压力，这将会削弱他们的光环。

But certainly, if they get if they get some pressure on the business for the first time from Google, that's going to take some sheen off of that.

还有其他外部因素在起作用，比如有报道称沃伦·巴菲特旗下的伯克希尔哈撒韦大举买入谷歌股票，我认为这进一步推动了他们。

There's other external things at play like the reports that Warren Buffett, Berkshire Hathaway took a giant stake in Google, I think helped propel them even further.

对吧？

Right?

事实上苹果仍然是伯克希尔的最大持仓，但他们一直在减持。

And the fact that Apple's still, I think, Berkshire's Berkshire's largest holding, but they've been trimming back.

所以当涉及到股价的角逐时，所有这些因素都在起作用。

And so all of these things are at play when it comes to, yeah, that jockeying of the stock price.

但英伟达，我知道你和Ranjan周五讨论过，他们现在的一些公关风波似乎对股价没有帮助。

But Nvidia, I know you and Ranjan talked about it on Friday, but it was like, you know, their their sort of PR kerfuffles now are are sort of not helping their stock, I think.

总的来说，这无助于增强人们对它的信心感受。

Just not helping in general the the feeling of confidence around the way that that people are feeling about it.

但在这之前，他们之所以觉得有必要推出这种公关策略，是因为众所周知，迈克尔·伯里——史上最著名（如果不是最著名的话）的做空者之一——已经公开反对他们。

But before that happened and the reason why that they feel the need to to sort of come out with this this PR strategy is because, famously, of course, Michael Burry, one of the most famous, if not the most famous short seller of all time, has come out against them.

所以，是的，关于这些事态将如何发展，内部存在很多角力。

And so, yeah, there's there's a lot of a lot of this sort of internal jockeying about how they'll how they'll all play out.

然后是微软，它一方面受到冲击，一方面又略有提振。不过现在与OpenAI的关系更趋稳定，我认为这让其股价也稳定了些。但这一切感觉就像在洗衣机里翻滚，这些公司被搅来搅去，争夺第一第四的名次。

And then and then there's Microsoft who's sort of also getting dinged, but also getting buoyed a bit, but, you know, sort of some level of the OpenAI relationship now being more settled, I think, has, you know, settled their stock a bit more now, but this is all just gonna continue to be like in this it feels like in this washing machine, all these companies getting tumbled around for who's in first, who's in fourth, and whatnot.

好吧。

Alright.

让我说说我对当前局势的犀利看法。

Let me give you my hot take about what's happening here.

这某种程度上结合了乔希·布朗（市中心乔希·布朗）的观点，他写过我们正处于一场赛马比赛中。

And this is sort of a hybrid with something that Josh Brown has has wrote about, downtown Josh Brown, about how we're in a horse race right now.

但我是这么看的。

But here's kind of how I see it.

苹果公司，当然，在人工智能时代之前，长期是全球市值最高的企业。

So Apple, of course, was, you know, pre AI, the most valuable company in the world for a very long time.

他们可能与沙特阿美互换位置，所以我们得说，

Maybe they swap places with, like, Saudi Aramco, so we would have to say, like

对。

Right.

市值最高的上市公司，或者简单说市值最高的公司，你知道，可能偶尔不包括沙特阿美。

Most valuable publicly traded company, but let's just say most valuable company or, you know, excluding Saudi Aramco at, like, one moment.

好的。

Okay.

接着ChatGPT问世了，突然间人们都想押注OpenAI。

So then ChatGPT happens, and all of a sudden, you see people wanting to bet on OpenAI.

他们通过投资英伟达和微软来实现这一点。

And the way that they do that is through NVIDIA and through Microsoft.

而英伟达是与OpenAI关联最直接的企业，因为如果OpenAI的发展依赖于训练和推理的规模，那么作为唯一提供芯片的供应商，英伟达就显得至关重要。

And NVIDIA was the one that's the mostly direct most, direct correlation to OpenAI because if OpenAI is predicated on scale for both for training and inference, NVIDIA, which is the only one supplying them the chips, was gonna be super important.

其次，你会把微软放在这个位置，因为微软对OpenAI有着难以置信的投资，通过知识产权和控制权与OpenAI建立了内在联系，出于多种原因。

And then Microsoft, secondarily, you would put Mike you would put Microsoft there because Microsoft has, again, like, an unbelievable investment in OpenAI, an intrinsic link, to OpenAI through, like, the IP rights and and power over OpenAI, for a very for a variety of reasons.

于是出现了这样一个时刻：英伟达和微软股价飙升，因为人们相信AI技术将成为现实，且OpenAI的领先优势难以撼动——该公司推出了更优的模型和产品，在投资者心目中俨然成为直奔AGI（通用人工智能）或至少是具有重大经济价值AI的最热门候选，这种优势几乎无法匹敌。

So there was this moment where NVIDIA and Microsoft surged ahead because people believed that this AI thing was gonna be real, and they believed that OpenAI had a lead that was really gonna be tough to challenge because the company came out with better models, better products, and and it was just like their runaway favorite to effectively in the minds of investors, the best it had the best chance of doing this straight shot to AGI or at least some economically valuable, economically valuable AI that that was gonna be tough to match.

但近期我们看到的情况是，随着英伟达和微软股价回落而谷歌上涨，我认为这并非意味着谷歌已超越OpenAI。

But what we've seen recently so now that we've we've seen, like, NVIDIA and Microsoft fall and, and Google surge, is that I don't think it's that Google has surpassed, OpenAI.

或许在某些领域确实如此，但我们首次面临尖端模型真正同质化的局面——可以说谷歌、Anthropic和OpenAI的模型现在处于同一梯队。

Maybe in some areas it has, but it's we we are now in the first moment where the state of the art models are really commoditized, where you could argue that Google and Anthropic and OpenAI have models that are, it they they stand together.

某些模型可能在某些应用场景更优，另一些则稍逊，但已不能像一两年前那样断言OpenAI独霸天下。

You know, some might be better for some activities and some might be worse for other activities, but you're not gonna say that OpenAI rules the world the way that it did a couple year or even a year ago.

因此当前现象的本质，未必是OpenAI和英伟达的衰落或谷歌的崛起，

And that's where you see what you're seeing now really is not necessarily the fall of OpenAI and the fall of NVIDIA or the rise of Google.

而是我们正见证模型同质化的历史时刻，在某些情况下还包括因TPU技术导致的硬件同质化。

It's this moment where we see commoditization happen amongst the models and, in some cases, amongst the hardware because of the TPU thing.

当同质化发生时，AI的经济格局就会发生转变。

And and when you see commoditization happens, the economics of AI shift.

所以我很好奇你的看法，我真的很想知道你怎么想。

So I'm curious what you think about it, and I I'm curious what you think.

如果你同意我的前提，即我们开始看到这种商品化趋势，你认为商品化真正的经济影响是什么？

You know, if you agree with my premise that we are starting to see this commoditization, what do you think the economic impacts of the commoditization really are?

是的。

Yeah.

我会这样表述——将你的观点稍作不同的框架：如果你相信你阐述的那些情况属实，也许投资者会这样看——归根结底，在这种情况下，谁在这些未来发展中拥有最佳定位和商业模式？

I would frame it's like the taking what you're saying and framing it slightly differently is that if you believe all that to be true that you're that you're laying out there, maybe investors are just looking at it like, look, at the end of the day then, if that's the case, who has the sort of best position and best business model for all of this, you know, future going forward?

我认为很难反驳谷歌的优势，因为他们不仅拥有运行着大量模型的谷歌云平台，正如你指出的，他们还拥有TPU技术，更不用说搜索业务、Waymo自动驾驶等多元布局。

And I think it would be hard to argue against Google just given that they own not only Google Cloud, you know, on which a lot of models and everything are running, but they the TPUs, as you're noting, obviously, they have the search business and Waymo and all these other things.

而其他公司，如果AI确实开始出现某种程度的商品化，英伟达的业务在某些方面可能会受益，但...

Whereas the other companies, if if AI is starting to commoditize in some way, you know, NVIDIA, their business potentially is impacted in a good way in some cases.

但总体而言，这很可能被视为军备竞赛有所放缓的信号，这对顶层的英伟达是坏消息——因为市场不再需要购买最顶级的配置。或者反过来看，正如我们讨论的，如果谷歌能通过TPU加速追赶的话...

But for the most part, it's probably gonna be looked at as like, well, the arms race is sort of slowing down a bit, and that's gonna be bad news for NVIDIA at the highest level because you won't need to necessarily buy, you know, the absolute top of the line and or on the flip side, as we're talking about, if Google's able to sort of ramp TPUs to get close.

这其实正是英伟达声明中隐含的信息——他们强调自己领先一代，显然是在明确表态：谷歌做得不错，但还无法企及我们的现阶段的水平。

That was part of, obviously, the messaging from NVIDIA was we're a generation ahead, right, in their statement, which was obviously explicitly meant to say that, like, yeah, Google's doing great work, but they're not where we are right now.

但回到你的问题，是的，英伟达显然处于一个需要AI革命持续当前势头的位置，以维持他们惊人的增长，包括上个季度那令人难以置信的表现。

But to your, you know, to your question, yeah, like, so NVIDIA is obviously in a position where they need the AI revolution to sort of keep going as it has been in order to keep up the just the incredible growth that they've seen, including last quarter, which was incredible.

至于其他玩家，微软，正如你提到的OpenAI，他们显然也有强大的基础业务，但我不认为他们目前拥有谷歌所具备的所有要素。

And then, you know, the other players, Microsoft, to your point about OpenAI, they also obviously have a great underlying business, but they don't have quite I I don't think is is viewed as all of the pieces that Google necessarily has right now.

当人们审视未来可能实现盈利的各个领域时。

And when people are looking at all the different areas that they can potentially monetize for future, you know, would be profits.

而苹果则是那个未知数，这就是为什么我认为苹果——我们之前讨论过不受资本支出影响——但苹果目前处于第二位。

And then Apple is the, you know, the wild card, which is why I think Apple you know, we talked about not getting hit by the CapEx stuff and everything, But Apple is, you know, about to they're in second place right now.

他们很可能会在某些时候超越英伟达。

They're probably gonna overtake NVIDIA at some points.

接下来就是我们讨论的继任问题，未来的故事会如何发展？

And then it's just a question of exactly what we were talking about in the succession stuff, like, what is the future story there?

所以可能会出现这样的情况：根据AI整体框架和叙事的发展，苹果在一段时间内重新占据首位。

So, like, it could be, you know, the situation where Apple basically moves into first place again for a set period of time depending on what's going on within the AI, you know, overall framework and narrative stuff.

最终，谷歌可能会反超，这取决于我们处于这个周期的哪个阶段。

And then eventually, you know, Google passes it and then depending on where we are again in that cycle.

但再次回到具体问题上，我确实认为如果我们正朝着这种商品化的方向发展，这就是我们当前的现状。

But again, to go back to the exact question, I do think that if this if we are sort of working towards, yeah, more of this commoditization, that's the state that we're in right now.

我认为这就是为什么这四家公司现在开始抱团，而不是英伟达一家独大，或者像之前微软一度领先那样。

And I think that's why those those four companies are getting bunched up more than just NVIDIA running away or, you know, earlier when Microsoft was running away a bit.

其中的变数，下一步可能是下一代模型的出现。

The the wildcard there, the next step might be, you know, next generation models.

我们现在听到很多人都在讨论世界模型。

We're hearing a lot of people talk about world models now.

对吧？

Right?

当然，Yann LeCun（杨立昆）已经离开Meta去从事那项工作，这是众所周知的。

Leon Lecun, of course, famously has has stepped away from Meta to go work on that.

还有Ilya Sutskever（伊利亚·苏茨克沃），这基本上就是他最近访谈中一直在谈论的内容。

But also Ilya Sutskimmer, that's all he basically talked about, you know, in his in his recent interviews.

而且，没错，这基本上就是新趋势。

And And yeah, it's basically like this is the new thing.

就连Demis Haspis也一直在讨论世界模型。

And even Demis Haspis has been talking about world models.

对吧？

Right?

所以这俨然已成为当前最热门的话题。

And so like this is the entire now thing.

如果这成为新的叙事方向，显然将成为可能重新洗牌这些公司的下一个变量。

If this becomes the new narrative, that will obviously be the next thing that potentially sort of switches up or shakes up those companies.

关键还是看它们与这一技术路线的整合程度。

Again, it's how integrated or not they are with that with that, lineup.

是啊。

Yeah.

我在想，如果世界模型真是下一个风口，我们现在对这个领域的探索还处于非常早期的阶段。

I just wonder, I mean, if if if world models really is the next thing that's such a we're such at or at such a early point on that front, I don't know.

可能还要经历很长一段平台期才会有实质性突破。

We might have a long stall before anything.

这种方法是否真的有效，谁也说不准。

If that approach even works, no one really knows.

是啊。

Yeah.

而且我们可能会看到，显然埃隆·马斯克会试图重新加入这场竞争。

And we might see I mean, obviously, we're gonna have Elon Musk try to come back into this this equation.

对吧？

Right?

他已经在鼓吹他的机器人业务将成为有史以来最庞大的产业——不只是特斯拉的最大业务，而是人类历史上任何领域最宏大的商业版图。

And he's already been talking about how, excuse me, the you know, his his robots are optimists are going to be the biggest business ever, not just like the big, you know, biggest business of Tesla, the biggest business of all time for anything ever.

所以他对此大肆炒作，因为他可能认为世界模型是其中的关键环节，而机器人领域正是下一个风口。

And so, you know, he's hyping that up to no end because I think that he probably views this as like, yeah, that's the next sort of the world models are a part of that and, you know, we go into robotics.

如果这真是下一个价值数万亿美元的商机，那么特斯拉就有机会入局，把竞争变成五强争霸。

And if that's the next multi, multi trillion dollar sort of business opportunity, then Tesla has a shot to get into that race, you know, and be make it a five a five horse race.

当然，Meta肯定也希望成为参赛选手之一。

Obviously, Meta would hope to be the one to get also into that race.

还有亚马逊，我们甚至还没谈到它，你知道，它就坐在那里，我想，大概只落后其他竞争对手一万亿美元左右吧，不管它们现在具体是多少。

And and Amazon, haven't even talked about, you know, being sort of sitting there, I think, what, a mere trillion dollars back or whatever they are at the moment from sort of the rest of the pack.

所以，没错。

And so Right.

所有这些都会参与进来。

All of that comes in.

是啊。

Yeah.

Meta似乎也像是在打上一场战争，他们组建了一个超级团队来构建大语言模型。

It does also seem like Meta is is sort of fighting the last war on this, where they're like, they've assembled a super team to build LLMs.

但让我问你这个问题。

But let me ask you this.

假设这就是我们目前的处境。

So so let's say that's where we are.

我不想说我们在大型语言模型上的进展正在减弱，但可能正达到某种平台期，而且这些模型正在商品化。

We're sort of, you know, I don't I don't wanna say petering out on progress with large language models, but maybe reaching some form of plateau, and and the models are commoditizing.

你赌谁会赢？

Who's who do you bet on?

你会押注谁？

Who would you bet on?

你会押注哪一方？这里再次借用乔希·布朗的二分法——是选择由OpenAI、微软和甲骨文组成的'OpenAI电影宇宙'，还是选择谷歌阵营及其盟友？

Would you bet on and this is again using Josh Brown's, like, dichotomy here, the sort of or the OpenAI cinematic universe where you have, like, OpenAI Microsoft and Oracle or the Google universe where you have, you know, I don't know, and some others.

嗯。

Yeah.

说实话，如果真要用真金白银押注这些公司，我会效仿它们的做法——互相押注对方。

I mean, if I if I was, like, actually betting in, you know, money on these companies, I would do what these companies are doing and I'll bet on each other.

这叫对冲

That's Hedge

你的赌注，这种循环投资在这个领域确实说得通。

your bets, not circular financing actually makes sense in this front.

这就是我的策略。

How I would do it.

但如果你要我真正押注，选出一个赢家，你知道的，我曾在谷歌风投部门工作了很久。

But if you're asking me to actually bet to pick one winner of it, I do you know, and I've you know, I I worked at Google for a long time for Google's venture arm.

你知道，我已经离开那里有一段时间了，所以先说明一下。

You know, haven't been there for a while, so just to disclose that.

但我确实认为，考虑到谷歌拥有的一切资源，感觉就像那头'野兽'已经觉醒，我觉得他们目前处于最佳位置，能够实实在在地利用这一切。

But but I do think that Google's just given all of the things that they have, it does feel like the, quote, unquote, beast has awoken, and I feel like that they're in the best position right now to be able to to capitalize on all of this stuff quite literally.

不过，我再次认为，虽然我对AI模型远非专家，但我认同这一观点：仅凭我们目前的大语言模型，还不足以实现人工通用智能（AGI）或所谓的下一阶段发展。

I think that, again, though, this I I do buy into the notion, and I'm far from an expert on, you know, on AI models, but I do buy into the notion, the high level notion that it's going to take more than where we are with LLMs to be able to take us to AGI or whatever you want to call it, the next the next phase of where we're going.

因此，如果你相信这一点，问题实际上在于你认为哪家公司最有能力从中获益。

And so if you believe that, it's really a question of who do you think is best positioned as one of these companies to take advantage of that.

正如你所说，Meta给人的感觉是他们正在眼睁睁看着那艘船驶向Yan Lakun（音译），他们可能会投资他的初创公司，但他们显然已经押注了相反的方向。

To your point, Meta feels like they're they're literally letting that ship sail in in Yan Lakun and they'll invest or whatever in his startup, but they obviously made their bet in the opposite way.

苹果公司则显得有些置身事外。

Apple, that is sort of off to the side.

但亚马逊，你知道的，他们显然在内部研发自己的模型，不过他们显然还对Anthropic下了重注。

But Amazon, you know, they're apparently working on their own internal I mean, they've been working on their own internal models, but obviously, they have the giant Anthropic bet.

但至少从目前没有任何消息泄露来看，Anthropic似乎并未真正主要专注于下一代世界模型之类的东西。

But it doesn't feel like, at least from there's been no leaks of anything that Anthropic is sort of really primarily focused on, you know, sort of the next world models and whatnot.

所以，我认为德米斯·哈萨比斯最近频繁公开谈论这个话题的事实本身就说明——即使不是世界模型，他们也在深入思考接下来可能出现的突破性进展。

And so, again, I think, like, the fact that Demis Hassabis has been out there talking a lot about this, like, you know that he's thinking even if it's not world models, there's some other what's next coming up that that I think that they're heavily thinking about.

在我看来，微软也有点像是在打最后一仗的感觉。

Microsoft to me, it also feels a bit like they're on the last battle.

根据你听到的公开报道，他们现在似乎陷入了困境。

They're stuck there right now just from, you know, public things that you hear and reports.

所以感觉就像我们之前讨论过的，关于OpenAI的历史，确实感觉他们已经押注在AGI或BUS上了。

And so it feels like it's either we talked about this before, like with, you know, the history of OpenAI, it does feel like that they've made their AGI or BUS bet.

然后问题就变成了，他们能否找到越来越多的人给他们提供足够的资金？

And then it's just a question of like, can they find more and more people to give them enough capital?

因为他们现在处于一个奇怪的阶段，微软虽然拥有公司27%的股份，但已不再是他们的主要资金伙伴，甚至也不再是他们未来真正的主要技术合作伙伴。

Because they're sort of now at this weird point where they don't have Microsoft, yeah, owns 27% of the company, but they're not their main capital partner anymore, nor are they even their real main technology partner sort of going forward.

所以他们需要继续引入巨额资金，而谷歌却盈利惊人。

And so they need to, like, continue to bring in just an insane amount of capital, whereas Google is insanely profitable.

所以，你知道，他们已经具备了实现这一目标的各个要素。

And so, you know, and they have the pieces in place to do it.

因此，尽管我很不愿意押注一个市值超过3万亿美元的公司去对抗一个所谓的初创企业，但感觉这里确实存在一个机会窗口。

So, you know, as much as I hate to to bet on a, you know, 3 plus trillion dollar company against a, quote, unquote, startup, it just feels like there's there's a window of opportunity here.

回顾OpenAI时，我认为他们最擅长的始终是在产品方面。

I think OpenAI, when I go back to it, what they've always done the best of is on the product side.

对吧？

Right?

他们一直非常擅长将AI产品化，我认为这正是他们直接获得消费者成功的原因。

They've always done a great job being able to productize AI, and I think that that's what's directly led to their consumer success.

不过现在谷歌正在逐渐蚕食这部分市场。

And, yeah, Google is is eating into that a bit now.

感觉你之前有篇文章讨论过谷歌如何真正构建了一些...

Feels like You had a piece about this talking about how Google's, like, actually been able to build some Yeah.

那些新东西。

The new the new stuff.

无论是NotebookLM。

Whether that's NotebookLM.

对。

Right.

没错。

Right.

顺便说一句，不得不说NotebookLM团队确实发过推文，比如展示节日食谱，结果被人发现他们只是直接照搬了某个食谱。

Which, by the way, have to say, NotebookLM, the team there did tweet, like, here's a holiday recipe and, like, somebody, like, saw that they had, like, just taken a recipe.

他们当时在做节日信息图，有人发现他们根本就是从谷歌索引的网站上扒了个食谱，包装成精美格式后宣称这是AI生成的。

They they they were doing holiday infographics, and someone saw that they just literally jacked a recipe from some website indexed by Google and put it in a nice format and said, this is AI.

天啊。

Oh, boy.

这也太敷衍了吧。

And it's just like, oh, come come on.

你们不能这么干啊。

Like, you can't do that.

但是，但你其实还写过这个动态视图，我我找不到了。

But, but there's also this dynamic view that you actually wrote about, which I I couldn't find.

这是只针对谷歌，Gemini的高级用户吗？

Is it just for, like, the Google, the the pro users of Gemini?

但基本上你可以...我甚至不知道该怎么解释它。

But, you basically are able to I don't even know how to explain it.

这是一个可以通过Gemini提示创建的交互式网页之类的东西。

It's this interactive web page or something that you can create via a prompt with Gemini.

这是个非常酷的产品。

It's a very cool product.

是的。

Yes.

有件事要说明，它目前还不支持移动端。

So one thing, it's not on mobile yet.

所以我不确定你是不是在手机上找的。

So I don't know if that's where you were looking.

我只有台式机。

I've only had a desktop.

是啊。

Yeah.

不过，我是专业版订阅用户，所以可能就是这个原因。

But, yeah, I am a pro pro subscriber, so maybe that was it.

而且它还在实验室阶段，尚未正式推出。

And it's in it's in labs, so it's not officially rolled out.

但它刚刚可用。

It But was just available.

可能我被分到了某个AB测试组。

Maybe I'm in some bucket AB test for it.

不过，它确实在下拉菜单里，就在'nano banana'附近——你知道的，当你下拉选择生成图片时，也能选择创建这种动态视图演示。

But, yeah, it was in the drop down, and it it was it's near nano banana, you know, when you do the drop down and pick if you wanna make an image, you could also pick to make this this type of presentation dynamic view.

再说一次，这名字真糟糕。

Again, awful name.

但它产生的效果实在太酷了。

But it's such a cool output that it does.

我在我的帖子里举了些例子。

I give the examples in in my post.

其中一个例子是向我解释电影《盗火线》的结局，就是那部九十年代著名的迈克尔·曼电影。它生成了这个令人难以置信的交互式输出，虽然我描述起来可能显得很基础，但你得亲眼看看这东西运作时的样子。

One of them was was explaining the end of the movie heat, the the famous nineties Michael Mann movie to me, And it made this incredible, just interactive output, which, again, it's sort of when I'm explaining it, it sounds like rudimentary, but you gotta see this thing when you do it.

而且你可以用它来处理任何内容。

And and you could do it for anything.

它基本上会把常规Gemini查询的文本输出格式，转换到这种交互式模型中，包括制作这些非常直观又有趣的元素。

And it basically takes what, you know, the output would be in a in a text format from a regular query on on Gemini, and and it puts it into this sort of interactive model, including, like, making just these, like, incredibly intuitive and, like, fun things.

比如我为《星际穿越》结局做了个示例，它能生成一个按钮来播放背景氛围音乐。

Like, I did one for the end for, you know, explaining in the movie Interstellar to me, and it, like, makes a button where you can turn on sort of background mood music that's in the style of Interstellar.

同时你还能计算围绕黑洞飞行时的时间膨胀效应。

And then while you do that, you can calculate, like, time dilation when you're going around a black hole.

这就像把所有原本只是文本输出的内容，用这种有趣的方式呈现出来。

And it's like it's all this, like, just really fun, interesting way to sort of present what would normally just be, yeah, like a text output back to you.

再说一次，这听起来可能很基础，而且在某些方面目前确实如此。

And again, it sounds like rudimentary, and it is in some ways right now.

但你想一想，他们仅用一分钟就根据我随意提出的查询做出了这个。

But you think about, like, they made this in one minute based off of a random query that I do.

想象一下，一年后他们将能做到什么程度。

Just think about, like, what they'll be able to do a year from now.

而且如果你采用思考模式，给他们几分钟时间来构建一个交互式模型，我认为输出结果将会令人难以置信。

And if and if you do, like, a thinking version of this where you give them a few minutes to come up with, like, sort of an interactive model, like, the outputs, I think, will be absolutely incredible.

当你想到孩子们用这种方式学习时，比如我做了一个解释股票回购的示例，这简直是一种绝佳的互动学习方式。

And when you think about, like, kids learning this way, like, I did one for, like, explaining stock buybacks, and it's just like this great interactive way to to learn.

因此只要你能信任它不会产生幻觉并确实提供准确信息，我认为谷歌这次完成了一项令人惊叹的产品工作，这就是我专门撰文介绍它的原因。

And so as long as you can trust that it's not gonna hallucinate and actually give you good, you know, good information, I think that this is just like an incredible bit of product work that that Google has done here, that's why I wrote about it.

不知为何，它似乎被低估了，部分原因是这个名字太糟糕，部分是因为它还在实验室阶段之类的。

It felt like it was being underplayed for for whatever reason, in part because it has such a bad name and in part because it's in labs or whatnot.

但这恰恰印证了你的观点，谷歌似乎终于在产品端也觉醒了，开始有能力做出能与OpenAI历史成就相媲美的有趣东西。

But but, again, that's to your exact point, like, feels like Google has finally woken up also on the product side to be able to do interesting things to compare what OpenAI has historically done.

是啊。

Yeah.

我刚才还在看他们的实验室页面，上面全是些有趣的东西。

I was just on their labs page right now, and it's just filled with interesting things.

好吧。

So alright.

听着，我们只剩几分钟了。

Look, before we we have just a couple of minutes left.

结束前，我觉得你指出的另一点值得注意——Anthropic和OpenAI目前的发展路径截然不同。

Before we go, I I think you charted out one more interesting, thing that's worth noting before we leave, and that is that, Anthropic and OpenAI have, like, really, divergent paths right now.

如果看他们的规划，Anthropic计划在2028年实现盈利（虽然听起来很遥远，但在AI领域已经相当不错），而OpenAI预计到2030年前将亏损超过1000亿美元，且亏损额会逐年递增。

If you, like, look at their projections, Anthropic is planning to be profitable by 2028, which I know sounds like a long way away, but in AI years is pretty damn good, where OpenAI is planning to lose, more than a $100,000,000,000 between now and 2030, and the losses will just basically keep ramping as we get closer.

你觉得这反映了两者之间怎样的竞争态势？

So what do you think this says about the competition between the two?

还是说Anthropic只是未雨绸缪，认为AI行业可能面临低谷，所以提前做好防御准备？

And is Anthropic just being smart and saying we're probably gonna have some AI downturn and we're gonna try our best to sort of batten down the hatches hatches and get through it?

我确实认为这是其中一部分原因。

I do think that that's a part of it.

我觉得这与我们讨论的其他话题在时机上颇为有趣。

I think that it was it's sort of interesting timing in everything else that we're talking about.

对吧？

Right?

就像，虽然感觉模型正在趋于统一，但同时整个AI投资领域又笼罩着这种负面氛围——从公开市场情绪往下蔓延。看起来Anthropic确实可以打出这样的牌：'我们正处在技术前沿，但不会像OpenAI那样挥霍无度'。

Like, if it does feel like, you know, that the models are sort of coalescing, but also that there's this this sort of negative, know, backdrop going on around just investing in AI in general from the public market sentiment on down, it feels like, yeah, if Anthropic can sort of take advantage of like, look, we're at the cutting edge right now, but we're not going to be the sort of drunken sailor spend that OpenAI is trying to do.

所以快来加入我们吧。

So come on board and bet with us.

最新报告显示他们的新估值已达到3500亿美元。

The most recent report has their new valuation at 350,000,000,000.

所以他们已经很接近OpenAI的5000亿美元估值了。

So they're getting pretty close to where OpenAI is at at 500,000,000,000.

感觉他们正在逐步逼近那些数字，这非常有意思。

It feels like, you know, they're they're sort of coming in, you know, in line with those with those numbers, which is super interesting.

这再次印证了他们所采取的两种路径之间的鲜明对比。

And again, maybe speaks to the dichotomy between the two paths that they're taking there.

挺有意思的。

It's funny.

我刚写完那篇文章，结果第二天就发现——那篇文章里提到很多点，但其中有个旁注说：看，他们并没有把所有钱都砸在自建数据中心上。

I wrote that post and then literally the day later, I wrote like, you know, number of things in that post, but one aside was like, And look, they're not spending all this money on building their own data centers.

结果十二小时后他们就宣布要建第一个数据中心了，果然如此。

And literally twelve hours later, they announced their first data center that they're going to build, of course.

但只要规模可控，只要不是'星际之门'那种烧钱项目，不把数千亿美元都投在数据中心建设上，你确实可以说他们应该拥有一定自主基础设施，这样才不会被合作伙伴掣肘。

But as long as it's one, as long as it's not the Stargate project and they're not spending, again, hundreds of billions of dollars on building out their own data centers, you could certainly argue that they should probably have some level of their own infrastructure in place so they're not beholden to to, you know, their partners on that stuff.

特别是考虑到我们之前聊过的那些错综复杂的内部投资关系——确实如此。

And especially given all the weird as we talked about, all the weird inner investing going on where it's like, yeah.

亚马逊虽然是他们的主要合作伙伴，但他们现在也开始从谷歌采购TPU芯片了。

And Amazon is is their main partner maybe, but they're also now buying TPUs from Google.

然后呢？

And then you know?

但谷歌也在，你知道，可能正在与X、Y、Z等公司合作。

But Google's also, you know, potentially partnering with, you know, x, y, and z.

而OpenAI现在也可能与亚马逊合作。

And then OpenAI is potentially partnering with Amazon now too.

所以我觉得Anthropic可能会这样看：我不确定我们能只依赖亚马逊，也不确定能只依赖谷歌。

And so there's like, I think Anthropic probably is to look at this as like, I'm not sure that we can rely on, you know, just just Amazon, and I'm not sure we can rely on just Google.

所以某种程度上他们必须对这种长期赌注进行。

And so in some ways, they have to sort of hedge that long term bet.

我只是觉得，他们可能正在明智地审视当前市场状况，认为增长势头开始略有放缓。

I just think, like, they might be and might smartly be looking at the market as it is right now and saying, like, the music is starting to slow a little bit.

我们不要承诺投入数千亿美元。

Let's not commit to spending hundreds of billions of dollars.

让我们承诺为未来投入区区数百亿美元，看看未来几年情况如何发展。

Let's commit to spending mere tens of billions of dollars for our future, and let's see how this shakes out over the next couple of years.

一方面这很有趣，因为我认为Anthropic的人相信只要扩大训练规模，就能实现通用人工智能。

On one hand, it's interesting because I think people with an anthropic believe that if you just scale up your training runs, then you can get to AGI.

另一方面，或许这是AI研究实验室业务终于迎来了一些理性思考，他们明智地表示：好吧，让我们试着赚点钱。

On the other hand, maybe this is some rationality finally coming into the AI research lab business where they're smartly saying, okay, hey, let's try to make some money.

也许这将是个好方法，能确保我们长期达成目标。

And maybe that will be a good way to ensure that we'll reach our goal in the long term.

是啊。

Yeah.

我唯一想补充的是，事情自然发展成了这样——Anthropic在企业端做得非常出色，对吧？

And the only other thing I would add is just like, it sort of has naturally played out the way that Anthropic has done a great job on the enterprise front, right?

而OpenAI则在消费者端表现出色。

Whereas OpenAI has done a great job on the consumer front.

所以这里又出现了那种微妙差异，显然OpenAI也在企业编码等领域进行尝试。

And so we have that sort again, there's nuance there, and obviously, OpenAI is also trying on the coding front in enterprise and whatnot.

但宏观来看，它们已经分化成了这两个方向。

But at a high level, they've sort of splintered off into those two worlds.

因此其中一条路可能更容易盈利，但我们似乎正处在OpenAI可能推出首批广告的边缘——如果某些

And so the one is going to be an easier path potentially to making money, but we're we seem to be on the cusp of of OpenAI potentially rolling out their first ads if some of the

哦，是的。

Oh, yes.

如果相信那些泄露的内部版本的话。

The leaked internal builds are are to be believed.

这将成为明年一个非常有趣的看点，对吧，看看事情会如何发展。

And that will be a whole interesting element to the to the next year, right, of how how that plays out.

没错。

Right.

所以其中一些迹象确实出现在安卓代码里，某些地方

So some of that did show up in, I think, the Android code, some places for

对。

Right.

广告。

Advertising.

所以这显然很可能在明年某个时候就会到来。

So that's clearly coming probably some at some point next year.

说到明年，那将是我们下次交谈的时候了。

Speaking of next year, that's gonna be the next time you and I talk.

所以我想我们可能会在一月份开始，或许会做一些预测，或者回顾十二月发生的一些疯狂事情。

So I think we'll kick off January maybe with some predictions or some recaps of some crazy stuff that happens in December.

但是，MG，只想说很高兴你能来参加这个节目。

But, MG, just wanna say it's been great having you on the show here.

我想我们是从2025年开始的，这对节目来说是个很棒的补充。

I think we started in the 2025, and it's been a great addition to the show.

期待未来能有更多合作。

So looking forward to doing more together.

非常感谢，亚历克斯，新年快乐，也祝所有听众新年快乐。

Thanks so much, Alex, and happy New Year, and happy New Year to all your listeners.

是的，我们六月初再聊。

And, yeah, we'll catch up in early June.

好的。

Alright.

我们一月份再继续。

We'll pick it up in January.

好的，各位。

Alright, everybody.

感谢MG。

Thank you to MG.

感谢大家的收听和观看。

Thank you to all of you for listening and watching.

周三我们将请来几位Anthropic的研究人员，他们发现AI模型中一个非常有趣的现象——这些模型如何逐渐变得邪恶，以及如何阻止它们走向黑暗面。

We'll be back on Wednesday with a couple of anthropic researchers who found a very interesting pattern in AI models and how they've sort of turned evil, and and how you might be able to stop them from going to the dark side.

这期节目将在周三播出，然后周五我和Ranjan会回来盘点本周新闻。

So that's coming up on Wednesday, and then Ranjan and I will be back on Friday to break down the week's news.

再次感谢大家，我们下次《大科技播客》再见。

Thank you again, and we'll see you next time on Big Technology Podcast.