本集简介
双语字幕
仅展示文本字幕，不包含中文音频；想边听边看，请使用 Bayt 播客 App。
据报道，OpenAI的高层管理人员在IPO计划的开支问题上存在分歧。
Top OpenAI executives are reportedly at odds over its spending in IPO plans.
苹果一直在为Siri制定计划，但至今仍没有实质成果。
Apple keeps making plans for Siri and, well, nothing to show for it yet.
Meta能否找到下一个爆款产品？
And can Meta find its next hit?
这将在M环节中揭晓。
That's coming up with M.
G。
G.
这段内容之后，我们将连线Spyglass的西格勒。
Siegler from Spyglass right after this.
欢迎收听《大型科技》播客。
Welcome to Big Technology Podcast.
今天是本月的第一个星期一，还有M。
It's the first Monday of the month, and M.
G。
G.
西格勒今天和我们在一起，像往常一样为我们剖析科技界正在发生的事情。
Siegler is here with us in his traditional spot to break down what's going on in the world of tech.
今天我们有很多话题要讨论，包括OpenAI高层的更多人事变动以及这意味什么。
We have a lot to talk about today, including more turnover at the top of OpenAI and what that means.
我们还会稍微谈谈OpenAI和Anthropic似乎正朝着同一款产品发展，然后我们再回到这些讨论中的一个常见主题——苹果公司，以及它对Siri的计划是否真的能实现。
We'll also talk a little bit about OpenAI and Anthropic seemingly headed toward the same product, then we will return to a common theme in these discussions, Apple, and whether that company's plans for Siri will ever materialize.
最后，为什么Meta似乎始终找不到下一个爆款产品。
And then finally, why Meta seems not to be able to find its next hit.
M。
M.
G，很高兴见到你。
G, great to see you.
欢迎回到节目。
Welcome back to the show.
谢谢，亚历克斯。
Thanks, Alex.
伦敦这里阳光明媚，外面天气真好。
Super sunny here in London, so it's nice outside.
本周密歇根狼獾队打进了全国冠军赛，所以我很高兴我们现在就录这个，而不是等到明天，那时候我可能得熬到凌晨四点。
Got Michigan Wolverines in the national championship game, this week, and so I'm glad we're recording this right now and not say tomorrow when I might be, staying up till about four in the morning.
天啊。
Oh, goodness.
这一周肯定会在多个方面都很忙，尤其是OpenAI团队必须再次弄清楚高层发生了什么，因为就在我们录制的这段时间里，有最新消息传出：OpenAI首席执行官萨姆·阿尔特曼和首席财务官莎拉·弗里尔之间可能存在分歧。
Well, it's definitely gonna be a busy week on a number of fronts, and certainly, the OpenAI team is going to have to figure out what's happening once again, in the C Suite because there's breaking news that we have coming, oh, you know, basically within our our recording window here that, there is potentially some discontent between OpenAI CEO, Sam Altman, and CFO, Sarah Friar.
这是来自相关信息的报道。
This is from the information.
萨姆·阿尔特曼已承诺OpenAI在未来五年内投入6000亿美元，并私下表示，尽管预计公司将在实现盈利前烧掉超过2000亿美元，他仍希望在第四季度尽快上市。
Sam Altman has committed OpenAI to spend 600,000,000,000 in the next five years and privately said he wants to go public as soon as the fourth quarter despite expectations his company will burn more than $200,000,000,000 before it starts generating cash.
幕后，他的首席财务官莎拉·弗里尔表达了对CEO这一极其雄心勃勃计划所蕴含的紧张局势和风险的担忧。
Behind behind the scenes, Sarah Friar, his chief financial officer, has voiced concerns that reflect the tensions and risks inherent in the CEO's extraordinarily ambitious plan.
她今年早些时候告诉一些同事，她不相信公司在2026年之前会准备好上市，因为还需要完成大量的程序性和组织性工作，而且其支出承诺也带来了风险。
She told some colleagues earlier this year that she didn't believe the company would be ready to go public in 2026 because of the procedural and organizational work needed and the risks from its spending commitments.
她还表示，不确定OpenAI在未来几年是否真的需要投入如此巨额资金购买AI服务器，或者其增长放缓的收入能否支撑这一承诺。
She also said she wasn't sure whether OpenAI would near need to pour so much money into obtaining AI servers in the coming years or whether its revenue growth, which has been slowing, would support the commitment.
而似乎正是因为这一点，Altman已将她排除在一些与公司财务计划相关的讨论之外。
And seemingly as a result, Altman has excluded her from some conversations related to the company's financial plans.
如今，OpenAI发布了一份声明，称Altman和Friar之间一切安好，但这似乎又是OpenAI高层的一场新风波。
Now OpenAI has a statement between Altman and Friar saying everything is hunky dory, but this does seem like yet more drama at the top of OpenAI.
一方面，我预计任何有经验的CFO都会与OpenAI这样极其雄心勃勃、完全不符合以往任何模式的计划产生分歧，所以我无法判断这究竟是公司内部的重大问题，还是仅仅是一家以史无前例速度增长的公司，在财务与产品战略交汇时必然出现的典型摩擦。
On one hand, I would expect this type of disagreement between any CFO who's done any sort of financial job and, OpenAI with its, like, unbelievably ambitious plans that don't really fit any model previously, so I can't tell whether or not this is, a big deal and another problem within the company or whether, basically, this is just kind of business as usual for a company growing at an unprecedented pace that's gonna deal with some, you know, typical tension at that juncture where the finance and the product strategy may meet.
你觉得呢，M？
What do you think, M.
G？
G?
所以，是的，我最初的想法和你刚才说的一样，心想：天哪。
So, yeah, my first initial thought, same as what you just said, it was like, oh my god.
我无法相信一位CEO和CFO会意见不合。
I can't believe a CEO and a CFO are at odds.
这种情况发生的几率有多大？
What are what are the odds of that?
这种事情在每家公司都会发生，因为显然，CFO常常得扮演坏人的角色，提醒大家要控制开支，或者给任何未来的计划泼点冷水。
Like, that happens in every company, of course, because, obviously, the CFO often hot has to be the bad cop telling, you know, slowing down spend or or sort of putting a a dose of reality into, you know, any sort of, future plans.
但在OpenAI，这家以不按常理出牌著称的公司，事情总是走向极端。
But at OpenAI, not a normal company famously, things are always sort of to the extreme.
我认为这次的情况特别独特，因为我们讨论的是IPO。
And I think this one is obviously unique in the case that we're talking about the IPO.
当然，自从去年起就一直有传闻说，OpenAI和Anthropic至少已经开始为今年可能上市做准备了。
You know, this had been rumored, of course, dating back to last year that that basically OpenAI and, of course, Anthropic are sort of in starting to put the wheels in motion at least to go to go public potentially this year.
我曾在年终预测中预测，OpenAI和Anthropic都不会在2026年上市。
I actually predict as one of my year end predictions, I predicted that neither OpenAI nor Anthropic will go public in 2026.
几周前，这个预测看起来还挺不靠谱的，好像上市真的要发生了。
That was looking fairly bad, I would say, a few weeks ago, like it like it was gonna happen.
但你知道，现在出现了这样的报道，其实并不令人意外。
But, you know, there's reports like this now coming out that are not too surprising.
再次强调，这是CFO的事。
Again, it's the CFO.
当然，她可能会比公司其他人都更倾向于踩一踩刹车。
Of course, she's probably gonna be, you know, if anyone putting the brakes on a little bit more than sort of others at the company.
但CFO对于一家公司上市并完成融资显然是至关重要的。
But, you know, the CFO is obviously vital to a company going public and and being instrumental in that raise.
特别是莎拉·弗里尔，她曾参与过多家上市公司，担任过董事会成员，深知该如何操作。
And Sarah Friar in particular, who's been involved with public companies and and on the board of companies and knows what to do here.
但我想说的是，当所有关于OpenAI和Anthropic可能上市的传闻出现时，未来六七个月甚至更久的宏观环境仍存在太多不确定性，正如我们之前讨论的，任何事情都可能让IPO计划泡汤。
But, again, I just think when I was you know, when all these these rumors have come out about OpenAI and Anthropic potentially going out, like, there's so many wildcards still at play over the next even, you know, six, seven, eight, nine months macro wise as we've talked about before, like where anything could derail like an IPO from happening.
但我觉得，既然SpaceX已经正式启动上市进程，而且听说他们想在六月，或许是为了埃隆·马斯克的生日上市，
But I do think, you know, now that SpaceX is formally, you know, put the wheels in motion to go out, and it sounds like, you know, they're angling for June, for maybe for Elon Musk's birthday.
这无疑施加了更多压力，毕竟他作为OpenAI的联合创始人，如今却与公司、尤其是与萨姆·阿尔特曼产生了严重分歧。
And, you know, that puts a little bit more pressure, of course, given his position with OpenAI as technically a cofounder, but now at, you know, at odds, obviously, with the company and with Sam Altman in particular.
你知道，这又增添了一层复杂性。
You know, that adds another layer to this.
但感觉莎拉·弗里尔似乎在纵观全局，关注这些泄露的财务数据。
But it does feel like that maybe Sarah Friar is looking at sort of the overall picture and looking at these leaked you know, what these leaked financials.
当然，她们对实际数字有更深入的了解。
Obviously, they have they have even better insight into what the actual numbers are.
并且她只是在观察市场，觉得如果OpenAI上市，会发生什么还不明朗。
And just looking at the market and being like, it's unclear what would happen if OpenAI went public.
你会想，哦，这是一家AI公司。
You think like, oh, it's an AI company.
有太多热议了，人人都在关注。
There's so much buzz like everyone.
但再说一遍，SpaceX已经准备大肆宣传它的AI概念了，对吧？
But again, SpaceX is already going be out there touting the AI angle, right?
埃隆会大力强调太空中的数据中心、xAI以及所有这些内容。
Elon's going to play up the data centers in space and xAI and all of that.
所以，我们可能会在公开市场上，让AI炒作的势头已经先被削弱了一些。
So we're going to have sort of some of the wind taken out of the AI hype sales in the public market already.
然后，如果OpenAI上市后，你看到如此巨大的资金消耗，公开市场会如何反应？
And then if OpenAI goes out and you see just this massive, massive burn, how are public markets going to react to that?
这有点未知，但我认为她可能想表达的是：我们现在正处于彻底改变业务的阶段。
It's sort of unknown, but I think like she might be saying, look, we're in the midst of changing a lot about the business right now.
众所周知，我们已经暂停了这些非核心项目，同时可能转向更专注于Anthropic过去在企业市场的强项，以及当前Cloud Code和Codex的发展。
Know, famously, we're putting a pause on these side quests and at the same time, sort of maybe pivoting to sort of focus more on what historically has been Anthropic strength in enterprise and obviously with what's going on with Cloud Code and Codex.
因此，我们甚至无法完全预见未来一年的财务状况会是什么样子。
And so we don't have a full vision into what the next year is gonna look like even financially.
是的。
Yeah.
我的意思是，无论谁坐上这家公司的CFO位置，恐怕都会心脏病发作，因为OpenAI内部普遍相信这项技术正在呈指数级发展，这或许确实如此，但作为CFO，你不能真的用指数思维去思考问题。
I mean, CFO, you know, no matter who it is sitting in this company would probably be having a heart attack because the belief within OpenAI is that this technology is moving on an exponential, and it may well be, but as a CFO, you don't you're not really able to think in those exponentials.
你必须考虑的是：这家公司会不会破产？
You have to think of, like, am I gonna is this company gonna go bankrupt?
我们是不是投入太多了，以至于最后无法兑现承诺？
Like, are we investing so much that we're gonna end up, you know, sort of not being able to meet our commitments?
我们已经看到OpenAI最近退出了几项承诺。
And we've already seen OpenAI pull out of a few commitments recently.
这是一个引述，这个信息报道中的一个挺有意思的引述。
This is a quote, it's kind of an interesting quote, in the in this information story.
有一位与她和阿尔特曼密切合作的人说，弗莱尔的工作很难做。
Friar has a hard job, said said someone who works closely with her and Altman.
她为一位雄心勃勃的创始人工作，这位创始人希望在支出上尽可能地突破极限。
She's working for a founder with big ambitions who wants to push the envelope as hard as he can on spend.
对我来说，这看起来像是公关人员说的非正式言论。
I know that to me seems like a non background quote from a comms person most likely.
但接着有个开头，让我再说一件事，我几乎忽略了这一点，那就是在这份报告的开头，他们提到——我再读一遍。
But then there's an intro let me just say there's one interesting thing here though, which I almost overlooked, and that is that in the at the beginning of this report, they they I'll I'll read it again.
弗莱尔表示，她不确定OpenAI在未来几年是否需要投入如此多的资金来获取AI服务器，还是其放缓的收入增长足以支撑这些承诺。
Friar said she wasn't sure whether OpenAI would need to pour so much money into obtaining AI servers in the coming years or whether its revenue growth, has been slowing, would support the commitments.
收入增长正在放缓。
Revenue growth has been slowing.
我之前没看到过这方面的报道。
I haven't seen that reported previously.
所以如果收入增长在放缓，就更难相信你们正处于指数级增长阶段，也许这正是所有问题的根源。
So if the revenue growth is slowing, it's harder to believe you're on an exponential, and maybe that's the source of all this.
是的。
Yeah.
我的意思是，很难确切知道他们指的是什么，或者措辞到底是什么。
I mean, it's hard to know exactly what they mean or what, you know, what the wording exactly is there.
对吧？
Right?
因为，好吧，你可能会说，是的，收入增长相对于之前的增速确实在放缓，这属于大数定律的现象，对吧？
Because, okay, you might say, yeah, the revenue growth is slowing relative to the speed at which it's been Like it's a law of large number stuff, right?
也就是说，增速在下降，但仍然增长得非常快。
Like it's coming down, but it's still growing very fast.
很难准确理解他们这句话的含义。
It's hard to parse exactly what they mean by that.
但与此同时，即使从宏观层面来看，如果你已经知道这类指标正在放缓，而你又希望将这家公司推向公开市场，你就必须牢牢确立未来的增长前景和关键驱动因素的叙事。
But at the same time, even at the high level, like if you already know that sort of things are slowing down a bit and you're, again, angling to take this company into the public markets, like you've really got to lock down the narrative in terms of like what both growth is going to look like going forward and sort of what the key drivers are going be.
而且，这家公司本身正处于产品转型和商业模式转型的过程中。
And again, the company itself is in the midst of this, like, product pivot and and sort of business model pivot in a way.
因此，他们自己也不清楚未来一年的增长前景会是什么样子。
And so it's it's unclear that they know exactly what what the growth picture is going to look like over the next year.
所以，如果我是莎拉·弗里尔，我可能会说：嘿，我们刚刚完成了这笔1220亿美元的融资，这是迄今为止最大的一轮融资，远超任何IPO，包括SpaceX即将传闻的IPO，后者目标估值仅为700亿或800亿美元，对吧？
And so if I'm Sarah Friar, I'm probably saying like, Hey, we just raised this $122,000,000,000 round, by far and away the largest round ever raised, by far and away larger than any IPO, including SpaceX upcoming rumored IPO, would target a mere 70,000,000,000 or $80,000,000,000 right?
而OpenAI在私募轮中筹集的资金已经超过了这个数字。
And OpenAI has raised more than that in a private round.
所以，如果我是莎拉·弗里尔，我会说：瞧，我们刚把这事办成了。
And so if I'm Sarah Friar, say, well, look, we just got this done.
这为我们争取到了X个月的运营缓冲期。
This buys us X amount of runway.
我们并不确切知道那是什么，但根据这些消耗数字，它并不是无限期的。
We don't know exactly what that is, but it's not indefinite given these burn numbers.
而且，这确实让事情稍微慢了下来。
And, you know, it just sort of slowed things down a little bit.
让我们看看我们正在投入的所有Codex工作，以及我们正在开发的这个新超级应用。
Let's see how all the codex work that we're putting in, this new super app, you know, that we're working on.
让我们给它几个月时间来显现，这样在正式启动之前，我们能对实际的财务状况有更清晰的了解。
Let's see you know, give this some months to to play out so we have a better picture of what our actual financial picture will look like before sort of we start the wheels in motion.
但同样，萨姆·阿尔特曼和公司内部其他人可能正在审视这个整体局面，不仅关注SpaceX，而且Anthropic可能是主要的担忧。
But, again, Sam Altman and maybe others within the company are probably looking at this, you know, overall picture and looking at not only SpaceX, but again, Anthropic is is probably the main concern here.
因为如果Anthropic在OpenAI之前上市，并且能够更清晰地描绘出盈利前景，这对OpenAI来说将更加不利。
Because if Anthropic were to go public before OpenAI does and has a sort of is able to paint that better picture in terms of profitability, that's going to be even more problematic for OpenAI.
所以我理解这里存在的两方面，那种推拉关系。
And so I get sort of the, you know, the two sides here, the push and pull as it were.
但无论如何，考虑到当前宏观环境，他们今年尝试上市似乎风险极高。
But again, it seems very risky for them to try to go out this year regardless of all the macro stuff going on.
是的。
Yeah.
如果你要这么做，你真的需要你的CEO和CFO保持一致。
And you're gonna really need your CEO and CFO to be in sync if you're gonna do it.
而这，再次说明了信息的来源。
And this is, again, the reporting from the information.
最近几个月，阿尔特曼在与OpenAI主要投资者之一的领导层讨论服务器支出时，没有邀请弗莱尔参加。
In recent months, Altman left Friar out of a conversation about server spending with leaders at one of OpenAI's top investors.
考虑到之前关于同一话题的讨论曾包括她，她的缺席显得格外明显且尴尬。
Her absence was notable noticeable and awkward given the previous conversation on the same topic included her.
另一位也出席了会议的人，抱歉。
A different person who also attended who atten sorry.
另一位在今年早些时候参加过阿尔特曼主持的OpenAI高层会议的人表示，弗莱尔未被邀请是不同寻常的。
A different person who attended a senior level meeting at OpenAI with Altman earlier this year said it was unusual that Friar was not invited.
此外，这再次来自相关信息：在一个大型公司中，CFO几乎总是直接向CEO汇报，抱歉，是向CEO汇报，这本是常规做法，但这次却不同。
Also, this is again from the information, in an unusual move for a large company, where CFOs almost always answer directly to the CFO to the CFO sorry, to the CEO.
弗雷尔自八月起不再直接向阿尔特曼汇报，而是开始向斐济·西莫汇报，西莫此前加入公司担任OpenAI应用业务负责人。而正如我们所知，在上周末的新闻之后，斐济目前处于医疗休假状态。
Friar stopped reporting directly to Altman in August and instead began reporting to Fiji Simo, who had joined as head of OpenAI's applications business, who, as we know, after, last Friday's news, Fiji is now on medical leave.
那么，弗雷尔现在到底被安排到哪里了？
So leaving Friar where exactly?
我的意思是，你刚才说的这些内容里信息量太大了。
I mean, there's there's so much in just what you said right there.
那我们从后往前开始梳理吧。
So get started going backwards.
当斐济·西莫加入的消息公布时，我当时就觉得有点奇怪，因为所有这些汇报线都转到了她那里。
When the news came out that Fiji Sima was joining, I did think it was odd at the time even that, yeah, all of these reports were moving over to her.
你知道，她确实有CEO头衔，但这并不是公司的CEO。
You know, she had the CEO title, but it's not the CEO of the company.
对吧？
Right?
当时她的头衔是应用业务的CEO，现在又改成了AGI什么的CEO。
It's it was at the time CEO of applications, and now they've changed it to CEO of AGI something or another.
我不完全确定AGI部署到底是什么。
I'm not a 100% sure what it is AGI right deployment.
是的，AGI部署。
Yes, AGI deployment.
AGI部署。
AGI deployment.
完全正常，不管那意味着什么。
Totally normal, whatever that means.
所以当时觉得有点奇怪，是的，首席财务官莎拉·弗里尔竟然转到了这个领域。
And so it seemed a little weird at the time that, yeah, Sarah Friar, the CFO, was sort of moving in that.
现在他们会说：看，我们只是需要让萨姆腾出时间。
Now they would play it as like, Look, we just needed to free up Sam's time.
他需要去融资。
He needs to do fundraising.
但显然，融资与财务密切相关，你可能会想，他们应该在这件事上保持完全一致、紧密配合。
Though obviously finance is pretty important with fundraising, you'd think like, again, they might want to be perfectly in sync and in line with one another talking through that.
但这就是为什么我一直以来心里都隐隐觉得，如果最终由斐济·西莫出任首席执行官，我一点也不惊讶，而他们只是让萨姆·阿尔特曼留在那里，因为他在融资和更高层次的使命目标方面依然至关重要。
But that's sort of why I've always sort of had in the back of my head that I would not be shocked if Fiji Simo is eventually sort of overall CEO, and they just sort of have Sam Altman there because he's vital still, obviously, from a fundraise perspective from sort of these higher level mission oriented goals.
但你知道，如果和当维吉·西莫的健康状况好转，能够更积极地参与日常运营时，显然，每个人都希望她早日康复，尽快结束这次休假回来。
But, you know, if and when Vigi Simo, you know, is is in a place with her health where, you know, she can be sort of more instrumental on the day to day, obviously, you know, every everyone hopes that she gets better and she's able to come back quickly from this leave of absence.
你可以想象这样一种情况：她已经让那些通常向首席执行官汇报的人直接向她汇报，相当于她担任副首席执行官。
You can see a world in which she already has all these people who normally report to a CEO reporting to her as the sort of sub CEO.
所以我们只能拭目以待，看这条路径会如何发展。
And so we'll see where that path goes down.
但没错，萨姆和莎拉·弗莱尔之间这份报告中的分歧，似乎是其中最关键的部分。
But yes, the the sort of fracture in this report between Sam and and Sarah Friar seems like the big element of it.
显然，如果你在开财务会议时，首席财务官却没有参与，而且你特意排除了她，这非常不利。
Obviously, that's that's very damning if you have financial meetings and the CFO isn't involved and you're specifically not including the CFO.
这暗示了什么？
What does that suggest?
现在，这可能暗示着，谁知道呢？萨姆·阿尔特曼可能在说：你看，我正在尝试寻找各种新的融资方式，就像他过去提到的那样，也许我只是需要一个能和我头脑风暴的人，而这个人不一定是我们的CFO，因为她肯定会说不，或者会引导我们走更传统的路，而我真的很需要跳出框架思考。
Now, it might suggest that, who knows, Sam Altman is saying like, Look, I'm trying to come up with all these new methods of financing, as he's talked about in the past, And maybe I just need someone to brainstorm with that's not necessarily our CFO because, of course, she'll say no or she'll lead us down a more traditional path, and I really need to think outside the box.
我正在想办法解释，为什么CFO不需要参加这个她通常都会参与的重要财务会议。
I'm trying to come up with some way in which you could say like, yeah, the CFO doesn't need to be in this very important financial meeting that she's normally involved in.
但事实上，这可能是个坏兆头，或者背后另有隐情，但无论如何，从外部来看，这看起来都不太好。
But the reality is like, it might be a bad sign or, you know, there might be something else going on, but it doesn't look good regardless optically, you know, from the outside.
当然，公司会说一切正常，就像他们过去对微软和其他一切事情那样，总是说‘没什么可看的’。
And, of course, the the company is gonna say everything is hunky dory just like they historically has all have always said with Microsoft and and, know, everything else, like, you know, nothing to see here.
但正如我们所知，OpenAI总是有事情可挖。
But there's always something to see with OpenAI as we've learned.
没错。
Absolutely.
而且有趣的是，这次会议的内容竟然泄露了。
And it's also interesting that that meeting leaked.
是谁泄露的？
Like, who leaked it?
很可能不是OpenAI干的。
It probably wasn't OpenAI.
这看起来像是投资者泄露的，也许是有人对最近这1220亿美元的融资过程感到不满。
It seems like that was leaked by the investors, maybe someone who wasn't happy with how the latest $122,000,000,000 funding round went down.
我不确定。
I don't know.
我的意思是，这当然都只是猜测，但这也是一个有趣的细节。
I mean, obviously, this is all just speculation, but that is an interesting wrinkle as well.
一位顶级投资者泄露了这些信息，
A top investor leaking this stuff is
顺便说一句，这很奇怪。
By the way weird.
关于这1220亿美元融资的整个公关策略也很奇怪。
The entire the entire press strategy around the $122,000,000,000 round was weird too.
我知道，他们确实——正如我们刚才提到的——创下了有史以来最高的融资纪录。
It's like, look, I get that they had you know, it's it's again, we've as we just mentioned, it's like a record raise.
这是一个惊人的数字。
It's an incredible number.
但他们刚刚才宣布了1100亿美元的融资。
But they also had just announced the 110,000,000,000 raise.
对吧？
Right?
而现在，他们又觉得有必要再搞一轮，让莎拉·弗莱尔出去跟大家沟通。
And now and then they felt the need to do a whole another cycle with Sarah Friar going out there and talking to folks.
这感觉就像是以最防御性的方式宣布了历史上规模最大的一轮融资。
It sort of felt very it felt like the most defensive way to to announce the biggest fundraise in history.
你知道，显然他们在多个方面都受到了攻击。
You know, obviously, they're under attack on a few fronts.
你知道，我们谈到了Anthropic，还有谷歌等其他各种事情。
You know, we talked about Anthropic and and everything else going on Google and and etcetera, etcetera.
但即便如此，整个这一套操作还是太奇怪了。
But still, that whole cycle was just weird.
是的，又增加了120亿美元的投资，这确实令人惊叹。
Yeah, an upgrade of 12 more billion dollars coming in, which again is incredible.
这比除了OpenAI之外的几乎所有融资都要大。
That's bigger than almost every fundraise ever, not done by OpenAI.
光是这两个数字之间的差距就令人震惊。
That in and of itself, the delta between those two numbers.
但还是感觉像你刚宣布了01/2010，现在又宣布了01/2022，仿佛你只是从这些事情里拿到了两个字节的苹果。
But still, it was like you just announced the 01/2010, now you're announcing the 01/2022 and like you're getting, you know, two bytes of the apple out of these things.
我知道。
And I know.
如果你退一步看，OpenAI目前的种种情况确实有些奇怪。
There's weird optics around this if you just take a step back around a lot going on in OpenAI.
嗯，M。
Well, M.
G。
G.
TBPN解决了这个问题。
TBPN fixes this.
你知道的。
You know that.
是的。
Yeah.
我们还没谈到那方面。
We haven't even gotten into that.
我的意思是，是的。
I mean yeah.
但确实，目前有很多奇怪的事情在发生。
But, like, yes, there's there's so many weird things going on at the moment.
关于收入方面，最后再提一件事。
One last thing about about the revenue side.
这个信息故事里有一张图表。
There's this chart in this the information story.
对于正在收听的朋友们，我简单描述一下。
And, for those of you on audio, I'll just kinda narrate it a little bit.
它对比了OpenAI和Anthropic的收入，你可以看出为什么它们想在Anthropic之前上市，因为去年年底，Anthropic的年化收入看起来大约是100亿美元，而OpenAI则超过200亿美元。
It looks at OpenAI versus Anthropic revenue, and you could sort of see why they'd wanna go public before, before Anthropic, because late last year, Anthropic was doing looks like about, $10,000,000,000 in annualized revenue, and OpenAI was over 20,000,000,000.
但现在Anthropic已经追上来了。
But now Anthropic has closed that.
它们的年化收入达到190亿美元，而OpenAI最新的报告显示收入为250亿美元。
They're at 19,000,000,000 in annualized revenue, while OpenAI or the latest report has 25,000,000,000.
所以，如果这种情况持续下去，Anthropic会超过OpenAI的收入吗？
So if things continue at this rate, does Anthropic surpass OpenAI in revenue?
这对它们的IPO故事意味着什么？
And what does that do to the IPO story?
我想我确实让Claude做过这个预测。
I think I actually asked Claude to do that extrapolation.
我正努力回忆几周前的事，当时它确实预测两者会交叉。
I'm trying to remember exactly a few weeks ago, and it did have them crossing.
依稀记得，大概是在明年某个时候。
I think at some point next year, if memory serves.
这显然是一个不断变化的目标。
And it's obviously a moving target quite literally.
但即使是这些数据也恰恰说明了为什么OpenAI正在进行这种产品和商业模式的转型，对吧？
But even that speaks to exactly like why OpenAI is doing this sort of product and in a way business model pivot because of those very numbers, right?
他们也看到了这些情况。
They see them too.
他们注意到Anthropic正在迎头赶上。
Like they see that Anthropic is gaining steam.
我们经常谈到，是的，Claude正在获得越来越多的用户使用，而这在过去并不是他们的强项。
We talk a lot about like, yeah, Claude is getting consumer usage, which they hadn't had historically or hadn't had a strong point there historically.
这很重要。
And that's big.
但关键在于，是的，Claude在代码使用方面的增长正促使企业选择Anthropic而非OpenAI，这正是你所说的在缩小差距的原因。
But the key thing is, yeah, this Claude code usage driving these enterprises to sign up for Anthropic versus using OpenAI and how that is closing that gap, you're talking about.
因此，无论这对一家准备上市的公司来说多么不利，从整体上看，这对公司本身也很不利，因为这正是他们进行转型的全部原因。
And so regardless of that being bad from a public company narrative when you're trying to go public, it's bad for the company overall because that's entirely why they're pivoting, it would seem.
对吧？
Right?
这就是他们为什么要做所有这些事，放弃那些旁支项目，关停Sora，只专注于核心，因为他们看到了这个曾经远不如他们的竞争对手带来的真正威胁。
That's why they're doing everything and killing off these side quests and killing off Sora and just trying to focus because they see this real threat from what had historically been their much smaller competitor.
是的。
Right.
那我们来谈谈这一点，因为这也与这里的一些高管离职有关。
And so let's let's go to that because that also plays into some of this executive turnover here.
让我读一段Spyglass关于你写的OpenAI的历程的内容。
Let me read from Spyglass about that something that you wrote about OpenAI's odyssey.
你说，当OpenAI在尝试拓展到所有潜在市场时，Anthropic却专注于自己的核心市场。
You said, while OpenAI was doing a a million things to expand to any and all potential markets, Anthropic was focused on their core market.
游戏已经变了，OpenAI必须随之改变。
The game has changed, and OpenAI needs to change with it.
幸运的是，他们还有第二位CEO，弗莱尔·西莫，他在动态公司拥有丰富的经验。
Good thing they have that second CEO, Friar Simo, with a range of experience at dynamic companies.
很难想象阿尔特曼能在这里建立起所需的纪律。
It's hard to imagine that Altman could instill the discipline needed here.
他真的能对乔尼·艾夫说不吗？
Is he really going to be able to tell Johnny Ive no?
不能。
No.
他做不到。
He's not.
但很容易看出西莫能做到。
But it's fairly easy to see how Simo could.
这可能并不是她被聘任的真正原因，而且她显然已经离开去专注于健康和其他计划，但她现在就在那里，而且目前处于休假状态。
This may not be exactly why she was brought in and she'd clearly left to focus on health and other initiatives, but here she is, and now she's on leave.
她确实就在那里，是的，不幸的是。
Here she was, yeah, unfortunately.
但，但，是的，我的意思是，这正是目前OpenAI内部以及再次审视Anthropic时所有问题的核心。
But, but, yeah, I mean, that is the sort of crux of everything going on right now within OpenAI and and just looking again at at Anthropic.
而他们最终在这里所做的，感觉现在是一场全力冲刺，要把这个所谓的超级应用搭建起来，整合OpenAI和ChatG的所有元素。
And what they end up doing here, it feels like it's a full on sprint now to get this quote, unquote super app, you know, in place, bring in all the the elements of of OpenAI's and and ChatG.
特别是BT，他们的优势在于，看看他们能否利用这一优势，结合庞大的消费者用户基础，再加上他们在编解码器方面所做的所有工作，以及他们在Atlas浏览器上的尝试——听起来这也是其中的一部分，把这些都整合进来。
BT in particular, their strength, and sort of see if they can use that strength with the big consumer installed base and mixed with the sort of all the work they've been doing on codecs and then whatever they're trying to do with Atlas, the browser, it sounds like that's gonna be a piece of it as well, you know, bring this in.
他们能否借此遏制Anthropic在过去几个月里积累的势头？
And can they use that to sort of cut off the momentum that Anthropic has garnered over these past several months?
而这正是这场冲刺的真正目标。
And that's really what the sprint is all about.
但同样，与此同时，我要回到一点：这不仅仅是他们集中精力做这些事情。
But again, at the same time, I go back to, it's not just that they're sort of doing all of this focus.
更重要的是，这与他们过去所必须遵循的策略完全不同——过去他们并不直接面向消费者销售，而是专注于企业市场。
It's also that they're this is a different sort of muscle than they've needed to sort of adhere to in the past where, again, they're not selling directly to consumers, they're talking about selling into enterprise.
据各方面消息，他们一直做得不错。
And by all accounts, they've been doing a good job.
对于他们来说，Codex似乎是一个快速增长的业务，但仍然远不如Cloud Code的规模。
It seems like it's a fast growing business, Codex, for them, but it's still not nearly as big as Cloud Code is.
他们还有 Cursor 在外面，以及无数其他潜在的竞争对手。
And they've got Cursor out there and they've got And myriad other would be competitors.
因此，接下来的几个月对 OpenAI 来说，将比以往任何时候都更加引人注目。
And so these next few months are going to be even more fascinating than they always are for OpenAI.
过去几年，这家公司 arguably 是世界上最具看点的公司。
The fact that these past few years have been This has been arguably the most interesting company in the world to watch.
而现在，接下来的几个月似乎更是至关重要，值得密切关注。
Now these past these next few months seem like they're gonna be absolutely vital to watch.
嗯。
M.
G，我问你一个问题。
G, let me ask you.
OpenAI 要把计算机使用、浏览器使用、ChatGPT 和 Codex 全部整合成一个超级应用，这明智吗？
Is this wise, the fact that OpenAI is going to build this super app that will have computer use or browser use, ChatGPT, and Codex altogether.
Anthropic 也有他们自己的超级应用版本，也就是 Claude、Cowork 和 Claude Code 的整合。
And Anthropic also has their own version of the super app, which is, you know, Claude and Cowork and, Claude code altogether.
你认为，追求这种AI代码助手——即使用户没有技术背景，也能替他们完成一切——真的是正确的应用场景吗？
Do you think that this is the right use case to be running after, basically, this AI code jockey that can just basically go do everything for you even if you're not technical.
我的意思是，我对这个问题的看法一直摇摆不定，我想公司们也是如此。
I mean, I I go back and forth a little bit on this and as as I think the companies do.
对吧？
Right?
这就像是著名的‘捆绑与拆分’说法，对吧？
It's like the famous bundling and unbundling quote, right?
就是当年的吉姆·巴克斯代尔说的。
The Jim Barksdale back in the day.
就像公司们总是这么做，比如Meta。
It's just like companies always do this from Meta.
Meta已经多次这样做了，对吧？
Meta famously has done this multiple times now, right?
他们曾经把所有产品都整合在Facebook里，然后又开始把它们拆分成各自独立的产品。
Where they have all these products inside of, back in the day, Facebook, and then they start to unbundle them into their own sort of disparate products.
然后他们后来意识到，哦，实际上我们应该把这些产品重新整合在一起。
And then later they've realized like, Oh, actually we should bring them all back in together.
所以公司总是不断地整合与拆分。
And so companies just bundle and unbundle.
而OpenAI，感觉上只是比以往更快地在做这件事。
And OpenAI, again, is just doing it a little bit faster, it feels like, than sort of historically has been the case.
我觉得这是一个不错的策略？
I think it's a good strategy?
我觉得现在对他们来说，这可能是正确的做法，因为他们拥有庞大的ChatG用户群体。
I think it's probably the right thing to do right now for them because, again, they have this massive user base with ChatG.
BT。
BT.
他们拥有自认为具有竞争力的Codex产品。
They have what they feel like is a good competitive product in Codex.
如果他们能把这两者结合起来，我认为在这样一个世界里是有道理的：如果你真的相信Codex不仅仅是为了开发者编程，而是未来代理式软件的方向，既面向消费者也面向企业，所有功能一体化，那么这样做是有意义的。
If they're able to bring the two of them together, I think that it makes sense in a world if you truly believe that Codex is not just going to be coding for developers, if you believe it's going to be the future of sort of agentic software, that it's going to be meant for consumers as well as enterprise and everything sort of all in one, then I think it makes sense to do that.
我认为，Claude 现在已经内置了这一点，对吧？
And I think that, you know, Claude right now sort of already has that baked in, right?
比如，它们现在至少都包含在同一个 Mac 应用程序包中。
Like, they're all part of the sort of the same, at least Mac app payload right now.
我认为在移动端还有其他一些变量需要考虑，比如如何让这一切在手机上运行，以及它可能是什么样子。
I think there's other sort of variables at play with regard to mobile, like how you make this all work on mobile and what that might look like.
但至少目前来看，我认为这样做是明智且合理的策略。
But at least for now, I think it probably makes sense and I think it's sound strategy to do it.
不过，显然存在一个风险：你的应用可能会变成像 Microsoft Office 那样臃肿不堪，到处都是下拉菜单，甚至下拉菜单里的下拉菜单。
There is a risk, though, obviously, that both like your app just becomes this bloated mess of Microsoft Office and you've got dropdowns and you've got dropdowns of dropdowns.
ChatGPT 过去就遇到过这个问题，比如模型选择器，他们不得不简化界面。
And ChatGPT has had this problem in the past, right, with the model picker and they had to simplify things.
所以当你把两个产品，甚至像我们提到的，再加上 Atlas 这第三个产品整合在一起时，会不会让消费者突然觉得这产品像个拼凑起来的怪物，一个弗兰肯斯坦式的怪胎？
And so when you're bringing in two products together and potentially three, as we mentioned with Atlas in the mix as well, like does it just become the sort of thing consumers all of a sudden don't like because it's just like this Frankenstein of a product, Frankenstein's monster of a product?
因此，这样做是有风险的。
And so there's risk to it.
但再说一遍，我认为他们的策略是利用ChatGPT庞大的用户基础，以抢先一步或试图遏制他们所看到的Anthropic的增长。
But again, I think that their strategy is to basically leverage that ChatGBT massive user base in order to, again, get out ahead or try to stop the the growth of Anthropic that they're seeing.
对。
Right.
这很有趣，因为过去OpenAI瞄准的是消费者市场，而Anthropic瞄准的是企业市场。
And it is interesting because there used to be, like, OpenAI was going after consumer, Anthropic was going after enterprise.
现在他们都在争夺同一个市场。
Now they're going after the same thing.
这确实将体验提升到了一个全新的层次。
And it really does take, the experience to a different level.
对吧？
Right?
它让AI从一个聊天机器人，变成了一个能帮你工作或处理个人事务的个性化助手。
It goes from AI being a chatbot to AI being this, like, personalized agent for you that can help you with work or your personal life.
所以我想问的是，你相信这个愿景吗？
And so I guess what I'm asking is, do you believe in that vision?
我的意思是，这个愿景足够吸引人，能让公司彻底转向吗？
Like, do you think that is appealing enough to pivot the entire company, toward?
如果真是这样，那你基本上就会看到 OpenAI 和 Anthropic 直接对决。
And if it is, I mean, you're basically gonna see OpenAI versus Anthropic head to head.
那你觉得谁会赢？
So who do you think wins?
首先，我认为这确实是一个方向正确的选择。
So first, I do think that it is again, I think it's directionally the right bet.
我们现在已经看到了足够的早期迹象。
I think we've seen enough now of the early signs.
说实话，我不确定是 Claude Code 在多大程度上推动了这个决定，更多是因为 Claude Cowork。
And I think it was frankly I'm not sure how much it was Claude code that drove this decision as much as it's Claude Cowork.
对吧？
Right?
这个衍生产品瞄准的是更偏向计算机使用、代理型的应用场景，更适合普通人的日常使用，而不仅仅是开发者写代码——OpenAI 一定看到了这一点，再结合 OpenCLaw 的发展，然后就想：看吧。
The the offshoot that sort of goes after the more, again, computer use, agentic use cases that are more meant for regular day, everyday usage of of, quote, unquote, regular people, not just, developers coding, where OpenAI must have looked at that mixed with what's been going on with OpenCLaw, and just been like, look.
我们需要成为每个人做这件事时的首选平台。
We need to be the one go to shop for everyone to do this.
我们已经有了ChatGPT。
We already have chatg.
但它已经成了AI领域中像‘克林耐纸巾’那样无处不在的代名词。
Bt, which has sort of become the Kleenex of the ubiquitous brand within AI.
如果我们不在这些代理式编码和代理式服务上有所作为，就有失去这一优势的风险。
And we have a risk of losing that if we're not there with these sort of agentic coding and agentic services.
而且，即便现在已有如此多的势头和炒作，我仍觉得这些技术要真正主流化、产生实质性影响，还为时过早。
And again, it feels like it's early even now, like with all the momentum and all the hype around these things, it still feels like it's too early for these things to really take off in meaningful way on the mainstream.
但我认为，每个人都在试图切入其中的一小部分。
But I think that you're seeing everyone sort of go after little pieces of it.
我们之前谈过OpenCLaw。
Again, we talked about OpenCLaw.
我觉得背后确实有一股潮流，但同时，它对普通大众来说还是太晦涩了，难以理解。
It felt like there's a movement behind that, but at the same time, it's sort of too obtuse, I think, for sort of everyday regular people to wrap their heads around.
Claude Code和其他工具，比如Perplexity，甚至微软等公司，现在正努力让这些产品更贴近普通用户的日常使用体验。
Claude Code comes in and others, Perplexity and even Microsoft and some others are now trying to make it more you know, regular user consumer friendly.
这些关于让智能代理代表你、在你的电脑上自由行动的总体想法，同时将它们隔离或置于独立的沙盒环境中，以确保更安全，避免人们轻易陷入各种不良行为的陷阱。
These general ideas about using agents on your behalf, on your computer, and letting them, like, sort of roam free and sort of trying to, yeah, sequester them or put them in their own little sandbox to make it a little bit safer so people aren't doing all the bad things that you can easily go down the path of.
所以你问，我认为OpenAI或Anthropic谁更有希望实现这一点？
And so you ask, do I think that OpenAI or Anthropic has a better shot of sort of doing this?
再次强调，OpenAI得益于ChatG的广泛使用。
Again, OpenAI with thanks to just ChatG.
B t在用户基础方面拥有巨大的、巨大的优势。
B t has a huge, huge advantage in terms of just installed user base.
当他们在未来几个月内将这款新产品——这个超级应用——强行推到每个人面前时，他们将有机会超越Anthropic目前所拥有的势头。
And when they're shoving this new product, you know, this new super app in in everyone's face presumably in in the next couple of months, they're gonna have the opportunity to overtake certainly the momentum that Anthropic has.
话虽如此，Anthropic在这里做出了绝对正确的选择，而OpenAI似乎没有，至少就我们目前的处境而言。
That said, Anthropic made the absolute right bet here, whereas OpenAI seemingly did not, at least for where we stand right now.
在过去几个月里，他们所做的一切都令人难以置信地出色，他们成功地将这家原本看似专注于企业市场的公司和产品，转向了可能成为AI未来——至少在相对短期之内——的一些方向。
And everything they've done over the past few months has just been insanely impressive of how they've angled, you know, what seemed to be this this enterprise oriented company and product towards being able to do some things which may or may not be the future of AI, at least in the the relative short term.
所以我给了你一个没直接回答的答案，但没错。
So I gave you a non answer, but but yeah.
我的意思是
I mean
很多事还待定。
A lot of it is TBD.
是的。
Yeah.
目前，我真心觉得 Anthropic 会率先做得更好。
Right now, I would bet honestly on Anthropic doing it first, doing it better.
但关键在于 OpenAI 什么时候推出这个超级应用，以及世界是否已经准备好接受这些代理式工作流。
But it really comes down to the timing of when OpenAI gets this super app out there and if the world is ready for sort of, yeah, all this this agentic workflows that come in.
我觉得现在还为时过早。
And I still think it's early.
所以我认为他们还有时间赶上 Anthropic 到目前为止所做的所有工作。
And so I think that they'll have some time to be able to catch up to all the work that Anthropic has been doing.
顺便说一下，我们再回头看看数据，当我看到《华尔街日报》发布的那些图表时，简直难以置信，Anthropic 在计算资源投入远低于 OpenAI 的情况下，竟然能达到如今的地位。
Which by the way, just going back to the numbers for a second, when I looked at the published charts, Wall Street Journal published those charts, it's sort of incredible that Anthropic is where they are given the relatively much smaller spend on compute versus OpenAI.
对吧？
Right?
根据这些未来的预测，这种情况还会持续下去。
And that's still going to be the case from these projections going forward.
当然，在某些情况下，你可以说 OpenAI 的模型稍微好一点。
And it's not like in some cases, of course, you could argue OpenAI's models are a little bit better.
但据所有报告称，Claude 在大多数使用场景的排行榜上都位居前列。
But by all accounts, Claude is near the top of most things, most of the leaderboards for most use cases.
而且，他们似乎是在远低于 OpenAI 的投入下做到这一点的。
And, they've done that without nearly the amount of spend, it seems like from OpenAI.
但这对 Anthropic 来说会是个问题，因为看起来他们服务的需求已经超过了他们在计算资源上的投入意愿。
But but that's gonna hurt Anthropic because it seems already like the demand for their services is outpacing what they're willing to invest in from a compute standpoint.
从开发者们的反应来看，我就是有这种感觉。
That's just the sense I get from seeing the developers react.
他们刚刚做了一件事，撤回了部分来自OpenCLaw类型的实例的代码。
They just did this thing where they pulled back some of the clawed code from OpenCLaw type instances.
这很有趣，因为Anthropic的态度是我们要谨慎发展，而OpenAI则是我们要快速发展，因为我们看到需求正朝这个方向增长，这可能正是最初导致CFO出现问题的原因。
And if you're it it is interesting because Anthropic's like we are gonna build with caution and OpenAI is like we're gonna build because we see the demand going this way, which is probably what's causing those issues with the CFO in the beginning.
但如果你到了需求超出你提供服务能力的地步，而你提供的服务和OpenAI一样，人们就会转向OpenAI。
But if you get to the point where the demand is outstripping your ability to deliver the service and you're delivering the same service as OpenAI, people are gonna go to OpenAI.
是的。
Yeah.
而且这是真的。
And that and that's true.
我们现在就正在看到这种情况。
And we're seeing that right now.
对吧？
Right?
我认为即使今天，Anthropic也出现了一点服务中断，我觉得他们正以非常不自然的方式被拉伸着。
I think even even today, sort of Anthropic had had a bit of downtime, and I think that they're they're probably just getting stretched in ways that are very unnatural to them.
而且，之前当我们讨论这两家公司走上不同道路时，确实如此，我写过这个角度，就像两条路分叉了一样，因为当时感觉Anthropic完全愿意让OpenAI承担那笔巨额资本支出，而他们自己则专注于核心业务，只花必要的钱。
And again, before, when we were talking about when these two companies were going down separate paths, was basically Yeah, I wrote about that angle a little bit, like the two roads diverged because it did feel like basically Anthropic was totally fine with letting OpenAI take on that CapEx spend, that insane CapEx spend, while they sort of, yeah, just sort of focused on their meat and potatoes and what they needed and of they were just gonna spend for what they needed.
但没错，正如你所说，反过来，现在OpenAI在纯算力方面可能处于相对更有利的位置。
But, yes, to your point, the flip side of that is that now OpenAI may be in a the the relatively better position from a from a pure capacity standpoint.
对吧？
Right?
但我知道，再说一遍，我更多是在谈论他们如何在支出更少的情况下依然完成模型训练，这真的令人难以置信。
But I you know, again, I I'm sort of talking more about, like, the way that they've been able to not spend as much and still do the model training, you know, is has been insanely impressive.
是的。
Yeah.
毫无疑问。
Without a doubt.
对的。
Yeah.
但你说得对。
But you're right.
比如，如果算力最终变得极其重要，那Anthropic在这方面的故事是什么？
Like, yeah, if if capacity ends up mattering far more, what is what is Anthropic's story there?
因为，再说一遍，这是一家也正试图上市的公司。
Because, again, this is a company also trying to go public.
如果他们突然意识到，我们严重受限于服务器资源，必须立即与所有新兴云服务商和大型科技伙伴达成合作。
And if they all of a sudden realize, like, we are badly, badly server constrained, we need to start cutting deals left and right with all of the neo clouds, with all of the big tech partners.
现实是，现在每个人都在受限。
The reality is like just everyone's constrained right now.
那么他们将如何应对呢？
So how are they going to be able to do that?
顺便说一句，每次这种情况发生时，比如去年年底，OpenAI就遇到过需求超过其算力供应能力的问题。
By the way, every time that does happen, like OpenAI had this issue at the end of last year where the demand outstripped their ability to deliver the compute.
他们确实出去做了这些合作，但那时你就被动了，因为要为数据中心支付巨额费用，你的利润率就会下降。
They did go out and they made those deals, but you're kind of over a barrel then and your margin goes down because you're paying so much for the data centers.
是的。
Yes.
这就是去年年底的情况。
And that was the story of sort of the end of last year.
我认为今天其中一个故事中提到了这一点。
And I think it's mentioned in one of those stories today.
对吧？
Right?
他们的利润率并没有朝着错误的方向发展。
Like, their margins are not going the wrong way.
它们实际上一直在朝着错误的方向发展。
They've been going the wrong way.
这听起来主要是因为人们没有充分认识到推理成本会飙升。
And that's it's largely sounds like it's just the under appreciation for how inference costs would sort of go through the roof.
而这一点确实是这样。
And that's yeah.
再说一遍，这是对这些技术的需求和使用率造成的。
Again, that's demand and utilization of these things.
展开剩余字幕（还有 369 条）
我可以分享一下我对这件事走向的看法吗？
Can I share, like, one perspective of the way I think this will go?
你知道这些GPT-5系列模型会给你一个答案，我觉得它们在这方面已经进步了一点，但基本上你提出的每个问题，它们都会试着为你做点有用的事情。
So you know how, like, the these GPT five series models, they'll give you an answer, and I think they've gone a little bit better at this, but, like, basically every question you've had, they've tried to do something useful for you.
所以它们会说，让我给你生成一个PDF，或者给你一个五步计划并把它变成图片，或者，你知道，把你那个商业点子拿过来，帮你完善并以文档形式交付给你？
So they've been like, let me build you a PDF of that or give you a five step plan and I'll make that an image or, you know, why don't I take that, you know, idea that you had for a business and then flesh it out and deliver it to you in a doc?
我认为，随着所有这些技术融合，最终结果可能会是：让我为你构建一个应用程序，能够帮你处理一些你想做的事情。
I think as all this stuff merges, that will end up being either a, let me build you an app that will be able to, you know, sort of handle some of these things that you wanna do.
比如，我们就拿健身这个例子来说。
Like, for instance, let's just take the fitness example.
你知道的？
You know?
好吧。
Okay.
你有一些健身想法，现在要回来和我讨论。
You have some you're key coming to back to me with these fitness ideas.
我认为我可以为你开发一个应用，帮助你实现这些目标。
I think I can build you an app that will sort of, you know, help you accomplish those goals.
或者更进一步，让我用我所在领域的一个例子来说明。
Or even take it a step further, let's say you're asking I'm just gonna use an example from my world.
假设你问它如何将一个视频分成章节，这正是我上周和格雷格·布罗克曼讨论过的事情。
Let's say you're asking it for help about, like, how to break a video into chapters, and this is something I spoke with Greg Brockman about last week.
你知道，它不会只是把字幕稿分成章节，而是会直接进入Adobe Premiere，为你开始剪辑。
You know, instead of just taking the transcript and breaking it into chapters, this thing will now wanna go into Adobe Premiere and start the edit for you.
它会再进一步，因为它知道你真正想做什么。
It's gonna just take that one step further because it knows what you're trying to do.
它了解你的兴趣，并且具备编程和执行任务的能力，因此可以主动提出这些建议，而不仅仅是给你一个PDF或商业计划。
It knows your interest, and having the ability to code and go out and get stuff done, it can become proactive about suggesting these things then do more than just give you that PDF or the business plan.
这很有趣。
That's interesting.
当你提到这个时，我想到了教阿曼达·菲什的那个想法，嗯。
When you're talking about that, my mind goes to the, you know, the, if teaching Amanda Fish, you know, idea and, like Mhmm.
基本上，人工智能的意思是：我们知道你想做什么，所以我们直接帮你完成，这样你就不用一次次回来，一步一步、慢慢地操作了。
Basically AI is saying, like, we know what you're trying to do, so we're just gonna go ahead and do that so you can do this and you don't have to keep coming back over here and, you know, doing it step by step and and slowly.
听你这么一说，我脑子里还想到另一件事——考虑到隐私等问题，这有点敏感，但你想一下，当年YouTube、Netflix以及其他类似服务在扩展时，他们意识到与其一遍遍重复做同样的事，不如把那些知道会被多人请求的内容提前保存下来。
One other thing that that jogged in in my head while hearing you talk through that, I do wonder if there's a way, and this is sort of delicate given privacy and whatnot, but like you think about how back in the day when YouTube and Netflix and just the sort of other services along those lines had to scale, And they sort of realized, like, rather than sort of doing the same thing over and over again, they were able to sort of save up some of the content that they knew would be sort of pulled by it was going to be asked for by multiple people.
于是他们只需保留一个版本，然后分发给所有需要的人，对吧？
And instead, they would just serve they've had one version of that, and then they would distribute it out, right, to whoever asked for it.
那么，对于这些反复被问到的AI想法和问题，有没有可能——也许他们已经在某种程度上这么做了——
Like, is there a way in which AI for some of these, you know, for some of these ideas and queries that are just getting asked over and over again, and maybe they're already doing this to some degree.
他们可以不再每次都临时处理，而是回头参考自己过去积累的、关于以往问题的更大记忆库，这其实是一种很好的成本节约机制，对吧？
But they can stop sort of trying to do everything on the fly and instead sort of go back to their own sort of broader memory of of things that have been asked in the past and sort of reference that up just as a way for like a They would never frame it this way, but it's like a good cost saving mechanism, right?
当你不需要一遍遍重复执行相同的查询时——抱歉，意思是，你不需要反复运行模型、反复进行推理，因为你已经拥有了大量用户曾经提出过类似问题的数据集。
Like when you don't have to do the same sort of queries over Or sorry, you don't have to run basically the models and do the inference over and over again because you already have a corpus of data that people that have asked for this.
我的意思是，考虑到绝大多数用户可能默认开启了允许AI训练使用我对话内容的开关。
I mean, considering that I would imagine the vast majority of users have the default, toggle on to allow the AIs to train on my conversations.
换句话说，无论有意还是无意，我们大多数人其实都给了AI这样做的许可。
I mean, basically, you know, wittingly or not, most of us are giving these things permission to do this.
你说得对。
Exactly what you said.
是的。
Yeah.
走着瞧吧。
We'll see.
走着瞧吧。
We'll see.
对。
Yeah.
就像你说的，我会变得越来越明显，我同意你的看法。
It's gonna get like like you said, I agree with you.
事情会很快变得非常有趣，尤其是在未来几个月，我们会看到这一切逐步展开。
It's gonna get very interesting very quickly, especially over the next few months as we see this all play out.
你知道，IPO的异常表现，尽管这会带来新的变数，我们也能看到更多关于收入数据的信息。
You know, IPO anomalous standing, although that will add a new wrinkle to it, and we'll be able to see even more data on the revenue numbers.
是的。
Yeah.
而且，正如我们之前讨论的，SpaceX已经秘密提交了申请。
And I guess so if so as talked about, SpaceX has has confidentially filed.
听起来他们计划在六月进行。
Sounds like they're targeting June.
所以，如果OpenAI和Anthropic真的打算上市，他们都表示目标是第四季度，至少报道是这么说的。
So if OpenAI and Anthropic really wanna go out and they both said they're targeting q four, at least that's those are the reports.
他们并没有明确说过，但这些是内部报告中的说法。
They haven't said that, but those are the, you know, the internal reports say that.
你知道，我们关注的是夏季末、初秋这段时间，他们那时必须提交申请。
You know, we're looking at the end of, you know, summer, early fall when they would have to sort of file.
因此，这些就是我们要关注的目标，看看届时他们的业务状况如何。
And and so those are the targets that we're gonna be looking for and seeing what how their businesses look at that point.
好的。
Okay.
我本来想说他们应该都等等，但他们不会。
I was gonna say they should both wait, but they won't.
但你觉得他们应该等吗？
But do you think they should wait?
也许他们现在就该发布。
Maybe they should just go out now.
我只是觉得他们应该赶紧出去。
I just raised get get out there.
去筹钱吧。
Raise that money.
嗯，
Well,
我的意思是，这里未明说的真相是，我们可以反复讨论他们是否应该等。
I mean, that is the That is the unstated thing here, is that we can go back and forth about whether they should or not.
但现实是，基于这些数字，OpenAI 必须上市。
But the reality is, like, given these numbers, OpenAI is going to need to go public.
他们需要资金，Anthropic也需要。
Like, they need the capital and Anthropic as well.
他们基本上已经榨干了所有可能的私人资本。
Like, they basically have tapped into all of the possible private capital.
我肯定外面还有一些其他资金来源。
I'm sure there's some other things out there.
还会有一些其他小规模的战略投资进来。
There'll be other little strategic things that come in.
我说的小，当然是指数十亿美元规模。
And I say little, I mean billions of dollars, of course.
你觉得呢
You think
考虑到伊朗目前的局势，海湾国家的资金还靠谱吗？
the Gulf money is still in play given what's going on in Iran right now?
这是个好问题。
That's a good question.
而且实际上
And actually
因为我原本以为他们是下一个，是最终BOSS，但UIPO我们还没听到他们提到过
because I thought that they were the next one, the final boss, then UIPO, but we haven't heard them mention it
完全没提到。
at all.
其中有一些。
Some of it in there.
对吧？
Right?
MGX
MGX
就是它。
is the one.
所以那里已经有不少资金了。
So there is already quite a bit of money in there.
但确实，我是说，萨姆·阿尔特曼和达里奥一直在那边花时间，是有原因的。
But, yeah, I mean, Sam Altman and Dario have been spending time over there for reasons.
对吧？
Right?
他们正在建立这些关系。
They're establishing these relationships.
所以，是的，这可以说是最后一件事。
So yes, that would be sort of the one final thing.
但正如你所指出的，鉴于该地区目前的动荡局势，这有点棘手。
But as you noted, given all the turmoil in the region right now, it's a little bit tricky.
但与此同时，还有其他报道，比如今天就有新闻提到中东财富基金正在向华纳兄弟交易注资，我们之前讨论过这个，对吧，关于派拉蒙和华纳兄弟的交易。
But at the same time, there's other reports like there was headlines about today just about the debt that the Mideast wealth funds are putting into the Warner Brothers deal, which we previously talked about, right, for Paramount and Warner Brothers.
所以显然，他们仍在积极进行这类活动。
And so they're still apparently active you know, doing that.
因此，看起来他们仍然在正常开展业务。
And so it seems like they're still open for business.
但归根结底，OpenAI 和 Anthropic 最终还是得上市，才能获得所需的大笔资金——不是一次性投入，而是因为它们目前融资的规模已经超过了任何一次IPO。
But still, at the end of the day, at some point, OpenAI and Anthropic are going to have to go public in order to just access the amount the sheer amount of capital and not meaning in one big slug because, again, they're already raising rounds that are bigger than any IPO.
但当你成为一家完全流动的上市公司时，你可以接触到各种不同的金融工具和机制，这些都大不相同。
But it's just like all different sort of instruments that you have access to when you have know, when you're a fully liquid public company and the different mechanisms you can use.
请记住，尤其是 OpenAI，一直很难建设自己的数据中心，今天的一些报道也提到了这一点，因为它们没有利润。
Because remember, again, like OpenAI in particular has had a really hard time, and this was in some of the reporting today as well, with regard to trying to build their own data centers because they don't have profits.
通常这类项目是通过债务融资的，但它们在筹措债务方面遇到了巨大困难，原因在于公司的财务状况。
You typically do that with debt, and and they're having a hell of a time raising debt because of the financials of the company.
所以，如果它们最终上市了，即使在一段时间内仍不盈利，至少也会拥有更多途径和手段来推动这些项目落地。
And so, you know, if and when they go public, while they might not be profitable for some time, at the very least, they'll have they'll have more avenues and more levers to pull in order to to put some of those wheels in motion.
好的。
Alright.
我们先休息一下吧，因为还有很多内容要聊，包括 Siri 的最新动态，以及 Meta 试图在广告之外拓展业务的最新进展。
Let's let's go to a break because we have plenty more to talk about, including what's going on with Siri and then the latest that we have on Meta's attempt to diversify its business, beyond advertising.
所以，广告后我们马上继续。
So let's do that right after this.
如果你仔细想想，大部分工作其实并不难。
If you think about it, most work isn't actually hard.
它们只是重复的状态更新、任务分派，一遍又一遍地回答同样的内部问题。
It's just repetitive status updates, routing tasks, answering the same internal questions over and over again.
这些事情悄悄地消耗了你团队每周的大量时间。
These are the things that quietly eat up your team's hours every week.
这就是Notion全新自定义智能代理的用武之地。
That's where Notion's new custom agents come in.
Notion是一个为团队打造的AI驱动的协同工作空间。
Notion is an AI powered connected workspace for teams.
Notion将你的笔记、文档和项目整合到一个顺畅运行的统一空间中。
Notion brings all your notes, docs, and projects into one space that just works.
它无缝、灵活、强大，而且用起来真的很有乐趣。
It's seamless, flexible, powerful, and actually fun to use.
内置AI后，你花在切换工具和应用上的时间更少，而用于创造优质工作的时光更多。
And with AI built in, you spend less time switching between tools and apps and more time creating great work.
现在，借助 Notion 的新自定义代理，那些过去需要数小时才能完成或根本不会发生的工作，现在都能自动运行了。
And now with Notion's new custom agents, the busy work that used to take hours or never actually happened at all runs itself.
这里有趣的是，这些代理不仅仅是对提示做出响应。
What's interesting here is these agents don't just respond to prompts.
它们基于触发器和计划运行。
They run on triggers and schedules.
因此，一旦设置好，它们的运作方式更像嵌入式系统。
So once they're set up, they operate more like embedded systems.
立即前往 notion.com/bigtech 试用自定义代理。
Try custom agents now at notion.com/bigtech.
全部都是小写字母。
That's all lowercase letters.
前往 Notion.com/bigtech 今天就开始试用自定义代理。
Notion.com/bigtech to try custom agents today.
当你使用我们的链接时，你就是在支持我们的节目。
And when you use our link, you're supporting our show.
这是 notion.com/bigtech。
It's notion.com/bigtech.
Notion.com/bigtech。
Notion.com/bigtech.
如果你们车队的司机明天出了事故，没有视频记录，你能证明到底发生了什么吗？
If a driver in your fleet got in an accident tomorrow, can you prove what actually happened without the footage?
这要难得多。
It's much harder.
所以你的保险费率飙升，还得自己承担费用。
So your insurance rates spike and you're stuck paying for it.
这就是为什么这么多车队选择萨姆拉的AI智能行车记录仪。
That's why so many fleets choose Samsara's AI powered dash cams.
清晰的视频证据、实时警报，以及帮助预防事故发生的事前指导工具。
Clear video evidence, real time alerts, and coaching tools that help prevent accidents before they happen.
萨姆拉AI将事故率降低了近75%。
Samsara AI helps reduce crash rates by nearly 75%.
例如，丹佛市和县的虚假索赔减少了50%，整体安全事件减少了94%。
For instance, the city and county of Denver saw a 50% reduction in false claims against them and a 94% reduction in safety events overall.
这是每位运营经理都需要的可见性。
This is the kind of visibility that every operation manager needs.
不要等到下一次事故发生才采取行动。
Don't wait for the next accident to take action.
前往 samsara.com/bigtech 申请免费演示，了解 Samsara 如何为您的运营带来可见性与安全性。
Head to samsara.com/bigtech to request a free demo and see how Samsara brings visibility and safety to your operations.
那就是 samsara.com/bigtech。
That's samsara.com/bigtech.
Samsara，更智能地运营。
Samsara, operate smarter.
开始一件新事情不仅很难。
Starting something new isn't just hard.
它令人恐惧。
It's terrifying.
做这件事投入了这么多工作，你却不敢确定它是否能成功，要迈出这一步确实很难。
So much work goes into this thing that you're not entirely sure will work out, and it can be hard to make that leap of faith.
当我刚开始做这个播客时，我不确定会不会有人听。
When I started this podcast, I wasn't sure if anybody would listen.
现在我知道，这是个正确的选择。
Now I know it was the right choice.
当你有Shopify这样的合作伙伴支持时，情况也会好很多。
It also helps when you have a partner like Shopify on your side to help.
Shopify是全球数百万企业的电商平台，占美国所有电子商务的10%。
Shopify is the commerce platform behind millions of business around the world and 10% of all ecommerce in The US.
从Allbirds和Cotopaxi这样的知名品牌，到刚刚起步的新品牌，都在使用它。
Household names like Allbirds and Cotopaxi to brands just getting started.
Shopify提供数百个即用模板，帮助你打造与品牌风格一致的精美在线商店。
With hundreds of ready to use templates, Shopify helps you build a beautiful online store that matches your brand style.
让你的推广效果仿佛背后有个营销团队在运作。
Get the word out like you have a marketing team behind you.
您可以轻松地在客户浏览或漫步的任何地方创建电子邮件和社交媒体活动。
Can You easily create email and social media campaigns wherever your customers are scrolling or strolling.
是时候用 Shopify 将那些‘如果’变为现实了。
It's time to turn those what ifs into with Shopify today.
立即前往 shopify.com/bigtech 注册您的每月 1 美元试用。
Sign up for your $1 per month trial at shopify.com/bigtech.
前往 shopify.com/bigtech。
Go to shopify.com/bigtech.
就是 shopify.com/bigtech。
That's shopify.com/bigtech.
我们回到《大科技》播客，与 M. G. Siegler 一起。
And we're back here on big technology podcast with M.
G.
G.
Spyglass 的 Siegler。
Siegler of Spyglass.
你可以在 spyglass.org 找到 Spyglass，这是我必读的网站之一。
You can find Spyglass at spyglass.org, one of my must reads.
我强烈推荐你去查看并注册。
Definitely recommend you go check it out and sign up.
好的，M。
Alright, M.
G。
G.
所以，我给这一部分起了个标题叫‘Siri 依然严重出问题’，因为苹果似乎不断公布更多关于 Siri 未来功能的宏大计划。
So, you know, I kinda titled this section as Siri Seriously Still Broken, because it does seem like we keep getting more and more, big plans from Apple about what Siri can be.
但事实是，现在已经是 2026 年 4 月，距离 ChatGPT 发布差不多就要四年了，而 Siri 依然糟糕透顶。
But, yes, here we are in April 2026, nearly it's gonna it's gonna be, I mean, couple months away from four years after ChatGPT has launched, and the thing is is still terrible.
那我们先聊聊最新动态，然后再探讨一下苹果的愿景，以及他们是否真能实现这些目标。
So, let's just talk about the latest, then we can kind of riff on on what the vision here is and whether they'll be able to accomplish it.
你提到了马克·古尔曼在彭博社的报道，其中提到苹果正在为 iPhone 测试一款独立的 Siri 应用。
You highlighted some reporting from Mark Gurman at Bloomberg that where where Gurman said that Apple is testing a dedicated Siri app for the iPhone.
该应用的主要界面将以前的对话以列表或带有文本预览的圆角矩形网格形式显示。
The app's main interface will display prior conversations in either a list or a grid of rounded rectangles with text previews.
用户可以固定喜欢的聊天、保存旧对话、跨对话搜索，并开始新的聊天。
Users can pin favorite chats, save older conversations, search across interactions, and start new chats.
对话视图类似于苹果信息应用中的线程。
The conversation view resembles a thread in Apple's messages app.
抱歉，你刚才说。
You also sorry.
古尔曼还指出，正在测试的一个新设计是将Siri放在屏幕顶部的动态岛区域。
Gurman also highlighted that one new design in testing, places for Siri is at the top of the screen within the dynamic island.
还有一些报道称，它将使用手机上的操作按钮。
There are also some reports that it's gonna use the action button on the phone.
所以这很有趣。
So it is interesting.
看起来苹果基本上正在接受这样一个想法：将一个基于谷歌的聊天机器人应用——即Siri聊天机器人应用——直接集成到操作系统中，无论是在动态岛上方以独立应用形式呈现，还是通过操作按钮，或以上所有方式的组合。
It seems like Apple is basically kinda warming to the idea that it's going to bake a chatbot app or Siri, the chatbot app built on top of Google, directly into the operating system, whether that's above the Dynamic Island in its own app, using an action button, or a combination of, all of the above.
但又一次，你知道，Siri其实还没准备好做这件事。
But yet again, you know, Siri is not really ready to do that.
那你觉得这一切意味着什么？
So what do you think this is all about?
你认为这代表了什么？
What do you think this means?
你觉得这会走向何方？
Where do you think it's going?
你认为苹果有能力实现这一点吗？
And do you think Apple Apple is capable of pulling it off?
我觉得这是个好迹象，毕竟根据古尔曼的报道，他们打算推出一个独立应用。
I think this is a good sign, the fact that they're gonna do a standalone app, you know, per Gurman's reporting.
其实我去年就写过，当时我觉得他们不这么做简直不可思议。
I actually had written about this last year that it seemed wild to me that they weren't going to do it.
对吧？
Right?
因为之前的报道说，Siri 将会像过去十五年那样继续作为后台服务存在，你只能通过说‘嘿，Siri’之类的指令来唤醒它，而我不想这么做，因为那样会触发我所有的设备，或者通过某种软件或按钮来调用。
Because the previous reporting had said that Siri was going to sort of remain as it has been over the past fifteen years where it's sort of in the background and you invoke it by either saying, you know, hey, so and so, which I don't want to do because I'll trigger all of my devices right now or using some sort of software or button or whatnot.
但现在，如果他们推出一个独立应用，感觉他们必须这么做，因为这已经成为标准——多亏了 ChatGPT，现在人人都习惯了这种方式。
But now if they have an app, it felt like they needed to do that because that is basically the standard, thanks to ChatGPT, that everyone is used to now.
就像你会说：‘我要用 AI 来做这件事。’
Like you go to you say like, Oh, I'm going to use AI to do this.
至少，你会打开一个网站来使用它，但在手机上，你肯定是要打开一个应用的。
And you either At the very least, you're loading up then a website to do it, but on a phone, you're definitely loading an app.
你不会在后台跟 Siri 说话。
You're not talking to Siri in the background.
所以，这看起来像是苹果过去常常会以惨痛方式学到的经验——他们先发布产品，却没有推出应用，用户会感到困惑，不知道 Siri 的应用在哪里，然后他们才最终推出应用。
And so it felt like this was the type of thing that historically Apple would learn the hard way, meaning like they'd roll it out, they wouldn't have the app, people would be confused and not understand like where Siri's app is, and then they would eventually launch the app.
但这次，感觉他们提前行动了，意识到这才是市场的现实，尽管过去许多报道都曾贬低过聊天机器人和这类应用的想法。
But instead, it feels like they're getting ahead of it by doing that and sort of recognizing this is the reality of the market, even though they've historically said Historically, by many reports now in the past have disparaged the idea a chatbot and an app for those things.
他们曾认为 AI 会是某种更宏大、更宏伟的东西，再次运行在后台，成为每个应用的一部分，而不需要成为一个独立的应用。
They thought that AI would be some bigger, more grandiose thing where it's, again, running in the background and a part of every app and doesn't need to be its own standalone thing.
但现在的市场现实并非如此。
But that's not the reality of the market right now.
现实是，人们会想要这个聊天应用，并希望有一个存储空间来查看他们之前问过的问题，而如果不开发一个应用，他们根本无法做到这一点。
The reality is people are going to want this chat app and they're going to want a repository to see what they've asked before, and they had no way to do that without building an app.
所以我认为这是一个好兆头。
And so I think it's a good sign.
你知道的？
You know?
是的。
Yes.
一切仍在继续推迟，但现在很明显，他们只是在等待即将到来的WWDC大会。
Everything continues to be delayed, but it's obvious now that they're just waiting for WWDC coming up.
他们已经公布了明年六月的举办日期，这将是他们上一次宣布所有计划后的两周年纪念日，当时他们因无法按时交付而不得不撤回所有功能，这可谓臭名昭著。
They announced the dates of it coming up next June, which will be the two year anniversary of when they last announced, you know, everything that was coming before famously, infamously having to pull back all the things that that they just couldn't ship on time.
因此，随着时间推移，我越来越乐观，相信这次他们会做对——他们显然已经与谷歌达成了协议，使用Gemini，而Gemini将成为Siri模型的基础。
And so I'm getting more optimistic as time is going on that I think that they're going to do it right this time, that they have obviously the deal in place with Google to use Gemini and Gemini is going to underpin you know, the the Siri models.
现在又有新报道称，他们将允许其他第三方进入，显然他们已经将ChatGPT纳入其中。
And now the new reports that they're gonna allow other third parties potentially to come in there, including obviously, they've already had ChatGBT as a part of it.
我觉得目前它的位置太隐蔽了。
I think it's way too buried for what it's been right now.
它在Siri里面藏得太深了。
It's way too buried within Siri.
但如果他们把它提升为一个核心功能，并且支持Claude、Perplexity或其他任何类似应用，让用户可以直接在基于Gemini的Siri基础上无缝接入这些服务，那就好了。
But if they make it more of a first class citizen and along with Claude, if you wanted Claude, if you wanted Perplexity, if you wanted any number of other sort of apps, they'll just allow them to plug right into it on top of their baseline Siri powered by Gemini.
我认为这才是正确的做法，我也越来越乐观了。
I think that's the right approach, and I'm getting optimistic.
我知道这听起来很傻，毕竟已经十五年了，我们每年都被骗一次，但我认为他们现在开始逐渐看清了自己应该走的方向。
I know this sounds foolish given the fifteen years, given we've been fooled every single year for those fifteen years, but I think that they're starting to sort of come into focus on what this should look like for them.
有趣的是，他们的观点——认为这必须是一个聊天机器人应用——现在正与其他人的看法互换，那些人觉得AI应该是一种神奇的、深度融入所有编程中的存在。
Interestingly, their perspective of, yeah, it's gotta be a chatbot app kind of is now changing places with the others, which are like it should be an AI that is sort of magical and baked into all programming.
嘿，这都是小步前进。
Hey, it's baby steps.
这是苹果。
This is Apple.
他们总是会落后一步，而这正是人工智能真正的风险，对吧？
They're always going to be late to the game, and that's the real risk with AI, right?
即使从更大的层面来看也是如此。
Even at a bigger level.
人工智能的发展速度远超其他技术，而苹果历来都是在竞争对手之后才进入市场，然后所谓地‘做对’、‘正确地’完成它。
The AI moves so fast relative to other technology, and Apple historically has come into markets later than competitors and sort of, quote unquote, done it right, done it correctly.
因此，这里的情况可能是，他们做出了有史以来最精美、最出色的AI聊天应用，让其他应用相比之下显得更粗糙、更幼稚等等。
And so that might be the case here where they make the most beautiful, the best sort of AI chat app that anyone's ever done, it makes the others sort of look much worse, more childish or whatnot.
但其他人已经转向了这些完整的智能代理服务套件，而苹果则不得不重新回到起点，尽快赶上来。
And then everyone else, though, has moved on to these agentic full on service suites AI, and Apple then has to sort of go back to the drawing board and try to hustle to come up with that as quickly as possible.
对。
Right.
我的意思是，这挺有意思的。
I mean, is interesting.
苹果仍然没有做好聊天机器人，而OpenAI和Anthropic已经迈向下一步了。
Apple still hasn't gotten chatbot right, and OpenAI and Anthropic are already on to the next thing.
你在你的文章中恰好强调了这一点。
And you you sort of highlight this in your story.
你曾经提到，最大的问题在于，苹果将核心工作外包给谷歌，可能永远无法在纯技术层面赶上，这阻碍了硬件真正迈向下一阶段的转型。
You used to talk about the biggest issue or the bigger issue remains that by outsourcing the core work to Google, Apple is might never be able to catch up from a pure technology standpoint, and this encumbers the true what's next shift in hardware.
我认为这正是我们在这档节目中一直争论的问题：苹果是否真的不该在AI开发上进行大规模投入？如果只是把谷歌的技术整合进产品，而不必承担巨额研发开支，它到底会失去什么？
And I think this kind of points we've been having these, like, debates back and forth on the show for a little bit about whether Apple was actually right and not making these big investments, in AI development, and sort of what does it lose if it just implements like Google's technology into its products and doesn't have to do all that big R and D spend.
但我认为真正的答案是，如果你不投入研发，你就会置身于一个正在超越你的技术世界；当你好不容易跟上第一步时，你的竞争对手可能已经走到第三或第四步了，你能否参与这场游戏，完全取决于他们的意愿。
But I think the answer really is that if you don't do the R and D spend, you're in a you're working in a technology world that's going to be passing you by, and by the time you catch up on, you know, step one, the your competitors can be at step three or four, and it's basically at their whims, as to whether you're even able to participate in that game.
这就是问题所在。
That's the problem.
是的。
Yeah.
我认为这一切都是正确的。
And and sort of so I think that that's all correct.
而且这还会带来第二层影响，那就是，他们可能在产品和模型方面确实落后了。
And there's also the sort of the second order effect of that, which is that, look, they may or may not be behind on on the product side and and with the models.
但即使只是因为不参与这一领域的竞争，虽然从财务上看这可能很有道理，但他们却面临一个真正的风险：公司内部永远无法培养出真正参与AI竞争所需的基因和心态。
But what if even just by not sort of competing in that space, while it may make a lot of sense financially, they run the real risk that they just never get the right DNA within the company and the right mentality to really be able to compete in AI.
如果你相信AI是未来渗透一切的技术，那么他们很可能根本缺乏这种正确的思维，总是像你所说的那样，落后一步，或者根本找不到合适的人才来执行这项工作。
And if you believe that it's the future technology that sort of permeates everything, there's a real risk that they just don't have, again, that right mindset and they are constantly, as you know, running a step behind and or, you know, they don't quite have all the right people in place to be able to execute on this.
正如我们刚才提到的，这种技术的演进速度比以往任何技术都要快。
And as we just talked about, like, this is technology that's moving faster than any other technology that's come before it.
因此，问题不仅仅是落后，而是他们会越来越远远地被甩在后面。
And so it's not just being behind, it's that they will fall increasingly further and further and further behind.
我真的认为这是一个实实在在的风险。
And I do think that that's a real risk.
这也回到了苹果长期以来秉持的一个理念：如果你想在计算和硬件领域证明自己，你就必须做软件。
And it's also like it goes back to the notion that Apple has long sort of adhered to, which is that if the whole mentality of if you want to prove yourself in computing and in hardware, you've got to Sorry, in software, you've got to make hardware.
那么，如果要在计算的下一阶段证明自己，你也必须自己打造AI呢？
And it's like, what if in order to prove yourself in this next phase of computing, you also have to sort of make your own AI?
这是其中一部分。
That's a part of it.
而苹果目前正将这一点传递给谷歌。
And Apple is sort of seeding that to Google right now.
你当然希望他们有自己的内部项目在推进。
And you'd hope they have their own projects, obviously, that they've been working on internally.
你希望这只是一个临时的权宜之计。
You'd hope that this is just a stop gap measure.
但与此同时，如果他们因为现在只是在持续输出这些工作，而永远无法弥合这一差距怎么办？
But at the same time, what if they can never bridge that gap because they're just, you know, going and seeding that work right now?
这几乎完美地引出了我们接下来计划讨论的话题：他们是否应该收购Anthropic？
And this is almost the perfect setup for what we had planned to speak about next, which is should they acquire Anthropic?
最近在讨论中有人提出，苹果或许应该尝试收购Anthropic，而你过去一个月也写过一些相关内容。
Now it's been floated, in the discourse, shall we say, that maybe Apple should try to acquire Anthropic, and you wrote about it a little bit over the past month.
我们之前在节目中很少深入讨论这一点，只是觉得达里奥根本不可能把这家公司卖给蒂姆·库克，成为苹果的子公司。
I we never really spoke about it too much in-depth on the show just because it seemed like Dario just I couldn't imagine Dario selling that company to Tim Cook and being a subsidiary of Apple.
而且现在财务上也可能根本行不通，苹果是一家价值3.8万亿美元的公司，而Anthropic很可能将以一万亿美元的估值上市。
And now also the finances might just be unworkable, or Apple's a $3,800,000,000,000 company, and Anthropic is probably gonna go public, at a trillion.
所以，你知道，收购Anthropic可能需要动用苹果超过四分之一的市值。
So, you know, maybe more than a fourth of Apple would have to go towards, buying Anthropic.
但让我们先把这一点放在一边，抛开这些怀疑，好吧。
But let's let's throw that aside, throw that skepticism aside and just Yep.
现在轮到你来谈谈，这样的收购可能会是什么样子。
Give you the floor here to talk about what an acquisition like this could look like.
是的。
Yeah.
但需要明确的是，这更像是一种思想实验。
And this was, to be clear, like, more of a thought experiment.
我之前已经做了铺垫。
I I couched it.
对吧？
Right?
我的意思是，我不认为这件事会发生，但我的出发点是，正如我们在之前的节目中讨论过的，Anthropic 和美国国防部之间的关系。
Like, that I don't think this is going to happen, but I did think my jumping off point was the, as we've talked about in previous episodes, the Anthropic and DoD situation.
这对我来说，似乎是一个独特的时间点，对吧？
That to me made a unique moment in time potential here, right?
Anthropic 正受到政府的攻击，而且至今仍在被攻击。
Where Anthropic was under attack from the government and still is.
所以，他们会不会更愿意接受某种收购，也就是被一家大型科技公司收购，以保护他们，确保他们能继续实现自己的使命？
And so would they be potentially more amenable to doing a type of acquisition, meaning getting acquired by one of the big tech companies in order to protect them and ensure that they can continue to do what their mission is.
如果你看看整个行业格局，哪家大型科技公司最有可能完全认同这种愿景？
And if you look around the landscape, like who's the only real of the big tech companies that sort of would fully buy into maybe that vision?
看起来苹果和 Anthropic 的理念最为契合，这有点讽刺，因为苹果正是目前唯一一家没有投资 Anthropic 的大型科技公司——他们根本没有投资任何一家。
Like, it does seem like Apple and Anthropic are the most aligned, which is sort of ironic because Apple is the one company that I think is not an investor now of the big tech companies in Anthropic because they're not an investor in any of them.
他们曾经走过与 OpenAI 合作的道路，但最终没有成行。
They famously went down the path with OpenAI, but didn't end up doing it.
所以，确实感觉两者之间存在一些非常自然的契合点。
And so it does feel like there's some real natural alignments there.
现在，达莉亚在那里工作吗？
Now, Daria work there?
你知道吗，他会敢卖掉一家可能估值达到万亿美元的公司吗？
Would, you know, would he dare sell a company that's potentially going to go public at a trillion dollar valuation?
再次强调，鉴于他们目前面临的风险，这是一个独特的时间点。
Again, it's a unique moment in time given given the risk that they face right now.
另一方面，对于苹果来说，我们刚才说的所有这些，对，就是这样。
And on the flip side, for Apple, everything we just talked about, yeah, that's that's exactly it.
感觉他们必须做点什么。
It feels like they need to do something.
存在一种可能性，他们需要进行一项真正改变格局的交易。
There's a world in which they need to do something that's a true game change deal.
这并不是说，像他们时不时做的那样，花一亿美元去收购一些人才，那些交易也不错。
And it's not, you know, making these $100,000,000 purchases here and there, which they do from time to time of talents, and those are good.
当然，过去几十年里，苹果在执行这一策略方面做得和任何人一样出色。
And, obviously, Apple has has run that that game plan as well as anyone over the past couple of decades.
但这次更像是——我不太想提起收购NeXT并请回史蒂夫·乔布斯的例子，虽然那笔交易相比现在这个规模要小得多。
But this would be more like you know, I I hate to invoke the the idea of buying next and bringing Steve Jobs on, which was a much smaller deal relative to what this would be, of course.
但那确实是一次改变游戏规则的交易，真正重塑了公司的基因。
But it's still, that was a game changer deal that really changed the DNA of the company.
而这个假设性的理论交易也会朝着类似的方向，我认为它远不止是引入一个出色的产品或有潜力的业务那么简单。
And this hypothetical theoretical deal would be along those lines in that I think it would be about much more than bringing on a great product and a great potential business.
它关乎改变公司的基因，真正让公司为迎接人工智能未来的下一阶段做好准备。
It would be about changing the DNA of the company and really getting them geared up for the next phase of whatever the AI future is.
而目前，就我们刚才讨论的所有内容来看，很难想象他们在没有某种根本性的思维转变、没有彻底改变文化的情况下，如何实现这一点。
And right now, it's hard to see for everything we just talked about how they do that without some sort of real mentality shift and something that totally changes the culture to do that.
所以，这就是我的观点。
So that's my argument.
你知道，我本来想说，我实在难以想象达里亚会成为苹果的产品经理，但话说回来，我们知道他们一直在寻找蒂姆·库克的接班人。
You know, was going to say I couldn't really imagine Dario as a product manager at Apple, But then again, we know they've been looking for a successor to Tim Cook.
他？我不知道。
He he I don't know.
这永远不会发生。
That would never happen.
他显然是AI部门的负责人。
He would he would be the head of AI, obviously.
是的。
Yes.
你知道的，他会是约翰·G。
You know, he'd be the g John G.
G。
G.
AI主管。
AI officer.
但他还会接受这个职位吗？
But would he still be okay with that?
我不知道。
I don't know.
我的意思是，再次强调，你会拥有苹果无限的资源。
I mean, again, you would have Apple's unlimited resources.
他们根本不需要担心融资问题。
They would never have to worry about fundraising.
当然，如果他们与苹果达成了合适的协议，就根本不必担心这个问题。
Well, presumably, if they if they struck the right deal with Apple, they would never have to worry about that.
就像苹果的利润如此之高，完全可以投入这笔资金。
Like Apple has more profits that they could funnel into this.
而且，如果你相信谷歌有DeepMind，微软有OpenAI，现在他们也都在内部开发自己的版本，比如我的项目，还有他们自己的超级智能项目。
And again, if you believe Google's got DeepMind, Microsoft had OpenAI, now they have their you know, version that they're working on internally, the my stuff and and their own superintelligence stuff.
亚马逊显然也在内部研发自己的超级智能项目。
Amazon's obviously working on their own internal superintelligence stuff internally.
那么苹果，他们真的能只选择合作吗？
And so Apple, can they really afford to just partner?
我的意思是，他们真的负担得起吗。
Like, yes, literally, can afford it.
这就是为什么这是一笔很棒的交易，因为他们没有像所有竞争对手那样把钱都花在资本支出上。
That's why it's like a great deal in that they're not spending all the money on CapEx that all of their their rivals are.
但仍然存在真正的风险。
But there's still a real risk of that.
每个人都说，哦，苹果最终会笑到最后。
And everyone points to like, oh, Apple's gonna get the last laugh here.
但如果不走这条路，公司面临着实实在在的长期风险。
But there's a real, real longer term risk for the company by not going down this path.
完全同意。
Totally agree.
好的。
Alright.
在我们结束之前，我们 definitely 应该花几分钟聊聊Meta。
Before we go, we should definitely spend a couple minutes on Meta.
他们一直没能推出下一个爆款，我们可以谈谈Reality Labs，还有这家公司内部的AI进展，我的意思是，有时候确实会出现延迟，需要几年的研究才能把东西整合起来。
They just haven't been able to put together the next hit, and we we can talk about Reality Labs, clearly whatever's happening on AI, within that company, I mean, maybe there's, you know, sometimes you can have delayed progress where you need a couple of years of research to put something together.
他们没有一个有竞争力的大型语言模型。
They don't have a competitive LM.
而且，这和苹果面临的是同一个问题。
And again, it's the same problem as Apple.
他们现在落后了不止一代，而是两代，与前沿技术相比。
They're now, you know, not one but two generations behind what the cutting edge is.
那么，你认为这反映了Meta如今的状况如何？这又说明了扎克伯格怎样的处境？他们无法推出下一个爆款产品，也确实未能摆脱对广告的依赖。
And so what what do you think the state of what do think that says about where Meta is today, where it said what it says about Zuckerberg, the fact that they are unable to come up with that next hit and certainly have not been able to diversify away from advertising.
是的，我的看法有些转变了。当他们最初宣布这个规模庞大的交易时，这实际上标志着LAMA战略的终结，以及他们未来的新方向。我当时写了一篇文章，问：Meta能买下AI的未来吗？
Yeah, I've swung around a little bit here because when they first sort of announced the scale deal that sort of, you know, was the effective end of the LAMA strategy and, you know, their new path going forward, I wrote a post just saying like, Can Meta buy the future of AI?
我的观点是不能，我觉得他们本质上是在收购雇佣兵，而要真正做成这件事，需要一支像Anthropic那样的真正信仰者团队。
And my thesis was no, that I felt like they were basically buying up mercenaries and it would require a team of true believers in the mold of sort of Anthropic, I think was the main example I used to be able to actually do this.
但随后发生了很多事，显然在接下来的几个月里发生了大量变化，尤其是埃隆迅速启动了XAI，并如此快速地建成了那个数据中心。
And then all this stuff happened, a lot happened obviously in the subsequent months, but like certainly with Elon spinning up XAI and how we got that data center up so quickly.
我开始觉得，也许你真的可以单纯用钱来解决这个问题。
And basically, felt like maybe maybe you could just throw money at the problem.
对吧？
Right?
而且你基本上能快速跟上进度。
And and you could basically get up to speed quickly.
快进到几个月前，有新的报道指出，尽管XAI花了这么多钱，模型从各方面看也不错，但还不够好。
Fast forward, you know, a couple months ago, there's new reporting that, well, actually, looks like x AI for all that money spent and the models by all accounts are are good, but it's not good enough.
这无关紧要。
It doesn't matter.
对吧？
Right?
那意义何在呢？
Like, what's the what's the point?
XAI到底在做什么？
Like, what's what's XAI doing?
还有Meta，花了这么多钱，十亿美金，现在却不得不推迟他们的新项目，而且从公开表态来看，这项工作显然被刻意淡化了。
And and Meta, all that money spent, a billion spent, and now they have to delay they delayed, you know, potentially their their new work, and it seems like that work definitely feels like it's being couched in the public comments about it.
这就像说，这会是一个不错的起步，但无法与Geminis和前沿模型竞争。
It's like, oh, it's going to be a good first step, but it's not going to be competitive with the Geminis of the world and the cutting edge models.
所以现在似乎存在一种情况：你就算花光全世界的钱，也还是追不上前沿水平。
And so there's also a world now, it seems like, in which all of that money You could spend all the money in the world and you still can't catch up to what the leading edge is.
或者即使达到前沿水平也无济于事，因为ChatGPT和现在的Claude with Claude Code已经占据了心智优势，它们会持续陷入这种不断改进的循环中。
And or even being at the leading edge, it doesn't matter because the ChatGeeBTs of the world and now Claude with Claude Code, they have the mindshare advantage where they're just going to be locked into these cycles where they just keep improving.
而这种自我改进机制让它们变得越来越好、越来越快，就像当年谷歌在搜索引擎领域凭借优势碾压Bing和雅虎一样，因为这本质上是一种自我强化的过程。
And it's sort of the self improvement stuff just makes them better and better, faster and faster, just like Google leveraged back in the day with search engine where they were able to just outcompete the Bing's of the world and Yahoo's of the world because, again, it just was self fulfilling in a way.
所以我们现在似乎正进入这样一个阶段。
And so it feels sort of like that we're into that.
而如今，Meta的情况也是这样，是的。
And now Meta, you know, is yeah.
有很多报道提到他们正在启动一些新的小型超级智能项目，组建新团队，还有各种其他举措。
There's all these reports about these little new sub superintelligence projects that they're spinning up and these new groups and all this other sort of stuff.
而且，这些报道中，他们公开宣称一切都很棒，我们正在取得进展，诸如此类。
And and, again, the reports, you know, that they say on the record, everything is great, and we're and we're making progress and and yada yada.
但你所暗示的那篇我写的文章提到，他们在过去十多年里推出新产品的记录并不好，元宇宙显然是一个典型的例子。
But, like, the the post that you're sort of alluding to that I wrote about where they weren't you know, they just they don't have a good track record over the past decade plus of launching new things, obviously, with the the metaverse being, you know, the prime example.
但还有其他几个项目，比如他们的加密货币项目Diem，以及许多其他完全没有任何进展的项目。
But several other projects, they had a crypto project, Diem, and and, know, they've had many, many other things that just have gone nowhere.
当然，他们还曾推出扎克伯格的重大倡议——专注于加密通信，这本应成为社交网络的未来，也成为公司的未来。
And and, of course, they had their Zuckerberg's big initiative of going after encryption, and that was gonna be the future of of social networking, and that was gonna be the future of the company.
但现在他们正在撤回这一计划，因为再次说明，除非收购某些东西，否则他们似乎无法获得任何进展。
And now they're pulling back on that because, again, they don't get they don't seem to get traction unless they acquire something.
而我们现在处于一个棘手的境地：除了Scale，他们还能收购谁呢？这似乎是一个有问题的、充满争议的交易。
And we're in the tricky position where who were they going to acquire besides scale, and that's a seems like, you know, maybe a problematic, problematic deal that they did there.
是的。
Yeah.
我觉得这个方案行不通。
I don't think that worked.
你知道吗，这很有趣，因为我回想起Meta过去举办的开发者大会，扎克伯格曾展示过一个十年路线图。我记得2016年坐在观众席上时，他几乎准确地描绘了技术发展的路径，而人工智能正是这条路线的终点。
You know, it's interesting because I was thinking back to, like, the old developer conferences that Meta used to hold, and Zuckerberg put, like, a ten year road map, and I remember being in the audience in 2016 specifically, and he was basically right on the road map that there was gonna be some technology development, and then AI was sort of at the end of the tunnel.
而我们现在看到，他们和苹果一样，也拿不出什么像样的成果。
And here we are, and yet they don't have they also, like Apple, don't have much to show for it.
不仅如此，超级智能实验室已经出现了一些人员离职。
And not only that, Super Intelligence Labs, like, has already seen some departures.
我们不知道王大仁在公司里到底在做什么。
We don't know what Alexander Wang is really doing within the company.
据我听到的风声，他和Meta管理层在产品方向和未来规划上确实存在一些分歧，你们打算怎么处理他？
You know, from the chatter I've heard, you know, there's definitely been some disputes with him and, and Meta leadership about the product, you know, the the path forward and what are you gonna do to him?
你们刚刚花了数十亿美元，把他和他的团队招揽进来。
You just paid more, you know, billions of dollars to bring him and some of his team on board.
而现在，同样根据报道，公司可能正面临一场大规模裁员，可能在我们下次交谈之前就会发生。把这些事情综合起来看，你会发现，他们可能只剩下几次机会，能真正把AI做对了。
And now, and this is again just according to reports, but the company might be staring down the barrel of a massive layoff that could be en route between now and the next time we speak, and you put it all together, and you're just like, they got maybe a couple more shots at the AI, getting this AI thing right.
除此之外，他们看起来有点像当年的雅虎——广告业务是他们唯一做对的事，也只能靠它撑一段时间。
Other than that, they they look like, they look kinda like a potential Yahoo, right, where, like, they they got advertising right, and it's the one thing they do right, and they're gonna live off it for a while.
但凡了解马克·扎克伯格的人都知道，这几乎是他最大的噩梦。
But, you know, anything we know about Mark Zuckerberg is that that has effectively been his greatest nightmare.
是的。
Yeah.
这个类比挺有意思的。
That's that's an interesting analogy.
我之前没这么想过，但我能理解你为什么这么讲。
I hadn't thought about that, but I could see why you draw that.
不同的是，扎克伯格还在那里。
The difference would be that Zuckerberg is there.
就像你记得的，杨致远曾试图重返雅虎，当年在史蒂夫·鲍尔默时代，雅虎差点以四百多亿美元的价格卖给微软，而扎克伯格对公司的控制是全方位的，从股票到管理权都掌握在他手中。
And while, like, Jerry Yang tried to come back, as you'll recall, to Yahoo, and they almost sold to Microsoft for 40 some billion dollars back in the day with in Steve Ballmer days, Zuckerberg has total control of the company, right, from the stock on down.
他掌控着一切。
He is in control.
这既是优势，也可能是一种劣势——他们的一切都跟着扎克伯格走，Meta的命运也随他而动。
And so it's a strength and probably a weakness in ways, right, where, like, they do as he goes, so goes Meta.
所以他带领公司走向了元宇宙，也许他在这一点上走偏了。
And so he leads them down the metaverse path, and maybe he maybe he was wayward in that.
对吧？
Right?
也许正如你所说，他分心了，他其实早就有路线图，知道AI即将到来。
And maybe he he was distracted to your point of, like, they he had the roadmap for knowing AI was coming.
谷歌也是，顺便说一句，桑达尔已经谈论这个话题十多年了。
So did Google, by the way, and Sundar has been talking about that for for a decade plus.
对吧？
Right?
不同的是，谷歌真正执行了，而Meta至今还没有，尽管他们引进了杨立昆，尽管当年他们曾试图收购DeepMind但失败了。
The difference is Google actually execute upon it and Meta so far has not, even though they brought in Yan Lakun and even though they tried to buy DeepMind back in the day and sort of failed at it.
而且，再次说明，这导致了杨立昆的加入，他们似乎已经具备了正确的要素，但不知为何，始终无法聚焦于此。
And again, that led to Yan Lakun coming in potentially and sort of they had the right pieces in place, but they have not been able to sort of, for whatever reason, zero in on it.
有一部分我很好奇，回到那篇帖子，他们其实很擅长收购。
And some of me part of me just wonders, again, going back to that post, it's like they're good at they're great at acquisitions.
他们是历史上最擅长收购的公司之一。
They're one of the best companies ever at acquisitions.
但当他们不做收购时，其实并不怎么样。
But when they're not making acquisitions, like, they're not that great.
比如，他们在模仿Snap之类的产品上还挺拿手的。
Like, they've been they've been good in sort of copying Snap and stuff like that.
对吧？
Right?
从某种意义上说，你甚至可以把这称为收购另一个创意。
Even in a way you could call that an acquisition an acquisition of another idea.
在这里，他们既没能收购可能改变游戏规则的公司，从而真正进入AI竞赛，也没能像现在这样成功复制其他公司的策略。
Here, they have not been able to both acquire potentially the game changing company that gets them into the AI race in a meaningful way and or clone as they're obviously trying to do now the game plan of others.
那什么才能改变这种情况呢？
And so what changes that?
现在，扎克伯格和Meta的其他人都可能会说，是那些眼镜的问题。
Now, Zuck and and everyone else at Meta would probably say it's it's the glasses.
对吧？
Right?
比如，他们在下一代设备 paradigm 转变的早期阶段已经站稳了脚跟。
Like, that they have a foothold in in the early part of of, you know, the next paradigm shift in terms of new devices.
咱们走着瞧吧。
Like, we'll see.
我的意思是，他们现在确实很成功，但相比 iPhone 等主流产品，规模还是太小了，而且仍然依赖 iPhone。
I mean, they're certainly successful right now, but it's super small compared to a lot certainly compared to the iPhone, and it still relies upon the iPhone.
那么，他们接下来还会推出什么产品呢？
And so, like, what else are they going to come out with?
听起来像是一个腕戴设备，他们一直在研发，还有其他一些我肯定正在秘密进行的 Skunk Works 项目。
You know, sounds it like a wrist device, you know, they've been working on and and several other, I'm sure, Skunk Works projects that they have right now.
这正是他们的希望所在，否则，他们又只是在跟那些已经在 AI 领域更成熟的对手竞争。
And that's gonna be the hope for it because otherwise, they're just competing, again, with all of these sort of already more established players in AI.
所以，他们需要推出真正能改变游戏规则的产品，而不是一味追赶，而是真正引领潮流。
And so they need to come up with something that actually changes the game, and they're not just chasing, and they're actually leading.
而且，眼镜目前可能是他们最好的机会，但等到谷歌、苹果和其他公司推出竞品时，情况会如何发展，我们拭目以待。
And, again, the glasses are probably the best place they have right now, but we'll see what happens when Google and Apple and everyone else comes out with their competing products.
这会非常有趣，因为除了技术层面之外，别忘了苹果在全球拥有庞大的零售网络。
Like, that's gonna be fascinating because beyond even the technology angle of it, like remember, Apple has a retail footprint all around the world.
他们能够轻松推广任何推出的产品。
They're gonna be able to move any sort of product that they put out there.
如果定价合理，而不是采用Vision Pro那种定价策略，他们在多个方面都将超越Meta。
And if it's fairly priced, if it's not the Vision Pro pricing strategy, they're gonna have an advantage over Meta in a number of ways.
这实际上变成了一场竞速，看谁的策略更能有效地将这些新兴设备推向市场。
And it really becomes, you know, trying sort of a foot race as to, like, what's the better strategy in terms of, you know, getting these newfangled devices out there.
我知道我们超时了，但我只想就此结束一下。
I know we're over time, but I just wanna end on this.
塞巴斯蒂安·马拉比写了一本关于谷歌DeepMind负责人戴密斯·哈萨比斯的新书，书中有一段摘录刊登在《华尔街日报》上，你刚才谈到Meta在收购方面的出色表现，以及扎克伯格在投资上的反复无常，这段内容让我立刻想到了它。
Sebastian Malaby has this book about Demis Asabas, the head of Google DeepMind coming out, and there was this excerpt published in the Wall Street Journal, and it just made me you made me think about it when you talked about how good Meta has been at buying and about how Zuckerberg has sort of been all over the place in his bets, and this anecdote, just brings it together perfectly.
当时戴密斯·哈萨比斯还在西海岸，DeepMind尚未被收购，他与谷歌的拉里·佩奇共进午餐，扎克伯格得知后便邀请他共进晚餐。
So Demis Sabas was on the West Coast when he when DeepMind was still independent to have lunch with Larry Page from Google, and Zuckerberg heard about it and invited him, to dinner.
这段内容来自摘录。
This is from the excerpt.
哈萨比斯抵达扎克伯格在帕洛阿尔托的家时，对他进行了一次微妙的测试。
Arriving at Zuckerberg's Palo Alto home, Hassabis administered a subtle test on him.
两人讨论了人工智能的潜力，扎克伯格表现出了适当的兴奋。
The two men discussed the potential of AI, and Zuckerberg expressed appropriate excitement.
但随着晚餐继续，哈萨比斯提到了其他热门技术，比如虚拟现实、增强现实和三维打印。
But then as the dinner continued, Hassabis brought up other hot technologies, virtual reality, augmented reality, three d printing.
扎克伯格对所有这些技术都表现得同样兴奋。
Zuckerberg sounded equally excited about all of them.
这让我明白了我需要知道的一切，哈萨比斯后来说。
That told me what I needed to know what I needed to know, Hassaba said later.
Facebook提供了更多的钱，但我想要的是真正理解为什么人工智能会比所有这些技术都更重要的那个人，于是他选择了谷歌。
Facebook offered more money, but I wanted somebody who really understood why AI would be bigger than all those things, and then he sold to Google.
天啊。
Oof.
这真让人难受。
That's that's painful.
这顿晚餐代价高昂，因为他没有花在可能收购公司未来的关键人物上。
That is an expensive dinner in what he didn't get to spend on on potentially acquiring the future of his company.
这当然恰恰印证了未来的趋势：据报道称，OpenAI的创始人也想让马克·扎克伯格把他招入麾下，或许还提供了更高的报酬，但他最终选择了OpenAI。
And it, of course, harkens exactly right to the future where by the reporting, the founder of OpenClaw, you know, wanted Mark Zuckerberg wanted to bring him on board as well and and maybe offered more money as well, and he went to Open AI.
而且，历史再次重演。
And and, again, history repeats itself.
所以，是的，这看起来并不太好。
So, yeah, not a not a great look.
网站是spyglass.org。
Website is spyglass.org.
嗯。
M.
G。
G.
西格勒当然是我们的嘉宾。
Siegler, of course, is our guest.
每个月的第一个星期一。
First Monday of every month.
M。
M.
G，很高兴见到你。
G, great to see you.
谢谢你能再次回来。
Thanks for coming back on.
谢谢，亚历克斯。
Thanks, Alex.
有空再聊。
Talk soon.
好的。
Alright.
很快再跟你聊。
Speak with you soon.
感谢大家收听和观看，我们下次再见于《大科技播客》。
Thank you everybody for listening and watching, and we'll see you next time on Big Technology Podcast.
关于 Bayt 播客
Bayt 提供中文+原文双语音频和字幕，帮助你打破语言障碍，轻松听懂全球优质播客。