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。
欢迎各位收看《The Informations》TITV。
Welcome, everyone, to The Informations TITV.
我是阿卡什，来自普华永道。
My name is Akash PwC.
今天是6月8日，星期一，纽约尼克斯队今晚将进行第三场比赛。
It is Monday, June 8, and the New York Knicks are in game three tonight.
我们今晚会全程观看。
We're gonna be watching.
在开始今天的节目之前，我想先提一下《The Informations》今天发布的一则报道。
Before we get to today's show, I wanna highlight some reporting that the information published today.
谷歌和英伟达正在考虑把英特尔作为备用芯片制造商。
Google and NVIDIA are considering Intel as its backup chip manufacturer.
这是因为台积电目前难以应对爆炸性的需求。
This comes as TSMC is struggling to keep up with overwhelming demand.
更多详情请访问我们的网站。
You can read more about that on our website.
今天我们也在关注苹果的全球开发者大会。
We're also tracking Apple's Worldwide Developers Conference today.
我们会了解苹果计划如何在其软件生态中推出更多人工智能功能。
We'll learn about how the company plans to implement more Apple intelligence features in its software ecosystem.
我们也在等待SpaceX的重磅公开亮相。
We're also awaiting the blockbuster public debut of SpaceX.
这家公司计划在本周五在纳斯达克上市。
The company plans to list on the Nasdaq this Friday.
本周三，6月10号，我会采访几位主导《The Information》SpaceX报道的记者：西奥·韦特、弗莉塔·PwC和科里·温伯格。
This Wednesday, June 10, I'll sit down with the reporters who have been leading the information SpaceX coverage, Theo Wayt, Vlita PwC, and Cory Weinberg.
这场活动仅限订阅用户参加。
The event is exclusive to subscribers.
你现在就可以去theinformation.com/events报名。
You can RSVP right now at theinformation.com/events.
今天节目中，我们会深入探讨这次IPO中最有趣的一个角度——负责SpaceX财务的那个人。
Today on the show, we will dig deep into one of the most interesting angles of this IPO, the man leading SpaceX's finances.
布雷特·约翰逊即将公开发布一篇关于这位CFO的专访，我们马上就会谈到他。
Brett Johnson will publicly publish a profile about the CFO that we're going to talk about very shortly.
接下来我们会转个话题，聊聊AI的自我迭代升级，它到底是什么，为什么这么重要，为什么各大公司都在砸钱投入。
We're then going to shift gears and talk about recursive self improvement in AI, what it is, why it's important, why companies are pouring money into it.
节目最后，我们会和普华永道聊一聊，在AI和代币优化的时代，预算该如何预测。
We're going to close out the show with a conversation with PwC about budget predictability in the era of AI and tokenmaxxing.
今天这期节目内容非常精彩，咱们直接进入正题。
It's going to be a great show, so let's get right on into it.
先请我们的埃隆·马斯克记者西奥·韦特来帮我们铺垫一下背景。
To get us started, I want to bring on our Elon Musk reporter, Theo Wayt, to help set the scene for us.
西奥，周一好。
Theo, happy Monday to you.
这周是SpaceX的重头戏。
It is SpaceX week.
你最近怎么样？
How are doing?
我挺好的。
I'm doing well.
你呢？
How are you?
我也不错。
I'm doing well.
好的。
Okay.
那你给我们说说具体情况吧。
So give us the rundown here.
现在到周五之间还有不少时间。
Now and Friday, there's a lot of time between then.
逐天给我们讲讲。
Give us the day by day.
我们该关注哪些事？
What should we be watching for?
你直接说吧。
Just go ahead.
这一周大部分时间我们都在紧张地等着周四和周五，因为那时候才会真正有大动作。
So most of the week is kind of just sitting here nervously awaiting Thursday and Friday, which is when most of the action will be.
接下来几天，大家都在下单呢。
The next few days, you know, everyone's getting their their orders in.
根据需求，SpaceX会在周四确定IPO的定价，然后周五股票正式以SPCX代码开始交易。
And based on demand, you know, SpaceX will finalize the IPO pricing on Thursday, and then the actual stock will start trading under SPCX on on Friday.
SpaceX说每股定价是135美元，但这个数字还可能变动，具体多少得等到周四才能知道。
And SpaceX has said the pricing will be a $135 a share, but that could still change and and fluctuate, and we'll know the actual number on Thursday.
所以，说白了，我们现在就是干等着，心里七上八下的。
So so, yeah, mostly, we're just anxiously sitting around waiting.
明白了。
Okay.
不过在等的这段时间，我想聊聊你上周末发的那篇人物专访。
Well, while we wait, I wanna unpack a profile that you published over the weekend.
你深入分析了首席财务官布雷特·约翰逊。
You did a deep dive into CFO Brett Johnson.
他是SpaceX财务方面主导IPO的关键人物。
He is the man leading the IPO from a financial perspective for SpaceX.
能跟我聊聊布雷特·约翰逊这个人吗？
Tell me a little bit about Brett Johnson.
他和埃隆·马斯克有什么不同？
I mean, how does he differ from Elon Musk?
他是个非常低调的人。
So he's he's a very under the radar figure.
我们之所以是第一个为他做深度报道的媒体，正是因为他是那种特别内敛的理想型CFO。
I mean, there's a reason that we were the first people to really write a profile of him, which is that, you know, he he is he's the ideal CFO and that he's very, you know, understated.
他话不多。
He's quiet.
大家都说他稳重可靠，这跟总是备受关注的埃隆形成了鲜明对比。
People describe him as as steady and, you know, reliable, which doesn't you know, obviously, is a gigantic contrast with Eon who gets all the attention.
但他一直挺过来了，你知道的。
But he's he's been, you know, a survivor.
他早在2011年就加入了公司，那时候大多数人连SpaceX是干什么的都不知道，而他一直是公司唯一的首席财务官。
He's he joined way back in 2011, when most people didn't know what SpaceX was, and he's been the first and only CFO of of the company.
所以他是个极其重要的人物，而且马上也要成为亿万富翁了。
So he's, you know, an incredibly important figure, and and he's about to be a billionaire too.
明白了。
Okay.
他确实是个关键人物，但你之前接触过和布雷特·约翰逊共事过的人吧？
So he's an important figure, but what I mean, you know, you spoke to people who were in Brett Johnson's orbit.
能跟我们说说他这个人吗？
I mean, tell us a little bit about him.
他是不是像我不好说的那样，跟埃隆·马斯克一样充满戏剧性？
Is he as I don't want to say chaotic, but is he as extra as Elon Musk on every level?
或者听起来，他其实没那么张扬？
Or, I mean, it sounds like he's a little bit less less
我觉得不会。
I would say no.
他其实挺能适应的，比如当埃隆·马斯克说，咱们得收购xAI，哪怕那家公司几个月后就要上市，还在疯狂亏钱。
I I would I mean, he's willing to to, know, roll with the punches, like, when when Elon Musk says, you know, we need SpaceX to acquire xAI, which is losing a ton of money just a few months before we're going public.
换作别的公司，这种操作简直太疯狂了。
I mean, you know, in a normal world, that would be a very extreme, thing to do.
但在埃隆·马斯克的世界里，这事儿再正常不过。
And in Elon Musk's world, it's just, you know, pretty typical.
他愿意跟着这种节奏走，结果也确实挺划算的。
And and he's, you know, shown willingness to go go along with that kind of thing, and that's, you know, paid off.
说到上市流程，我那些负责IPO的同事和我都听说，他在跟投行打交道时特别有底气，敢提要求。
In terms of the actual IPO process, you know, my my colleagues who cover the IPO and and I have have both, you know, heard about him being pretty, know, willing to throw SpaceX's weight around and working with bankers.
大家都想参与这个项目，所以他能强硬地要求分权、压低费用，诸如此类。
Like, they you know, everyone wants to be a part of this deal, and therefore, he can make pretty aggressive demands from them around, you splitting up responsibilities and having lower fees and that kind of thing.
他在跟埃隆沟通这方面，做得挺到位的。
He's a good conduit for Eon in that way.
他当初是怎么拿到这份工作的？
How did he even get this job in the first place?
这工作听起来可不好找。
This seems like a really tough job to get.
是的。
Yeah.
他在九十年代的时候，曾在一家会计公司工作。
So he, you know, in the nineties, he worked in an accounting firm.
后来去了芯片公司高通。
He worked at chipmaker Qualcomm.
还短暂待过一家数字媒体公司。
He worked at this digital media company for a second.
之后在博通一干就是很多年。
Then he worked at Broadcom for a long time.
2008年，他去了家叫MindSpeed的芯片公司，正好赶上经济大衰退刚开始。
In 2008, he lands at this chipmaker called MindSpeed, which he started that job, like, right at the beginning of the great recession.
股价跌了差不多百分之七十五。
The stock went down, like, 75%.
他裁了不少人，压缩成本，还卖掉了不少资产。
He did a bunch of layoffs and cost cutting and sold off assets.
当时情况挺艰难的，但他确实帮这家公司从深坑里爬了出来。
He it was a pretty tough situation, but he kind of helped this company dig out of a pretty deep hole, essentially.
到了2011年，他接到一个猎头的电话，是关于一家叫SpaceX的公司，他压根没听说过。
Then in 2011, he gets a call from a recruiter for this company called SpaceX, which he's never heard of.
他一开始还以为是家做存储服务的，就是给你提供额外空间的那种。
And he thinks at first that it's a storage company, you know, offering you extra space.
猎头说，不是的。
And the recruiter says, no.
是一家航天公司。
It's a space company.
我们造火箭。
You know, we make rockets.
他说：‘我对这个一窍不通，但我看过《星球大战》。’
And he says, oh, I don't really know anything about that, but I've seen Star Wars.
这就是他后来常讲的那次初谈的经过。
It's the story that he's he's told about that first conversation.
但他还是接了这份工作，留了下来。
But he takes the job anyway and sticks around.
没过多久，猎鹰九号火箭的性能就明显提升了。
And pretty soon, you know, the Falcon nine rocket is working better.
星链项目启动了，SpaceX在接下来的十五年里迅速成长为一家巨头公司。
Starlink launches and and, you know, SpaceX becomes a pretty gigantic company, you know, in the in the subsequent fifteen years.
所以他一入职就当上了首席财务官。
So so he gets hired as CFO right away.
说真的，这可是他在SpaceX唯一做过的一份工作。
Mean, that that's that's the only job he's ever had at at SpaceX.
是的。
Yeah.
他是第一个这样的人。
He's the first one.
这是他唯一做过的工作。
It's the only job he's had.
跟其他那些大型科技股相比，SpaceX想跟它们并列，这事儿可真不寻常。
It's it's you know, compared to all the other gigantic big tech stocks that SpaceX wants to be traded alongside, it's pretty unusual.
那到底是什么让他这么受马斯克青睐？
And what is it about him that Elon Musk loves so much?
要赢得马斯克的信任，尤其是为史上最大规模的IPO，可不是件小事。
Earning Elon Musk's trust, let alone for the biggest IPO ever, is not a small task.
你有没有听过他接受过的采访？
You know, does he did you listen in any interviews with him?
你跟了解他的人聊过吗？
You talk to people that know him?
他到底有什么特别的地方，能让马斯克这么看重？
Like, what is it about him that is so special to Eon?
我的意思是，他就是能灵活应对各种突发状况。
I mean, I think he's just been willing to roll with the punches, basically.
明白了。
Okay.
而且把账目管得很紧，我想是这样。
And keep the books tight, I guess.
你刚才提到的那记重拳——xAI的合并，现在进展如何？
One of the punches you just talked about, the xAI merger, how is that going?
关于他怎么整合团队、挖掘协同效应，咱们知道多少？
What do we know about, you know, how he's managing the integration there and finding synergies, as they say?
是的。
Yeah.
xAI确实在大力削减开支。
I mean, xAI has definitely been, you know, cutting costs.
他们已经裁掉了一些人。
They they've done, you know, some layoffs.
他们已经裁掉了大量研究人员。
They've lost a lot of researchers.
我们之前已经详细讨论过这件事了。
You know, we we've covered that pretty extensively.
实际上，研究团队那边的情况并不理想。
Like, the actual research lab part is not going so great.
我同事格蕾丝·凯上周也讲过一个相关的例子。
My my colleague, Grace Kaye, had a good story about that last week as well.
但从约翰逊的角度来看，他一上来就拼命想尽快创造大量收入。
But, you know, from from Johnson's perspective, like, he's coming in and and trying to generate a ton of revenue as quickly as possible, basically.
是的。
Yeah.
这就导致了SpaceX几周前和Anthropic达成的那些巨额算力租赁协议。
And that has meant these gigantic compute rental deals that SpaceX announced with Anthropic a few weeks ago at or several weeks ago at this point.
还有上星期五SpaceX和谷歌宣布的另一笔交易，而Anthropic那笔合同每月高达十亿美元。
And then there was also this deal with Google that SpaceX announced on Friday, which the Anthropic deal is 1,250,000,000 a month.
谷歌的这笔交易是每月九亿二千万。
The Google deal is 920,000,000 a month.
这些数字显然大得惊人，这意味着SpaceX今年在AI收入上会报告出惊人的增长。
You know, these are obviously gigantic numbers, and it will mean that SpaceX gets to report huge growth in in AI revenue this year.
对。
Mhmm.
但这和OpenAI或Anthropic报告的AI收入完全是两种性质。
But it's a totally different kind of AI revenue than OpenAI or Anthropic will be reporting.
我的意思是，这并不是来自人们为Grok订阅支付高额费用。
I mean, this is not coming from people paying a lot of money for Grok subscriptions.
这笔收入来自xAI在建设数据中心上投入了巨额资金，现在正想办法怎么利用它们。
This is coming from xAI having poured so much money into building data centers and now trying to figure out what to do with them.
这其实就是说，
And that's just you know,
完全是两码事。
it's totally different.
你一直报道马斯克帝国的人事变动。
You've reported a lot on the turnover in the Musk empire.
你觉得布雷特·约翰逊能像现在这样一直干下去吗？
Do you think Brett Johnson lasts in this role for as long as he has been in the role already?
他啊，就像我说的，任职时间比任何一家‘七巨头’的CFO都长。
I mean, know, he's he's like I said, he's been in the role longer than any, you know, mag seven CFO out there.
在特斯拉换了四任CFO的这段时间里——包括一个离职后又回来的——他一直在SpaceX。
He's he's been at SpaceX during a period in which Tesla has had four different CFOs, including one that left and then came back.
说实话，他能干这么久已经很不寻常了。
I mean, there it's unusual that he's been in the job so long already.
如果他再干两三年，哪怕等到SpaceX上市之后，这依然算是非常了不起的了。
So if he lasts, like, two or three years, that'll still be you know, after SpaceX is public, that'll still be pretty pretty gigantic.
我采访的一些人提到，私企CFO和上市公司CFO完全是两码事。
And, you know, some of the people I talked to for the story just made this point that, you know, being a being a CFO of a private company is a very, very different drop than being a CFO of a public company.
一旦上市，你就得参加那些高度规范的财报电话会，因为是SpaceX，成百上千万人都在听。
And when you're public, you have to do all these, you know, tightly choreographed earnings calls where millions of people are gonna be listening because it's SpaceX.
你得参加那些会议，每一句话都会被别人细细揣摩。
You have to go to these conferences where your every word is watched in a different way.
这完全是两码事。
So it's a totally different job.
我觉得咱们得看看他能不能适应过来。
And, you know, I I I think we'll we'll see if he's able to adapt or not.
没错。
Right.
我想问问你关于上周末你发表的另一篇文章。
I want to ask you about another story that you published over the weekend.
你写了关于SpaceX的‘帮派’，也就是那些已经离开但仍然持有公司股份的前员工。
You wrote about the SpaceX mafia or the alumni of the company who are no longer there, but still very much hold shares in the space giant.
你能说说你重点写了哪些人吗？那些离开后自己创业的知名前员工。
Tell me a little bit about who you focused on here, notable alumni who have gone on to found their own companies.
是的。
Yeah.
说实话，现在SpaceX的这个圈子已经不小了。
I mean, the SpaceX mafia is quite big at this point.
就像你说的，很多前SpaceX员工创办的公司，估值都超过了十亿美元。
Like you said, there there's a ton of, you know, billion dollar plus companies founded by former SpaceX people.
我聊过一个挺有意思的人，叫莫顿，他在SpaceX负责火箭研发和测试的软件工作。
You know, one of the people I talked to who I thought was interesting was this guy, Morton, who worked on basically software for building and testing rockets at SpaceX.
现在他说，要把在SpaceX学到的经验用到其他领域，比如能源公司和别的航天企业开发软件。
And he now says, you know, we're gonna apply the lessons we learned there to building software for a whole bunch of other things like, you know, energy companies and and other aerospace companies.
他们最近刚融了12.5亿美元，公司还特别年轻。
So that they just raised that, I I think, 1,250,000,000 recently, and it's a it's a very young company.
我还联系了K2 Space的人，这家公司也有大量前SpaceX员工，正在做卫星。
I also talked to someone from those company k two space, which has a ton of SpaceX alums and is building satellites.
总的来说，投资人特别愿意投前SpaceX的人，因为他们已经证明了，在太空领域真的能赚到钱。
You know, there there's just this in general, there there's this gigantic appetite among investors to invest in former SpaceX people because they have, you know, just shown that you can do things in space that can actually potentially make a return.
现在这个领域简直有大把资金在涌进来。
There's just a ton of money being thrown at this area.
没错。
Right.
除了这些人都怀有宏伟的登月目标之外，你有没有发现他们之间还有其他共同点？
Other than the big moonshot ambitions that all these people seem to hold, Did you get any sense for any other through lines that connect them?
有没有什么共同的性格特质？
Are there any personality traits?
他们平时会经常聚在一起吗？
Do they all hang out together?
他们是不是都属于某种特定类型的创业者？
Is there a specific type of founder that they have taken?
会不会他们都想成为下一个马斯克？
Maybe they're all trying to be the next Elon.
也就是说，这是不是他们之间的纽带？
Like, is that is that what connects them?
我的意思是，大家普遍觉得，他们在SpaceX养成的敬业精神，是其他公司的人所不具备的。
I mean, there there's definitely this perception that they have a a work ethic from from being at SpaceX that that other people from other companies don't have.
而且我觉得，确实如此。
And, you know, I I I think that there yeah.
他们愿意熬夜，突然飞去德克萨斯，特别拼地去搞定某个具体环节。
There there's this willingness to, you know, pull all nighters, fly to Texas randomly, you know, go be super scrappy and figure out a particular part.
确实有一种强烈的专注劲儿，全身心扑在工程问题上，很多SpaceX的人这么说，投资人也买账。
Like, there there's definitely this, like, intensity and and kind of single-minded focus on engineering problems that, you know, a lot of SpaceX people say they have and and investors, you know, respond to.
我聊过一个人，Rebel软件公司的斯科特·莫顿，他说很多SpaceX员工之间有种纽带，源于一起参与火箭发射的经历。
You know, I the one person I talked to, you know, Scott Scott Morton from Rebel, the software firm, you know, he kind of described this this bond that a lot of SpaceX people have based on, you know, the experience of working together on a rocket launch.
你看着火箭升空，而每个人几个月来都在埋头打磨某个微小部件或某个细节流程。
Like, you you you watch something take off and everyone has spent months, you know, grinding away at a tiny component or a particular tiny process.
只要其中一个人负责的那一点点出问题，整个任务就全完了。
And if any tiny part, you know, that that one person has been focused on fails, the whole thing falls apart.
正是这些高强度的经历，让这些人之间形成了独特的联结。
And there's this, like, collective bond that all these people have based on those intense experiences.
很多人即使离开了，也一直保持着联系。
And a lot of them are you know, even when they leave, they they keep in touch.
他们会参加彼此的婚礼。
They go to each other's weddings.
他们都在一个群聊里。
They're all in group text.
他们会一起去喝一杯。
They go to happy hours.
我觉得，这简直就像一个非常紧密的圈子。
Like, it's a very tight knit community, I I I think.
不过，随着公司越来越大，我们还得看看这种纽带还能不能保持下去。
And and, you know, as it gets bigger, we'll see if they, you know, continue that that that bond or not.
但另外我问过不少前员工一个问题，就是上市之后，大家突然变得那么富有，会不会就不再拼命干了。
But, you know, another question that I had when I was talking to a lot of these alums is, you know, whether the IPO is just gonna make people so rich that they'll stop working.
我的意思是，很多这些人其实根本不想停下来。
I mean, a lot of these people even want
你觉得呢？
What do you think?
那你有得到答案吗？
Did did you get an answer there?
我觉得吧，确实有一小部分人可能会去买辆酷车、一艘帆船之类的，然后就歇着了。
You know, I I think there's the it seems like there's a minority that's gonna just, like, buy a cool car or a sailboat or whatever and, like, hang out.
但很多人特别坚持说，他们加入SpaceX不是为了钱。
But a lot of people were very insistent that they, you know, they didn't join SpaceX for the money.
几年前，谁也没想到公司估值会涨得这么离谱。
It was less clear, you know, years ago that it was gonna increase in valuation so gigantically.
很多人说，钱可能反而让他们能去搞个更疯狂的创业项目，不用到处拉投资，更能坚持自己的初心。
And a lot of them insists that, you know, maybe the money will enable them to, like, do a crazier start up or, you know, have to raise less money and get to be more true to their own vision for something.
但很多人也说，他们不会就此停手不干了。
But a lot of them say, you know, they're not gonna they're not gonna stop working.
他们不会停下，还是会继续去解决那些工程难题。
They're not gonna stop, you know, trying to solve engineering problems.
对了，你名单里提到的贾斯汀·洛佩兹，他是Base Power的COO，我们节目之前也请过他。
Well, and and I will say, I mean, you know, one one of the folks that you had on your list, Justin Lopez, he's the COO of Base Power, and he's someone who we've had on the show.
我的意思是，他们所追求的目标可一点都不小。
I mean, look, these ambitions are not small that they have embarked on themselves.
所以他们早就准备好要走一条漫长的路了。
So they very much have set themselves up for long journeys ahead.
这真是个很棒的故事，我建议大家都去了解一下。
But it was a great story, and I encourage everyone to check it out.
谢谢你来参加节目，Theo。
I want to thank you for coming on, Theo.
我猜这周至少每天都会再找你聊上一聊。
I imagine I will be talking to you, if not every day this week, nearly that much.
这位是Theo Wayt，我们《The Information》的马斯克记者。
That is Theo Wayt, our Elon Musk reporter, here at The Information.
在人工智能领域，你可能会越来越多听到一个关键术语，叫‘递归式自我改进’。
One of the most important terms in AI that you might start hearing more about is called recursive self improvement.
今天《AI简报》通讯里，我的同事Rocket Drew就专门写过这个话题。
My colleague Rocket Drew wrote about that in today's AI Agenda newsletter.
我想请他来给我们讲讲，这到底是什么，为什么现在这么受关注。
I want to bring him on to tell us what it is and why it's getting so much attention right now.
火箭，欢迎再次做客我们的节目。
Rocket, welcome back to show.
很高兴你今天能来。
It's great to have you here.
嗨，阿卡什。
Hey, Akash.
谢谢邀请我。
Thanks for having me.
好的。
Okay.
递归式自我改进。
Recursive self improvement.
这个术语，我记得你们一年多前发布的AI术语表里还没有吧。
This was not on the glossary of AI terms that you published, I think, a little over a year ago now.
我觉得这个条目可能需要更新一下。
I think that one probably needs to be updated.
这个术语是什么意思，Rocket？
What does this term mean, Rocket?
是的。
Yeah.
没错。
That's right.
如果我们现在出个2026年的版本，它肯定排在第一位。
I think if we did one for 2026, it would be at the top of the list.
名字是有点长，但最近我们到处都能听到。
Mean, it's a mouthful, but we're hearing it all the time these days.
简单说，就是AI模型已经强大到能自己创造出下一代版本。
It basically means AI models that are capable enough that they can create their own successors.
比如说，Claude Opus 4.8就能生成Claude Opus 4.9。
So say, Claude Opus four point eight could create Claude Opus four point nine.
到那时，人类在AI模型研发的整个过程中就完全被排除在外了。
And at that point, humans would be out of the loop entirely on the research and development process that goes into creating AI models.
现在我们经常听到这种说法。
We're hearing it a lot.
Anthropic刚发布了一篇长文，提到他们观察到的一些迹象，表明公司可能正接近或即将达到递归自我改进的阶段。
Anthropic just released a big blog post talking about some of the evidence they're seeing that Anthropic is approaching or nearing this point of recursive self improvement or RSI.
现在有很多初创公司正在融资，声称自己正在开发能训练其他AI模型的系统，甚至开发能设计用于制造AI芯片的AI模型。
There are a lot of startups that are raising funding saying they're working on AI models to train other AI models or even AI models to design the chips that will then be used to create other AI models.
当然，AI在写代码方面进步很快，但目前还有一些研究环节是AI还做不到的，研究人员在这方面依然更有优势。
Of course, AI is making a lot of progress at writing code, but so far, there are parts of the research process that have eluded the capabilities of AI models where, for now, researchers are still better.
人类研究者比AI模型更强。
Human researchers are better than the models.
这其实和你刚才聊的SpaceX话题有个有趣的关联。
This is actually a fun connection with the SpaceX topic you were just on.
就像Theo说的，人们花大量时间盯着火箭的每一个小零件，生怕哪里出问题。当年挑战者号航天飞机爆炸，就是因为O型密封圈失效了。
To Theo's point about people, you know, working on these rockets, spending a lot of time obsessing over little components that could break, the reason the Challenger space shuttle famously exploded is because, the O rings failed.
火箭上这个看似简单的部件，在低温下就出问题了。
This, like, very simple component on the rockets, they failed in the cold.
结果整个火箭直接炸了。
The whole rocket blew up.
这就引出了‘O型环自动化’这个经济学术语，指的是整个生产流程被最薄弱的一环卡住了。
And this is where we get the term O ring automation, which is an econ term for a process, a production process that bottlenecked by the weakest link effectively.
所以人们在讨论，在递归自我改进的过程中会不会出现类似情况：AI或许能写出很棒的代码，但研究过程的关键瓶颈，可能就在于人类得提出最好的研究想法，或者决定该探索哪些方向。
So people talk about whether we'll see O ring automation here in recursive self improvement where, sure, maybe the AIs can write lots of great code, but maybe the research process is bottlenecked on this one step, which is humans coming up with the best research ideas or deciding which research areas to explore.
但这个观点的核心是
But the idea behind this
让人人都得亲力亲为。
how people should do it all themselves.
好吧，让我插一句。
Okay, so let me jump in here.
我想先总结一下几点。
So I want to recap a couple things.
第一，我刚想到一件事。
Number one, it just hit me.
你叫Rocket，但你并不是专职的火箭记者吗？
Your name is Rocket, and you're not a Rocket reporter full time?
算了，别管这个了。
Mean, did No, you ever think what about Okay, the well put that aside.
我很高兴听到你对火箭的了解几乎面面俱到。
I mean, I'm glad to hear that you seem to know just about everything you need to know about rockets.
所以如果你有兴趣拓展领域，正好有个专属话题等着你。
So I think if you ever want to expand, there is a beat waiting for you.
但我们还是回到递归自改进这个话题吧。
But let's go back to recursive self improvement.
核心问题是，我们能不能让模型达到足够好的水平，从而自己训练下一代模型？
So the idea is that can we get the models to a point that they are good enough to then train the next generation of model?
或者说，自己设计出下一代芯片。
Or basically, design the next generation of a new chip on their own.
这就是所谓的递归自我提升临界点。
That's what this recursive self improvement threshold is.
对吧？
Is that right?
没错，正是这样。
That's exactly right.
一旦达到这个临界点，人工智能的整体发展速度可能会急剧加速。
And there's a sense that once we hit that threshold, the overall pace of AI progress could explode.
我想我们都清楚，人工智能现在已经发展得非常快了。
Mean, I think all of us are familiar that AI has been going very fast already.
但你可以想象，如果把所有动作缓慢、容易出错的人类，换成运行迅速、错误更少、又能协同并行完成大量任务的AI模型，那么AI的进步速度可能会比现在更快。
But imagine you take all of these slow humans that make mistakes and you replace them with AI models that run very quickly, make fewer mistakes, and can accomplish a lot of tasks in a way that's coordinated but in parallel, you could imagine that the pace of AI progress goes even faster than it's been.
但我们现在已经达到那个临界点了吗？
But have we hit that threshold yet?
还没有。
No.
答案肯定是否定的。
The answer is definitely no.
它们写了大量代码，但还没能写完所有代码。
They're writing a lot of code, but they're not writing all the code.
而代码正是它们取得最大进展的地方。
And the code is the place where they're making the most progress.
整个研究过程包括提出假设、设计实验来验证这些假设、然后实施实验——到目前为止，模型在这部分进步最大。
So the overall research process involves coming up with hypotheses, designing the experiments to test those hypotheses, implementing the experiments that's where the models have made the most progress so far.
但接下来还要解读实验结果，用这些结果来更新假设，然后整个循环继续下去。
But then interpreting those results and using that to update your hypotheses and then, you know, this whole loop.
写作这一块。
The writing
所以，这就是警报。
So, this is the alarm.
Anthropic 正在发出警告，说我们正逐渐接近一个临界点——模型可能开始设计下一代模型了。
So, Anthropic is sort of sounding the alarm in saying that, hey, we are getting to that threshold where the models could design the next generation of models.
我们还没到那一步，但一旦到了，可能会带来一些影响。
We're not there yet, but once we get there, there might be some implications here.
没错。
That's right.
他们觉得这已经近在眼前了。
They think it's in view.
他们觉得这就在前方了。
They think it's on the horizon.
当然，有些人可能不同意，但不管是一年、两年，还是十年后，大家都能感觉到，这个想法确实有点让人担忧。
Now, think some people would disagree with that, but, you know, depending on if it's a matter of, you know, one year away, two years away, or even ten years away, I think people have a sense that there is something kind of alarming about this idea.
这事儿听起来有点像《终结者》。
There's something kind of terminator y about it.
等等，意思是人类以后完全不参与下一代模型的开发了？
It's like wait, so humans aren't going to be involved at all in the process of creating the next model?
我们只是让现在的模型去训练下一代，然后下一代再训练下下一代。
We're just going to trust our current generation of models to train their successors and then those will train their successors.
人类到底在什么时候介入并引导这个过程呢？
At what point are humans getting involved and steering this process?
这也是 Anthropic 对此感到担忧的原因。
So that's the reason Anthropic is concerned about this possibility.
他们指出，最终可能会出现一个拥有自身目标和欲望的模型，然后下一代模型继承这些相同的目标或欲望，甚至是更奇怪的目标，导致整个过程脱离我们的控制。
They say you could end up with one model that has its own goals, its own desires and then maybe the next generation of model inherits those same goals or desires, or ones that are even weirder and the whole process could kind of run away from us.
因此，他们在发布的那篇文章中提倡的一点是，希望至少让 AI 实验室有能力对 AI 能力的进展进行某种协调性的放缓或暂停。
So, as a result, one of the things that they're advocating for in this post that they put out is that they would like the ability, at least the ability, for AI labs to do some kind of coordinated slowdown or pause in the progress of AI capabilities.
他们并不是说我们必须踩刹车。
They're not saying we have to pump the brakes.
他们只是说，你知道的，刹车装置应该得装上去。
They're just saying, you know, the brakes should be in there.
所以这是一个
So that was one
原因。
of reasons.
你觉得Anthropic提倡这个，但其他实验室真的支持这个所谓的暂停吗？
How likely do you think that I mean, Anthropic is advocating for that, but are other labs on board with this so called pause?
是的。
Yeah.
对。
Yeah.
我要说明一点，暂停这个想法并不是Anthropic突然想出来的。
I want to be clear that, like, the idea of a pause is not something that Anthropic just came up with.
很多人早就讨论过这个问题了，也一直敦促Anthropic采取这样的立场，而他们现在终于这么做了。
People have been talking about it for many years, and people have been pushing Anthropic specifically to adopt this position, which they're now taking.
他们的意思是，如果其他人都愿意暂停，那我们或许也会考虑跟着暂停。
Which is that, well, if everyone else is willing to pause, maybe we would be interested in pausing.
实际上，今年早些时候在达沃斯，谷歌DeepMind的首席执行官德米斯·哈萨比斯也说过类似的话。
Actually, Demis Hassabis, the CEO of Google DeepMind, said something similar earlier this year at Davos.
有人在采访中问他，如果别人都暂停了，你们会不会也考虑在AI竞赛中来个协调性的缓和？
He was asked in an interview, if everyone else paused, you know, could we do some sort of coordinated detente in the AI race?
他还表示自己对此持开放态度。
And he said that he was open to it as well.
至于OpenAI，他们也在致力于实现递归式自我改进，也就是打造自动化的AI研究员。
OpenAI for their part is also working towards recursive self improvement, having an automated AI researcher.
但他们的许多研究人员，甚至一些管理层，至少对暂停的提议感兴趣。
But plenty of their researchers and even some of their leadership are at least interested in the idea of the pause.
不过，就像我之前说的，大家对这种暂停究竟何时才真正有必要，还存在分歧。
Though, like I said, there's disagreement about when such a thing would actually be needed.
而且，这看起来似乎涉及一些博弈论的问题。
Well, and it seems like there's a bit of game theory here.
必须有人真正停下来，这一切才可能奏效。
Somebody has to actually pause for any of this to really work out.
我的意思是，他们最好坐到一起，共同决定：我们就这么干吧。
And I mean, it sounds like they just all need to get in a room together and collectively decide we're actually going to do it.
对。
Right.
对。
Right.
但他们
But they
都得达成一致。
all have to decide.
所以现在大家都在努力确保能监测到是否有谁违背承诺，并验证其他人是否真的如约行事——这在只有Anthropic、OpenAI和DeepMind这几个主要玩家的时候还相对容易做到。
And so there's work going into making sure that they could detect if any of the other ones defected and verify that the other ones are actually following along like they said they would, which is easier to do when it's just, you know, Anthropic OpenAI and DeepMind.
但如果牵扯到政府，还要协调中国和中国实验室一起暂停，情况就复杂多了，而任何真正全面的暂停都必须包括这些。
It gets more complicated if you have government in the picture and they're also trying to coordinate a pause with China and with the labs in China, which any, you truly comprehensive pause would have to do, of course.
那我问你一个问题。
So, let me ask you this.
在讨论中，‘通用人工智能’这个说法好像渐渐淡出了视野，尤其是微软和OpenAI已经不再把AGI作为界定版税何时停止或继续的依据了。
The term AGI kind of drifted out of focus in the conversation, certainly as Microsoft and OpenAI have basically removed that from the You know, the definition around AGI has been removed from the negotiating table in terms of when the royalty payments stop and or keep going, stuff like that.
所以，这现在就是新的‘通用人工智能’吗？
Is this just the new AGI?
对我来说，这听起来就像个词，我们根本说不准什么时候能真正达到。
Like, that's what it sounds to me, is it's just a term that we don't really know when we'll reach it.
我们只是在朝着这个方向努力。
We're kind of working towards it.
很多人都对它感到担忧。
It's something people are kind of fearful of.
但这只是个可以随意解读的概念，还是真有个明确的标准，比如我们确实实现了自我递归改进？
But is it just open to interpretation, or is there actually a measurable thing that, No, we have hit recursive self improvement.
现在这已经很危险了。
This is dangerous now.
是的，这是个很好的问题。
Yeah, that's a great question.
让我为‘自我递归改进’这个说法说几句，其实长期以来，人们说AGI的时候，心里想的本来就是这个意思。
Okay, let me say one thing to defend, I guess, the merits of this term recursive self improvement, and that's that when people said AGI, for a long time what they really meant was recursive self improvement all along.
所以某种程度上，这其实是对AGI概念的细化。
So in some sense, this is a refinement of AGI.
我来解释一下为什么是这样。
Let me explain why that is.
对吧？
Right?
AGI 的概念是指一个能完成人类所有认知任务的模型。
The idea of AGI was a model that's capable of doing all of the cognitive tasks that a human can do.
这自然也包括了所有创建 AI 模型所需的任务。
That necessarily includes all of the tasks that go into creating an AI model.
对吧？
Right?
这就涵盖了我之前提到的整个 AI 研究流程中的每一个步骤。
That includes all of the steps of that AI research process that I outlined.
所以 AGI 本身就隐含了递归自我提升，这也是为什么 AGI 如此有分量，吸引这么多人关注的原因。
So AGI implies recursive self improvement, which is part of the reason why AGI had so much oomph or people were paying so much attention to it.
人们觉得，一旦达到那个临界点，可能就会触发这个递归自我提升的循环。
People thought that if we reach that threshold, that might be what kicks off this recursive self improvement loop.
现在人们意识到，智能其实是参差不齐的。
Now people realize that intelligence is jagged.
目前AI在很多方面还做得不好，也许短期内都很难做得好。
And there's all sorts of things that AI can't do very well now and maybe won't be able to do well anytime soon.
但递归自我提升这个关键节点，我们可能更快就能达到。
But the recursive self improvement milestone could be something we hit sooner.
所以我觉得，某种程度上，这其实是一种细化。
So I think that's why people in some sense, you can think of it as a refinement.
另一种说法是，AGI这个概念已经完成它的使命了。
Another way to defend this is that, you know, AGI kind of served its purpose.
几年前，当我们觉得AGI还很遥远的时候，这个词很有用，那时候我们只是需要一个大致的词来形容那些能力很强的AI系统。
It was a useful term when we were like a few years out, like a few years ago, when it still seemed more distant and we just sort of like roughly needed to gesture at like very capable AI systems.
现在AI在某些领域已经非常强大，甚至能媲美或超越人类表现，也许这个词已经完成了它的任务，不再适用了，所以我们需要把它细化成更具体的表述。
And now that, AI systems are much more capable on some domains, rival human performance or exceed human performance at some domains, maybe that term has served its purpose and is no longer useful, so we need to refine it to something more specific.
这就是我对这个问题的看法。
Now that's my defense of it.
我觉得你说得对，这个词确实很模糊，等我们越来越接近那个临界点时，会发现还得进一步细化它。
I think to your point, it is absolutely true that this term is really fuzzy, and it's we're going to realize again as we get closer to that threshold that we're going to need to refine it further.
结果可能会是这样，好吧。
It's gonna turn out that, okay.
当然。
Sure.
Opus 4.8 制造了 Opus 4.9。
You know, Opus four point eight created Opus four point nine.
但制造这些模型所用的芯片，背后有大量人力参与，而且下一代模型的芯片、运行这些系统的数据中心，也依然离不开人。
But, like, there were a lot of humans involved in creating the chips that went into that model, and humans are still involved in creating the chips that are, you know, gonna run for the next generation model and maintaining the data centers where these things that, When you take a more holistic view, there are still going to be humans involved in Which is what you
你提到的，正是当前所有这些进展中最关键的人力瓶颈——芯片。
mentioned is one of the key human bottlenecks to all this, is the chips.
芯片的研发高度依赖人类，这一点毋庸置疑。
The chips is very it's very much something that you need humans to develop.
所以，Rocket，我最后一个问题是：我们离自我递归提升还有多远？
So the last question I have for you, Rocket, then is how close are we then to recursive self improvement?
Anthropic 发布了一篇博客文章。
Anthropic has put out this blog post.
我们有没有什么可衡量的标准，能说离那个目标还剩一年、三年，或者两个月？
Do we have any kind of measurable criteria saying we are a year away, three years away, two months away?
这是个好问题。
It's a great question.
我觉得大家的预测差距太大了，我不想明确站队。
I feel like estimates differ so wildly that, like, I don't want to stake a particular position.
我的保守说法是，进展显然在加快。
My cop out answer is that it's clearly going faster.
对吧？
Right?
就算我们还没达到那个明确的节点，趋势也很清楚：模型在研究过程的每个阶段都能做越来越多的事，能处理的任务也越来越复杂、周期也越来越长。
Like, even if we don't hit that discrete point, the trend line is clear that models are capable of doing more and more at all stages of that research process and the complexity and the time horizon of the task that they can accomplish is getting longer and longer.
我觉得这一点已经很明显了，目前还没看到任何放缓的迹象，虽然未来可能会变。
So I think we see that pretty clearly, and there's no sign yet that that's going to stop, though it could in the future.
展开剩余字幕（还有 314 条）
很好。
Great.
Rocket，非常感谢你今天来做客。
Well, Rocket, I want to thank you for coming on.
一如既往，这位是Rocket Drew，我们《The Information》的AI与机器人记者。
As always, that is Rocket Drew, our AI and robotics reporter here at The Information.
我们的下一位嘉宾曾经把一家公司卖给了亚马逊云服务。
Our next guest has previously sold a company to Amazon Web Services.
现在他正在创办另一家初创公司，叫Eon，做云备份的。
He is now working on another startup, Eon, in the cloud backup space.
这家公司上次估值是40亿美元，我来邀请Ofir Ehrlich上场。
The company was last valued at $4,000,000,000 I want to bring on Ofir Ehrlich.
Ofir，欢迎来到节目。
Ofir, welcome to the show.
很高兴你能来。
It's great to have you here.
非常感谢。
Thank you very much.
很高兴能来。
It's great to be here.
我们刚才在聊自我迭代改进，我相信这现在也是你特别着迷的话题。
We were just talking about recursive self improvement, which I'm sure is a topic that you also are geeking out about nowadays.
好的。
Okay.
你对这个怎么看？
What are your thoughts on it?
我们从这里开始吧。
Let's start
。
there.
它
It
说实话，就在ChatGPT爆火前一天，要是有人跟我说将来会有软件能写诗，我肯定不信。
feels like, you know, for the last, a day before the Che GpT moment, would have told me that, you know, we're going to have a software that can write poems.
我当时觉得Siri已经是最好的了。
I thought that Siri is the best of breed.
结果一夜之间，事情就变了，可前一天你根本看不出任何苗头。
All of a sudden, sprottingly happened that the day before that, you couldn't see that.
我觉得这正是技术呈指数级增长时的常态。
And I think that's what's happening always with exponential growth in technology.
等你意识到它来了的时候，往往已经晚了。
You don't see it happening until it's already too late.
记得刚开始的时候，大概两年前吧，我也不确定这东西到底会走向哪里。
And you know, when we started, when this started to be a thing, maybe two years ago, I wasn't sure where is it going?
那会儿我跟不少人讨论过这个问题。
Had numerous discussions with people about that.
有人说，我们已经碰到天花板了。
People saying, Well, we've reached the asymptote.
它永远不会更聪明。
It will never be smarter.
它没法自己进化。
It cannot evolve itself.
归根结底，还是得靠人，人依然不可或缺。
At the end of the day, humans will be better, will be needed.
回到你之前说的，也许人类永远都需要，但可能只是用来造芯片，或者像《黑客帝国》里那样，给机器生产电池。
And going back to your previous discussion, maybe humans will always be needed, but maybe to build the chips and maybe like in the movie Matrix, to build the batteries so that we can run the machine.
所以我也说不准，但感觉每天比前一天都更离谱。
So I don't know, but it feels every day becomes crazier than the previous one.
没错。
Right.
其实上周我们刚聊过，评估模型效果的时候，模型有时候知道自己在被测试，所以得考虑这一点。
We were just talking last week actually about how the models when you're evaluating the effectiveness of models, sometimes the models know that they're being evaluated, and so you have to account for that.
这又给那个矩阵式的场景增添了一层复杂性。
Mean, this would be another level to that matrix sort of scenario.
这太惊人了。
It is amazing.
我觉得，想要阻止进展——这么说吧——暂停的难点在于，竞争太激烈了，能获得的回报也太多，感觉就像一场竞赛。
I think that the main issue, the problem with trying to withhold progress, let's call it this way, the problem with pausing is that there's so much competition, so much to gain, and it feels like a competition.
大家表面上在合作，但一旦觉得自己有了明显优势，就会立刻转向全力冲刺。
It's a cooperation until you have something to gain, you think you have a significant edge, and then you pivot into making that happen.
你看到模型在自我进化。
So you're seeing the models evolve themselves.
你看到人类有动力推动这些模型自我进化，最终让自己变得可有可无。
You're seeing humans have the incentive to make these models evolve themselves and to make us eventually redundant.
我也不知道，最后会不会真变成《终结者》那种情况，
There's a question of, I don't know if it's a Terminator scenario at the end of the day,
还是说，
or it's
机器人其实该改名叫‘天网’，或者我们正走向一个乌托邦：什么都不用干，全由机器人包办一切。
where, you know, the robots maybe Anthropic or OpenEye should change Net to Skynet, or it's a utopian world what we're going through when we don't have to work because robots will do everything for us.
但你觉得暂停真的可能发生吗？
But do you think a pause is likely?
我不觉得有人会真的同意暂停。
I don't believe a pause is I don't believe the people
因为这事儿得有个危机才能推动。
Because this feels like there would have to be a crisis.
我是说，我想象的是金融危机会成为导火索。
I mean, you know, I'm imagining financial crises.
银行以前之所以能坐到一起，不就是因为出了大事吗？
I mean, the only reason the banks really ever got together was when there was a crisis.
我觉得，只有当真出了急事，德米斯、达里奥、萨姆，还有米拉和马拉迪这些人在思维机器公司的人，才会被迫坐到一块儿。
I mean, I'm imagining that is what we would you would need some acute event to have Demis and Dario and Sam and even Mira and Maradi at Thinking Machine.
他们只有在非得联手的时候，才会真的聚在一起。
They all have to get into a room together, but it's only going to happen when they need to.
我不确定从每个人的角度看，他们会不会在同一时间达成一致。
I'm not sure that from each of their perspective, I'm not sure that it will happen on the same time.
我不确定会不会真有那么一个时刻，所有相关方都一致认为出现了危机。
So I'm not sure that there will actually be a moment in time where all the relevant stakeholders think the same that there is a crisis.
回顾过去的金融危机，通常都会看到像债务危机这样的全球性事件。
Looking back at financial crises, usually you would have seen something like a debt crisis, something that's global.
从每个人的角度来看，我很难预见到类似过去金融危机那样的客观危机。
I think it's very hard for me to anticipate an objective crisis from each of those people's perspective, unlike financial crisis in the past.
我以前觉得市场迟早会调整。
I used to think that there's going to be a correction in the market.
我到现在还是觉得，市场终归会调整，毕竟市场有周期嘛，诸如此类。
I still think at some point there'll be some correction in the market because, you know, market cyclist and so on and so forth.
我们曾经想过，可能是多种因素叠加造成的。
And we thought maybe it will be a combination.
应该是债务问题加上其他一些因素共同作用的结果。
It should be a combination of debt and something.
比如数据中心过度建设，同时又过度杠杆，导致产能严重过剩。
So, data centers, overbuilding data centers at the same time with leveraging that too much, not enough to use capacity.
但现在一切都爆发了，竞争这么激烈，需求这么大，我真的说不准最先会出什么事。
But now with everything exploding now, so much competition, so much need, I just don't know what's going to happen first.
好吧，那先别管什么临时补救措施了？
Okay, well, what just a hack?
我的意思是，先把资本市场放一边。
I mean, put the capital markets aside, okay?
金融影响是一回事，但递归式自我提升呢？我们说的是那种知道自己正在被评估的AI。
The financial implications are one thing, but recursive self improvement, I mean, we're talking about AI that knows when it's being assessed.
我觉得最可能引发突发危机的，反而是某种技术漏洞，这种事说不定真能把所有人都拉到同一个房间里讨论。
A hack seems like the most likely acute crisis that could then get everyone in a room together, no?
我不觉得一旦被黑客攻击击垮，就会给别人创造机会。
I don't believe that once demised by a hack would be someone else's opportunity.
别只想着公司，还要想想国家层面。
Think not just companies, but also nations.
想想中国和美国之间的博弈，不只是企业之间的竞争。
And think China versus The United States, not just companies.
再想想开源社区和那些大型公司之间的对比。
And think the open source world versus all those significant companies.
所以，一个关键问题是，包括开源贡献者在内的所有利益相关方，会不会真的把这当成一个问题来对待。
So it's a really good question whether all the stakeholders, including open source contributors, will actually go and look at this as a problem.
你已经能看到，我们之前关注的是备份，但就在最近几个月——之前可没有——越来越多和我交谈的人告诉我，他们公司或者他们自己开发的软件，不小心删掉了生产环境里的数据，是个会自动删除数据的程序。
And you're already seeing, we're focused on backup and we started seeing in the last few months, just last few months, not before, a significant amount of people I'm speaking with tell me either my company or personally I created a software that deleted data, an agent that deleted data from production that they did not intend to.
这其实也相当于一种黑客行为。
So that could be also the equivalent of hacking.
你甚至可以说，根本不需要黑客攻击。
You could say that you don't need a hack.
只要删掉某个至关重要的东西，所有人都会惊呼：天哪，这也太可怕了。
You could just delete something truly paramount that everyone will say, Oh my God, it's horrible.
但我看不到全世界会因此统一认定某件事是必须停下来的理由，毕竟继续前进的动机太多了。
But I don't see all the world decides that one thing is the reason for everyone to stop, and there's so much incentive to go on.
真的吗？
Is that true?
假设你就在AWS工作，公司被AWS收购了，你还是他们灾难恢复团队的顶级负责人，怎么样？
So even if, let's say, you were at AWS, your company was bought by AWS, you were one of the top leaders there for disaster recovery, okay?
我明白你讲的国家间对抗这部分，但我们就假设一下，像AWS这样的超大规模云服务商，以前也不是没出过事，对吧？
And I hear you on the nation versus nation part of this, but let's just say that a hyperscaler, AWS it's happened before, right?
我的意思是，AWS过去确实出过问题。
I mean, AWS has had issues in the past.
但这些问题我们都能解决，他们也都能解决。
We've been able to fix They've been able to fix it.
所有大厂都是这样。
All of them.
我的意思是，别光说AWS，还有GCP、Azure，也都一样。
But I mean, look, let's just say that it's not just AWS, it's GCP, it's Azure.
我也不确定。
I don't know.
也许他们真解决不了。
You know, maybe they can't fix it.
可能一拖就是一星期，你知道的。
Maybe it goes on for a week, you know.
那会不会导致某种暂停呢？
Mean, could that then trigger a pause of some kind?
我真不这么认为。
I honestly don't believe it.
回顾过去，云计算相比人工智能，曾经是一场小革命。
It's always looking back at cloud is a, used to be a mini revolution compared to what happens in AI.
刚开始搞云计算的时候，大家可能都忘了，那时候标准特别多，只要有个标准出来，大家就赶紧跟上。
And in the beginning of cloud, people probably don't remember, were lots of standards, And every time someone could go to the standard, would do that.
所有人都想跟着AWS走，因为他们是领头羊。
And everyone wanted to follow AWS because they were the leaders.
我们做过灾难恢复和迁移，而且就是在AWS上做的。
And we did disaster recovery and migration, and we've done that in AWS.
在那之前，我们还和GCP做过OEM合作。
And before that we added OEM with GCP.
所以我们比任何人都更早看到企业从本地部署转向云端的过程，亲眼见证了他们在这段旅程中的真实状态。
So we think we saw more than anyone else alive, or companies moving from on premise to the cloud and seeing how their journey looks like in their eyes.
但当时发生的那场变革，跟现在AI领域正在发生的一切相比，根本算不上什么。
Now, the revolution that happened there is nothing, nothing compared to what's happening in AI right now.
但当你去观察——
But when you look for a-
规模要小得多。
It's smaller.
你说它规模更小。
You're saying it's smaller.
不，现在要大得多。
No, today it's way bigger.
没错，没错。
Right, right.
云计算
The cloud
那时候比现在小多了。
was smaller than what we're seeing now.
明白了。
Okay.
是的。
Yes.
云服务那时候虽然小，但也已经规模惊人了。
Cloud was smaller and it was still giant.
你能看到那些正在做迁移的人。
And you could see people that were doing the transition.
他们真的非常害怕迈出这一步。
They were really, really afraid of doing that.
他们明白自己必须这么做，别无选择。
They understood that they have to do it because they have no choice.
现在呢，员工和公司转向人工智能，不只是因为能解决很多问题。
Today, you see employees and companies, they don't just move to the cloud, to AI because it solves a lot of problems.
确实如此。
It's true.
人工智能太厉害了。
AI is amazing.
这是一项很棒的技术。
It's great technology.
但你跟他们聊的时候，他们会说，他们想转型是因为这很时髦。
But when you're speaking to them, they're telling you they either want to move because it's cool.
你不玩AI，就显得落伍了，迟早会被淘汰。
You're not a cool kid if you're not using AI, or you will become irrelevant.
一年前、半年前，我们还看到很多人坚决不用AI。
We saw a lot of people insisting on not using AI a year ago, and a half a year ago.
你知道后来怎么样了吗？
And you know what happened to them?
他们被辞退了。
They got fired.
看看那些公司，为了给AI预算腾空间，一下子裁掉百分之几十的员工。
See all of those companies firing tens of percent of their people to make room for the AI budget.
同样地，别见太多人了。
Also the same, don't meet so many people.
在你走之前，我再问你一个问题。
Let me ask you one more question before you go.
你之前提过这一点，但现在你做的生意是帮人用上AI。
You touched on this earlier, but you know, you are in the business now of helping people adopt AI.
以前你做的是帮人上云。
You were in the business of helping them adopt the cloud.
有些地方还在从本地部署转向云端，这事还没完。
The on prem to cloud transition is still happening in some cases.
当然。
Of course.
但上云那会儿，得建一大堆基础设施，现在AI的情况也差不多。
But it did require a lot of infrastructure build out, and that is similar to what we're seeing right now with AI.
你觉得现在这种基础设施建设，是高估了还是低估了AI的实际需求？
Do you think that the infrastructure build right now is overestimating or underestimating how much demand there will be for AI?
我觉得问题不在于建了多少算力，而在于太分散了。
So I think that the problem is not the amount of capacity being built, but the fragmentation.
你会看到一些所谓的云平台、新云平台或者数据中心，有的算力根本不够用，而另一些则空置率很高，GPU堆了一大堆，却没人用。
So you're going to see some, let's call it clouds or neo clouds or data centers that don't have enough capacity in a lot of vacant data centers that with lots of GPUs leveraged according to those GPUs and not enough usage.
这种碎片化的情况非常棘手。
The fragmentation is very hard.
到目前为止，你看到的都是集中化趋势。
Until now you've seen consolidation.
云计算的意思就是，别搞那么多数据中心了，直接找亚马逊、微软、谷歌、甲骨文、IBM这些大公司，他们能以更大规模做得更好。
Cloud meant that instead of using many, many, many data centers, you're just going to say, Hey, let's go to this company, Amazon, Microsoft, Google, Oracle, IBM and others that can do that better than me at scale.
但现在你看到一个很有趣的现象。
Now you're seeing a very interesting phenomenon.
趋势居然在倒退。
You see it going backwards.
你现在看到一些小公司说，我们要自己建数据中心。
You're seeing smaller companies saying, We'll build our own data centers.
我们觉得自己能做得更好，但我觉得现在这种想法并不合理。
We'll somehow know how to do it better, which I don't think right now makes sense.
他们不仅要建自己的数据中心——这是大家都在做的事，还得去买GPU，而GPU现在非常紧缺。
And they need to do not only build their own data centers, which is what everyone is doing, they need to buy GPUs, which is scarce.
作为一家小公司，做这些事比大公司难多了。
It's harder for you to do it when you're small company as opposed to a big company.
租场地、找电力，已经成了大问题，还得搞市场推广、做销售，让人来试用你的基础设施。
No, rent space, find electricity, became a really big problem, and also do marketing, also do sales, get people to try and use your infrastructure.
这太难了。
It's very hard.
我觉得，如果市场要调整，也不是说整个市场都会下滑。
And I think that if you're going to see a correction in the market, it's not that all the market will go down, in my opinion.
而是很多这样的小型数据中心根本没多少需求，最终可能会陷入债务危机。
It's that a lot of these smaller data centers simply wouldn't have demand and they'll have a debt crisis in some sense.
这就会成为问题。
And this will be the problem.
我认为解决办法将是更多的整合。
And I think that the resolution will be more consolidation.
我觉得几年后，等这种碎片化局面出现，市场会重新整合，回到少数几家主要玩家手中。
I think that you're gonna, in a few years, after you're gonna see this fragmentation, you're gonna see defragmentation going back to a few key players.
当然不会是现在这些公司，但最终会剩下几家关键企业。
Not certainly it's gonna be the same place that we have today, but a few How key
我们离那一天还有多久？
long away are we from that?
是2026年还是2027年？
Is that 2026, 2027?
你什么时候会开始看到问题显现？
When do you start to see the cracks here?
好吧，我来跟你说说问题在哪。
Okay, so I tell you the problem.
2025年的时候，人人都说2026年会是关键一年。
In 2025, everyone was saying 2026 is going to be the year.
现在大家都说，哦，2026年真不错。
Now everyone's saying, Oh, 2026 is great.
那些IPO啊，说不定还有太空里的数据中心之类的。
All those IPOs, maybe there are going to be data centers in space, etcetera.
所以每当所有人都觉得是泡沫的时候，反而就不是泡沫了——2025年就是这么回事。
So whenever everyone thinks it's a bubble, so it's not a bubble, so that's what happened in 2025.
但现在大家都觉得稳了。
But now everyone thinks they're safe.
没人觉得这是泡沫。
Nobody thinks it's a bubble.
所以现在，也许我们现在就像2021年11月那样，只不过这次有真正的技术支撑，不是光靠空投资金，但要说准了，真不容易。
So maybe now, maybe now, we're at Maybe this now we're November 2021, but you know, with real technology behind it, not just free money, but very, very hard to say.
我估计未来三年内会迎来一次调整，但我认为之后会立刻反弹并重新整合。
Would estimate what in the next three years we're gonna see a correction, and I think that it's immediately gonna jump back up and consolidate.
就是这样
That's how
除非人工智能的发展突然停顿下来。
I'm Unless there's somehow a pause in AI development.
如果把所有人聚在一起，大家都会说：我懂你的意思，我觉得我们都同意，短期内不太可能开什么会。
If everyone gets which in a room and says I hear you, you know, I think we're in agreement that I don't think there's going to be a meeting anytime soon.
不过，奥菲尔，非常感谢你来参加。
But Ofir, I want to thank you for coming on.
这次对话非常精彩。
It was a great discussion.
这位是尤恩公司的首席执行官兼联合创始人奥菲尔·埃尔利希，欢迎来到TITV。
That is Ofir Ehrlich, the CEO and co founder of Eon here on TITV.
接下来的环节，我们将由我们的赞助商普华永道带来。
Our next segment is with our sponsor, PwC.
在人工智能时代，预算的可预测性是个大问题，普华永道深入观察了全球众多大型企业是如何应对这些变化的。
Budget predictability in the era of AI is a huge issue, and PwC has a close-up look at how many of the world's biggest corporations are handling those dynamics.
我想请达拉斯·多伦来聊聊，他是我们科技、媒体和电信业务团队的负责人，听听他在这一领域观察到些什么。
I want to bring on Dallas Dolen, a leader at the firm's tech, media, and telco practice, to talk us a bit through what he is seeing in that arena.
达拉斯，欢迎再次做客我们的节目。
Dallas, welcome back to the show.
很高兴你今天能来。
It's great to have you here.
能来太好了，科什，你好啊。
It's great to be here, hey, Kosh.
见到你真好。
Good to see you.
达拉斯，我们节目里一直聊到‘代币最大化’这个话题，我猜这在你们公司高管层的讨论中也经常出现。
So Dallas, we've been talking about tokenmaxing here on the show quite a bit, and it's a firm I'm sure is coming up a lot in your boardroom discussions.
我想问问你，你接触的那些高管们，对‘代币最大化’这件事到底怎么看？
My question for you is where do executives that you speak with land on the issue of tokenmaxxing?
他们支持这个做法吗？
Are they fans of it?
他们是想结束它，好省点钱吗？
Are they wanting it to end so that they can save some money?
你详细说说看。
Walk me through it.
当然。
Sure.
对。
Yeah.
要不我先讲个会计笑话吧，就是关于EBITDA的定义。
Well, maybe I'll just start with an accounting joke, which is, you know, the definition of EBITDA.
我不知道你有没有见过这个说法，EBITDA里有两个T。
I don't if you've seen this, but EBITDA with two Ts.
第一个T是税，第二个T是代币。
The first one is tax and the second one is tokens.
这正是现在很多公司新采用的衡量标准
And that's the new measure that firms
这个我还真没听过。
can Haven't heard that yet, actually.
明白了。
Okay.
就是这样。
There you go.
周一这天出了个突发新闻。
Breaking news today on a Monday.
不过，确实，我
But, yeah, I
我的意思是，我
mean, I
我觉得，有几个因素在起作用。
think, you know, there's a couple factors that are coming into play.
最主要的就是使用量的不确定性。
I mean, the biggest one is unpredictability of usage.
当然，我们之前也聊过几次的，就是关于投资回报率的问题。
And of course, the conversation we've had a couple of times, which are around the ROI bit.
当CFO们试图预测下一季度甚至未来几年的财务状况时，他们会考虑人员编制和整体人力容量，再把技术成本加进去，变量多得数不清，但最关键的问题还是：他们用的是什么设备？
So, when you have the CFOs trying to do some level of predictions of, you know, whether it's the next, you know, set of, financials you for a quarter or better, maybe you're trying to predict out a couple of years or a few years and looking at, you know, headcount and, you know, where you're gonna be from an overall human, you know, capacity point of view, you start layering in the technology costs and there's so many variables that come into play, but the number one thing is gonna be, hey, what sort of machines are they using?
这些设备用在哪儿？
Where are they using them?
还有，我需要多少软件？
And how much software do I need?
过去我们预测这些其实挺简单的，说实话。
And historically we had a really, you know, I'll say easy ish way of predicting this, right?
因为那时候所有软件都是按座位授权的。
Because everything was on a seat license basis.
比如我需要给每个人配一台笔记本，再让他们都能用上公司最重要的那十款应用，算起来一目了然。
So, if I needed a laptop and I needed everyone to have access to the top, let's say, 10 applications from a business point of view, I could do that math really quick.
但现在情况变了，硬件方面的差异也大得多。
Now, we're finding that there's such high degree of variability, including in the hardware, right?
比如上周，我们开始看到英伟达和微软发布关于AI笔记本电脑的消息。
We started to see these announcements last week, for example, with Nvidia and Microsoft on the AI enabled laptops.
这些设备的价格会高得多。
These things are gonna be much more expensive.
再加上计算成本，情况就更复杂了。
Then you layer in the compute costs on top of it.
我觉得这几乎成了不可能完成的任务。
I think it's becoming a near impossible task.
上周我跟一位CFO聊过，他们预测成本时的误差范围高达400%。
And one CFO conversation that I had last week was they are trying to predict within a range of certainty of approximately 400%.
也就是说，他们预计的计算成本可能是实际花费的1到4倍。
So, one time to four time the number, what is gonna be their cost of compute.
你该怎么在这么大的不确定性下运营一家公司？
How do you run a business in that, in about a 400% sensitivity analysis?
这等于说，我预算了1000美元，但心里已经准备好接受最高5万5千美元的开销。
That's saying that for every $1,000 I think I'm going to spend, if I've budgeted for 1,000, I'm okay with $55,000.
对。
Right.
400%到100%就是五倍。
404100% is 5,000.
对。
Right.
没错。
That's right.
我觉得这
I think it's
那你给企业什么建议呢？
how do you so what what advice do you give to companies?
我的意思是，这种情况还会持续多久？
I mean, how how long is this gonna last?
有意思的是，
Well, here's the funny thing, right?
我们已经聊了很久关于纯AI公司的话题了。
We've been talking a lot about AI native companies now for a while.
你想啊，要是真做纯AI公司，是不是就不用操心什么基础设施或者其他那些东西了？
It's like, what could you build AI native and not to, you know, not have to have any infrastructure or any of these other things?
那些100%纯AI驱动的公司，其实要敏感得多。
The companies that are 100% AI native are actually far more sensitive, right?
所以，对这类成本的敏感度分析，远高于那些依赖硬件或人力的公司——毕竟他们得靠这些来推动业务、维持日常运营，让灯一直亮着。
So, sensitivity analysis to these types of costs is dramatically higher than a company who might be hardware or human reliant, if you will, in terms of the ability to drive the business and to run the day to day and keep the proverbial lights on.
所以你会发现，那些刚起步、技术上更依赖AI的公司，受到的影响差异会特别大。
So, think what you're gonna see is a high degree of variability in terms of impact between those who have maybe newer out the gate, by the way, in terms of technology and reliance on AI.
但你到底该怎么经营一家公司呢？
But how do you run a business?
我觉得你得接受一个现实：到最后，你根本没法准确预测某一项支出具体会落到什么水平。
I think you have to get very comfortable that at the end of the day, you cannot predict within a level of certainty on that specific line item where you're gonna be.
那这项支出会是最大的吗？
Is that the highest line item?
不，这只是一个需要关注的支出项目。
No, it's gonna be just a line item to be concerned about.
这才是最大的问题。
And that's the biggest concern.
我明白你讲的长期趋势，就是得学会适应不确定性和保持灵活。
What are the long term, so I hear you on the long term, you get comfortable being uncomfortable and being flexible.
但如果回到‘令牌最大化’这个趋势，你觉得这种状况能持续多久？
But if we go back to sort of the tokenmaxing movement, how long do you think that lasts?
你觉得迟早CFO们会说：嘿，你们得像管理其他预算一样，对令牌使用实施更严格的审批流程。
Do you think at some point CFOs are going to say, Hey, look, you actually have to apply for approvals for token use in a much more governed fashion than it's being done right now.
你觉得这种变化会到来吗？
Do you see that coming?
当然会，百分之百会。
Yeah, 100%.
我觉得接下来的发展方向是，至少我已经看到很多客户，甚至一些技术服务公司都在做：把技术成本计入实际使用这项技术的部门的损益表里。
I think the way that's going to happen, at least what I'm already going to see, you know, with a number of my clients and even seeing around, you know, the services companies too, is they are moving the cost of technology into the P and L of the group who's, you know, who's operating the technology itself.
所以，现在不再由首席信息官或首席技术官来管了，而是直接落在了业务部门的损益表上。
So no longer is it sitting with the CIO or CTO, it's actually sitting with the P and L within the business.
不管你是在制造部门、财务团队，还是资金团队，责任都在那儿。
So if you're in manufacturing, if you're in the accounting team, if you're in the treasury team, it's sitting there.
这实际上给团队负责人施加了压力，得开始做出真正明智的决策。
That's actually putting the pressure, if you will, on the manager of that team to start making, quote, good decisions.
当然，‘明智’是相对的，但关键是要在计算资源、人力投入和其他变量之间，做出合理的资源配置。
Of course, that's a relative concept, but still good decisions around exactly what sort of compute and what sort of monetary allocation you're gonna make to the computational piece versus the human piece versus the other variables that come into play.
投资回报率成了必须考虑的一部分。
ROI becomes a part of the story.
还需要对技术使用进行非常明确的规划。
Also, really deliberate planning as it relates to what you're going to do with technology.
对很多公司来说，高层管理团队和董事会最该谈的，可能就是从顶层开始的这场对话。
Probably starting from the top is gonna be the most important conversation that's gonna take place within the C suites and at the board level for a number of these companies.
现在，我想问你一下，我们之前在节目中聊过的编程对安全的影响。
Now, let me ask you about what we were talking about earlier on the show, the security impacts of coding.
我这么问是因为，现在‘tokenmaxing’这个趋势鼓励大家去‘氛围编程’，自己动手开发应用。
And I ask this because, I mean, with the tokenmaxing movement is an encouragement that people should be vibe coding, they should be building their own apps.
我们节目里也请过一些人聊过，这些非专业开发者自己做的程序，往往更容易出现安全漏洞，甚至可能造成严重混乱。
And we've had folks on the show talking about these citizen developers building their own programs that are innately more susceptible to security breaches or might cause havoc themselves.
所以你觉得，安全风险会不会是公司限制‘氛围编程’、或者给它加更多管控的原因之一？
So do you think that security breaches could be one reason that companies actually pull back on vibe coding or put more guardrails around it?
绝对是的。
A 100%.
我接触过的不少企业里，首席安全官们也都提到，这正是他们放缓这类技术部署的主要原因，尤其是在受监管的行业里。
And I think even within the organizations that I've talked to, the CSOs and talking to our own CSO, that is actually the reason for a slower deployment of some of this technology, especially within regulated companies.
你看银行、保险公司、医疗行业，还有会计事务所和律师事务所，都是这样。
You look at the banks, you look at the insurance companies, you look at medical, you look at, you know, the accounting firms and the law firms.
会有尝试，但你也看到新闻里报道过，对吧？
You're gonna see experimentation, but you also see, we've this seen in the news, right?
新闻里已经多次报道过，滥用AI技术，比如在法律文件里出错，会带来巨大的负面后果。
The news cycle has shown extraordinary downside for when you misuse some of the AI just in some of the legal filings, for example.
这太丢脸了。
It's embarrassing.
也许现在还不算违法，但你想想，如果有人因为搞 vibe coding，不小心给公司制造了严重漏洞，首席安全官根本没法接受。
It's not maybe breaking the law necessarily just yet, but now start thinking about it if the person has created, because they're vibe coding and created some sort of significant vulnerability in the company, CSOs can simply not live with that.
首席执行官也一样，根本没法接受。
CEOs can't live with that.
所以，高层的指令会是两方面的。
And so, the directive from down on high is gonna be twofold.
第一，确保你给员工用的工具都配有安全防护机制。
One of them is make sure the technology you're giving people have security controls around it.
第二，最好只选用那些你信任的厂商，他们的产品从一开始就内置了安全设计。
And the second one's gonna be, hey, only probably go with the technology companies that you trust who are actually building it into the first layer of what they're able to do.
这样一来，实际上会缩小人们使用这项技术的范围，包括能用什么数据、哪些员工被允许使用等等。
That may shrink actually the zone, so to say, in which people are allowed to use this technology, including on what data and which individuals within an organization are allowed to use it too.
我觉得这很自然，对吧？
I think it's just the natural tendency, right?
局势又开始往回摆了。
The pendulum swings back the other way.
从之前那种开放自由的状态，一下子又回到了极度重视安全的阶段。
You go from Open Claws and Clawbots all the way back to being heavily concerned about the security side of it.
当然，现在的网络安全公司都在全力转向人工智能管理方案。
And of course, by the way, I mean, the cyber companies out there are all going to AI managed solutions.
帕洛阿尔托、谷歌、微软这些大厂都在这么做。
We're seeing this from Palo Alto and from Google and Microsoft and the rest of the gang.
他们拼命推这个方向，因为这事儿必须百分之百可靠，对吧？
They're pushing really hard on that agenda item because they need it to 100% of the time, right?
不再像以前那样，有人坐在房间里盯着屏幕，观察漏洞和潜在攻击了。
It's no longer like when you have people, you know, in the room watching the screens and seeing like where the vulnerabilities and where the potential attacks are.
现在一切都由系统自动处理了。
It's happening 100% Last of the
那我问你一个问题。
question for you then.
那你觉得这对AI实验室和模型销售公司会带来什么影响？
So, what impact do you think this has on the AI labs businesses and the companies that are selling the models?
是的，说实话，这些方向最终会怎么发展，现在还不好说。
Yeah, look, I mean, still think it's yet to be written as far as where all these ones play out.
大家还得继续努力，打造最好的模型，同时也要全力开发集成技术，尤其是针对特定行业和垂直领域，对吧？
And people need to continue to work really hard to build the best models and work really hard to build the best integrated technology, especially into specific industries and to specific verticals, right?
所以，人们会把安全机制融入其中，以最大化它们在特定领域的影响力。
So, people will build security in to allow it to maximize the impact that they're gonna have in a given area.
比如Harvey这样的公司，就会在法律领域持续这么做。
There's companies like the Harvey's that will continue to do that, for example, in the legal arena.
我觉得这将是未来我们能看到的趋势。
I think that's gonna be something that we see going forward.
上周在纽约科技周，我跟很多人聊过，发现情况确实如此。
And I saw that a lot in the conversations I had at New York Tech Week last week.
现实是，所有人都在加倍、三倍、甚至四倍地投入，争取打造出独特且高度定制化的产品，抢在别人前面率先推出。
The reality was everyone is still doubling and tripling, quadrupling down on their ability to build something that's so custom and unique and also be the first out of the gate to get there.
我觉得资金投入还会继续。
So I think there'll be continued funding.
创业者们也会持续加码。
There'll be continued effort on the part of founders.
这事儿还远没定论。
It's really yet to be written.
现在还只是初期阶段。
This is early stages.
目前各方正在建立管控机制，避免冒太大风险。
Again, the controls are being put in place to ensure that we're not taking too many risks.
你上周也参加了科技周活动吧。
You were here for Tech Week last week.
你有看尼克斯队的比赛吗？
Did you catch any of the Knicks games?
我看了尼克斯那场比赛。
I watched the Knicks game.
这是个真实的故事。
This is a true story.
等一下。
Wait a second.
你来自达拉斯。
You're from Dallas.
稍等。
Hold on.
你真是来自达拉斯。
You're you're from Dallas.
你是尼克斯队的球迷吧。
Your name is you're you're from Knicks fan.
不是。
No.
这跟……没有任何关联。
This is there's no affiliation with with
和圣安东尼奥没有任何关系。
No affiliation with San Antonio.
对。
Yeah.
体育方面。
Sports.
只是因为我名字取自一部电影。
Just it's just that I've been named after a movie.
我名字是取自那个外星人托马斯。
I was named after Thomas, the alien.
虽然和德克萨斯没关系，但我当时在鲍厄里街一家酒吧外面看比赛，有四百多人涌到街上。
So no affiliation with Texas, but I watched the game from outside of a bar on Bowery Street where like 400 people were streaming out into the streets.
我得说，这让我对人类社会重拾了希望。
And I gotta tell you, it gave me renewed hope in civilization and society.
阿卡什，看到这么多纽约尼克斯球迷聚在一起，以二十年来从未有过的方式尽情欢庆，那场面实在太震撼了。
Akash, like it was the most powerful thing to see all these New York Knick fans come together and enjoy themselves in a way they haven't enjoyed themselves in over two decades.
我真为纽约人这次全力支持球队的表现感到无比自豪。
I actually couldn't be prouder of the way that New York showed up and is supporting their team.
说实话，我也没特意盼着马刺赢，但这次纽约能赢，可能是好久以来最让人振奋的事了。
And yeah, I don't sit here and root against the Spurs necessarily, but it could be the best thing to happen in New York in a long time.
接下来几天，尼克斯在麦迪逊广场花园拿下这个冠军，可能性很大。
It'd be the Knicks taking this thing in the next couple of days in Madison.
说实话，我也不太确定该怎么说。
Well, I thought, you know, I don't know.
我本来以为你可能是小牛队的球迷，但现在看，又好像不是。
Maybe Mavericks fans have I figured maybe you were a Mavs fan, but maybe not.
也许吧。
Maybe
不是。
Negative.
说不定德克萨斯才是另一个情况。
Maybe Texas is the other.
总之，尼克斯队今晚有比赛。
Anyway, the Knicks are playing tonight.
谢谢你的夸奖。
Thank you for your kind words.
我们就不打扰你们继续为尼克斯队加油了。
We will let you continue to cheer for the Knicks nation.
还有达拉斯，一如既往感谢你来参加节目。
And Dallas, I want to thank you for coming on as always.
每次和你聊天都特别愉快。
It's always a great conversation.
这位是来自普华永道的达拉斯·多伦，正在TITV为您直播。
That is Dallas Dolen from PwC here on TITV.
今天的节目就到这里。
That does it for today's show.
提醒一下，我们周一到周五每天上午十点准时直播。
A reminder, we are on this stream Monday through Friday at ten a.
嗯。
M.
太平洋时间，下午一点。
Pacific, one p.
嗯。
M.
东部时间。
Eastern.
如果那时没法收听，节目内容可以在theinformation.com、我们的YouTube频道，或者你常用的播客平台收听。
If you can't make it then, episodes are available on theinformation.com, on our YouTube channel, or wherever you get your podcasts.
别忘了在X、Instagram和TikTok上关注我们。
Make sure to follow us on social media, on X, on Instagram, on TikTok.
我已经开始期待明天的节目了。
I am already excited for our next show tomorrow.
祝你周一剩下的时间愉快。
Have a great rest of your Monday.
为纽约尼克斯队加油！
Go New York Knicks.
那先这样，再见了。
Bye bye for now.
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。