20VC：Anthropic从微软和英伟达融资300亿美元 | 英伟达核心业务受TPU威胁 | 分析山姆·奥特曼的"战争模式" | Sierra达成1亿美元年度经常性收入：能否支撑100亿美元估值？ | Lovable实现2亿美元年度经常性收入及传闻中的60亿美元融资轮

OOP（面向对象编程）哪里出问题了？

What about the OOP doesn't work?

嗯，团队会给你无穷无尽的借口。

Well, you get you get endless excuses from the team.

明白吗？

Okay?

因为这是事实。

Because it's true.

因为你们有几十年的技术债务和上百个功能，却还向大客户承诺交付这个迟迟未完成的功能。

Because you have decades of technical debt and a 100 features and you just promised a big customer you'd have this feature that still doesn't come out.

而现在，Num Nuts、Lemkin或Steppings还要我们做更多。

And now, Num Nuts, Lemkin, or Steppings wants us to do even more.

所以除非进入像Facebook移动团队那样的超激进模式，否则什么都做不成。

So you can't get anything done unless you're in hyper aggressive mode, like mobile for Facebook.

我只是不确定对团队喊话是否有效。

I just don't know if telling the troops works.

我只是...我从未见过它奏效，但你必须进入超级激进模式，因为我认为他的意思是如果我们以目前的速度发展，我们将会落后。

I just I've I've never seen it work, but you do gotta go into hyper aggressive mode because I think what he's saying is if we if we evolve at our current pace, we're gonna fall behind.

就这么简单。

It's just that simple.

这就是他的意思。

That's what he's saying.

我们会落后的。

We're gonna fall behind.

而当我与初创公司会面时，尤其是当你后期投资因为资金已经投入的时候。

And when I meet with startups, especially when you when you invest later because your money's already in.

对吧？

Right?

当他们不处于超级激进模式时，我不想说我会失去信心（以防有人在看），但我确实会失去信心，因为你根本不会有任何进展。

When they're not in hyper aggressive mode, I don't wanna say I lose confidence in case anyone's watching, but I lose confidence because you're just not gonna get anywhere.

尤其是在今天大家都在AI领域落后的情况下，我需要看到（你们的行动）。

I need to see especially today when folks are falling behind in AI.

对吧？

Right?

当大规模的公司都落后时，我不在乎你的说辞。

When you have companies at scale falling behind, I don't care about your talk.

我不想听你的试点计划。

I don't wanna hear about your pilot.

我要感受到你的团队正处于极度激进模式。

I wanna smell that your team is in hyper aggressive mode.

如果是这样，我认为你还能翻盘。

If it is, I think you can come back.

如果我是

What is If I

如果我没感受到这种状态，你就毫无机会。

don't smell it, you have no chance.

有哪些迹象或感觉能表明团队处于极度激进状态？

What are signs or smells that suggest hyper aggressive tea?

创始人们会听到这些

Founders will be listening to this

管理层中存在一种加快行动步伐的紧张氛围。

going The management team there's a tension in the management team to move faster.

管理团队中的每个人都表示，我们正在加快产品交付速度。

Every single person in the management team says, we are shipping product faster.

我在更努力地销售。

I am selling harder.

我出差更频繁了。

I'm getting on the road more.

我正在获取更多销售线索。

I am generating more leads.

你管理团队的每个人都在全力以赴，他们执行得更快了。

Every single person on your management team is sweating it, and and they're executing faster.

你可以看到他们参加董事会会议或其他任何场合时，每个领域都充满干劲。

You can just see they come to your board meeting, they come to whatever, and there is velocity in every area.

我们只是不知道，你在董事会上真正看到效率提升的情况有多少次？

We just don't you how many of times do you truly see velocity increase in a board meeting?

并不常见。

Not that common.

对吧？

Right?

这是超级激进模式。

That's hyper aggressive mode.

不知怎的，大家在这点上达成了一致。

Everyone is aligned on this somehow.

这才造就了一位伟大的CEO。

This is makes a great CEO.

这对CEO来说非常难做到。

This is very hard for a CEO to do.

一切都在对抗惯性。

Everything is is anti inertia.

即使在早期初创阶段，一切也会慢下来。

Everything slows down even in an early stage startup.

对吧？

Right?

然后当你进来时，大家都在流汗，他们就像，你知道的，我们交付的故事点是上个季度的两倍，软件产出也是两倍，但我们仍然落后，我很恼火。

And then when you come in and everyone's sweating, and they're like, you know, we shipped twice the story points, twice the software as last quarter, and we're still falling behind, and I'm pissed.

如今在优秀工作领域，我看到的还不够多，而这正是你需要的。

These days in the area of great job, I just don't see it enough, and that's what you need.

你需要让你的团队变得极度激进。

You need to make your team hyper aggressive.

这是一种不自然的状态。

It is an unnatural state.

如果我推动你，这是不自然的。

If I push you It is unnatural.

如果我同时推动你们两个，消费者会青睐谁？

If I push you both, who will win the consumer?

是OpenAI以及他们显然将在未来几年内推进的应用发布吗？

Is it OpenAI and the app rollout that they are clearly going to do over the next few years?

还是谷歌会保留现有的消费者群体、搜索层和应用层？

Or will Google retain the consumer, the search layer and the app layer that they have today?

我认为谢尔盖为谷歌重新带来了极度进取的精神。

I think Sergey brought back hyper aggressiveness to Google.

他在采访中表达得很明确。

He's clear in the interviews.

他，你知道，他直接对我说，我刚来时，我们甚至不被允许使用自己的工具。

He, you know, he just to me say, I came in, we weren't even allowed to use our own tools.

我们不被允许使用自己的编码工具。

We weren't allowed to use our own coding tools.

我们不被允许使用自己的芯片。

We weren't allowed to use our own chips.

他说，我一周内就废除了这些规定。

He's like, I got rid of that in a week.

我听到了一些争论。

I heard arguments.

我在一周内就解决了这个问题。

I got rid of it in a week.

这正是你需要的。

That's what you need.

谢尔盖必须展现出伟大的领导力。

Sergei had to come as great as the leadership is.

我认为谢尔盖必须回归并采取极度激进的策略。

I think Sergei had to come back and make it hyper aggressive.

所以不要认为这是唯一的关键，但你需要这种态度。

So don't think it was the only point, but but you need that.

你需要这种态度。

You need that.

我看到的大多数未能成功运营的初创企业，都无法重新进入极度激进模式。

And I just most of the startups that I see that aren't found to run are unable to get back to hyper aggressive mode.

对他们来说这根本不可能。

They're just it is impossible for them.

我并不是说萨提亚没做到过，但看看我们的投资组合。

I'm not saying it hasn't happened with Satya and and and but look at our portfolio.

在你的投资组合中，有多少非创始人能切换到超激进模式？

How many folks in your portfolio without a founder can switch into hyper aggressive mode?

所以你是在说谷歌？

So you're saying Google?

我得说这是20风投。

Over I have to this is a 20 VC.

有

Have

得在...之间选择

to Will pick between the

谷歌留住了消费者吗？

Google retain the consumer?

是的。

Yes.

他们现有的应用套件，或者说OpenAI是否会持续蚕食谷歌的核心业务，我们将看到一个更加分散的市场格局？

The app suite that they have or will OpenAI continue to eat more and more and more of Google's core business and we will see a much more shared landscape?

哦，听着。

Oh, listen.

我和X平台上大多数人的想法一样。

I think the same thing most people on X do.

我认为谷歌的产品很棒。

I think the Google products are great.

我每天都在使用它们。

I use it every day.

我也用AI。

I use AI.

实际上，我根本不在乎用的是AIO概览、ChatGeeb还是Euthropic，但我其实更喜欢AIO概览，因为不用离开谷歌。

In fact, don't even care whether I use AIO Overviews or ChatGeeb or Euthropic, but I actually prefer AIO Overviews because I don't have to leave Google.

这是个很棒的产品。

That's a great product.

Gemini三Pro也是。

So is Gemini three Pro.

这些都是很棒的产品。

These are great products.

没有理由。

There's no reason.

你没有给我任何离开的理由。

You're not giving me any reason to leave.

它们确实是优秀的软件。

They're truly great software.

我来回答这个问题。

I'll answer this question.

我确实想回到领导力的话题，因为我想稍微聊聊这个，也许能学到些什么。

I actually do wanna come back to the leadership thing because I wanna chat a little bit about that and maybe learn something.

但回到第一个问题，关于谷歌的问题，如果你之前听过关于守望先锋的讨论，六个月或九个月前，人们会说天啊，谷歌要完了。

But going back to the first question, the Google question, if you listen to the Overwatch stuff a while ago, Google six, nine months ago, it's like, my God, Google's dead.

我当时完全被守望先锋的讨论影响，显然他们在云计算和模型方面执行得非常好，而核心搜索业务的实际下滑幅度远没有那么大。

I was all overwatched, obviously executed really well in cloud and the model and fundamentally the core search business has not declined nearly as much.

《华尔街日报》有一篇采访，我记得是搜索业务负责人，她谈到了搜索业务的情况。

There was an interview with I think the head of search in the Wall Street Journal and she talked about how search yes.

虽然某些功能在变化，但随着人们提问增多，搜索量实际上在上升。

Some things are going but search volume is going up as people ask more questions.

所以很明显，搜索业务的衰退不会突然发生。

So it was it's clear that the decline of search isn't gonna be precipitous.

这是我的明确结论。

That's my clear takeaway.

我们明白了

We got that

这个判断有误。

one wrong.

确实如此。

Exactly.

那次判断失误了，但他们仍能获取部分搜索业务。

Got that one wrong, and they're going to be able to get some search.

与此同时，我确实认为市场上已出现了一个全新且独立的消费者类别——为高级AI付费订阅，而我认为JaiGPD在该领域占据了有利位置。

At the same time, I do believe there has emerged a new and separate consumer category of paid subscription for advanced AI, and I think JaiGPD is a good position in that place.

拥有大约8亿用户，其中5%是付费用户。

Have what, 800,000,000 users, 5% of them are paying.

我不认为这些用户会‘回归’谷歌。

I don't see all those users, quote unquote, going back to Google.

我认为他们已经在注意力经济中开辟了自己的细分市场——而且是个大市场。

I think they've carved their niche, big niche, in the attention economy.

我不认为这是非此即彼的选择。

I don't see it as an either or.

即使搜索业务的增长放缓。

Even if the search business slows down growth.

我认为谷歌和ChatGPT可以构建一个可防御的大型消费者业务，规模虽不及谷歌，但能在更宏观层面从每个人那里获得几美元收入，对吧？

I I think Google I think ChatGP can charge can build a defensible large, enormously large consumer business to minor extent to the extent of Google and to a more macro extent to the you know, a few dollars from everyone, right?

我是说，考虑到广告收入，我们正接近一个临界点——在ChatGPT出现之前，谷歌、Facebook和亚马逊的数字广告支出已占广告总支出的极高比例。

I mean, do think the advertising dollars, we're getting close to the point where the digital advertising spend across Google, Facebook, and Amazon, even before ChatGipedia exists, is becoming a super high percentage of the total ad spend.

所以如果ChatGPT要分一杯羹，要么整体市场需要扩大，要么有人必须让出部分份额。

So if ChatGPT is going to get their share of that, either we're going to have to grow the pie overall or someone's going to have to lose a little bit.

归根结底，我认为这不是非此即彼的问题。

Net net, I don't think it's an eitheror.

我认为谷歌已经证明，关于其衰落的报道被严重夸大了。

I think Google has proven reports of that debt are greatly exaggerated.

我在泄露的备忘录中看到，Altman提到增长可能在一两年内降至5%，显然分析师称这是灾难性预测。

I saw in the memo, leaked in the memo, that Altman made a reference to growth falling to 5% in a year or two, and obviously the therapists call that catastrophizing.

当你假设如此糟糕的情况时——比如这家公司年增长率从500%骤降至5%，那将是史上最灾难性的增长下滑。

When you postulate something that's so awful, like this company is growing zero to 500% year on year, if it went to 5% year on year, that would be the single most catastrophic growth decline in history.

但就可能性而言，抛开人为因素，我认为Chachi P会继续打造这款引人注目的消费产品，占据差异化市场，同时谷歌搜索也不会'归零'。

But in terms of likely outcomes, independent of the human dynamic, the likely outcome for me is Chachi P continues to build this compelling consumer product, can command differentiated market, while at the same time Google search does not quote unquote go to zero.

蛋糕在扩大，美国科技巨头七雄从世界其他地区吸走了更多的利润。

The pie expands, the great American tech mag seven suck more of the world's profit dollars out of the rest of humanity.

加油，团队。

Go team.

好的。

Okay.

说到扩大的蛋糕，你还有什么要补充的吗？

Well, speaking of expanding pies, do you want to add anything or no?

没有。

No.

没有。

No.

我们来聊聊蛋糕吧。

Let's hit pie.

所以

So

说到扩张

speaking of expanding

我要回去了。

I did I'm gonna go back.

关于领导力这件事，正如我所说，对我而言，因为我不是那种喜欢夸夸其谈的领导者类型。实际上我对Jason的评论很感兴趣，我正在反思这一点，因为我往往更倾向于退居幕后，这也是为什么我觉得自己只能是个平庸的CEO，但作为投资人还算称职。

The leadership thing is, as I say, for me, because I'm not a good wah wah type leader, I'm actually was interested in Jason's comment, I'm reflecting on it because I tend to be more the step back person, which I think is why I'd be a mediocre CEO and I'm okay as an investor.

这是个很棒的问题。

It was a great question.

你知道有哪些迹象吗？

What are the signs you know?

作为投资人，我什么时候会审视自己投资的公司然后感叹：天啊，这家伙干得太猛了。

As an investor, when do I look at my companies and go, god, this guy's running hot.

对吧？

Right?

我一直在思考这个问题，这确实是个非常棒的问题。通常每年至少有一两次我会想：CEO先生（我心里确实有几位这样的对象），你是不是把团队逼得太紧了？

And I've thinking about it, it's a super question, and it's generally when at least once or twice a year I think to myself, mister CEO, and I have one or two in mind, are you pushing your team too hard?

虽然我不喜欢战争比喻模式，但你说得对，只有那些极度狂热、极度执着的人才能真正追求卓越。我回想我那些最优秀的公司，偶尔CEO会突然情绪失控。

Even though I don't love the war metaphor mode, you are correct, it's only the wildly intense ones, the wildly driven ones that really have excellence, and I'm reflecting on my very best excellent companies, and every once in a while the CEO would just lose his shit.

他或她会对进展缓慢感到极度沮丧，对胜利的渴望会产生这种紧张感。

He or she would be just so exasperated by the lack of progress and the drive they felt to win that you'd have that tension.

团队规模越大，人员越多，沟通就必须越简单，因此你可能需要使用更简单的比喻，比如'这是场战争，我们必须赢'——因为这是激励五千人朝同一方向前进的必要手段。

And the bigger the team, the more people you have, the more simply you have to communicate, and therefore you probably use simpler metaphors, and therefore you probably have to go for some kind of it's a war, we gotta win, Cause that's what it takes to motivate 5,000 people and point them in the one direction.

呃，我不确定这是个观点。

Well, I don't know if it's a thought.

仔细想想，你说过有时会建议CEO们对团队温和些。

Thinking on it, you know, I actually don't think you you said you sometimes advise your CEOs to to go easier on your team.

但我认为对真正优秀的人根本不需要温和对待。

I don't think you have to go easier on anyone great ever.

所有伟大的领导者、副总裁、高管都是如此——而这样的人很罕见。

Everyone great, every great leader, every great VP, executive, and they're rare.

非常罕见。

They're rare.

他们的干劲丝毫不逊色于CEO，甚至在某些方面更为拼命。

They're going just as hard or in some ways harder than the CEO.

CEO的工作更为艰巨。

The CEO's job is harder.

许多高管并不明白身处高位有多难。

A lot of executives don't get how hard it is to be at the top.

但在各自的领域里，比如CTO、CRO，他们比CEO还要拼命。

But in their own domain, right, CTO, CRO, they are going harder than the CEO.

如果现在是2021年，人们都会说我讲这种话很毒舌。

And so if this was 2021, people everyone would say I'm being toxic saying this.

其实不是。

It's not.

好吧。

Okay.

我们要展望2026年了。

We're going to 2026.

我不认为你会对团队施加过大的压力。

I do not think you can push your team too hard.

我认为你应该根据业务需求来推动他们。

I think you should push them as hard as the business needs to go.

如果他们离开，那很好。

And if they leave, it's great.

这太棒了，因为他们无法带你到达目标。

It's terrific because they're not gonna get you there.

也许Sam正在这样做。

And maybe that's what Sam's doing.

他正在告诉他的团队，自满情绪已经滋生。

He's saying to his there is complacency is set in.

每个人都在一次性销售2000万美元。

Everyone's getting to sell 20,000,000 at a time.

人员流动率很高，这曾一度奏效，但现在竞争加剧了。

Turnover is high, and it worked for a while, and now the competition is up.

也许他正在对20个人说，我需要你们更加努力，可能一半人会离开，另一半会坚持下去。

And I maybe he's talking to 20 people, and he's saying, I need you guys to go even harder, and maybe half will leave and half will will do it.

我认为这可能就是全部信息。

I think that may be the whole message.

但老实说，对于创始人，我的首要建议是不要对你的团队心慈手软。

But I honestly think to founders, my number one a bit of advice is do not go easy on your team.

爱护他们，支持他们，不要挑剔他们，不要在他们擅长的领域拖后腿，让他们在擅长的领域自由发挥，但要更加鞭策他们。

Love them, back them up, don't pick at them, don't back where they're good, let them run where they're good, but push them even harder.

最优秀的人总会迎难而上。

The best ones will always step.

他们可能会哭闹或崩溃几天，但最终会振作起来。

They might cry or lose it for a couple days, but they will step up.

或者他们会离开。

Or they'll go.

你说得对，杰森。

And you're right, Jason.

一般来说，在你最好的公司里，你说得完全正确。

Generally, in your best companies, you're exactly right.

有些人会崩溃退出，因为他们就是承受不了。

Some people crash out because they just can't take it.

我的观点是，你如何知道引擎是否在全速运转？

My point is, how do you know the motor is running at full speed?

答案是你会偶尔听到齿轮摩擦声。

The answer is you hear an occasional gear grind.

而你作为董事会成员的职责就是检查这些摩擦声，然后说，哦，好吧。

And your job as a board member is to check-in on the gear grind and go, oh, okay.

我可以忍受那种齿轮摩擦声。

I can live with that gear grind.

这没关系。

It's okay.

他们正在做正确的事。

They're doing the right thing.

让他们都坚持到2026年吧。

Let them all go for twenty twenty six.

放手让他们去吧。

Let them go.

因为他们不会带你走向任何地方。

Because they're not gonna get you anywhere.

他们不会带你走向任何地方。

They're not gonna get you anywhere.

那些觉得太难的人。

The folks that it's too hard.

比如，工作太难了。

Like, the job's too hard.

对吧？

Right?

说真的，在我目前增长最快的投资组合公司里，上次董事会会议上，首席营收官站起来说，我想明年超额完成扩展计划。

I mean, literally, in my my my fastest growing portfolio company now, the last board meeting, the CRO set stood up and said, I wanna do more than the stretch plan next year.

冷静地。

Calmly.

不是在‘我想做得更多’这种表面说辞下，而是给出了数据驱动的理由。

Not in the front of I wanna do more than that and gave a data driven reason why.

没人要求他这么做。

No one asked him to do it.

在这个失败是常态的时代，你需要团队有这样的自觉。

You want that out of your team in the age of you're just going to fail.

我在想。

I wonder.

我要表示反对。

I'm going disagree.

你可以推到近乎妄想的地步。

You can push to the point of delusional.

你可以推到刚好低于那个临界点的程度。

You can push to just one level just below that.

但我其实认为这是一种技巧，就是将团队逼到不切实际的地步，然后稍微收敛一点，因为你看，如果整个团队都散了就无效了，实际上这可以衔接你刚才关于战争话题的讨论。

But I actually think that's the technique to just really drive the team to the point where this is not realistic and then just take it down one notch because it look, it doesn't work if you lose your entire team and actually it could segue off the you you made this discussion on the war thing.

正如你指出的，在OpenAI董事会变更后的过去两年里，你已经失去了OpenAI的整个团队。

As you point out in the last two years since the OpenAI board change, you have lost an entire team at OpenAI.

呃，并非所有人。

Well, not all of them.

你还留住了格雷格。

You you kept Greg.

是啊。

Yeah.

你不想失去整个团队。

You don't wanna lose your entire team.

所以问题在于，在不让整个事情崩盘的前提下，你能施压到什么程度？

So the question is how hard can you push while not blowing the whole thing up?

你也不希望团队离开时又造出个Anthropic（注：AI公司），更别提什么思维机器那些了。

You'd rather not create an anthropic when your team leaves too, let alone thinking machines and all that.

你肯定不想逼走两位顶尖人才，结果他们却打造出你的头号竞争对手。

You don't wanna force two of your best people out and they build your top competitor.

我们之前讨论过战争模式。

We said about war mode.

目前看来，没有哪个领域比客户服务和客户支持市场的竞争更激烈了。

There is no category more competitive today, it would seem, than the customer service, customer support market.

Salesforce前联席CEO布拉德·泰勒创立的Sierra公司，上周宣布其ARR（年度经常性收入）在两年内突破1亿美元。

Brad Taylor, formerly co CEO of Salesforce, started Sierra, reached a 100,000,000 ARR within two years, it announced over the last week.

上一轮融资估值已达100亿美元。

The last round was at $10,000,000,000.

该公司九月份刚获得Green Oaks投资的3.5亿美元融资。

It raised $3.50 by Green Oaks in September.

按当前ARR计算其估值已达100倍。

It's 100x ARR as of today.

我们该如何分析这种ARR快速突破1亿的扩张模式？

How do we analyze this very fast scaling to a 100 in ARR?

这个价格合理吗？

Does it justify the price paid?

这是否表明客户支持市场的规模将远超我们之前的预期？

And does it show that customer support is gonna be so much of a bigger market than any of us previously thought?

嗯，我来告诉你

Well, I'll tell

从市场角度我可以告诉你，我们在这个领域有大量投资。

you what I can tell you from the street, having a lot of investment in this space.

这不是对Sierra、Decacon或其他公司的批评，而是整个企业级市场目前都在过度承诺其实际交付能力。

It's not a criticism of Sierra or Decacon or others, but the whole category, the enterprise side of it is massively overselling what they can deliver today.

今年年初Replit和Lovable也是如此。

Now, so did Replit and Lovable at the start of this year.

当时Replit、Lovable、微软等公司都说：给我们一行代码，我们就能为你打造一个销售团队。

That everyone Replit, Lovable, Microsoft, everyone said, give us one line and we will build you sales force.

明白了吗？

Okay?

如今这说法更接近、更贴近事实了。

That is closer, much closer to truth today.

在年初时，这几乎等同于欺诈。

In the early year, it bordered on fraud.

而现在它正在成为现实。

Now it's becoming reality.

我认识的许多人购买了新一代AI支持工具，但要么根本没部署，要么勉强部署，或者根本没有AI在运作。

So many folks I know have bought next generation AI support tools and have deployed them at all or barely deployed them or there's no AI working.

所以并不是说它不可能实现。

And so it's not that it can't happen.

就像它终将实现一样，我对此深信不疑。

Like it will, like I believe in it.

而且很少有比支持服务更能被AI颠覆的领域了。

And there's few categories AI can disrupt more than support.

但我告诉你，如果你深入调查，会发现很多部署甚至还没开始，或者半途而废、训练不足或功能残缺。

But I will tell you if you dig deeper, you'll find a lot of deployments haven't even happened or half there are untrained or broken.

所以这很大程度上是为了让首席行政官和其他人感到兴奋。

So it's a lot of getting CAOs and others excited.

而如今这就像今年早些时候的Vibe Coding一样被过度吹捧了。

And this is as oversold today as Vibe Coding was earlier this year.

但我并不是说它不会迎头赶上。

But I'm not saying it won't catch up.

我不是在冷嘲热讽。

I'm not being cynical.

我只是觉得，伙计，这被夸大了。

I've just, man, it's oversold.

很明显，支持服务与编程并列是LLM最大且最明显的两大市场之一。

It is clear that, yes, support alongside coding is one of the two largest and most obvious markets for LLM.

因此这是一个完全可以运作的巨大类别。

So this is kind of a big huge category that can totally work.

基于此，布拉德·泰勒显然才华横溢，而Sierra在高端市场尤其享有盛誉。

So you start with that, Brad Taylor is obviously wildly talented and Sierra has a great name, especially at the high end.

实际上我认为这个领域正在发挥作用，Lilly曾与某人进行过参考通话，对方表示，我们评估了Sierra，评估了所有这些其他公司，在LLM出现之前，我投资过一家做客户支持的公司，那时的解决率大约是23%。

I actually think the category is working and Lilly did a reference call with someone and they said, look, we evaluated Sierra, we evaluated all these other names And before LLMs, and I had an investment in a company that was pre LLM customer support, our resolution rate was around 23%.

换句话说，我们能解决23%的来电，也许最多30%。

In other words, we could solve 23% of calls, maybe up to 30.

而使用LLM这些现代技术，能解决60%的问题，这意味着在通话数量基础上，你可以大幅减少客服需求。

With LLMs, with these new modern things, can solve 60%, which means on a number of call basis, you can significantly reduce your customer support.

所以我认为这其中存在显著价值。

So I think there was some meaningful value there.

我承认有时可能被夸大了，但如果这都不是LLM在企业市场的用武之地，那就没有什么是了。

I'm sure at times it's overstated, but if this isn't the market for LLMs in the enterprise, then nothing is.

我们先把这点放一边。

So let's leave that to one side.

真正的问题在于，从现在开始的数学计算是怎样的？

The real question I think is, how does the math work from here?

比如它值1亿，却以100亿交易，这需要什么条件？

Like it's a 100,000,000 and it's trading at 10,000,000,000 and what would it take?

我在脑子里简单算了一下，去年你们从1000万增长到了1亿，实现了10倍增长。

And I just kind of laid it out in my head, it's like you went from 10,000,000 to a 100,000,000 in the last year, so you 10x.

假设明年增长5倍就是5亿，后年3倍就是15亿，再下一年2倍就是30亿。

So say you 5x next year, that's 500, 3x the year after, 1 and a half billion, two x the year after, 3,000,000,000.

等等，我们什么时候需要达到30亿年经常性收入来着？慢点说？

Wait, when do we have to get to 3,000,000,000 ARR again, slow down?

然后

And

接着会增长到45亿再到50亿。

then you go four and a half billion and then five.

所以未来五年你们将经历10倍、5倍、3倍、2倍、50%的增长，然后略微放缓至20%，五年后就能达到50亿。

So over the next five years, you've got a 10x, 5x, 3x, 2x, 50% growth and then slow down a little to 20% growth, and you're at 5,000,000,000 in five years.

作为对比，目前Salesforce最大的云服务Service Cloud规模是80亿，你们要从这个市场拿下50亿。

For context, Service Cloud today, which is the largest cloud within Salesforce is 8,000,000,000, you're getting 5,000,000,000 out of that marketplace.

到那时如果按Salesforce的5-6倍估值倍数计算，你们价值250亿，也就是现值的2.5倍。

At that point, if you're valued at the Salesforce multiple of five or six times, you're worth 25,000,000,000, so it's a two and a half X.

这凸显了你需要承担多大的增长才能使这笔账算得通，对吧？

What it highlights is the amount of growth you have to underwrite to make the math work, right?

所以当你从宏观层面看，就像我们上周说的，我相信在处理超高速增长市场时，关键就在于总可寻址市场（TAM）。

So then you zoom out level, it's something we said last week that I believe, when you're dealing with hyper growth markets, it's all about the TAM.

我刚才做的粗略计算是，如果这家公司的增速超过所有其他企业软件公司，五年后它就能达到50亿美元营收，这通过了当前服务云80亿美元规模的合理性检验。

The crude math I just did, which is if this company grows faster than any other enterprise software companies, in five years it's doing $5,000,000,000 which passes the sanity test of Service Cloud doing eight today.

这让我觉得，如果你只是下一代软件玩家，很可能无法实现这种增长，因为那将需要从服务云那里一美元一美元地争夺市场份额。

What that says to me is if all you are is a next generation software play, you're probably not going to get that growth because that would have to come dollar for dollar from Service Cloud.

这又回到Jason的观点，唯一能让这笔账成立的方式是吞下大量人力成本——因为你吃掉的是劳动力市场，不是200亿美元的服务软件市场，而是每年2000亿美元的客服人员服务市场。

It goes back to what Jason says, the only way this math works is if you eat a huge slug of the labor and because you eat the labor, instead of the $20,000,000,000 software market for services, you eat the $200,000,000,000 a year services market for customer support agents.

它必须从1亿美元增长到10亿美元才能表面上看估值合理，对吧？

It's gotta go from a 100 millionaire to a billion to justify the valuation on its face, right?

必须从1亿增长到每年10亿。

It's gotta go a 100 to 1,000,000,000 a year.

在企业级市场你能以这种速度部署吗？相比之下自助服务呢？

Can you even deploy in the enterprise at that pace, right, versus self serve?

我们拭目以待。

We'll see.

我们看看他们能否从现在的一亿增长到明年年底的十亿。

We'll see if they go from a 100 now to a billion at the end of next year.

我们应该能做个更好的计算表。

Should we we can make a Cal sheet better.

有哪家公司从一亿增长到十亿的吗？

Has any company done a 100,000,000 to a billion?

嗯，我想有的。

Well, think yes.

Anthropic。

Anthropic.

轻而易举。

Easy.

Cursor和Anthropic。

Cursor and Anthropic.

但是怎么

But how

这些收入中有多少是自助服务产生的？

much of that revenue is self serve too?

要在企业环境中实现这样的收入规模，还需要一定量的人力资本投入。

There's a certain amount of human capital you also need to deliver this amount of revenue in the enterprise.

对吧？

Right?

是的。

Yeah.

同意。

Agreed.

因为让我告诉你，如果布雷特·泰勒走进一家财富500强公司说'给我1000万美元，我能替换你们一半的客服团队，相信我'，他离开时绝对能带着1000万美元的合同。

Because let me tell you, Bret Taylor walks into a Fortune 500 company and says, give me $10,000,000 and I can replace half your support team, trust me, he will leave the building with a $10,000,000 contract.

马克·贝尼奥夫在这方面也很擅长。

Marc Benioff was great at this too.

当年企业需求蓬勃发展时，他常常走访大公司。

He would go back in the day when enterprise ask was booming, he would go to big company.

他会问：‘哈利，你最大的问题是什么？’

He would say, what's your number one problem, Harry?

哈利会说：‘唉，我搞不定这个网站和电商业务。’马克就会回应：‘给我1000万美元，我帮你搞定。’

And Harry would say, well, I can't get this website and this ecommerce, and and Marc would say, just give me $10,000,000 and I will get this for you.

这就是当时的价码。

That was the price.

而他在这方面简直神乎其技。

And he was magical magical at at it.

虽然马克已经是史上最强——毕竟他当过Facebook的CTO和Salesforce的联席CEO——但布雷特会比他更出色。

It, and And Brett's Brett's gonna be even better, even though Mark's the best there ever is, because he was the CTO of Facebook and the co CEO of Salesforce.

没人能比布雷特的条件更完美了。

He's got there's no one got a better package than Brett.

他能拿到1000万美元的支票，但明年能拿到1亿并完成培训和FDEs交付吗？

So he can get 10,000,000 checks, but can he get a 100 next year and deliver training and FDEs?

这将会很有趣，值得关注。

It'll just be interesting to watch.

对吧？

Right?

需要交付的内容很多。

It's a lot to deliver.

杰森，我认为你说得非常对。

I think, Jason, you're exactly right.

从100美元做到10亿美元的API业务，与从1亿美元做到10亿美元的业务截然不同，后者每份1000万美元的合同都意味着美国大企业里巨大的变革管理。

Going from a 100 to a billion on API business is very different than going from a 100,000,000 to a billion in a business where each $10,000,000 contract is a huge amount of change management in a large corporation in America.

我认为这项技术在企业中的扩散速度，将决定这家公司能发展多快。

And I think the physics of diffusing this technology into the enterprise will be the rate limiter on how fast this company can grow.

限制因素不会是原始需求，不会是CEO的才能，也不会是产品本身，而真正在于：你是否真的能在一年内部署好1.1亿美元或1001万美元的客户，每个客户都有自己独特的方案、动态和集成需求。

It won't be raw demand, it won't be the talent of the CEO, it won't be the product, it will literally be, can you really roll out, what is it, a $110,000,000 customers or a 1,001 million dollars customers in a year where each one of them has their own special sauce, their own dynamics, their own integrations.

我认为即使需求巨大，企业软件的增长速度也存在客观规律，尤其是当需要处理如此大规模的变革管理时。

I think there's a physics to how fast you can grow in enterprise software even with huge demand when that's amount of change management has to happen.

所以我认为这将是这里的速率限制。

So I think that would be the rate limit here.

是的。

Yeah.

人们还没有向企业销售过。

People haven't sold to the enterprise.

他们可能没意识到我说的是多么真实，说真的，我认为布雷特·泰勒能向任何人卖出1000万美元。

They might not realize how true I mean, literally, I think if Bret Taylor could sell 10,000,000 to anybody.

同意。

Agree.

不仅仅是因为他聪明或魅力非凡。

It's not just that he's so smart or charismatic.

实际上是因为企业变革本身就复杂。

It's that literally enterprises are complex to change.

他们在寻找世界上能解决他们问题的最佳人选。

They're looking for the best person in the world that can solve their problems.

比如，我面临的最大问题是什么？

Like, what's my big problem?

如果你能显著提升我的KPI并提供支持，而Brett正是那个人选，我会给你那1000万美元。

And if you can dramatically increase the KPIs and support for me and Brett is the guy, I will give you the $10,000,000.

对你我来说，要拿到1000万美元会很困难。

For you and me, it'd be hard to get $10,000,000.

好吧。

Okay.

你我可能得从一份1万或10万美元的合同开始，但说真的，我不是在开玩笑。

Well, you and I might have to start with a 10 k contract or a 100 k, but literally, I'm not being facetious.

Brett能拿到多张1000万美元的支票。

Brett can get multiple $10,000,000 checks.

尽管听起来有点傻，但对我来说1亿美元几乎不算什么。

So as goofy as this sounds, the 100,000,000 almost to me isn't impressive.

这并不令人惊讶，因为以他的背景，我认为他能做到这一点。

It's not impressive because I think he can will it out of the with his background.

明年用一千名全职员工做到十亿，他必须想办法更进一步并部署实施。

A billion next year with a thousand FTEs somehow he's gotta go higher and deploy that.

对我来说，那简直是魔法。

That to me, it was magic.

就像，那将是魔法般的，因为他们会给他一亿美元。

Like, that will be magic because they will give him a $100,000,000.

他甚至可能走进一家财富50强巨头企业，从类似Palantir这样的公司拿到一亿美元的合同。

He might even be able to walk into one of the biggest fortune 50 companies and get a 100,000,000 contract from one like Palantir.

如果他承诺摘月亮，他们也会给他一亿美元。

If he promised the moon, they would give him a $100,000,000.

关于Zoomart的评论你说得对，这是随着时间推移观察大量企业软件后逐渐领悟到的一点。

The Zoomart comment you're right about, and it's something you kind of learn as you see a lot of enterprise software over time is this.

大公司和大领导有大问题，而他们能求助的人选寥寥无几。

Big companies and big leaders have big problems, and there's only a small number of people they can take their problems to.

如果回顾过去四五十年的大型科技公司，那些在企业浪潮中占据主导地位的，无一例外都向美国最大企业的CEO和CIO们传递了一个信息：如果你在这个领域遇到问题，我们会让它消失。

And if you look at large technology companies over the last forty or fifty years, the ones that become dominant in an enterprise wave are the ones who have, one way or the other, project to the CEOs and CIOs of the largest companies in America, if you have a problem in this sector, we will make it go away.

就像思科，他们最初从路由器起家，但在接下来的十到十五年间，他们几乎收购了所有你需要的网络设备。

Like Cisco, they started off with routers, but over the next ten or fifteen years, they bought pretty much anything you needed for the network.

我认为当钱伯斯执掌时，他们的大局价值主张是：'嘿，首席信息官先生，

And I think when Chambers was running his thing, the big picture value proposition was, Hey, Mr.

如果您有任何网络问题，我们会直接收购所需资源并帮您解决。'

CIO, if you have any networking problems, we'll just buy whatever we need and make it go away.

因此，您可以放心地给我5000万或1亿美元。

Therefore, you can safely give me 50 or $100,000,000.

当年的IBM也是如此。

IBM back in the day, same thing.

杰森你说得对，我认为在AI新世界里，像布拉德·泰勒这样有才华的人，可以完成C3尝试但失败的事——就像Palantir正在做的——直接走进CEO办公室说：'老兄，你告诉董事会2026年的两大首要任务之一是实现XYZ公司的AI转型，而我能帮你达成。'

And you're right, Jason, I think someone as talented as Brad Taylor in the new world of AI can do what I think C3 tried to do and failed, Palantir is doing, which is walk into a CEO and say, dude, you told your board that a top two initiative for 2026 is make x y z corporation AI enabled, and I can help you with that.

给我一张2000万美元的支票。

Give me a $20,000,000 check.

我完全同意。

I absolutely agree.

我认为确实存在这种现象，这令人沮丧，因为由普通人经营的小公司无法做到这一点。

I think that is a thing, and it's frustrating because all the little companies who are run by ordinary folks can't do that.

但这就是作为一位成功且经验丰富的企业领袖二十年的魔力所在。

But that's the magic of being a successful proven enterprise leader for twenty years.

所以我同意。

So I agree.

我认为你可以接触到这类业务，这是一个非常有力的定位。

I think you can access that kind of business, and it's a very powerful positioning.

我认为这完全取决于物理规律。

I think it's all about the physics.

对我来说，作为投资者必须做出的假设，可以追溯到罗里你多期节目前说过的一句至理名言——我每天都会想到这句话——AI领域的核心赌注是：我们能否见证从人力劳动到软件支出的转型？

To me, the assumption you have to make as an investor with this goes back to a very wise statement that you said, Rory, many shows ago, which I think to you literally every single day, which is the bet within AI is will we see the transition from human labor to software spend?

如果我们看到呼叫中心被取代，所有相关成本消失，那么这些资金就会流向这些玩家，届时我们将迎来千亿美元规模的Sierra。

And if we see that and we see call centers go and all of the call center costs go, then it will go to these players, and then we have $100,000,000,000 Sierra.

问题在于，正如所有显而易见真理面临的情况一样——估值扩张会重新创造风险。

The question is, as is always the case when something is obviously true, valuation expands to recreate risk.

换句话说，存在一个临界点，你知道的，你可以整天做这些交易。

In other words, there's a point at which, you know, you do any of these deals all day every day.

真正的问题在于，在100倍运行率收入的情况下，你能让数字成立吗？

The real question is at a 100 times run rate revenues, can you make the math work?

而我认为，这要困难得多。

And that, I think, is a lot harder.

你准备好回答一个让Rory不舒服的问题了吗？

Are you ready for an uncomfortable question for Rory?

等我们讨论Lovable时，他可以把问题抛回给我。

And he can throw it back on me when we move to Lovable.

好的。

Okay.

你刚才提到了Intercom。

You mentioned Intercom there.

Intercom的营收大约是3041亿美元，估值25亿到30亿美元；而Sierra营收1亿美元估值100亿美元，但两者的增长率天差地别。

Intercom doing, say, $304,100,000,000 in revenue at 2 and a half you don't need to comment on these numbers, but 2 and a half billion, 3,000,000,000 in price versus Sierra doing a 100,000,000 at a $10,000,000,000 price, but the growth rates are wildly different.

你更倾向于选择哪一种情况？

How do you think about which one you'd rather be in?

我认为两者都能赚大钱，而且我对我在那家公司的投资感到非常满意。

I think both can make a ton of money, and I think I'm very happy with my bet in that company.

我认为他们在向AI优先世界转型方面做得非常出色，很少有SaaS公司能做到这一点，而且他们AI产品的增长率实际上可以与Sierra相媲美，当然这没有透露公司应该选择公开的细节。

I think they've done an amazing job in a way that very few SaaS companies have of transitioning to an AI first world and the growth rate of their AI product is frankly comparable to Sierra's without kind of revealing details that the company should choose to reveal.

所以我并不感到后悔，事实上，我总体上觉得自己很明智，能以非常吸引人的价格获得不错的AI增长率，同时还有现有的SaaS业务。

So I'm not sitting here, in fact, I feel broadly speaking pretty smart as having got some very nice AI growth rate alongside an existing SaaS business at a very attractive price.

所以我感觉非常棒，谢谢。

So I feel pretty damn good here, thank you.

但随着市场细分，赢家不会只有一个。

But there will be more than one winner as this market segments.

我要从更高层面说，不是针对这次投资，但如今既要支持现有客户群，又要投资纯AI原生客户，确实是个负担。

I will say at a higher level, nothing against this bet, but man, today having to support an installed base versus getting to invest just in AI native customers, it's a drag.

确实是个负担。

It is a drag.

我刚刚还在和投资组合公司开会，他们AI业务收入4500万美元，增长率约100%。

And I was literally with the portfolio company, it's got 45,000,000 of AI revenue growing about a 100%.

好吧。

Okay.

听着，这虽然比不上富勒，但也相当不错了。

Now listen, that's not to love Fuller, but it's pretty good.

行。

Okay.

但他们还有5000万美元的传统业务收入，增长率为零。

And then it's got 50,000,000 of pre AI revenue growing zero.

所以问题在于——我认为Fin和Intercom可能也面临同样情况。

So what do you and and problem is they are and I think this is probably true at Fin and Intercom.

它们虽然不可避免地相互关联，但本质上并非同一款产品。

They are inexorably linked, but they're not the same product.

这确实挺糟心的。

It kinda sucks.

我们可能在一年后改变主意，说天啊，那个装机基础是利用AI的最佳资源。

We may change our mind in a year and say, my god, that installed base was the greatest thing ever to leverage for AI.

但目前看来，要取悦几千个AI前客户感觉就像个累赘。

But right man, it feels like a drag to have to have a couple thousand pre AI customers to make happy.

这感觉就是个拖累。

It feels like a drag.

我郑重记录对此的强烈反对，我不同意。

Recording my profound opposition to that, I disagree.

实际上我认为这些市场的每个现有企业都有一个巨大优势和一个巨大劣势。

I actually think every incumbent in these markets has one huge advantage and one huge disadvantage.

我要明确地说，我认为优势在于客户基础、数据结构以及对你现有公司业务数据的访问权限。

And I'm gonna say it clearly, I think the advantage is the customer base and the data structures and the access to the data on your existing on what the company is doing.

因为我认为对于销售或客户支持这类工作，事实是在未来很长时间内都会是组合套餐——部分自动化加上部分人工客服，能够在这两者间无缝切换具有巨大优势。

Because I think for something like sales or customer support, the truth is it's going to be a combo package for a long time to come, some automation and then some human agents, and being able to move seamlessly between them both has huge advantages.

我认为每个大型现有企业面临的劣势是他们总是作茧自缚。

I think the disadvantage that every large existing incumbent has is they can't get out of their own fricking way.

因此他们无法在拥抱AI技术的同时充分利用现有资产。

And therefore they can't leverage the asset they have while at the same time embracing the AI technology.

我认为Intercom在这方面做得非常好。

I think that's frankly something that Intercom did really well.

嗯，我觉得情况比那更微妙。

Well, think it's more subtle than that.

我明白你的观点。

I hear your point.

理论上这是个优势。

In theory, it's an asset.

对吧？

Right?

理论上拥有现有客户群及其数据总比新AI公司指望获得这些数据要好。

In theory, it's great to have an installed base, all of its data, rather than an AI company, new AI company hoping to have that data, in theory.

问题不在这里。

The problem isn't that.

这确实是一项资产，所以我可能在十二个月后改变主意。

That is an asset, and that's why I might change my mind in twelve months.

问题在于所有的技术债务、功能债务，你向那几千名客户承诺过的所有功能——那些对你的闪亮新AI功能毫不在乎的五千名客户。

The problem is all the technical debt, all the feature debt, all the features you've promised those thousand customers, those 5,000 customers that don't give a rat's ass about your shiny new AI feature.

关键在于你必须维持他们的满意度，不能让他们流失，这会消耗你大部分的工程开发时间。

The fact you have to keep them happy, okay, and not let them deteriorate, that can consume the majority of your engineering development time.

确实如此。

It really can.

我很想听听欧文斯的看法。

I'd love to hear Owens.

我想听听他的真实想法。

I'd like to hear his honest thoughts.

我猜Intercom的做法——你应该比我更清楚。

I suspect what they did at Intercom, you would know better than me.

我猜他们做了必须做的事：因为工程师数量有限，他们无法兼顾太多。

I suspect they did what you have to do, which is they don't get so much because there's only so many engineers.

一百、五百、五十。

A hundred, five hundred, a 50.

现有客户会耗尽你所有的故事点和工程时间。

Your existing customers can consume all of your story points and all of your engineering time.

全部耗尽。

All of it.

你积压了十年的技术债。

You have ten years of debt.

如果你是一家新AI公司，你还没有数据积累。

And if you're a new AI company, you don't have the data.

这是个巨大的劣势。

It's a huge negative.

别误会我的意思。

Don't get me wrong.

但当风投们喋喋不休说原生AI客户更好时，那种吹捧简直让我作呕。

But the ability to run when VCs talk about endlessly about how native AI customers are better, I wanna gag with a spoon.

有时候它们更好，因为它们是真正为聪明人设计的。

Sometimes they're better because they're for really smart people.

有时候它们更好，仅仅是因为不需要支持上千个抱怨的客户。

Sometimes they're just better because they don't have a thousand complaining customers to support.

我只是说现在我觉得这是个负担，但一年后我可能会认为这是最棒的事情。

I'm just saying right now I feel like it's a liability, but in a year I may think it's the greatest thing.

比如看AgentForce，现在还处于早期阶段，对吧？

Like looking at AgentForce, right now it's early, right?

但Salesforce可能会成为下一个谷歌。

But Salesforce could be the next Google.

可能在18到24个月内实现。

It could be in eighteen to twenty four months.

我们可能会感叹，天啊，AgentForce太成功了。

We could be like, oh my god, AgentForce crushed it.

我们这些初创公司都不需要了，因为AgentForce做得太好了。

We all these startups, we didn't need them because AgentForce is so good.

我知道这听起来很疯狂，对吧？

I know sounds it crazy, right?

我可能会改变主意，但目前来看，我投资的那些拥有庞大用户基础的初创公司，感觉就像穿着灌了铅的鞋子一样沉重。

I may change my mind, but right now, the startups I've invested in with large install bases, feel like it's a frigging cement shoes.

我理解你的意思，但我还是要表示不同意。

I understand what you're saying, but again, I would disagree.

我认为最大的优势在于，你提到你有一家公司4500万美元的AI收入，而大量客户却完全不在乎这部分AI收入，这句话很能说明问题。

I think that the great advantage like, you made a comment on you have a company $45,000,000 of AI revenue and then a whole bunch of customers who don't care about the AI revenue at all, and that was the revealing sentence.

我不认为这是你在某些案例中看到的情况——当客户支持部门的负责人明显都知道他们将采用AI技术时，对吧？

I don't think that's the experience you're seeing in some of these cases where when you have an obvious anyone who is running a customer support organization knows they're going to be embracing AI, right?

而一家成功的公司介入并表示：'我们会帮助你完成这个转型'。

And a successful company goes in and says, we're gonna help you on that journey.

我无法想象一个客户支持部门的负责人在运营业务时会说：'不，我完全不会采用任何AI技术'。

It's unimaginable to me to think of a customer support executive who's running their business and saying, no, I'm not gonna do any of this AI stuff.

我就是特别喜欢花钱雇菲律宾人接电话。

I just really like paying people in The Philippines to answer phones.

但债务是真实存在的，对吧？

But the debt is is real, right?

而你能为他们做的也是实实在在的。

And how much you can do for them is real.

我是说，我为G2采访了Zendesk的CEO，每个Zendesk客户都配备了AI功能。

I mean, I did an interview for G2 with the CEO of Zendesk and every customer has AI at Zendesk.

我不是Zendesk专家，但他非常直率，因为他们现在是私有化企业。

I'm not an expert in Zendesk, but he was very direct because they're private now is right.

但Zendesk的基础版本（含AI功能）大约能实现20%的自动化。

But the base version of Zendesk, which you get with AI, it's about 20% automated.

这差不多是不经训练能达到的最佳效果了。

That's about as well as you can do without training.

要想达到Fin这种体验水平，或是Decagon、Sierra那样的效果，你得花两三周用FDEs等数据训练系统。

And to get this fin level of experience or better or Decagon or Sierra, you gotta train the thing for or three weeks with FDEs and the rest.

所以问题不在于双方是否都能受益。

And so, it's not that they both don't benefit.

只是我认为它既是资产也是阻碍。

It's just I think it's an asset and a hindrance.

我认为如果我们看看一些增长最快的公司，它们并没有沉重的负担。

I think if we look at some of our fastest growing companies, they don't have they don't have cement shoes.

我不是说Intercom是个例外，但他们确实没有沉重的负担，对吧？

I'm not saying, Intercom is the exception of rule, but they just don't have cement shoes, right?

我同意。

I agree.

坦白说，我认为这需要非常出色的管理，以及某种——天啊，真不敢相信——

I think it takes, frankly, I think really excellent management and a certain going back to the, oh God, I can't believe it.

回到那个‘战争模式’的评论（虽然我依然不太喜欢这个说法），我正在改变看法。

Going back to the war mode comment, which again, didn't love, I'm changing my mind.

这需要强大的产品领导力和清晰的愿景才能实现。

It takes a lot of lead product leadership and clarity of vision to make it happen.

你说得对。

You're right.

我也这么认为，是的。

And I think Yeah.

你看到贝尼奥夫正在努力解决这个问题。

You're seeing in Benioff trying to figure out how to do that.

对吧？

Right?

组织规模越大，难度就越高。

It's the bigger the organization, the harder it is.

我认为Intercom团队在这方面做得非常出色。

I think the team at Intercom has done a really nice job of doing that.

对吧？

Right?

这很难。

It's hard.

好吧，让我换个说法。

Well, let put me it differently.

Intercom做到了。

Intercom did it.

让我们明确一点，Owen做到了，这很棒。

Let's stipulate Owen did it, and it's great.

这就是为什么我认为我们大多数独角兽企业会失败，因为你必须和Owen一样优秀甚至更出色才能完成这种转型。

This is why I think most of our unicorns will fail because you gotta be as good or better than Owen to make this transition.

它们大多数都会失败。

Most of them are gonna fail.

当我看到这么多B2B领域的独角兽企业时，发现要兼顾两者实在太难了。

When I look at so many of these b to b unicorns, it's so hard to do both.

这太难了。

It's so hard.

Rory，我的年度经常性收入已经达到2亿美元了。

I've got 200,000,000 ARR, Rory.

我正以20%的速度增长。

I'm growing 20%.

我知道我必须做AI，但我几乎无法让团队跟上我的步伐。

I know I have to do AI, but I just I can barely get the team to do what I'm doing.

除非你具备这种侵略性、战时状态和卓越的CEO能力，否则根本无法兼顾两者。

Unless you have this aggressiveness, war mode, great CEO, there's just no way you can do both.

普通的私募公司一年能发布一个版本就不错了。

There's no your typical PE company can barely get a release out a year.

我同意。

I agree.

我认为有两个因素阻碍了你。

I think there's two things that stop you.

就像我们往常交谈时那样，杰森，你强调一个方面，我强调另一个，这很正常。

I think as is usual, when you and I talk, Jason, you emphasize one and I emphasize the other, which is okay.

通常面临的两大挑战，我想称之为管理挑战——这正是你强调的。

The two challenges you typically have is I want to call it the management challenge, which is what you emphasize.

哦，你既要处理所有这些旧事务，又要开展新业务。

Oh, you've got all this old stuff, you've to do this news.

两者兼顾很难，所以你只能做个管理者。

It's hard do both, so you've just got to be a manager.

但我要说的第二点是，前身问题在于两者之间必须有明显的关联。

But then the second thing I would argue, the predecessor thing is, there's got to be some obvious linkage between the two.

如果你在旧世界有个做X的SaaS公司，却没有明显的X+AI对应物，你就孤立无援了，然后决定在SaaS领域做新东西，那就完蛋了。

If you've got this SaaS company in the old world that does X, and there's no obvious X plus AI equivalent, you're just on an island and then deciding, oh, I'm going to do something new in SaaS, you're screwed.

如果你仅有的优势是'现有客户未来可能想买AI产品而现在买SaaS'，那你也完蛋了。

If all you have is I have customers who might want to buy AI in the future and they buy SaaS today, you're screwed.

唯一有可能成功的情况是两者天然契合。

The only time you have even a chance to do it is when both kind of organically go together.

我再给你举个这方面的例子。

I'll give you another example of that.

我认为Gong是家非常有趣的公司，在GPT之前做AI通话录音，他们现在正努力在这领域发展，并在基础上添加所有LLM功能。

I think Gong is a very interesting company that was pre GPT AI, call recording, and I think they're doing a decent job of navigating that terrain and adding all the LLM stuff on top.

但这并非因为他们是管理天才（也许确实是），更根本的原因在于从现状到目标存在清晰路径。

But it's not because they're management geniuses, maybe they are, it's more fundamentally because there was an obvious path from here to there.

在其他SaaS业务没有对应AI应用的领域，你就如同孤岛般孤立无援，确实无计可施。

In some other areas where the SaaS thing doesn't have an AI equivalent, you're on an island and then you're right, there's nothing you can do.

就算你是管理界的天选之子，也照样完蛋。

You could be God's gift to management and you're screwed.

我尽量不点名举例，但确实有些项目让人一看就觉得——宝贝，在这个AI优先的世界里，我们可能不需要你。

And I'm trying to think of examples of that without throwing anyone under the bus, but there are deals where you look and you just go, I'm not sure we need you in this AI first world, baby.

这可不是什么好事。

And that's not good.

你提到增长最快的公司及其与AI原生客户的关联性。

You said the fastest growing companies and how it's tied to kind of AI native customers.

与此紧密相关的是，比如Sierra达到收入里程碑，这能支撑其估值吗？

One that's really tied to this, like, Sierra hit revenue milestone, does it justify price?

Lovable实现了2亿美元年度经常性收入，短短四个月就翻了一番。

Lovable hit 200,000,000 in ARR, two x what it was in just four months.

传闻其新一轮融资估值达63亿美元，融资规模也是63亿美元。

Rumored $6,300,000,000 round in terms of new price valuation being $6,300,000,000.

当你看到增长率达到2亿美元时，这合理吗？

Is that justified when you see the growth rate hitting $200,000,000

Lovable的年度经常性收入？

in ARR for Lovable?

我不知道，但我要告诉那些批评者，这就像是一个古老的Saster教训，我真心认为在AI时代我们必须这么做。

I don't know, but I will tell you for folks that are critics, this is like an old Saster lesson I really think we have to do in the age of AI.

你必须细分你的客户群。

You've got to segment your customer base.

我的意思是，我假设在低端市场，流失率接近50%或更高，明白吗？

So what I mean is I assume at the low end, the churn approaches 50% or higher, Okay?

就是Venmo刚刚在音乐领域做的那件事。

The one Venmo just did in music.

叫什么来着？

What's it called?

我们使用

We use

它。

it.

Suno。

Suno.

对吧？

Right?

他们刚刚表示，在该级别上他们一年仅能留住20%的客户。

They just said they have they only retain 20% of their customers in the year at the level.

但我刚在Replic做了一个演示，展示了我在全员大会上的所有应用。

But I just did a presentation at Replic that showed all my apps at all hands.

我在场时，他们敲定了一笔相当可观的七位数交易。所以关键是，这种多年期七位数交易绝不可能出现客户流失。

And when I was there, they closed a pretty large 7 figure deal when I was So the point is, there's no way that's gonna churn as a multi year 7 figure deal.

从数学上讲，任何少于几年的情况都是如此，对吧？

Anything less than a couple years mathematically, is it?

我几乎想做的就是将Lovable和Replit进行细分，并说明：听着，这里的高端客户可能拥有140%到160%的重复收入率。

What I would almost wanna do is take Lovable and Replit and segment them and say, listen, there's a high end here that's probably got a 140 to a 160% RRR.

我对此相当有信心。

I'm pretty confident of it.

中间层按惯例来说，我猜加上追加销售等措施，留存率正接近100%。

There's a classic mid pack that my guess is with up sales and stuff is is approaching a 100% retention.

而底层部分的表现比我们习惯看到的要差。

And then there's a the part of the bottom is worse than we're used to.

可能只有40%或30%的留存率，但这并非没有先例。

It may have 40% retention or 30%, but that's not unprecedented.

几乎每个移动订阅应用——我们从RevenueCat得知——留存率都在20%到30%左右。

Like almost every mobile subscription app has like, we know from RevenueCat is like 20 or 30% retention.

所以我几乎想不仅设定柱状图标准，还要在底部划一条黑线，直接将其标记为营销支出。

So I just think I would almost wanna put a, not only set bar chart mind, but almost draw a black line through the bottom if I were to and just call that spend.

底层纯粹是营销人员。

Like that bottom is just marketers.

他们只是负责扩大声量，遍布社交媒体，我打赌未来两年内——对吧？

They're just folks to get the word out and they're all over social media and I'm betting in the the next two, right?

100%和160%的净收入留存率。

The 100% and the 160% NRR.

即使这只是Lovable公司的一半，我认为这也是个不错的赌注，对吧？

And even if that's half of lovable's, it's, I think it's a decent bet, right?

随着这些公司的发展，他们的净收入留存率会上升，流失率会下降，这纯粹是经典的企业级客户特性——大客户往往粘性更强。

As these guys grow, their NRR will go up and the churn will go down just because for classic enterprise reasons, those big customers are gonna be stickier.

他们就是会更有粘性。

They're just gonna be stickier.

我一直在考虑现金的机会成本，以及如何配置资金才能实现收益最大化。

I'm just always thinking about the opportunity cost of cash and where I put my money to make the most money.

为了

To

罗里，你是个交易员。

Rory's You're trader.

让罗里闭嘴。

To Rory's shut up.