全栈构建者与超杠杆化通才的崛起：微软首席执行官萨蒂亚·纳德拉

这个世界会对那些说‘相信我们’的科技公司和科技产品充满怀疑。

The world is gonna be very skeptical of tech and tech companies that say, trust us.

我们已经掌握了。

We've got it.

未来一定会非常辉煌。

The future is gonna be glorious.

这次你必须实实在在地带来好处，因为这件事太重要了。

You kinda have to deliver tangible benefits because it's too important this time around.

经济的比重这么大，不可能不这么办。

It's too much of the economy for it not to be the case.

真正的雄心，是把不可能变成可能。

True ambition is about making the impossible possible.

我从负责管理Azure网络的那些人身上获得了很大启发。

I take great inspiration from sort of the people who were managing the Azure network.

过去十五个月里，我们新增的Azure容量，超过了前十五年的总和。

We built in the last fifteen months more Azure capacity than we built in the first fifteen years.

我的意思是，这简直太疯狂了。

I mean, it's crazy.

太不可思议了。

Wild.

我们的任务不是去搞 Azure 的网络运维。

Our job is not to do Azure networking.

我们的任务是打造一个能自动完成 Azure 网络运维的智能系统。

Our job is to build the agentic system that does Azure networking.

对吧？

Right?

获取信息、自我学习、持续更新的方式，已经彻底变了。

The way to get to information, way to educate yourself, way to continuously keep yourself updated has changed so much.

也许下一个伟大的创业公司，会是那些重新设计大学教育模式的人——创造一种全新的教学方式，帮助人们完成课程，找到真正有价值的经济机会。

Maybe the next big start up could be someone who builds a new university, a new pedagogy even of how to get someone to go through a curriculum and find economic opportunity that's highly valuable.

让我们欢迎 Swyx、Saragoa、Alad Gil，以及微软董事长兼首席执行官萨提亚·纳德拉。

Please welcome swyx saragoa alad Gil and chairman and chief executive officer of Microsoft Satya Nadella.

你好，萨兰。

Hello, Saran.

能来到这里我特别兴奋。

I'm so excited to be here.

欢迎收听《NoPriars in Lane Space》与萨提亚·纳德拉的跨界对话。

Welcome to a crossover episode of NoPriars in Lane Space with Satya Nadella.

恭喜你完成了这么了不起的成果。

Congratulations on an amazing build.

不用客气。

No.

非常感谢，能和你们两位一起聊天真好。

Thank you so much, and it's great to be with both of you.

我经常听你们两个的播客。

I listen to both of you or both the podcast all the time.

能上这个节目真的很棒。

It's great to be on it.

非常感谢。

Thank you so much.

你一早上都在讲微软旗下这些令人惊叹的发布，持续了大概三个小时吧。

So you're just talking about these amazing announcements from across the Microsoft estate all morning for, I think, three hours.

你觉得最重要的反思或收获是什么？

What's the most important reflection or takeaway you have?

对我而言，最重要的可能是要把这件事看作一个生态系统，而不是单一模型或单一平台，对吧？

I'd say there are Perhaps the biggest one for me is let's sort of conceptualize this more as an ecosystem play as opposed to a single model or even a single platform, right?

我从小在微软长大，亲历过四次重大的平台变革，所以我一直认为，平台的价值不在于它自身能掌控什么，而在于它能为外界创造多少价值。

I mean, at least for me, having grown up at Microsoft, having seen whatever four major platform shifts, I sort of fall into that camp where a platform is defined by fundamentally its ability to create more value about the platform versus what's captured in the platform.

所以如果你看看现在发生的事，今天早上的主题演讲其实讲的是：无论是一家AI原生公司，还是一家传统企业，都能以平等的身份参与进来，推出自己打造的AI产品。

And so if you view what's happening right now, I think this morning's keynote was how can any company, whether it's an AI native company or a traditional enterprise company, participate as a first class participant where they can point to AI they create, right?

他们当然也会用别人的AI。

It's not that they don't use other people's AI.

这毫无疑问。

Of course, they will.

但对我来说，关键路径是什么？

But to me, what's the path?

具体该怎么做？

What's the recipe?

我该怎么着手？

How do I do it?

整个技术架构是怎样的？

What does the stack look like?

工具链又是什么样子？

What does the tooling look like?

到底什么才是有价值的？

What is valuable?

你该怎么实现这一点？

How do you do that?

就这么简单。

That's it.

这本来就是我们的职责所在。

That's sort of our job to do.

对。

Yeah.

生态系统战略非常复杂，对吧？

Ecosystem strategy is very complicated, right?

因为你得自己开发一些组件，和别人合作一些组件，还要维护它们。

Because you end up building certain components, partnering for certain components, supporting them.

你刚发布了这一整套模型。

You just announced this big suite of models.

能跟我们说说微软的训练策略吗？

Like, tell us a little bit about the training strategy for Microsoft.

是的。

Yeah.

我们做MAI模型的目标，正如穆斯塔法提到的，首先是打造一个优秀的技术传承。

So so the thing that we wanted to do with the MAI models was to build, and as Mustafa talked about, first of all, a great lineage, right?

从预训练开始，确保数据质量过硬，把各种变量都测试一遍，做到位。

Starting with pre training, with very good data quality, doing all the abolations, making sure.

因为现在要构建一个干净的模型谱系越来越难了，外面的东西太多，你必须把无关的全都剔除掉，才能做出一个出色的预训练模型。

Because in some sense, it's becoming even harder to build a clean lineage model just because there's so much stuff out there that you truly need to ablate out to be able to have a fantastic pre trained model.

事实上，很多开源权重模型的问题就在于，它们在一两个基准上表现不错，但实际应用中却不行。

In fact, that's one of the challenges of a lot of the open weight models is they look great on one benchmark or two, but they're not great on practice.

所以这也是为什么我们的FDE团队对这些MAI模型这么兴奋——一个只有50亿参数的小模型，到底是怎么做到这么强的？

So that's why, in fact, even in our FDEs are pretty gone really excited about these MAI models because how the heck can a small 5B model hill climb?

这其实又回到了我认为最关键的一点：找到那个认知核心。

And it goes back a little bit to what I think is ultimately the key thing to do, which is try to pursue finding that cognitive core.

对我来说，先从一个干净的谱系出发，再打造一套机制，让企业不仅能用它做通用模型，还能基于它构建自己的专业模型，围绕它搭建一套持续优化的框架。

So to me, starting with a clean lineage, then creating that ability for companies to be able to use this, right, not just as a generalist, but to create their own specialist by building this hill climbing scaffold around it.

对吧？

Right?

所以关键不只是模型本身，而是围绕它的一整套优化机制。

So it's not just the model, but you have a hill climb scaffold around it.

接下来你就要开始构建你的RLE了。

Then you will start building your RLE.

你得开始收集这些运行轨迹。

You will start collecting the traces.

最重要的是，你要有自己的私有评估体系，因为现在市面上那些评估标准虽然不错，也挺有意思，但其实没那么关键，毕竟都能被刷到满分。

Most importantly, you'll have private evals because we know all the evals out there are good, interesting, but they're not really that critical at this point because they all can be maxed.

所以关键在于，每家公司都该有自己的私有评估标准。

And so the point is each company will have its own private eval.

围绕我们的模型打造的这套端到端平台，这才是我觉得真正有意思的地方。

And so that end to end platform story around our models is sort of what I think is interesting.

还有一件事，莎拉，既然你提到了，我觉得确实出现了一个新的前沿方向。

And then the one other thing, Sarah, since you brought that up, is I do feel there's a new frontier.

大家总在说‘前沿’，问你是不是在前沿上做事。

Like, people talk about the frontier and are you operating at the frontier?

有意思的是，如果你加入一点时间维度，就能用上——比如我们展示的Land O'Lakes演示，就挺出色的。

Interestingly enough, if you add a little temporality to it, you can use, let's say And in fact, the Land O'Lakes demo we showed was pretty cool.

我们当时用了什么来着，GPT-5 0.5，对吧？

We used whatever, GPT-five 0.5, right?

然后你收集了一大堆推理轨迹，再用一个50亿参数的推理模型，效果反而更好了。

Then you collected a bunch of traces, and then you took a 5B reasoning model and achieved higher.

所以这又是‘站在前沿’真正意味着什么的另一个方面。

So that is another aspect of what it means to appear I mean, operate at the frontier.

是的。

Yeah.

我首先得恭喜你，短短两年内在微软内部建起了一个前沿神经网络实验室。

I I think I first of all, have to congratulate you on basically building a frontier neural lab inside of Microsoft in two years.

我想知道，你现在推行的这一整套AI战略，

I'm wondering, you know, you have all this AI strategy that you're rolling out.

如果能回到两年前，你会对自己说什么不一样的建议？

I'm wondering, what do you know now that you wish you would tell yourself two years ago?

等等。

Wait.

两三年前。

Two or three years ago.

和詹森的合作有三年了，MAI也有两年了。

Three years for the Jensen partnership, two years for MAI.

是的。

Yeah.

说实话，我经常回想起当初是怎么开始的，就是看到那个滑动损失的论文时特别兴奋，后来OpenAI那边来找我们合作，他们说：嘿。

I mean, I think the the thing when that I reflect quite a bit, right, which is sort of obviously, I got into all this when I got excited by the the skating loss paper and, you know, when, you know, even the OpenAI partnership came about when those folks said, hey.

他们打算投入大量计算资源和Transformer模型，这确实帮了大忙。

We're gonna really throw a lot of computer transformers, and they've helped.

我总忍不住感叹，这些模型的能力真的在不断攀升。

The thing that I always go back and say, wow, these things do have capability that they're climbing up.

这么说可能有点粗浅，但智力确实跟计算量的对数差不多成正比。

I mean, you know, this crude way of saying it is intelligence is log of compute kind of works.

不过现在回头看，我们可能低估了把这些技术真正落地、在现实世界中产生价值的复杂性。

Now, what I think we underestimated perhaps is the real world complexity of deploying these so that they actually deliver the value in the real world.

对吧？

Right?

所以，不管用什么基准来衡量结果，都挺有意思、也很重要。

So the outcomes as measured by any benchmark is interesting, important.

但真正的检验标准，是外面的人能做出只有他们自己才觉得有价值的独特事情。

But the true eval is when people out there are able to do unique things that they only can value.

而且这事儿是能清楚衡量的。

And it's very measurable.

没错。

Right.

我其实希望我们行业能更清醒地意识到这一点。

That I wish we had sort of even had more in our consciousness, right, which is as an industry.

因为现在，当人们说‘我不想受限于token上限’，这其实反映出我们还没意识到，作为行业，我们每一步都在用token创造价值。

Because right now, I think when people say, wow, I don't want a token max, it's an artifact of us not having thought ourselves as an industry that we are using tokens to create value every step of the way.

所以，我真希望我们早点想明白这点，不过好在我们现在总算走到了这一步。

So I think that's kind of what I wish we had gotten there, but I'm glad we are here.

您见过哪些使用场景为客户创造了最大的价值？

What are some of the use cases that you've seen that have created the most value for your customers?

我知道大家经常聊代码，我也觉得这确实带来了非常广泛的影响。

Because I know that people talk a lot about code, and I think it's pretty clear that that's something that's having very large scale impact.

除了代码之外，您还发现客户在哪些其他领域也受益明显？

Are there other areas that you find in common that your customers are really benefiting?

是的。

Yeah.

我觉得确实如此。

I think yeah.

说到这点，编程现在确实做得很好了，但有意思的是，Elad，我们居然还得讨论编程这件事，对吧？

To your point, obviously, coding is now got But it's interesting, by the way, Elad, to even talk about the coding, right?

因为编程太成功了，我们现在反而得重新设计开发环境。

Which is coding has worked so well that we now have to rebuild the IDE, right?

想想我们当初推出的东西，简直有点不可思议。

I mean, it's kind of nuts to see what we launched.

天啊，我这里有上百个智能代理会话。

It's like, oh my god, I have these 100 agent sessions.

认知负担太重了，作为人根本扛不住，现在必须 redesign 一个新界面。

Cognitive load, it transfers back to me as a human is so excessive that now I need a new UI.

顺便说一句，只靠聊天记录作为唯一载体根本行不通。

Oh, by the way, the chat as the only artifact is also impossible.

所以，我们需要一个画布。

So, that's why we need a canvas.

这其实挺有意思的，关于软件和界面到底该用在哪儿。

So, it's kind of interesting for all the things about where is software needed or where is UI needed.

哪怕在完全由智能代理主导的世界里，你还是需要这种东西。

You kind of need that even for code, right, in a fully agentic world.

不过话说回来，我们现在开始看到一些实际案例，比如之前展示的自动巡航功能，还有 clause 那个场景就挺典型。

But that said, one of the things that we are starting to see, we start seeing that co work, but even some of the work we showed with auto autopilot, right on what you see with clause is a good one.

因为很多人力其实都在做衔接的工作，如果你能用长期运行、稳定的智能代理来辅助这些工作，那像编程一样，这种需要判断和协调的任务也能被放大倍增。

Because if you sort of think about a lot of human capital is doing the glue work, right, if you now can augment that with tokensagents that are long running, durable, right, then your ability to scale, even what is still judgment and glue work, gets amplified like coding does.

我相信六个月后，我们都会惊叹：天啊，真没想到。

So you can I'm positive that six months from now, we'll all be saying, oh, wow.

整晚都有好多自动代理在替我做事，它们有我授权的权限，干了一大堆活。

Like all through the night, there was a bunch of stuff that all these autopilots that I have working on my behalf with my delegated authority, so to speak.

对吧？

Right?

就拿我的身份来说，它们已经帮我完成不少任务了。

I can sort of, given even my identity, did a bunch of work.

然后我肯定得靠新的辅助系统来问：你到底干了些什么？

Then, of course, I'll need my new ADE to say, what did you do?

我是不是做了这些工作？

Like, did I do this work?

诸如此类的问题。

And so on.

所以我觉得，工作流的压缩、任务的闭环，真正的价值就在这里面。

So I think that that's where compressing of workflows, completing of tasks, that's where I think a lot of the value gets created.

你提了一个特别有意思的观点，就是实际写代码的是那个智能体，而它周围还有一整套支撑系统。

I think you raised a really interesting point, which is there's the actual agent that's doing the code, then there's a harness around it.

这个支撑系统就是环境、上下文，是你作为开发者为代码智能体搭建的所有配套。

And that's the environment, that's the context, that's everything you're setting up as a developer around actually a coding agent.

对企业来说，这个支撑系统又该是什么样子？

What is the harness for the enterprise?

在更广泛的效率提升场景里，有没有类似的概念？你怎么看这个想法的普遍适用性？

Is there an equivalent concept for broader productivity work, or how do you think about that concept sort of generalized?

没错。

That's right.

某种程度上，你希望这个支撑系统能统一管理模型、数据和工具，让这三者形成一个闭环。

So, in some sense, you kind of want the harness to define the models, data, and the tools, and so that you have a loop across those three.

我们首先想确保的是，不管我们开发的是哪个产品——比如 GitHub Copilot、安全 Copilot、Emdash 展示的那些功能，甚至科学发现工具——都不重要。

And so, what we are trying to, first of all, make sure is each of our products that we build, right, whether it's GitHub Copilot or the security copilot, the stuff we showed with Emdash, or even the discovery for science, it doesn't matter.

所有这些产品本质上都是多模态的支撑平台，都支持工具调用，这样才能逐步展示功能，同时节省 token。

All of them are multimodal harnesses with tools access so that you can do this progressive disclosure of tools even so that they're token efficient.

然后你给它提供非常丰富的上下文。

And then you're feeding it with very rich context.

因为过去两年我们学到的另一个深刻教训是：天啊，要让计划高效执行，你得花多大功夫去准备上下文层，这才是关键所在。

Because that's sort of the other hard lesson we have learned in the last two years is, oh my God, the amount of work you need to do to prep the context layer such that your plan can execute in the most efficient way is where the magic is.

所以对我们来说，我们有一个 GitHub 框架，基本上用在了所有产品里。

So we have, in our case, we have the GitHub harness, which essentially we're using across all our products.

它在 Foundry 里也是可用的。

It's available in Foundry.

而且我们是开放的，你可以用你的 Lama 框架，或者任何开源框架，甚至你自己搭建的框架，用你的工具、多个模型和上下文来训练。

And we're open, like you can use your Lama harness, whatever, or you can use any open harness or any harness of yours and train with your tools and multiple models and your context.

这就是我们的核心主张。

And so that's the pitch.

因为现在很多人在讨论的是：嘿，如果你把框架、工具和模型一起训练，就能得到评估结果。

Because right now, a lot of dialogue is, hey, if I train the harness plus tools and the model together, you get evals.

而我们正在验证的是，最好的例子就是我们做 Emdash 时的做法。

And what we are proving out is, and the best example of that, is what we did with Emdash, right?

因为上线后，它发现了 Mythos 没有发现的漏洞和问题。

Because when it launched, it found bugs or vulnerabilities that were not found by mythos.

所以我认为，这已经证明了，一个多模态的工具框架在实际应用中确实可以表现得更出色。

And so there is existence proof, I would claim, that you can have a multimodal harness that can, in fact, be more performant in the real world.

独立前沿实验室的训练理念其实很简单：我们会拥有这些模型，做 API 商业，服务企业和初创公司，但真正带来大部分收入的，还是我们自己的产品——无论是生产力工具、代码助手还是搜索功能。

So premise behind the training at the independent Frontier Labs is really, you know, we're gonna have these models, We'll have an API business and we'll support enterprises and startups, but a first party product, be it productivity or code or search, drives the majority of revenue.

这和你描述的微软生态系统的价值逻辑不太一样，我觉得。

That's a different value equation than you're describing, I think, with the Microsoft ecosystem.

如果是这样的话，你能不能确认一下？毕竟你们既有自己的产品，也有赋能型产品。

If if that's the case, tell me if it's the case, because obviously you have first party products and you have enablement products.

在这种环境下，开发者的角色是什么？他们需要具备哪些关键技能？又该如何从中获取价值？

What is the role of the develop like, what's gonna be hard and the set of skills and the value capture the developer has in that world?

对。

Yeah.

我认为，一个平台构建者如果非常成功，完全也可以同时推出自己的第一方产品。

So I think that there's always gonna be the case that someone who is super successful as a platform builder can also have first party products.

Windows 就是这样。

It was true with Windows.

我们和其他公司做 SaaS 和云服务的时候，情况也一样。

It was true with the SaaS side and the cloud side as well with us and others and so on.

但关键在于，这不该成为别人也取得同样成功的障碍，对吧？

But the thing that is, is it should not be a limiter to other people achieving that same success, Right?

我认为这才是核心区别：这次围绕智能的网络效应之所以强大，是因为模型从数据中学习，但不需要海量数据。

That, I think, is the core difference, which is the network effects this time around, around intelligence are such because they learn from data and not really lots of data.

你只需要看几个样本，就能理解某样东西的新颖之处。

It's just a few samples that you have to see to understand what's novel about something.

所以问题的关键变成了如何保护这些优势。

So that's why the game becomes how to protect.

所以我才说，每家公司做私有评估，可能才是最大的知识产权。

So that's why I would say every company having private evals may be the biggest IP.

对吧？

Right?

我一直在想这个问题。

I think about it.

比如说，什么样的私有评估能让你即使使用前沿模型，也能不断优化，又不会留下痕迹？这可能是最重要的知识产权之一。

Like, what's that private eval that you can then use, even a frontier model, hill climb on and not leak the traces, maybe one of the biggest drivers of IP?

换句话说，另一个资产测试就是：你拥有一个私有的评估体系。

Like so in other words, another asset test is you have an eval that's private.

你正在用模型A。

You're using model A.

你能换成模型B，然后继续提升吗？

Can you switch it to model B and, know, climb up?

如果你能做到，那你就掌握了主动权。

If you can, then you're in control.

如果你做不到，那就没掌握主动权。

If you can't, you're not in control.

正因如此，选择什么样的评估框架变得至关重要。

And that's where even the harness decision becomes super important.

对吧？

Right?

所以，拥有一个开放的框架，让所有模型都能接入，用你的评估体系、上下文和工具来帮助你逐步优化，我认为这是AI原生初创公司、SaaS公司，乃至所有企业都需要的核心能力。

So therefore, having an open harness, letting all models come in, having your evals, your context, your tools help you hill climb, I think, is the skills that an AI native startup needs, a SaaS company needs, or every enterprise needs.

是的。

Yeah.

从本质上讲，你就像当年的微软，先做操作系统，再转型为云服务公司，而第三幕可能是成为框架或评估平台公司——不管你怎么定义这一系列相关概念。

I think in a very real way, you are Microsoft historically, as an operating systems company and then become a cloud company, maybe, like, the third act is that you're a harness or evals company, whatever whatever the the sort of conglomerate of concepts that you wanna put together.

我认为，让每一家公司都能获得前沿智能——我忘了你具体用的词是什么——这才是真正的使命。

And I think, like, enabling every company to have, like, frontier intelligence or what what I forget the the exact term that you use is the is the mission.

对吧？

Right?

没错。

That's it.

这正是你和我们共同打造的平台承诺。

That is that is the platform promise that you build with us.

你会为你的数据获得属于它的智能。

You will get your intelligence for your data.

没错。

That's right.

对我来说，如果这场开发者大会只能用一句话概括，那就是：每个人都能用前沿智能去操作前沿技术吗？

That to to me, that is the like, if there was one tagline for this entire developer conference is, can everybody operate at the frontier with their frontier intelligence?

对吧？

Right?

我觉得这特别重要，不然我真不知道怎么实现稳定平衡。

To me, that is so important because otherwise, I don't know how you achieve stable equilibrium, right?

也就是说，我怎么才能说，我的公司未来会有价值，因为我已经知道如何在不断进化的平台上持续积累优势了。

Which is how do I then go and say, well, my company is going to have a terminal value because I now know how to continuously compound on top of what's a platform that gets better.

对吧？

Right?

就像当年Windows刚出来的时候，Adobe、Autodesk都跟着做了，或者你看看Jensen说的那些。

So when, like, Windows obviously came out, Adobe built, Autodesk built, or even, like, take what Jensen said.

我们做了DX，他则在上面开发了CUDA。

We built DX, and he built, you know, CUDA on top of it.

对吧？

Right?

我老跟Jensen说，天啊，我真是错过了这个机会。

I mean, I always say to Jensen, god, I got the shorn head of that.

对吧？

Right?

要是早些意识到就好了。

I wish we had recognized it.

但不管怎样，能打造一个平台层，让别人在此基础上构建自己的智能层——这一点至关重要。

But, nevertheless, but that idea that you can build a platform layer that someone else can then extend out and build their own intelligence layer in this case, I think, is everything.

对吧？

Right?

如果没有这一点，办开发者大会还有什么意义？

Without it, why have a developer conference?

我可以直接过来，让你们所有人都跪拜在某个模型的神坛前。

I can just come and have you all sort of just worship at the altar of one model.

是的。

Yeah.

但这根本不是开发者大会。

But that's not a developer conference.

大会。

Conference.

后台我们讨论过，什么是公司的知识产权，或者公司真正的价值在哪里。

Backstage, we you had a discussion about what is IP or what is the the value in a company.

以前是看员工在公司积累的经验年限，现在变成另一样东西了——也就是评估和应用智能代理的实际经验。

It used to be the length of human experience at a company, and now it's this other thing, which is the evals, the experience in sort of applying agents to the company.

马库斯，你能不能再展开说说？

He I just want you to, like, flesh it out a bit, Marcus.

对。

Yeah.

这么讲真的很到位。

It's a great way to frame it.

对吧？

Right?

说到底，每家公司都会同时拥有这两种资源：人类资本依然极其重要，因为人天生擅长发现各种潜在问题，而这正是我们创造价值的核心。

Because, yep, at the end of the day, every company is gonna have both the human capital that is still gonna be super valuable because humans and their ability to find the gaps that exist at all times is going to be the way we all will create value.

没错。

Right.

我的看法是，无论代币价值怎么涨，未来的关键始终在于激发人类新的能动性和抱负。

I mean, so I'm definitely in the camp that this is going to be about expressing new forms of human agency and ambition, even as token capital goes up.

对。

Right.

假设一家公司既有大量代币，也有丰厚的人力资源。

So let's say any corporation has lots of tokens and a lot of human capital.

那问题来了：怎么把这两者结合起来，产生复利效应？

The question is, how do you compound the two?

如果你看看团队内部的情况，有一批智能代理在做事，也有一批人类在做事，它们之间的互动轨迹，其实特别关键，能反映出这家企业到底是怎么创造价值的。

So if you have a like, if you take in teams, I have a bunch of agents doing work and a bunch of humans doing work and the traces between those, that is really important context of how that enterprise is creating value.

这就回到一个问题：别去训练一个通用模型，而该专门训练那些熟悉公司老规矩的智能代理。

Then that goes back to train not a generalist model, but to taint train the company veteran agent.

对吧？

Right?

这种东西真的特别有价值，我觉得公司理应把它记入资产负债表。

That is super valuable again, which is when a company goes and says, it should in fact go on to the balance sheet is how I think about it.

对吧？

Right?

事实上，过去人类的无形知识没法记入资产负债表，因为根本抓不住那些只可意会的东西，但现在不一样了，这些智能代理通过长期积累的轨迹，已经能捕捉到这些了。

That's fact, they may be like, human capital was never possible to go put on a balance sheet because you didn't know how to capture the tacit knowledge, Whereas now, I think you can with the agents that have learned through the through time, through all the traces.

至少我们认为，未来就会朝这个方向发展。

So that's what at least we think will happen.

我觉得美国证交会迟早得出台针对代币专业能力的会计准则。

I think the SEC is gonna have to have accounting standards for token expertise.

你提到的是一种均衡状态，企业在这种状态下能积累持续的价值，并看到自身的长期价值。

You're talking about the equilibrium state and a stable equilibrium where companies have this compounding value and can see terminal value for themselves.

另一个挑战是，那些在某个行业或跨行业普遍存在的应用和工作流程，如何维持这种稳定的均衡。

Another challenge to, you know, the considered equilibrium of, okay, there are applications and workflows that are sort of common to a vertical or a horizontal.

这正是SaaS公司兴起的那一波浪潮。

And this was like the generation of SaaS companies.

微软也有很多SaaS产品。

And, you know, Microsoft has lots of SaaS properties as well.

但每个企业都有自己独特、无法复制的定制化需求。

And Then there are things that are very specific to every enterprise that they're differentiated against.

我肯定你听过很多关于‘软件即将终结’的讨论，毕竟现在这些工作流程生成起来成本已经很低了。

I'm sure you have heard much and participated in much of the debate about the end of software because all these workflows are cheap to generate now.

你觉得未来，企业内部开发的智能代理和供应商提供的代理，它们的均衡状态会有什么不同吗？

Do you think the equilibrium looks different between what agents get built in enterprises versus in their vendors in the future?

是的。

Yeah.

我觉得现在发生的情况是，我们以前有一种特定的方式来捕捉应用中的工作流程，对吧？

So I think what's happening there is, see, we we had a particular way we captured, I would say, workflow in apps, right?

因为我们建立了一个数据模型，对吧？

Because we built a data model, right?

我们把某些业务流程的部分进行了结构化。

We schematized some part of some business process.

然后我们编写了一大堆业务逻辑。

We then built a bunch of business logic.

没错。

Yep.

接着我们在上面加了一层用户界面。

And then we put a bunch of UI on top of it.

这基本上就是每家SaaS公司的做法——

So that's kind of what every SaaS company-

再加一点配置。

And a little configuration.

整整二十年，一直都是这样

For like twenty years, that

没错，就是这样。

was- Right.

就一直这么干。

And that was it.

有意思的是，现在你又可以重新讨论这种垂直整合的方式了，对吧？

So interestingly enough, now you kind of get to relitigate that vertical stacking, right?

我觉得，比如每个SaaS应用底层的数据模型其实特别棒，对吧？

So I still think, for example, that data model that you build underneath every SaaS application is super good, right?

干嘛要重新发明一遍呢？

Like, why reinvent it?

我的总账就该是总账。

My general ledger better be a general ledger.

我不需要每次都重新设计数据结构。

I don't need new schema creation.

事实上，这个实体关系结构非常稳健，我很愿意把它作为基础来使用。

In fact, that entity relationship is actually pretty good, robust thing that I want to feed.

而且你希望它保持稳定。

And you want it to be stable.

没错。

That's right.

对。

Yeah.

业务逻辑也是同样的道理。

Then same thing with business logic.

对吧？

Right?

就拿我们这个叫 Power BI 的产品来说。

If if you look at we have this product called Power BI.

对吧？

Right?

很多人创建了大量仪表板。

It is like dashboards galore people created.

这些仪表板背后其实是一个非常丰富的语义模型，对吧？

The beauty underneath that dashboard is a very rich semantic model, right?

有人花了很大功夫去设计仪表板，还做了所有计算指标。

Someone took the pain to create a dashboard and do all the measures.

你希望拥有这样的东西吗？

And you want that?

这其实就是业务逻辑，对吧？

That's business logic, right?

我希望这些能直接为我所用。

I want that to be available to me.

所以我觉得，SaaS商业模式的问题在于，我们过去是打包成一个整体的。

So I think the challenge of the SaaS business model is we packaged one way.

现在我们得学会把这些组件拆开，再以新的方式组合起来，探索新的商业模式，对吧？

We now have to learn how to unbundle these things and rebundle in new ways and discover new business models, right?

我的意思是，看看现在微软三月的情况，就是一个很好的例子。

I mean, if you look at it, what's happening today with Microsoft March is a great example.

对吧？

Right?

我们有个产品叫 WorkIQ。

We have this thing called WorkIQ.

实际上我们发现，这么说吧，这事儿还真有历史先例，对吧？

In fact, what we are realizing is, oh my god, like, if you look at it, in fact, there's a historical parallel to it, right?

我们最早卖的是 Exchange 和 SharePoint。

We sold first Exchange and SharePoint.

在 Teams 出现之前，我们还有个叫 Link Server 的东西，诸如此类的。

And, you know, before Teams, we had a thing called Link Server and what have you.

那时候我们以为，这些东西都会迁到云上。

And we thought, oh, that's all gonna move to the cloud.

但没想到，用云服务器的人数居然会增长十倍、甚至百倍。

But little did we realize that the number of people who will use servers in the cloud is 10x, 100x, right?

因为当时人们不再买服务器了。

Because people were not buying servers.

他们只买订阅服务。

They were just buying a subscription.

现在M365也在发生同样的事，有了WorkIQ，我们首次把公司里最重要却一直没被当数据库用的数据暴露了出来——它以前只被困在我们的应用里。

The same thing is now happening with M365 because with WorkIQ, we have exposed what is perhaps the most important database in a company that never got used as a database because it is only captive to our apps.

对吧？

Right?

所有的邮件都跑在上面。

Was all email operated on it.

Teams也是跑在上面。

Teams operated on it.

Word、Excel、PowerPoint、SharePoint，全都在上面。

Word, Excel, PowerPoint, SharePoint.

但现在，说真的，这真是WorkIQ让我最兴奋的一件事。

But now, like, this is one of the coolest things I get to do with WorkIQ.

我打开一个 GitHub 仓库，说：嘿，我上周参加了好几个跟这个仓库相关的设计会议。

I go to a GitHub repo and I say, hey, I attended a bunch of design meetings last week related to this repo.

你能把这些内容整理出来，告诉我该做哪些修改吗？

Can you capture all that and tell me what changes I should make?

你想想，这多厉害。

I mean, think about that, right?

它真能去翻所有会议记录，然后给你出一份修改代码的方案。

It literally can go look at all those transcripts, come back with a plan to change your code base.

对吧？

Right?

以前你根本不敢想，M365 能干这种事。

Previously, you could never have thought of using M365 for something like that.

所以现在在智能代理领域，价值创造的机会真的翻了十倍。

So the value creation opportunity now in the agent world is, in fact, 10x more.

但这要求我们得有相应的使用方式，比如在 M365 上怎么用。

But it does require us to have, for example, there's going to be usage around M365, right?

这甚至可能超过最终用户的需求。

Which is going to be perhaps more than even the end users.

我们还得重新架构。

And we have to even rearchitect.

我以前用来处理收件箱或邮箱的系统，根本没法用来服务智能代理。

Like, in fact, what I use to serve an inbox or a mailbox cannot be used to serve an agent.

所以我们现在做的就是这类调整。

And so that's sort of what we are doing.

我不相信这些领域会有永久不变的商业模式。

I don't believe in, like, permanent business models for any of these domains.

但短期内，你对按成果收费、按令牌收费、企业打包方案这几种模式，怎么看？

But in the near term, do you have a prediction between, you know, outcomes based pricing, token based pricing, enterprise bundles?

是的。

Yeah.

我是这么想的：我们以前一直用的是按用户收费的方式。

The way I think about this is always we've had like, let's even take the per user price.

对

Mhmm.

按用户收费其实只是因为有人要做预算，需要明确的数字。

The per user pricing is really an artifact of someone creating a budget needing certainty.

对吧？

Right?

因为这最关键。

Because it's the most important thing.

有人需要做预算。

Like, somebody wants a budget.

对

Mhmm.

他们需要按用户来算。

They need a per user.

而按用户收费，其实就是一组使用权限的设定。

And and per user is just a set of entitlements to usage.

对吧？

Right?

差不多就是这么回事。

That's kind of what it is.

所以做法是，先拿一些使用量，打包成按用户划分的套餐，然后卖订阅服务。

And so the way is is the first bundling will be take some usage, bundle it into per user stacks, and, you know, then sell subscriptions.

我觉得订阅模式肯定会保留下来。

So subscriptions, I think, are gonna be there.

按用户收费的方式也会继续存在。

Per user is gonna be there.

接下来最重要的会是按使用量计费。

Then the next big thing will be consumption.

人们会说，我想要按使用量付费。

So people will say, I want consumption.

而且也有可能，有人连订阅或使用量付费都不想付。

And it's also possible that people will say, I don't even want to pay for any of the subscriptions or the consumption's outcome.

但别忘了，大多数人喜欢的是结果，直到他们真得到了结果。

But remember, most people love outcomes until they have an outcome.

因为一旦你真给出了结果，就等于在交版税。

Because once you have an outcome, it's like giving away royalty.

对吧？

Right?

我接触过不少客户，他们一开始特别支持按结果付费。

I mean, I've to customers who love outcome based pricing.

我说好啊，全力支持，结果他们一听说：天哪，真的吗？

I say, I'm all in until they oh my god.

你这是在说什么？

Like, what are you talking about?

你要跟我分我的成果？

You're sharing in my outcome?

不，不是这样的。

No.

不行。

No.

不行。

No.

我希望你改回按用户收费，而不是按使用量计费。

I want you to go back to per user pricing, and I want you to consumption price.

对吧？

Right?

我觉得这场争论还会继续下去。

So I think that debate will go on.

但这些商业模式各有其适用的时机和场景，没有一种能通吃所有情况。

But and all all all of these business models have a particular time and a place versus one to rule them all.

如果你是SaaS供应商或者平台提供商，这种灵活性就特别重要。

And if anything, if you're a SaaS vendor or you're a platform vendor, having that flexibility.

说实话，我们在GitHub上也遇到过类似的情况，对吧？

And quite frankly, we face this with GitHub, right?

我们最近刚在 GitHub 上推出了按用户收费的模式。

We just recently announced upper user pricing on GitHub.

因为 GitHub Copilot 最初是按用户计费的，那时我们还没意识到代理工具的使用强度会这么高。

Because little GitHub Copilot was constructed at a per user level before we understood even the intensity of usage of agents.

那时候，它只是开发者用来辅助编码、完成任务的一种互动工具。

It was an interactive way for a developer to use CodeComplete, maybe task.

没人想到会有人一次性启动上万个代理，全天候运行。

It was not like, oh, I launched 10,000 agents that are going on all day.

所以这次调整就是为了解决这个问题。

So that is what the adjustment is about.

现在我们很清楚，用户订阅制还是会保留，但必须加上按使用量计费的机制。

So now that we really want, there will always be a per user, but there will have to be a consumption meter.

您怎么看待 SaaS 模式的长期可持续性？

How do you think about the durability of SaaS more generally?

我注意到，很多企业内部的团队都陷入了‘代理狂热’。

One thing I've observed is in a lot of enterprises internally, there will be teams that almost have agent euphoria.

他们对能构建的各种新东西兴奋不已，于是试图重写大量应用，或者跑去对他们的SaaS供应商说：我们不再用你们的服务了，或者我们正在考虑自己开发。

They're so excited about the explosion of things they can build that they're trying to rebuild a lot of applications or going to their SaaS vendors and saying we're not gonna work with you anymore or we're considering an internal project.

看起来六到九个月后，其中一些人会回来承认：其实我们根本没法把所有东西都重做一遍。

And it seems like in six to nine months, maybe some of those people will come back and say, actually, we we can't rebuild everything.

你觉得在这个领域里，什么才是持久的，什么又只是昙花一现？

How do you think about what's durable in this world and what isn't?

我觉得得等一个完整的预算周期走完，才能真正看清这种平衡是怎么形成的。

I think we have to go through one full budget cycle on this to really see the the sort of the emergence of the equilibrium.

归根结底，哪怕只是开发一个应用，也都有边际成本。

Because at the end of the day, there's marginal cost to even generating the app.

对吧？

Right?

其实说得简单点，只要自己开发和维护的边际成本高于直接采购，那就应该直接买。

In fact, there can be even a a simple way to say it, like, if you should always acquire something if the marginal cost of building and maintaining something on your own is higher.

对吧？

Right?