微软CEO萨提亚·纳德拉谈AI商业革命：SaaS、OpenAI与微软将何去何从？| 达沃斯现场直播

好了，各位。

Alright, everybody.

我们非常高兴邀请到唯一一位、真正的萨提亚·纳德拉，微软的第三任首席执行官，与我们的AI与加密货币负责人戴维·萨克斯进行一场即兴的炉边谈话。

We're thrilled to have the one, the only, Satya Nadella here, the third CEO of Microsoft for a, impromptu fireside chat with David Sachs, our czar of AI and crypto.

萨提亚，微软的第三任首席执行官，出生于印度。

Satya third, CEO of Microsoft, born in India.

这是一个多么了不起的故事。

What an incredible story.

他大学一毕业就来到了这里。

Came here right after college.

而且，你还特意回了一趟老家，去接你的妻子，把她带到这里。

And, you had a little round trip to pick up your wife in your book to to bring her here.

请简要告诉大家这件事是怎么发生的。

Tell everybody briefly how that occurred.

嗯，

Well,

你知道吗，这真是美国移民政策复杂性的绝佳例证，我觉得。

you know, so that's that's a great story of the the labyrinth that is the immigration policies of The United States, I think.

我和妻子在印度上大学时就认识了。

I my wife and I went to college together in India.

我来美国读研究生。

I came here for grad school.

后来我们结婚了。

We then got married.

我拿到了绿卡，但因为我们结婚了，她却无法来和我团聚。

I got my green card, and she couldn't come join because we got married.

所以事情是这样的，我不得不放弃我的绿卡。

So the story goes basically, I had to give up my green card.

有趣的是，我去德里的美国大使馆，问：‘申请放弃绿卡的队伍在哪里？’

So the funny thing is I went to the in American Embassy in Delhi, and I said, where's the line to give up my green card?

他们说：‘没有这样的队伍。'

And they said, there is no such line.

在九十年代做这种事简直疯狂。

That would be a crazy thing to do in the nineties.

所以放弃你的绿卡，申请H-1签证让她能来美国，这确实很奇怪，但最终一切都解决了。

So it was a strange thing to give up your green card, get an h one so that she could join, but it all worked out.

所以，这已经是个遥远的记忆了，但当时确实是一种变通的办法。

So, you know, it's a long lost memory, but it was, you know, a way to work around it.

我想问你一下，你先在GitHub上推出了Copilot，然后又在桌面端推出了Copilot。

I wanted to ask you, having launched a Copilot first with GitHub, then having the Copilot on the desktop.

你为微软做出了一个非常大胆的决定，将Copilot直接集成到Windows产品中，而我每天都在桌面端使用它，但那时它还不能真正识别文件系统或与应用程序交互。

You made a very bold move for Microsoft to put that in the Windows product, which I use every day on the desktop, which you did that before it really could recognize the file system and interact with applications.

当时反响平平，但你们现在却在不断加码。

Got a little bit of a lukewarm reception, but now you've been doubling down doubling down.

在我看来，知识工作者目前有三种使用模式。

And there seems to be, in my estimation, three modalities for knowledge workers.

埃隆在xAI正在打造他们所谓的‘人类模拟器’，如果你看了这周泄露的内容的话。

Elon's building at x AI, what they're calling a human emulator, if you saw that leak this week.

是的？

Yeah?

他们只是在为员工构建系统，并把他们放进聊天室和邮件里。

Where they're just building employees and just putting them into their chat rooms and email.

然后克莱德这周推出了CoWork。

Then you have Claude, came out with co work this week.

极其强大。

Incredibly powerful.

人们对此简直疯了。

People are kinda losing their minds over it.

我过去四十个小时一直在使用它。

I've been playing with it for the last forty hours.

真的令人印象深刻。

Truly impressive.

你对微软和知识工作者如何实际应用这一技术有什么愿景？

What's your vision for Microsoft and how knowledge workers will actually put this to use?

因为在玩转ChatGPT并获得一些有趣结果，与真正取得商业成果之间似乎存在差距。

Because there seems to be a gap between, you know, playing around with ChatGPT and getting some interesting results and getting business results.

是的。

Yeah.

所以我认为，要理解这些不同形态的最佳例证之一就是看编程，这显然是知识工作的一种形式，甚至可能是知识工作的最佳例子。

So I think it one of the most perhaps illustrative examples of trying to understand these various form factors is looking at coding, which is obviously a form of knowledge work or probably the best example of knowledge work.

如果你回顾一下编程的发展历程，它最初是从‘下一个编辑建议’开始的。

And if you think about the journey coding has been, it started with essentially the next edit suggest.

对吧？

Right?

那正是我真正开始相信这一代技术的时刻，我记得当时有个叫Codex的模型，那是在GPT-3.5之前。

That was the first time in fact my own belief in this entire generation of tech really sort of got formulated, where I started seeing I think there's you know, there's a codex model back in the day, it was pre g p t three five.

那时，下一个编辑建议开始以相当高的准确率运行。

That's when next edit suggestions started working with some real accuracy.

然后我们进入了聊天模式，接着是操作功能，现在发展到了完全自主的智能体，而这些自主智能体既可以是前台的，也可以是后台的，既可以运行在云端，也可以在本地运行。

Then we went to chat, then we went to actions, and now to full autonomous agents, and then the autonomous agents can be both a foreground, background, in the cloud, or local.

对吧？

Right?

所以这就是今天在编程时所有存在的交互形式。

So that's all the form factors that exist today when you're coding.

有趣的是，如果你仔细观察，你会发现你其实都在使用它们。

And interestingly enough, if look at it, you use all of them.

对吧？

Right?

并不是说只有一种交互形式。

It's not like there's only one form factor.

所以我认为这可能是另一个重要的教训。

So that's I think probably one of the other lessons.

例如，当我使用命令行界面时，我可以同时使用前台代理、后台代理，然后直接在 VS Code 里进行编辑，所有这些都在并行发生。

So for example, when I'm in a CLI, I can go a foreground agent, background agent, and then just literally go edit in Versus code right there, all happening in parallel.

这展示了这些交互形式是如何组合在一起的。

So that sort of shows how these form factors even composed.

所以你把这一点应用到知识工作上，正如你所提到的。

So then you bring that to knowledge work, to your point.

我们一开始是从聊天开始的。

We started with chat.

带有推理能力的聊天超越了简单的请求-响应，因为你现在可以看到思维链条的运作过程。

Chat with reasoning sort of goes beyond just request response, because you now have that chain of thought where you can see it work.

现在是他们的行动，对吧？本质上是通过计算机操作或通过技能和代理调用来实现，因此你可以执行操作。

Now their actions, right, essentially either through computer use or through basically skills and agent calls, so you can do actions.

这就是如今协作者的状态。

So that's kind of the state of the copilot today.

现在，我们可以从心智演化的理论角度来思考，对吧？因为如果你还记得，乔布斯对个人电脑或计算机说过一句绝佳的话，他说电脑是思维的自行车。

Now, there is a way to think about you know, the theory of the mind evolution, right, because you need, like, if you remember, know, Jobs had the best line, I would say, for PCs or computers was to say, if you it's a bicycle for the mind.

比尔也说过一句我很喜欢的话，那就是信息触手可及。

Bill had a line which I liked as well, which was information at your fingertips.

我们现在需要为人工智能时代如何使用计算机创造一个新的概念隐喻。

We kind of need now a new concept metaphor for how we use computers in the AI age.

你有吗？

You have one?

我最喜欢的一个说法来自Notion的CEO，你知道的，那个出色产品的负责人。

And the one I like actually came from the CEO of Notion, which I you know, that manager of Incredible product.

是的。

Yeah.

你还没买呢。

You haven't bought it yet.

我已经买了。

I've bought that.

但它既是管理，本质上是无限思维的管理者。

But it's both management, basically a manager of infinite minds.

这样想挺好的，对吧？

That's a nice way to think about it, right?

当你真正审视你所协作的所有智能体时，你其实需要理解我所说的另一个术语——我们宏观委派、微观引导。

When you sort of really look at all the agents that you are working with, you kind of need to understand what I in fact, the other term I like is we macro delegate and micro steer.

事实上，你确实需要这样。

In fact, you kind of need that.

在编程中，你某种程度上就是这样做的，对吧？

In in in coding, you kind of have it, right?

所以你会进行宏观委托，然后我可以在它工作的同时并行给出指令。

So you do a macro delegation, and then I can in parallel give it instructions while it is doing work.

这就是如今 Copilot 或类似工具的现状。

So that's sort of the state even today of Copilot or what have you.

你提到的这种交互形式之一，我非常期待，甚至在下周我们就会推出相关功能：当我使用 GitHub Copilot 时，并不意味着软件开发者是孤立工作的。

You bring up a little bit of One of the form factors I'm very excited about, and you'll see us even in the next week even do things, is while I'm sitting in GitHub Copilot, it's not as if software developers sit in isolation.

我工作的内容不仅仅局限于我的代码仓库。

It's not like the only thing I work on is my repo.

我还要参加会议。

I attend meetings.

我会撰写规格文档，或者实现他人撰写的规格文档。

I write specs or others have written specs that I'm implementing.

我需要让我的代码库与之保持一致。

I need to have my repo be consistent with that.

因此，这意味着我要么使用一个直接的MCP服务器，要么使用一个技能，以便能够调用我的工作智能——也就是Copilot，并将其整合进来。

So that means using either a straightforward MCP server or a skill, I want to be able to call into my work IQ, which is the Copilot, bring that in.

这就是未来知识工作将实现的组合方式。

That's the type of composition of knowledge work that will happen.

安全方面也是如此。

Same thing with security.

假设你是一名安全专业人士，你有大量的日志，你该如何真正地分析它们？

Say you're a security professional, you have lots of logs, How do you sort of really analyze them?

你把它们导入文件系统，然后在其上编写代码，创建仪表板等等。

You drop them into a file system, then write code on top of it, create a dashboard, what have you.

这些正是我们能够支持的知识工作类型。

Those are the types of knowledge work that we can enable.

在这里，我认为你又提到了另一点：你能否创建数字员工、数字同事之类的？

There, I think you bring up one more thing, which is can you create digital employees, digital coworkers, or what have you?

这一切都关乎凭证。

And it's all about credentials.

今天，你可以做到。

Today, you could.

你可以直接分配

You can literally assign

你们也在做这个吗？

Are you working on that as well?

是的。

Yeah.

事实上，我们推出了一项名为 Agent 365 的功能，用于赋予身份。

So in fact, we introduced something called Agent three sixty five as a way to give identities.

事实上，我们将如今为人类和其计算设备提供的身份和终端保护扩展到了智能体上。

In fact, extending the identities we have for humans today and the endpoint protection we have for their compute devices to agents.

所以，你可能会克隆我，让我在人力资源部或市场部工作，并在 Office 中拥有一个我的虚拟版本。

So So you might clone me working in the HR department or working in the marketing department and have a virtual version of me inside of Office.

没错。

That's correct.

所以那里有两种模式。

So so there are two sort of modalities there.

一种是给每位知识工作者无限的思维能力。

One is you give every knowledge worker infinite minds.

这是一种方式。

That's kind of one.

然后你创建独立于身份的无限思维，因为身份是必须准确把握的关键要素，否则系统无法正常运行。

And then you create even infinite minds independent of your identity because the identity is one of the key things you got to get right, even for it to work.

对吧？

Right?

所以对于

So for

权限和决策？

Permissions and decision making?

权限和决策，其中最关键的一点是，谁对谁做了什么，这通常是组织中最关键的查询问题。

Permissions, decision making, like one of the key things is who did what to whom is sort of the most important query in an organization.

归根结底，组织需要了解哪些工作完成了，这些工作的来源是什么，以及如何追溯它们，对吧？

At the end of the day, the organization needs to understand what work got done, and what's the provenance of that work, and how do you trace it back, right?

因此，你实际上需要的是：如果是由大量代理支持的人类，那就是人类通过其身份传递下来的宏观授权与微观引导。

So therefore, you kind of want either If it's a human with a lot of agents, then it's really macro delegation micro steering by the human whose identity was passed on.

所以，这涉及授权与独立身份之间的区别。

So it's delegation versus a separate identity.

而这种管理层次——产品管理——你们已经取消了，Alphabet也取消了，Meta四年前就开始在他们的组织中逐步取消。

And that was done by a level of management, product management that you've eliminated, that Alphabet's eliminated, Meta has started to eliminate in their organization four years ago.

你们的员工人数和现在微软一样，但在这段时间里，你们在收入线上增加了900亿美元，并且收入翻了一番。

You had the same number of employees you have at Microsoft now, but you put a $90,000,000,000 onto the top line of the revenue in that time, and you doubled your income during that time.

那么，这是怎么做到的？

So how did that happen?

这是这些岗位的自动化吗？

Is that automation of those jobs?

是你有点儿……

Is it you were a little bit Yeah.

人手过剩了吗？

Overstaffed?

解释一下

Unpack

你其实抓住了一个非常有趣的关键点，从某种意义上说，究竟需要发生什么样的重大结构性变化？

it's it's actually you're you're pulling on a very interesting thread, which is at some level, what's the big structural change that needs to happen?

事实上，我认为这可能是自个人电脑出现以来知识工作领域最大的变革。

In fact, I would say this is probably the biggest change in knowledge work since PCs.

我总是会思考，在个人电脑出现之前，工作是怎么进行的。

I always think about how did work happen pre PCs.

想象一下，像我们这样的跨国公司如何做预测。

Think about a multinational company like ours trying to do a forecast.

传真来回传递，内部备忘录不断发送，然后你才勉强做出一个预测。

Faxes went around, interoffice memos got sent, and then you kind of created a forecast.

然后，个人电脑突然变成了标准配置。

Then suddenly PCs became standard issue.

你放一个Excel表格，输入一些数据，通过邮件发送，每个人都能输入数据，这样就完成了预测。

You put an Excel spreadsheet, put some numbers, sent it in email, everybody entered numbers, you had a forecast.

所以，工作内容、工作产物和工作流程都发生了变化。

So the work, the work artifact, and the workflow all changed.

现在正在发生的就是这样的变化。

That's what's happening.

举个例子，在领英，我们过去有产品经理、设计师、前端工程师，还有后端工程师等等。

So for example, I'll give you at LinkedIn, we used to have product managers, we had designers, we had front end engineers, and then we had back end engineers and so on.

于是，我们将前四个角色合并，实际上扩大了他们的职责范围，称他们为全栈开发者。

So what we did is we took those first four roles and combined them, in fact, increased scope and said, they're all full stack builders.

我喜欢这一点，因为这是一种结构性的改变，能够提升这些职能之间的工作效率和协作流程。

So, I like that because that's a structural change that allows for us to increase the change both the work and the workflow between these functions.

我推测，由于不再需要四个人来回沟通想法，而是由一个人快速编码，速度自然会更快。

And I would assume the velocity because you don't have four people communicating that throughput of ideas, as such one person and vibe coding.

没错。

Exactly.

而且出现了一种新的工作流程。

And there's a new workflow.

所以，你可以想象，如果我们今天要开发一个AI产品，就会有一套完全新的工作流程。

So at the same time, as you can imagine, if we're to build an AI product today, there's a complete new workflow.

对吧？

Right?

它从评估开始。

It starts with evals.

对吧？

Right?

基本上，这是一个从评估到科研，再到基础设施的过程。

So basically, there's this eval to science, to infrastructure.

因此，这些评估由全栈开发者以及新型的产品经理来完成。

And so evals are done by these full stack builders and what have you, and product managers in the new form.

基础设施由后端的系统工程师构建，因为他们支持支撑产品的科学工作。

The infrastructure is built by the systems engineers at the back end because they support the science that supports the product.

所以某种程度上，出现了一个新的循环。

So in some sense, there's a new loop.

你必须进行结构性的改变。

And you have to structurally change.

因此，科技行业内部正在发生的许多变化，我认为将会非常巨大。

And so a lot of what is happening inside of tech is that change, which is I think going to be pretty massive.

同时，像我们这样的公司，我必须亲力亲为。

And at the same time, a company like ours, I have to do everything.

我不可能只是在未来直接放手不管。

It's not like I can just so go live in the future.

我必须确保我们在Windows上进行热补丁的质量做到极致，同时构建能提升Copilot质量的评估体系。

I have to make sure we're doing a fantastic job of doing hot patching on Windows is done with quality, while at the same time building the evals that are improving Copilot quality.

对吧？

Right?

所以这两者都必须成为首要任务。

And so both of those have to be first class.

我假设这是你职业生涯中最具挑战性的时刻，因为微软曾经在某些领域占据主导地位，形成双头垄断，但你当时面临的竞争远不如现在这么激烈。

I assume this is the most challenging moment of your career because Microsoft was so dominant, duopoly in some spaces, but you really weren't up against the competition level you're up against now.

我跟埃隆聊过，他说造车其实挺容易的，因为当时他面对的是传统汽车制造商，而现在你面对的竞争对手阵容完全不同。

I was talking to Elon, you know, and he was sort of saying, well, building cars was pretty easy because I was up against the legacy carmakers, and now I'm up against just look at the set you're up against.

是的。

Yeah.

这真是一个非常紧张的时期。

It's a it's a pretty intense time.

我的一贯看法是，每十年出现一批全新的竞争对手其实很有帮助，因为这能让你保持竞争力。

I mean, so the way I I I always think is it's always helpful when you have a complete new set of competitors every decade because that keeps you fit.

想想看，我1992年加入微软时，我们最大的生存威胁是Novell。

If you think about it, I joined Microsoft in '92 when I had Novell as the big existential competitor we had.

而到了2026年，你说得完全对。

And here we are in 2026, and you're absolutely right.

这是一个非常紧张的时期。

It's a pretty intense time.

我很高兴有竞争。

I'm glad there's the competition.

老实说，最终当我展望未来五年，科技在GDP中所占的比例会是多少？

Quite honestly, at the end of the day, when I look at it, as a percentage of GDP five years from now, where will tech be?

比例会更高。

It will be higher.

所以我们很幸运身处这个行业。

So we are blessed to be in this industry.

竞争确实非常激烈，但并没有一些人所说的那么零和。

It's a lot of intense competition, but it's not so zero sum as some people make it out.

高峰变得更高了。

Highs getting much bigger.

这个技术的总市场规模和影响力将会极其巨大。

The TAM and the just the impact of this tech is going to be so massive.

那么问题来了，我总是会回到这一点：微软的品牌身份是什么？我们拥有什么样的品牌授权？客户对我们有什么期待？

The question then, of course, is what is like, I always go back to what's the brand identity Microsoft has, brand permission we have, what do customers expect from us?

有时候我们会不自觉地过度思考，以为每个客户都希望从所有竞争对手那里得到同样的东西。

It's sometimes we kinda overthink somehow that every customer wants the same thing from all of the competitors.

而弄清楚这一点，其实是对彼得·蒂尔观点的一种不同解读：你必须通过真正理解客户对你有什么期待来避免竞争，而不是认为每个人都是竞争对手。

And finding that out, right, it's kind of a different take on the Peter Thiel thing, which is you've got to avoid competition by really understanding what customers really want from you versus thinking everybody's a competitor.

大卫？

David?

是的。

Yeah.

显然，在达沃斯这里有很多国家元首，也有许多《财富》500强公司的首席执行官。

So there are a lot of heads of state here obviously at Davos as well as CEOs of Fortune 500 companies.

我想你昨晚在晚宴上被问到一个问题，关于他们应该如何看待人工智能以及如何取得成功。

And I think you got asked a question last night at the dinner about how they should think about AI and how to be successful.

我记得他们用了“扩散”这个词。

And I recall they used the word diffusion.

我想请你进一步阐述一下这些观点，因为它们与我正在做的某些政策工作产生了强烈共鸣。

And I was wondering if you could expand on those remarks because that really resonated with some of the policy work I've been doing.

不。

No.

当然可以。

Absolutely.

事实上，你们所有人所做的，就是确保美国的技术栈在全球范围内被广泛使用并获得信任。

In fact, what you all have been doing to make sure in this context of the American tech stack is broadly used around the world and is trusted around the world.

因为在我看来，大卫，归根结底，你创造了技术，但真正的收益只能通过深度使用才能实现。

Because I think when I look back, David, to me, at the end of the day, you create the technology, but really the benefits come only by intense use.

事实上，我最喜欢的一项研究是经济学家——我认为是奥托·达特茅斯——他的名字叫迭戈·科门，他研究了工业革命期间发生了什么。

In fact, one of my favorite studies has always been this work that an economist, I think Otto Dartmouth did, his name is Diego Komen, where he studied basically what happened during the Industrial Revolution.

各国是如何领先的？

How did countries get ahead?

其简单的启示是：任何国家只要引进了最新的技术，并在此基础上进行增值创新，就能取得领先。

And the simple takeaway from that was any country that brought the latest technology into their country and then did value add technology on top of it.

所以就像不要重复造轮子，直接采用最新技术，然后在其基础上进行创新。

So it's like don't reinvent the wheel, bring the latest and then build on top of it.

这在我看来就是技术扩散时发生的情况。

That's to me what happens when you have diffusion.

因此，对于像人工智能这样的通用技术，它需要广泛传播，甚至就在我们美国国内。

So especially with general purpose technology like AI, it needs to spread, like right in our own country in The United States.

我们现在有了这项技术。

We now need, we have the tech.

问题是，它是否被应用于医疗、金融服务，是否被各行各业——大型企业、中小企业、公共部门——所使用。

The question is, is it being used in healthcare, is it being used in financial services, is it being used in every sector of the economy by large businesses, small business, public sector.

因此，在我们看到这种扩散和广泛应用之前，我们不会取得成功。

So to me, unless and until we see that diffusion and intense use, we're not going to have the success.

而我们现在正处于这个阶段。

And so that's the phase we are in.

它的传播速度正在加快。

It's diffusing faster.

因此，你所做的一些政策工作，总的来说，好消息是技术已经就位，围绕云计算和移动设备建立的基础设施使得这项技术得以广泛传播，对吧？

And so some of the work, policy work you have done, and in general, the good news here is the technology is there, the rails around cloud and mobile that were laid out make it possible for this thing to spread, right?

获取这些技术资源并不是不可能的。

It's not impossible to get the tokens.

关键问题是，有哪些应用场景？以及如何管理这一切中的变革？

The question is what are the use cases, and how do you manage the change in all of that?

至少在达沃斯，一个问题在于：这对西方和发达国家来说是一回事，那对全球南方国家呢？

One of the questions at least in Davos is it's one thing for the West and the developed nations, what about the Global South?

我认为全球南方国家也有巨大的机遇，坦率地说，因为在我看来，大多数全球南方国家的GDP中约有40%到50%来自公共部门。

I think Global South has a huge opportunity too, quite frankly, because to me, let's say 40%, 50% of the GDP of most Global South countries is public sector.

所以想象一下，这项技术如何改变政府将纳税人资金转化为公民服务的方式，如果能提高效率，那很可能直接带来几个百分点的GDP增长。

So just imagine this tech making a difference in how the governments really parlay the taxpayer money into services for citizens, and if there's efficiency gains, that's probably a couple of points of GDP growth right there.

因此，我非常乐观地认为，这种需求将会出现，我们作为美国，以及欧洲、亚洲、南美洲、非洲和其他所有地区，都应该推动这项技术的广泛部署。

And so I'm very optimistic that there's going to be a pull, and that we should as The United States, given the technology stack we have, in Europe, in Asia, in you know, in South America, in Africa, and everywhere else, get it to be broadly deployed.

我经常被问到的一个关于人工智能竞赛的问题是：你怎么知道你是否领先，或者美国是否领先于其全球竞争对手？

One of the questions I get asked a lot about the AI race is how do you know if you're winning or how do you know if The United States is ahead of its global competitors?

我给出的答案是市场份额。

And the answer I give is market share.

你知道，如果我们五年后环顾世界，发现美国公司、美国技术占据了80%的市场份额，那就说明我们做得不错。

You know, if we look around the world in five years and we see that American companies, American technology has say 80% market share, it means we did a good job.

如果我们五年后环顾世界，发现全球都在使用中国芯片和中国模型，那就意味着我们可能输了。

If we look around the world in five years and see that it's say Chinese chips and Chinese models that are being used all over the world, means we probably lost.

所以，归根结底，真正的检验标准是实践效果，用例才是关键。

So, you know, ultimately usage is the proof of the pudding is in the eating of it.

我的意思是，在这种情况下，判断你是否成功的标准就是市场份额和实际使用情况。

I mean, the in this case, the way that you know that you're succeeding is through market shares, through usage.

我同意这一点，但大卫，既然你曾在微软工作过几年，我始终牢记比尔·盖茨关于平台的一句话。

I agree with that, but David, since you even worked at Microsoft for a few years, one of the things that I'm very grounded on is always that Bill Gates line of a platform.

所以我一直思考的不仅是市场份额，还有生态系统的影响。

So one of the things that I always think about is its market share, but it's also ecosystem effects.

对吧？

Right?

你看，美国一直以来所做的不仅仅是关于我们的市场份额或美国公司的收入。

See, what The United States always has done is not just about our market share or even the revenues to US companies.

事实上，我在微软学到的一件事是，每当我访问一个国家时，我首先研究的数据是，比如在英国或瑞士等地，我们的渠道在瑞士创造了多少就业机会。

In fact, one of the things I learned at Microsoft is whenever I did a country visit, the data I would first study is in, let's say in The UK or in Switzerland or what have you, is what is the total employment created in Switzerland in our channel?

这曾经是我们国家报告中的头号指标。

That used to be like the number one thing in our country reports.

对吧？

Right?

以及总人数

And the total number

是指IT员工的数量、办公室职员的数量吗？

that be like the number of IT workers, the number of office workers?

渠道合作伙伴。

Channel partners.

渠道合作伙伴，我们使用独立软件供应商。

Channel partners, we use ISVs.

所以是那些存在的独立软件开发商的数量。

So number of ISVs who are there.

因此，我们过去会全面衡量一个国家中围绕平台的生态系统是如何建立的，而这正是美国一直以来所拥有的——事实上，包括在中国在内的美国技术栈，都是因为其他人围绕我们的技术栈构建起来的。

So we used to have a complete marker of how did the ecosystem around the platform get built one country And at a that is what The United States has In always fact, The US tech stack, including in China, got built because others built around our tech stack.

同样的事情将会发生。

The same thing is going to happen.

所以这就是为什么我认为你正在做的关于技术扩散的工作

So that's why I think the work you're doing around diffusion

对。

Right.

是真正扩大整个蛋糕的规模，增强对平台的信任，从而创造真正的经济机会。

Is about really increasing the size of the pie, the trust in the platform, so that there is true economic opportunity quite frankly.

嗯，

Well,

你说得对。

you're right.

我记得，大约十年前，我的公司Yammer被微软收购时，你曾提到过这些事。

And I remember actually you brought back some memories from this is about a decade ago when my company Yammer was acquired by Microsoft.

我们当时属于SharePoint团队。

We were part of the SharePoint group.

我记得那时的产品经理们非常自豪，因为SharePoint生态系统的收入——也就是非微软的咨询公司和实施服务商为企业部署SharePoint所获得的收入——大约是微软自身软件收入的七倍。

And I remember that the product managers there were very proud of the fact that the revenue from the SharePoint ecosystem, meaning non Microsoft, the the consulting community, the implementers who would go into companies to implement SharePoint, I think their revenue is something like seven times greater than Microsoft's own software revenue.

我认为总体来看。

And I think In aggregate.

总体来看。

In aggregate.

我认为比尔曾经说过一句话：只有当平台之上的收入达到你自身收入的若干倍时，你才算真正拥有了一个生态系统或平台。

And I think and I think Bill had a line about you're not an ecosystem or a platform until the revenue on top of your platform is some, you know, factor of your own revenue.

没错。

That's right.

我认为，当我们讨论技术扩散，以及希望美国保持领先地位时，真正重要的是：这并不意味着对世界其他地区不利，因为它们完全可以基于这些平台进行完全的构建。

And and I and I think I think what's really important about this is when we talk about diffusion and obviously want The United States to have this leading position, it doesn't mean it's bad for the rest of the world because they're able to build on top of those platforms 100%.

并创造更多价值。

And create even more value.

完全正确。

100%.

事实上，这才是最重要的观点。

In fact, that's sort of the most important point.

对吧？

Right?

这并不是关于美国科技和美国的收入。

So this is not that this is not about American tech and America revenues to The United States.

而是利用新平台在全球范围内创造机会。

It's actually creating opportunity using a new platform everywhere.

事实上，我记得在90年代我曾与SAP合作开发我们的数据库产品。

And in fact, I remember I worked on our database products in the '90s with SAP.

事实上，SQL Server和R3的组合在双方都取得了成功，人们经常谈论英特尔和微软，但另一个我成长过程中深受影响、并奠定了我世界观基础的，是我们与一家至今仍是巨头的欧洲软件公司所做的事情。

In fact, the combination of SQL Server and R3 were successful on both sides, there's a lot talked about Intel and Microsoft, but one of the other things that I grew up in, which has sort of been foundational in how I look at the world, is what we did with a European software company that is still a giant.

所以，没人知道下一个重要的AI应用会是什么、在哪里出现、会发生什么，但我一直抱着这样的态度：即使使用美国的技术栈，世界各地也可能涌现出新的科技公司，甚至可能进入全球前五大科技公司之列。

And so who knows what the next big AI app will be and where and what will happen, but I sort of go in with the attitude that there will be tech companies, maybe even top five tech companies that could emerge everywhere with even the American tech stack.

你完成了一些令人惊叹的收购，除了是一位技术专家，你还是一位出色的交易高手。

You have done some amazing acquisitions, you're quite a dealmaker on top of being a technologist.

这可能是你辉煌任期和巨大增长中最少被报道的方面。

It's probably the least reported aspect of your spectacular tenure and the massive growth you've had.

但你确实与OpenAI达成了交易，而对方的萨姆·阿尔特曼，可能是史上最精明也最具争议的交易者之一。

But you did a deal with OpenAI and probably one of the most savvy slash controversial deal makers of all time, Sam Altman.

这笔交易被外界解读为，你们会获得一笔巨额现金回报，而微软其实并不需要这笔钱。

That deal was looked at as you you you're you're set up to get a windfall in cash, which you don't need as Microsoft.

当然，能拿到这笔钱总是好事，我猜如果他们上市的话。

Always nice, I'm guess guessing if they IPO.

但你是否可能，这正是外界对这笔交易的批评——创造了一个微软最终的竞争对手？

But did you create potentially, and this was the criticism of it, an ultimate competitor to Microsoft?

你是怎么看待这个问题的？

And how do you think about that?

微软错过了史蒂夫·鲍默最大的遗憾——错失移动革命，那么你们如何能没有自己的Gemini、XAI或Claude呢？

And how can Microsoft, which missed Steve Baumer's biggest regret, missing the mobile revolution, how can you not have a Gemini, an XAI, a Claude, that is your own?

或者在你看来，你们是否已经拥有，因为你们拥有OpenAI的源代码？

Or in your mind, do you have that because you have the source code of OpenAI?

是的，我认为这是对的。

Yeah, I think that that's right.

所以当人们问，你们的基础模型在哪里？

So when when people say, hey, where's your foundation model?

归根结底，我们确实拥有这些知识产权。

I mean, the end of the day, we do have the IP.

但话说回来，你提到了几个不同的方面。

But that said, I think you bring up a couple different things.

其中之一是，对我而言，微软当前战略中最重要的事情是打造令牌工厂。

One is, to us, the most important thing, when I look at what is Microsoft's strategy today, one is we want to build token factories.

我们目前最大的业务是Azure，而Azure业务的市场规模由于未来的发展趋势将变得极其庞大，因此我们现在必须擅长构建这些令牌工厂，这意味着需要一套异构的基础设施，并像所有超大规模云服务商一样，通过软件最大化利用这些资源，以降低总拥有成本并提高利用率。

So our biggest business today is Azure business, and the Azure business, the TAM, given what's going to happen is so huge that we now need to be fantastic at building these token factories, and that means a heterogeneous fleet of infrastructure, and that every hyperscaler has always done, which is use software to make maximum use of it and for TCO and utilization.

所以这是其中一方面。

So that's one side of it.

然后还有应用服务器业务，对吧？你提到，如果每个人都将构建代理、拥有无限的思维、这些强化学习训练场、评估系统等等，那么就像每个平台都有一个应用服务器一样，这个领域也有一个应用服务器。

Then there's the app server business, right, which is, you talked about, if everyone's going to be building agents, have infinite minds, have these RL gyms, have evals, what have you, there's an entire, just like every platform is at an app server, this one has an app server.

我们正在用Foundry等做这件事，对吧？

That's what we're doing with Foundry and what have you, right?

所以这是一项应用服务器业务。

So there's an app server business.

在这一应用服务器中，现在结构上非常明确的一点是，任何构建应用或公司的人都将使用多个模型，而不是单一模型。

In that app server, one of the things that structurally now is pretty clear is anyone building any application or any company is going to use not one model, but all the models.

我为什么不呢？

Why would I not?

事实上，我会为任何特定任务协调使用多个模型。

Which is, in fact, I will orchestrate for any given task, even multiple models.

对吧？

Right?

在我们的医疗实践中有这样一个很好的工具，叫做决策协调器。

There's this one nice thing that came out in our healthcare practice called the decision orchestrator.

它证明了通过分配角色——比如调查员、数据分析师、领域专家——即使只是给模型赋予提示角色并进行协调，也能比任何单一的前沿模型取得更好的结果。

What it proves is that by assigning roles, right, so investigator, data analyst, domain expert, just giving even prompted roles to models, and then orchestrating them gets better results than any one single frontier model.

那么，我是否可以理解为，你对开源模型持乐观态度，认为大型语言模型将很大程度上商品化，而价值不会集中在这些模型上？

Am I right to read into that then that you're bullish on the open source models and think large language models will largely be commoditized, and that's not where the value will accrue?

事实上，我是这样看待这个问题的：

In fact, the way I think about it is that

就像什么？

Just like what

发生在苹果身上的事情一样。

happened Apple

顺便说一下，苹果也是这么想的。

thinks that too, by the way.

顺便说一下，你想一想数据库市场发生了什么。

By the way, you think about what happened in the database market.

我以前总觉得，所有东西都只是个SQL数据库，直到发现其实不是这样，对吧？

I used to be like, everything is just a SQL database, until it was not, right?

你想一想，有文档数据库，也有非SQL数据库，数据库的种类真是层出不穷。

There was, I mean, think about it, there are doc databases, there is no SQL databases, the proliferation of database.

谁会想到数据库市场会有如此丰富的多样性呢？

Who would have thought that the database market would have such a richness to it?

或者它竟然会变成开源的。

Or that it could ever be open source.

那正是我的想法。

That was my

我们谈谈Postgres，或者Mongo发生了什么，它是开源的，但甚至还有公司支持它。

We talk about Postgres or what has happened even with Mongo, which is open, but there are even companies that have backed it.

所以在我看来，这就是将会发生的事。

So to me, that's what's going to happen.

在我看来，模型就像数据库市场一样。

To me, a model is like the database market.

它确实有一些不同之处，但我总觉得，肯定会出现一些闭源的前沿模型。

It it it's it's got it's got differences, but I sort of somehow think that it's not there are definitely gonna be frontier models that are closed source.

你知道，也会有开源的模型达到前沿水平。

You know, they're gonna be open source models that are gonna be frontier class.

事实上，我认为在接下来的一年里，讨论的一个重要话题可能是：企业的未来会怎样？

In fact, if anything, I think in this next year, what will be probably a big part of the discussion is, what's the future of a firm?

企业应该能够将自己拥有的隐性知识，嵌入到自己掌控的模型权重中。

A firm should be able to take the tacit knowledge it has and embed it inside a weights in a model that they control.

对，所以当有人问我应该有多少个模型时，我会说：世界上有多少家企业，就应该有多少个模型。

Right, so when somebody asks me how many models should be there, I'll say as many models as firms in the world.

对，这是一种极端的说法。

Right, that's sort of an extreme way.

因为对我来说，这正是我理解知识经济如何转变为人工智能经济的方式。

Because because to me, that's how I think this this knowledge economy becomes an AI economy.

你是不是在秘密地——既然我们现在都在这里，可以说出来——开发一个能在Windows桌面运行的LLM，因为

Are you secretly, and you can say it here since we're on all in, working on an LLM to exist on the Windows desktop because that

你是。

you are.

你拥有它。

You have it.

比如今天，有一个PHY硅模型，完全驻留在NPU上，当然也使用GPU。

Like today, there's a PHY silica model, which is completely resident using NPUs, and of course using GPUs.

事实上，高功率设备最大的安装案例之一，是工作站的回归，这非常有趣。

In fact, the largest installation of high power in fact, it's one of the fascinating the workstation is back.

如果我去看看的话，我是最其中之一。

I'm one of the most if you went to see Yeah.

那太好了。

That's great.

或者微软，因为你有一个不错的桌面业务。

Or Microsoft because you have a nice desktop business.

当然。

Absolutely.

所以我们认为，这种形态，特别是我经常说，我职业生涯是从命令行开始的。

And so we and in fact, we think that that form factor, especially I mean, I I always say this, which is you know, I started my career on a command line.

谁知道呢？

Who knows?

我可能最终也会在命令行中结束我的职业生涯。

I may just end it in a command line.

但你当初在太阳公司，就是那个原始的51万美元的工作站吗？

But you are you at Sun, which was the original $510,000 workstation?

你有没有想过，未来你会在这里与客户会面，推广一台价值102万美元的台式机，它搭载了大语言模型和相应的硬件，你明白吗？

Do you see a time where you'll be meeting with your customers here and advocating a $1,020,000 dollar desktop machine that has an LLM and the hardware mean, you can.

你可以安装一张DGX显卡，打造一台极其出色的机器，而且模型方面，顺便说一句，我们只差一个架构调整，就能实现某种分布式模型架构，甚至是一个懂得如何自我分配的MOE架构。

You can put a DGX card and you can have just a fantastic machine and the models And by the way, we are one architecture tweak away from even having some kind of a distributed model architecture, even an MOE architecture that knows how to really distribute itself.

这种突破性进展可能会彻底改变混合人工智能的面貌。

That's the type of breakthrough that can completely change what hybrid AI may look like.

但我们绝对致力于并专注于将个人电脑打造成本地模型的理想平台，让这些本地模型能够处理大量提示信息，并调用云端资源。

But we're absolutely committed and focused on making the PC a great place for local models, and local models that then do even a lot of the prompt processing and call into the cloud.

对吧？

Right?

所以还有很多工作可以开展，这确实已经在进行中了。

So there's a whole lot of work that can happen, and that's sort of definitely something that's underway.

是的。

Yeah.

我认为云协作者已经展示了连接本地文件驱动器并加以利用的强大能力。

I think that the cloud co worker has kind of shown the power of tapping into the local file drive and be able to use that.

这引出了另一个观点。

That that brings another point.

你让我想到了Yammer。

You you got me thinking about Yammer.

对于不了解的人，要知道Yammer大约十五年前的成名之处在于，它率先在企业软件领域运用了许多消费者增长策略。

And for people who don't know, you know, Yammer's claim to fame, this is about fifteen years ago, was that it pioneered a lot of well, it used a lot of consumer growth tactics to attack enterprise software.

当我思考AI在企业中的采用时，你觉得未来一年它会如何普及？

I'm wondering as you think about enterprise adoption of AI, how do you think it's gonna spread over the next year?

我觉得我们正处在一个关键的时刻。

It feels like we're at sort of a critical point.

你觉得会是自上而下的吗？

Do you think it's gonna be top down?

会是CEO指示团队，给他们一个战略转型项目，然后他们去发布招标吗？

Is it gonna come from the CEO directing a team, giving them a strategic transformation project, and they're gonna do an RFP?

还是说，它会通过那些适应力强、在日常生活中已使用AI工具的AI原生员工，在企业内部自下而上地传播？他们开始把这些工具带入工作，并取得惊人的成果。

Or do you think it's gonna spread bottom up in the enterprise through AI native employees who are adaptable, who are using the tools in their own lives, and they start to bring these things to work and start accomplishing amazing things.

是的。

Yeah.

不。

No.

我认为，大卫，就像所有事情一样，这既是自上而下，也是自下而上的。

I think, you know, like all things, David, I think it's both the top down, bottom up.

对吧？

Right?

我之所以说自上而下，是因为如果我看看在客户服务、供应链或人力资源自助服务中应用AI的投资回报率，这些是IT和高管们能够轻易做出决策的简单项目，也是你目前看到AI真正落地的首批领域。

The the reason I say that top down is if I look at the ROI of applying AI in customer service, or in supply chain, or in HR self-service, those are the easy projects where IT and CXOs can make calls, and that's where you're seeing the first drop of real AI adoption.

但最终发生的是自下而上的变革，对吧？

But the bottom up is what ultimately will happen, right?

我的意思是，就连个人电脑也是如此，如果你回想一下，是律师们把Word带了进来，然后财务部门买了Excel，接着电子邮件出现，最终变成了标准配置。

I mean, even with the PCs, in fact, if you think back at it, the lawyers brought Word in, and then finance bought Excel in, and then email came, and then it became standard issue.

现在正在发生的就是这种情况。

That's what's happening right now.

所以，比如说，这些代理——当我谈到大家都在构建代理时，他们正在想办法创造这些改变工作流程、消除繁琐工作的工具。

So for example, these agents, when I sort of talk about everybody's building agents, they're figuring out a way to go create these things that are changing workflow and removing drudgery in their work.

这正是自下而上变革的开端。

That's sort of the beginning of what is a bottom up transformation.

事实上，我最兴奋的就是这种自下而上的变化。

In fact, the thing that I'm most excited about is this bottom up change.

甚至在微软，比如我们现在管理着全球大约500名Azure光纤操作员。

Even at Microsoft, for example, we manage something like 500 odd fiber operators around the world in Azure today.

顺便说一下，我自己之前也没意识到，很多这类工作被称为DevOps，但其实涉及的是物理资产。

And by the way, had not myself realized it, a lot of it, it's called DevOps, but it's a physical asset.

有些东西会被切断。

Things get cut.

当你提到DevOps时，意思是你真的在发邮件问别人：嘿，那根光纤断了是怎么回事？

And when you sort of say DevOps, that means you literally are emailing people and saying, hey, what happened to that fiber cut?

我们该怎么修复它？

How do we repair it?

所以来回沟通非常多。

So there's a lot of back and forth.

因此，负责我们全球网络的这个人，正如你所说，这些员工本质上就是数字员工，正在处理所有这些DevOps工作。

So this network, the person who runs our global network, has built, to your point about these person, they're just digital employees, essentially, that are doing all of that DevOps.

而且这是一种完全自下而上的方式，你看到这些工具就在那儿，就像：嘿，我有了构建智能体的新方法，它就在那里。

And that's And there's a completely bottoms up, where you see the tools, it's kind of like, hey, I have the new way to build agents, it's there.

我会用它来实现自动化，消除繁琐工作，提升效率和质量，而这最终是一种技能提升——而技能提升并不神秘，就是通过实践来实现的，对吧？

I'm going to use it to create levels of automation that remove drudgery, improve efficiency, improve quality, and that ultimately is a skilling thing, which is sort of the big issue, is, and skilling is not mystical, it's just by doing, right?