直播：黄仁勋谈构建智能时代的动力引擎

非常感谢你，詹森。

Thank you so much, Jensen.

我们现在正处在一场巨大的人工智能革命之中。

So we are in the middle of a massive AI revolution.

这场革命的规模和速度，可能甚至超过了工业革命。

It is probably bigger and faster than even the industrial revolution.

你曾指出，眼下正在发生的，是人类历史上最大规模的基础设施建设。

And you have called out what's happening right now as the largest infrastructure build out in human history.

这场建设的核心是人工智能工厂，而推动这一切的公司就是英伟达。

At the center of that build out is the AI factory, and the company enabling all of that is NVIDIA.

你能跟我们说说，什么是人工智能工厂？为什么它是未来十年任何企业最值得投资的方向？

Can you tell us what is an AI factory, and why is it the best investment for any enterprise in the next decade?

好吧，你对人工智能的理解可能有几种方式。

Okay, so you understand AI in a particular number of ways.

你对人工智能最熟悉的理解方式，大概就是通过聊天机器人，或者通过网页浏览器。

The way that you understand AI probably most is through a chat bot, through a web browser.

你正在和它互动。

You're interacting with it.

你给它一个指令。

You give it a prompt.

它就会给你回话。

It says something back to you.

即使你用AI已经有一段时间了，过去两三年里，你也明显看到AI的能力有了巨大提升。

And even those of you who have been using AI for some time, you've seen in the last couple, two, three years a very significant evolution improvement in the capabilities of AI.

两年前，大家还在谈论ChatGPT。

Two years ago you heard about ChatGPT.

ChatGPT本质上是一种能理解你输入内容的软件。

And ChatGPT basically is a computer software that understands the input you give it.

它能感知并理解信息。

It can perceive, understand information.

还能把信息翻译或转化成其他形式。

And it can translate and generate the information into something else.

好了，你可以给它一个指令，比如说：这是我给你的PDF文件。

Okay, so you can give it a prompt and you can say, here's this PDF I gave you.

请你现在帮我总结一下。

I would like you now to summarize it.

它以前是从文字到文字的处理。

It went from text to text.

你也可以告诉它：这是我给你的PDF文件。

You could also tell it, here's a PDF I gave you.

现在请你根据这个故事生成一张图片。

I would like you to now generate an image of that story.

这就变成了从文字到图片的转换。

It goes text to image.

你还可以从图片到文字，比如给它一张图，问：这张图里发生了什么？

You could use go from image to text, meaning you could give it a picture and what's happening inside this picture?

这就叫从图片到文字的转换。

It goes image to text.

这样说得通吗？

Does that make sense?

任何东西都能转成别的东西。

Anything to anything else.

两年前的AI主要就能做这种转换了。

And AI in two years ago was largely able to do this translation.

我们管这叫生成。

We call it generation.

生成式模型。

Generative models.

生成式AI。

Generative AI.

但在‘生成式AI’这个词里，真正关键的是，光会生成还不够，更宝贵的是能思考——既要理解，也要生成，但更重要的是思考。

But the thing that is very big deal inside that word generative AI is in order to do something even more valuable than generation, understanding and generating, is thinking.

不过，要是连话都说不出来，那也谈不上思考。

Well, you can't think if you don't generate words.

因此，生成式AI的基础让我们具备了产生内在思维的能力。

And so the foundation of generative AI gave us the ability to generate internal thoughts.

思考、推理、一步步地推理、解决问题。

Thinking, reasoning, step by step reasoning, problem solving.

它还让我们实现了另一件如今非常重要的事：生成智能来控制其他东西。

It also allowed us to do another thing that is now very important which is generate intelligence to control something else.

生成控制指令来使用工具。

To generate control to use a tool.

这说得通吗？

Does it make sense?

比如用浏览器、用电子表格、用Photoshop、用PowerPoint、用某个工具，或者用AutoCAD之类的。

To use a browser, use a spreadsheet, use Photoshop, use PowerPoint, use something, use AutoCAD, use another tool.

现在的工具都是数字的，但将来这些工具可能会变成机械的。

Now, that tool today is digital, but someday that tool will be mechanical.

如果我生成一条指令去控制一个机械系统，那这就叫机器人技术。

So if I generate a command to a mechanical system, that would be called robotics.

如果我生成指令来控制一辆有方向盘的汽车，那就叫自动驾驶。

If I generate commands for a machine with steering wheel, that would be called self driving cars.

这能理解吧？

Does that make sense?

好的。

Okay.

事实上，两年前你就已经看到了这些基础技术。

And so two years ago, in fact, you saw the foundations.

我们当时叫它ChatGPT。

We call it ChatGPT.

大家都说，哎，挺好玩的。

And everybody said, ah, it's fun.

觉得挺荒唐的，或者生成了一堆胡编乱造的内容。

It's silly, or it produced a whole bunch of crazy hallucinated text.

这些都没错，但那正是促成今天这一切的基础技术。

That's all true, but it was the foundational technology that led to all of this.

两年后的今天，我们已经有了具备自主行动能力的系统。

Two years later, we now have agentic systems.

这只是一个看待人工智能的角度。

Now, that's one view of AI.

我刚才描述的就是这个视角——人工智能到底能做什么？

I just described the view, which is what can AI do?

对吧？

Right?

现在你们都注意到了，无论是ChatGPT、Codecs还是Cloud Code，你们都能看到，它不再只是能理解，还能真正干活儿。

And so now all of you realize you see it from ChatGPT, you see it from Codecs, you see it from Cloud Code, you see that it's now able to not just understand, but it's able to do work.

能思考，也能做事。

Reason and do work.

两年前，当人工智能还能理解你、生成一些有趣又新颖、带点可爱的内容时，

Now, two years ago, when AI was able to understand you and generate information that was interesting, novel, a little cute.

需要写首诗？那会是个不错的选择。

Whenever you need a poem written, great way to do it.

对吧？

Right?

谁不想写一首乡村歌曲呢？

Who doesn't want to write a country song?

两年前是那样，但现在因为AI能干活了，它就有了价值。

And so that was two years ago, but now because it's able to do work, AI is valuable.

价值就在于它能生成信息、产出有用的工作成果，而且还能为此收费。

Valuable meaning it can generate information, it can generate useful work, and it can be paid for it.

我们愿意花钱请聪明人，喜欢那些什么都知道的人，但不会为他们的知识付钱。

Because we pay for, we're interested in having friends that are smart, we love people who are know it alls, but we don't pay them for it.

我们付钱的是那些真正做事的人。

We pay for people who do work.

这说得通吧？

Does that make sense?

好了，这就是过去两年发生的事。

Alright, which is what happened in the last two years.

AI从当初只是有这个能力，到现在AgenTek已经能产出真正有用的工作了。

AI went from having this capability to now AgenTek went from not very valuable to now producing useful work.

它产生的有用工作多到你我每天都在用，而且按小时给AI付钱。

So much useful work that you and I are doing this every day, we're paying AI by the hour.

对吧？

Right?

我们可能每小时付它三十块、二十块来做这些事。

And so we might pay them $30 an hour to do the work, 20 an hour to do the work.

如今，我们其实正在给AI花不少钱。

We're basically paying AI a lot of money today.

这是人类历史上增长最快的软件业务。

The fastest growing software business in the history of mankind.

因为它现在能做有用的事，而我们愿意为这些事付钱。

Because now it's doing useful work and we can pay them to do it.

当然，这只是看待AI的一种角度，就是它能做什么。

Now, that's one view of AI, which is what it can do.

但还有一种看待人工智能的角度，对理解康斯坦丁所说的话特别重要。

But one other view of AI that's really important to help reason through what Konstantine is saying.

比如，有些公司和人之所以能打造出成功的企业，甚至切入大型行业的核心，是因为他们看到这项能力时，觉得非常有意思。

So, example, the reason why some companies, some people are able to build great businesses and could maneuver themselves into the center of very large industries is because when they see this capability, this is very interesting.

一个值得思考的问题是，如果我们能做到这一点，这对下游产业会有什么影响？

One interesting thought is if we're able to do this, what is the implication to this downstream industries?

这个话题确实值得深入聊聊。

That's an interesting conversation we should have.

明白吗？

Okay?

现在人工智能能做到这些了，那医疗、金融、生命科学、制造、物流、运输这些行业会怎么样？还有更多领域。

So now that AI can do this, what happens to all the industries like healthcare and financial services, and life sciences, manufacturing, logistics, transportation, on and on and on.

电商、广告、未来的娱乐，只要想到人工智能现在能做这些，就能展开一连串关于它接下来还能做什么的讨论。

E tail, advertising, future entertainment, the conversations list you can have about now that AI can do this, what can it do as a result subsequently?

这确实是个值得探讨的话题。

That's an interesting conversation.

但你应该往上游看，从产业角度来说，这到底意味着什么？

But you should go upstream, meaning industrially, what does that mean?

所以你首先会意识到这一点。

And so the first thing you realize is this.

回到最根本的原理。

Go back to first principles.

我刚才跟你说过，人工智能其实就是软件，是由计算机生成的。

I told you just now that AI is software and is being produced by a computer.

那么，是什么让计算机具备了这种能力呢？

Now, what happened to the computer that has made it possible to do this?

关键在于，我们今天所知的计算机，其实大约六十四年前才出现。

Well, the big idea is, you think about the computer as we know it today, it really emerged about sixty four years ago.

当年IBM System/360的发布，是计算领域最重要的里程碑，而那时IBM是全球最有价值的公司。

IBM System March was the biggest announcement of computing and sixty four years ago IBM was the most valuable company in the world.

正是他们奠定了现代人对计算机的理解。

And they created the modern understanding of computers.

我们今天能描述的关于计算机的一切，早在1964年就已经说清楚了。

Everything that we can describe about computer was really described in 1964.

接下来整整四十年，基本没怎么变过。

For forty years, largely has remained the same.

这种计算方式叫作‘预录制’。

And what happened was that form of computing is called pre recording.

你写好一个故事，存成一个文件。

You write down your story, you save it to a file.

你亲手写一段程序，存成一个文件。

You write a program by hand, you save it to a file.

你拍张照片，存成一个文件；录段音乐，存成一个文件；现在你拍视频，我们虽然在直播，但有人迟早会录下来，存成一个文件。

You take a picture, you save it to a file, you record music, you save it to a file, you make a video right now, we're streaming right now, but somebody's going to record it, you're going to save it to a file.

等你以后想用的时候，再从硬盘里调出来。

And when you want to use it later, you retrieve it from the disk drive.

这样讲清楚了吗？

Does that make sense?

而这个检索过程是智能完成的。

And the retrieval process is done intelligently.

所以每个人看到的新闻内容都不太一样。

So that's why everybody's retrieval of a news story is a little bit different.

这叫推荐系统。

It's called a recommender system.

但说到底，我们今天所熟知的计算机本质上是个基于检索的系统，这也解释了为什么这些地方叫数据中心。

But basically, computers as we know it today is a retrieval based system, which is the reason why these buildings are called data centers.

它们存储数据。

They store data.

注意，它们不叫计算机中心，因为其实没怎么进行计算。

Notice they don't call them computer centers because it's not doing much computing.

它们只是存着数据，等你用手机点一下，再把相关内容调出来。

They just store data that you retrieve based on what you touch on your phone.

不过现在的情况是，如果你仔细想想我刚才说的，为了让AI按我说的方式运作，每次我对它说话，都得给它提供新的信息，我们称之为上下文。

Well, what happened now, if you look at what I just described, in order for this AI to work as I described it, every time I say something to it, I have to give it new information, we call it context.

我给它一个新的输入，这就叫查询。

I give it a new prompt, that's called a query.

它会先结合上下文和查询来理解，再进行推理，最终根据具体情况生成回应。

Between the context and the query, it will understand it first, reason about it and it will produce an output based on that context in that query, based on the circumstance.

到目前为止，你们能跟上吗？

Does that make sense so far?

好，给我点个头示意一下。

Okay, give me one nod, okay?

如果是这样，每次你使用AI时，内容都是全新生成的。

And so, if that is the case, every time you use the AI, the content is produced originally every single time.

我此刻对你们说的每一句话，都是实时生成的。

Everything I'm saying to you right now is being produced in real time.

因为我的讲解是基于这样一个前提：你们来自60个不同的国家、128个不同的家庭。

And it's because my explanation is based on the fact that I realize all of you come from 60 different countries, 128 different families.

你们每个人都有截然不同的背景。

You all have many different backgrounds.

你们当中有些人可能来自计算机行业。

Some of you probably came from the computer industry.

但大多数人应该不是。

Most of you probably did not.

所以我用足够深入的方式向你们解释这些信息。

And so I'm explaining the information to you in a way that is sufficiently deep.

但我的最终目标是帮你们学会如何做出下一次投资决策。

But ultimately my goal is this, so that you know how to make your next investment.

我之所以这么说，是想给你们足够的信息，让你们自己能独立判断，下次看到有潜力的机会时，就知道该不该投。

That's what I'm leading to and so I'm going to give you enough sufficient information that you can reason about it for yourself so that when you see something in the next time you go, that's worth investing in.

那会是一个巨大的产业。

That's going to be a big industry.

现在它值一万亿美元，看起来已经很大了，但跟它未来的发展相比，根本不算什么。

That's $100,000,000,000 right now and it looks really big, but that's nothing compared to how big it's going to be.

我会帮你们建立直觉，去解决这个问题。

I'm going to give you the intuition to solve that problem.

明白吗？

Okay?

所以我们现在在这里。

And so here we are.

过去六十年，计算机行业主要靠检索信息，但突然有一天，一切都变成了实时生成。

Went from a computer industry that was largely based on retrieval for sixty years and all of a sudden, one day, it's completely generated in real time.

我们管这叫智能。

We call it intelligence.

我此刻正在给你们演示的就是这个。

This is what I'm doing right now for you.

我正在展示什么是智能。

I'm demonstrating intelligence.

有上下文意识，他给了我一个提示，我就给出了回应。

Contextual awareness, he gave me a prompt and here comes my answer.

这样讲清楚了吗？

Does it make sense?

这是极其先进的人工智能。

Extremely artificial intelligence.

明白吗？

Okay?

那现在到底发生了什么？

So here, what's going on now?

我刚才只给了你一个词。

I just gave you one word earlier.

这叫生成式。今天的计算机已经彻底被重新定义了，现在它具备生成能力。

It's called generative The computer of today has been completely reinvented, it's now generative.

从今往后，你看到的每一个字符、每一个字母、每一个词、每一段视频、每一张图片、每一条广告、每一支电视广告、每一个新闻故事，全都不一样。

And every single toe, every single letter you see, every single word you see in the future, every video, every image, every ad, every TV commercial, every single time you read a story, a news story, every one of them will be different.

你看到的、我看到的、康斯坦丁看到的，都会完全不同，因为内容是专为你生成的。

What Konstantine sees, what I see, what you see will be completely different because it'll be generated for you.

因为你的兴趣、你的上下文、你是谁、你为什么问、你是怎么问的，全都独一无二。

Because your interest, your context, who you are, for what reason you asked, how you asked is completely different.

这样理解对吗？

Does it make sense?

所以，未来你看到的每一个像素、听到的每一个声音、每一段视频，都是全新生成的，而不是从库里调出来的。

And so therefore, therefore, every single pixel that you see, every single sound that you hear in the future, every video you see in the future will be originally generated, not retrieved.

这意味着，未来我们需要大量的生成器。

Which means in the future we need a lot more generators.

而这些生成器，正是我们做的东西。

And these generators is what we build.

我们平时就是干这个的。

That's what we build for a living.

这些是大型计算机，它们在生成智能。

These are large computers and they're generating intelligence.

接下来的问题是：

Now the next question is this.

那它究竟能有多大呢？

Well, how big could it be?

它究竟能有多大？

How big can it be?

事实上，如今我们为全球大约十亿人提供信息生成和智能服务。

And so it turns out, the amount of information, the amount of generation of intelligence, we do it for about a billion people in the world today.

现在我告诉你们，AI已经具备了自主行动的能力，也就是说，它能独立完成任务。

Now that I told you that AI has become agentic, meaning that it can actually do work by itself.

既然它能独立工作，那一个智能体就可以跟另一个智能体沟通：我这儿有活儿，咱们联手干吧。

Well, if it can do work by itself, then one agent can communicate with another agent and say, I have some work to do, let's team up together and let's do some work.

于是，这么多智能体就协同合作，帮你公司解决各种问题。

And now you have all these different agents and they're all working together to solve problems inside your company, say.

在我们公司，康斯坦丁很清楚，我们是自主型AI的重度使用者。

So inside our company, Konstantine knows that we're huge users of agentic AI.

眼下可能有几十万个智能体在后台跑着，各自干活、互相沟通、解决问题。

We have hundreds of thousands of agents probably running around right now that are doing work and they're talking to each other and they're solving problems.

所有这些智能体都受到严格管控，运行在隔离环境中，但它们彼此之间紧密协作。

All guardrailed, all sandboxed, all guardrailed and sandboxed, but they're all working with each other.

这意味着，未来我们今天供十亿人使用的互联网，很可能主要由几千亿个智能代理全天候运行，它们彼此通信、使用网络。

Which means in the future it is very likely that the internet that we use today for a billion people will likely mostly be several billion, call it a 100,000,000,000 agents working around the clock and they're using the internet and talking to each other.

那它们都在说什么呢？

And what are they saying?

比如，公司之间会互相协作，员工的代理也会与其他员工的代理配合工作。

So for example, there'll be companies working with companies, employees' agents working with other employees' agents.

会有自动驾驶汽车，它们都是智能代理。

There'll be self driving cars, which are agentic.

也会有机器人，它们同样是智能代理。

There will be robots, which are agentic.

所有制造系统，每一栋建筑，都会成为智能代理。

All the manufacturing systems, every building will be agentic.

到处都会是这些代理，它们都在使用互联网。

There will be agents all over the place, they'll be using the internet.

它们之间发出的所有指令，都会自动生成。

And all of those commands that they generate to each other will be generated.

它们所有的思考过程，也都是自动生成的。

All of the thoughts that they have to understand, all generated.

这听起来明白吗？

Does that make sense?

所以，整个世界将会被一层无处不在的计算网络包裹起来，时刻不停地产生智能。

And so basically the world is going to be this layer of computing that's going to cocoon the earth and it's going to be generating intelligence all the time.

我刚才说的这话，听起来可能有点荒谬。

Now I just said something that sounds ridiculous.

但事实上，它已经发生了。

Except, in fact it's already happened.

两次了。

Twice.

三百年前，一家德国公司，西门子，制造了一台机器。

So three hundred years ago, a company in Germany called Siemens produced a machine.

这台机器非常特别。

And this machine is a really interesting machine.

你走到这台机器跟前，点一把火把它烧了。

You go up to this machine, you light it on fire.

然后，一股奇妙的无形力量就会从另一端涌出来。

And then this incredible invisible force comes out the other end.

当时没人知道那是什么东西。

Nobody understood what it was.

现在我们管它叫电。

We understand it now as electricity.

世界上一共有多少发电设施？

How many power generators are there in the world?

我们把它叫做电网。

We call it a grid.

发电设施就像一层外壳，包裹着整个地球。

Power generation cocoons the planet.

我们管它叫电网。

We call it the grid.

当然，三十多年前，甚至更早的时候，美国这边诞生了一种网络架构，后来发展成了互联网。

And then of course, twenty some odd years ago, earlier than that, thirty five years ago, this networking scheme, networking matrix fabric was created here in The United States, eventually became the internet.

那它到底在哪呢？

And where is it?

它包裹着整个世界。

It cocoons the world.

所以现在，我们有了能源、通信和智能，它们都会像这样包裹整个世界。

And so now you have energy, communications, intelligence, and it will cocoon the world.

到时候，它就会变成一种日用品，到处都能用上。

And it'll just be a commodity, we'll use it all over the place.

而英伟达现在做的，就是这种全新的机器。

And so what NVIDIA does for a living is this new machine.

三百年前发明的那台机器，叫发电机。

The machine that was invented three hundred years ago is called the dynamo.

只要有任何东西在动，就能被这台发电机利用。

That dynamo, anything that moves comes in.

可能是瀑布，也可能是风、火、蒸汽。

It could be waterfalls, it could be wind, fire, steam.

把运动——也就是原子——转化成电子。

Transfer it from motion, atoms, right, to electrons.

从原子到电子。

Atoms to electrons.

然后我们把这些电子输入到我们的机器里，也就是NVIDIA。

We then take the electrons into our machine called NVIDIA.

电子进入我们的机器，进入这座工厂，输出的东西是数字。

Electrons now comes into our machine, comes in this factory, and what comes out are numbers.

这些数字，只要组合方式不同，就能变成语言、数学。

These numbers, depending on how you combine them, turns into language, math.

它还能转化成我们新掌握的一种语言——蛋白质。

It can also turn into a new language we've learned, proteins.

我们学会了人类生物学的语言。

We learned the the language of human biology.

我们学会了物理世界、气候和天气的语言。

We learned the language of the physical world, physics, climate, weather.

我们学会了三维世界、机器人和自动驾驶汽车的语言。

We learned the language of the three d world, robotics, self driving cars.

我们学会了各种不同形式智能的语言。

We learned the language of all kinds of different forms of intelligence.

但关键是，这两台机器相隔三百年，输入的是原子，输出的是电子；输入电子，输出数字，而这些数字可以被重新组合、重新构建，变成各种各样的智能。

But the point being, now, these two machines, three hundred years apart, atoms in, electrons out, electrons in, numbers out, and those numbers could be rejiggered, reformulated into all kinds of different intelligence.

我们打造的就是这个，这正是我们谋生的方式，所以我管它叫工厂——它在生产东西，我们叫它们令牌，但其实它们就是数字，这些令牌就是智能。

That's what we built, that's what we do for a living, and that's why I call it a factory because it's producing, we call them tokens but they're just numbers, tokens and these tokens are intelligence.

就是这样。

That's it.

我们做的就是这个。

That's what we do.

其实没那么难。

It's not that hard.

太棒了。

Brilliant.

现在你知道AI是干什么的了，也知道它是怎么构建的，以及它未来会有多大的规模。

Now you know what AI's for, and now you know how AI's built, and how big it's gonna be.

感谢大家今天参与。

Thank you all for joining today.

谢谢。

Thank you.

干得漂亮。

Good job.

问得好。

Excellent question.

我

I

我真的觉得这一点说得很到位。

I really felt that carried it.

你知道，那个提示词。

You know, the prompt.

你把它讲得清清楚楚。

You laid it up.

是的。

Yeah.

明白了。

Okay.

这是一场巨大的变革。

So this is a massive revolution.

对。

Yeah.

你刚才提到了三个转型方向。

And you you laid out three transformations.

能源转型，跟今天每个人都息息相关，现场很多观众都来自全球的制造和能源企业；通信技术，把我们所有人连在一起；还有就是智能。

Energy transformation, which touches everyone today, and a lot of the people in the audience are part of these manufacturing and energy producers globally, telecommunications, which connects all of us, and now intelligence.

在能源领域，你提到过发电机，还有电信。

And in energy, you talked about the generator, telecommunications.

我觉得，这在通信领域对应的大概是交换机，或者类似的东西，负责全球通信的路由。

I guess the comparable would be the switch or something along those lines for routing communications globally.

而在智能革命中，核心就是GPU。

And now in the intelligence revolution, it's the GPU at the core.

还有像H100这样的AI工厂，或者所有新系统，把所需的一切都整合在一起，就像维拉·鲁宾那样，会发生什么变化。

And the AI factory like the h 100 or any of the new systems that bring everything you need under the same hood, Vera Rubin, what happens.

这些工厂，你知道，这些发电机，我们每个单元叫一个机架，里面装了72块芯片。

These factories, so you know, these generators, each one of our units, we call it a rack, there's 72 chips inside.

今年我们生产了大约800万台。

We manufacture, call it 8,000,000 of them this year.

但72块芯片会组成一个机架。

But 72 of them go into a rack.

一个机架重达两吨。

That rack weighs two tons.

售价四百万美元。

It is $4,000,000.

包含一百五十万个零部件，是全球最昂贵的设备。

Has one and a half million parts and it's the most expensive piece of equipment in the world.

我们生产这些设备，就像生产手机一样。

And we manufacture them like, I guess it would be like phones.

我们大规模量产，然后运往世界各地的数据中心。

I mean we crank them out And they go into data centers all over the world.

我们是以大批量的方式制造这些机器的。

We build these machines in volume.

这些设备体积很大。

They are big devices.

这相当于你做力量训练的方式。

This is how you do your weight lifting.

我明白了。

I understand.

没有批量折扣。

No volume discounts.

明白了。

Okay.

你描绘了一个令人兴奋的世界图景，

So you laid out this picture of a very exciting world

是的。

Mhmm.

而我们就生活在这个世界里。

That we are in.

我们正处在一场革命的中心。

We're in the middle of this revolution.

我们确实身处其中。

We're Yeah.

你说得对，这已经持续了几十年，也可以说是好几年了。

You could say decades in, you could say years in.

是的

Yeah.

现在我们确实已经身处智能革命的主流之中。

Certainly now in the the mainstream of the intelligence revolution.

我们该如何参与进来？

How do we participate?

我相信在座的每个人都想参与到这场变革中。

I'm sure everybody here wants to participate in this revolution.

是的

Yeah.

非常好的问题。

Excellent question.

我们先从大型企业说起，然后再谈谈个人。

Let's start with big enterprises, and then let's get to individuals as well.

对

Yep.

人们怎么加入这场运动呢？

How do people join this movement?

刚才我给了你们两个思维模型。

And so now I gave you two mental models.

我再给你们讲一个思维模型。

I'm gonna give you one more mental model.

其中一个模型，其实就是我们刚才说的这个故事，希望接下来能完整讲完这四个阶段。

So so one mental model, they're they're really you know, we could talk about this story, hopefully, we cover all four phases.

我刚刚讲了AI能做什么。

I just talked about what AI can do.

我还讲了AI是怎么制造出来的。

I talked about how AI is made.

这些AI设备都放在这些工厂里。

And these things are in these factories.

每座工厂的功率达到一吉瓦，成本大约是五百亿美元，这应该是世界上造价最高的工厂了。

These factories are each gigawatt is about $50,000,000,000 So have you ever seen it's the most expensive factory in the world.

但这座价值五千亿美元的工厂，还能产出三到四万亿美元的智能价值。

But also, that 1 $50,000,000,000 factory also generates $300 $400,000,000,000 in intelligence.

所以它的产出价值简直惊人。

And so the production value is incredible.

投资回报的速度快得不得了。

The return on investment is extremely fast.

这就是工厂这部分的情况。

And so that's the factory part.

接下来我要跟你们说的，这一点在考虑投资时特别关键：人工智能的产业布局到底是什么样子？

The part that I'm going to tell you now, and this is very important when you think about investment, is what does the industrial layout look like for AI?

理解这个产业格局，可以把它想象成一个五层蛋糕。

And the way to think about the industrial view is think of it as a five layer cake.

最底层是能源，这一点我之前说过。

Now, told you on the bottom is energy.

最底层，还记得我提过的动力源吗？

The bottom is, remember I said the dynamo, Okay?

现在我们有各种交流发电机之类的设备，用于发电。

Now we have, of course, different AC generators and things like that, power generation.

所以最底层就是能源。

And so on the lowest layer is energy.

这是几十年来能源行业最大的增长机遇。

This is the single greatest opportunity in several generations for the energy industries to grow.

可能是近一百年来，全球许多国家的电网第一次有机会大规模投资。

The very first time in probably, I don't know, one hundred years since the energy grid in many countries could be invested in.

这是投资清洁能源的最佳时机。

This is the best opportunity to invest in sustainable energy.

只要你关心清洁能源——核能、风能、太阳能，不管哪种形式，甚至氢能，只要能发电，就一定能拿到资金。

If you care about sustainable energy, nuclear, air, or wind, solar, you name it, whatever form, hydrogen, whatever form, so long as it produces energy, it's going to get funded.

这说明现在有多好的时机，因为我们手上有整整一万亿美元。

And so that tells you something about how great of a time this is because we have a trillion dollars.

你想想，仅今年一年，市场就会投入一万亿美金到我接下来要讲的这五层架构里。

Just think, this year alone, we're going to put a trillion dollars from the market into this entire five layer cake I'm about to describe to you.

第一层是能源。

So the first layer is energy.

这就是西门子表现这么好的原因。

That's the reason why Siemens is doing so well.

三菱也做得非常出色。

Mitsubishi is doing fantastically.

通用电气，维诺瓦。

GE, Vernova.

我的意思是，所有公司都一样。

I mean everybody.

这个蛋糕的第一层是能源。

The first layer of the cake is energy.

第二层是芯片、电脑、网络、交换设备和硅光子技术。

The second layer of the cake is chips and computers and networking and switches and silicon photonics.

这样理解清楚吗？

Does that make sense?

这些都是计算机。

It's all the computers.

第三层我们称之为基础设施。

The third layer of the cake, we call it infrastructure.

土地、电力、建筑、资金、数据中心运营。

Land, power, shell, money, data center operations.

如今每一项都供不应求。

Every one of them in scarce supply today.

所以下一层就是基础设施层。

And so that's the next layer, is infrastructure layer.

而大家都能看到、以为是人工智能的那层，其实是模型层。

And then the layer that everybody sees, that everybody thinks is AI, is the model layer.

这样理解清楚了吗？

Does it make sense?

再下一层就是这个了。

That's the next layer.

它搭建在计算机和云基础设施之上。

It sits on top of the computers, cloud infrastructure.

这是我所知道的，近年来人类历史上最大的机遇——这么多市场驱动的投资正自然涌入这个生态系统。

And this is the greatest opportunity in recent human history that I know that so much market driven investment is naturally coming into the ecosystem.

现在正是打造的好时机。

This is a great time to build.

所以，这就是模型层。

And so now that's the model layer.

模型层包括OpenAI和Anthropic，但这一部分你绝对不能忽视。

The model layer is OpenAI, it's anthropic, but this is the part that you can't overlook.

这非常关键。

This is very important.

你听说过两家公司，都知道它们的名字。

So you have two companies that you know of, that you hear about.

不过别忘了，就像我之前说的，AI已经掌握了语言。

However, don't forget AI as I was explaining earlier has learned the language.

它学会的不只是语言，还有背后的意思。

It learns the language and the meaning.

语言，还有任何有结构事物的含义。

Language, the meaning of anything that is structural.

所以这一层最关键的是，我们总在谈语言，但别忘了，只要有结构，就能学会任何东西。

So that layer, what's really important is that we hear, we talk about all the language, but don't forget, you can learn anything with structure.

我来举个例子，说明什么是结构化的东西。

And so let me give you an example of something with structure.

今天我走进房间时，看到很多人，你们的出现其实都在我预料之中。

Today, when I walked into a room, I was explaining a lot of people and it wasn't unexpected the way you appeared.

但如果你们当中有人倒挂在天花板上，漂在半空，还有人身体分散成十七个部分，我能直接看穿你们，那这种场景就太难学了，因为每次都不一样。

Now, if some of you were hanging off the ceiling and floating in mid air and some of you, the one human but body parts in 17 different places and I could see through some of you then it's hard to learn that because every time it's different.

明白了。

Okay.

因为量子层面的东西实在太难掌握了。

Because it's hard to learn quantum things.

不过，有规律的东西我们是可以学会的，对吧？

However, things with structure we can learn, right?

人都有眼睛，所以这些你都能学会，明白吗？

People have eyes and so you can learn these things, okay?

所以，关于三维空间，我学了物理定律，坐下去的时候我就有信心了。

And so, three d, I learned the laws of physics, I sat down and notice, I sat down with confidence.

我有53%的时间能稳稳坐下，剩下47%的时间就直接穿过去了。

I 53% of the time I landed safely on the chair, the other 47% of the time I went right through it.

所以我不能百分百依赖它。

And so I can't trust it but 100% of the time.

你们明白吗？

Do you guys understand?

如果一件事是可预测的，那它就有规律，你可以从中学习，也能理解它的意义，对吧？

And so if things are predictable, then there's structure you can learn from it and you can learn the meaning of it, okay?

我们已经理解了蛋白质的意义，现在也在探索基因的意义，不只是测序，也不只是用CRISPR编辑，而是搞清楚这个基因到底意味着什么。

And so we learned the meaning of proteins, we're learning the meaning of genes, not just sequencing it, not just CRISPR editing it, but what is the meaning of that gene?

细胞的意义是什么？

What is the meaning of a cell?

细胞为什么会做它做的事？

Why does a cell do what the cell does?

当两个细胞相遇时会发生什么？

What happens when two cells coming together?

这就跟我是怎么理解一个词的意义一样，我也想这样理解细胞的意义。

And so this is no different than imagine I learned the meaning of a cell the way I learned the meaning of a word.

那当我把两个词放在一起时，会发生什么？

And what happens when I take two words and put them together?

这两个词会相互激活，变成另一种全新的意思。

These two words activate each other, turn into something else of another meaning.

明白吗？

Okay?

从计算机的角度来看，它根本不关心你处理的是细胞、蛋白质、词、图像还是汽车。

And so from a computer's perspective, it doesn't care if it's a cell, a protein, a word, an image, a car.