2026-04-15 Jensen Huang.TPU competition, why we should sell chips to China, & Nvidia’s supply chain moat
Refer To：《Jensen Huang – TPU competition, why we should sell chips to China, & Nvidia’s supply chain moat》。
00:00:00 – Is Nvidia’s biggest moat its grip on scarce supply chains?
00:00:00 – Nvidia 最大的护城河，是它对稀缺供应链的掌控吗？
Dwarkesh Patel
We’ve seen the valuations of a bunch of software companies crash because people are expecting AI to commoditize software. There’s a potentially naive way of thinking about things, which is: look, Nvidia sends a GDS2 file to TSMC. TSMC builds the logic dies, it builds the switches, then it packages them with the HBM that SK Hynix, Micron, and Samsung make. Then it sends it to an ODM in Taiwan where they assemble the racks. Nvidia is fundamentally making software that other people are manufacturing, and if software gets commoditized, does Nvidia get commoditized?
我们已经看到一批软件公司的估值暴跌，因为人们预期 AI 会让软件商品化。有一种可能比较天真的思考方式是：你看，Nvidia 把一个 GDS2 文件发给 TSMC。TSMC 制造逻辑裸片，制造交换芯片，然后把它们和 SK Hynix、Micron、Samsung 生产的 HBM 封装在一起。之后再把这些东西送到台湾的一家 ODM，由他们组装成机柜。从根本上说，Nvidia 做的是软件，而制造由别人完成；如果软件被商品化，那么 Nvidia 会不会也被商品化？
Jensen Huang
In the end, something has to transform electrons to tokens. The transformation of electrons to tokens and making those tokens more valuable over time is hard to completely commoditize. The transformation from electrons to tokens is such an incredible journey. Making that token is like making one molecule more valuable than another molecule, making one token more valuable than another. The amount of artistry, engineering, science, and invention that goes into making that token valuable, obviously we’re watching it happen in real time. The transformation, the manufacturing, all of the science that goes in there is far from deeply understood and the journey is far from over. I doubt that it will happen.
归根到底，总得有某种东西把电子转化为词元。把电子转化为词元，并且随着时间推移让这些词元变得更有价值，这件事很难被完全商品化。从电子到词元的转化，是一段极其惊人的旅程。制造那个词元，就像让一个分子比另一个分子更有价值，让一个词元比另一个词元更有价值。为了让这个词元变得有价值，其中投入了多少艺术性、工程、科学和发明，显然我们正在实时目睹这一切发生。这个转化过程、这个制造过程，以及其中所有的科学，远远没有被深刻理解，而这段旅程也远未结束。我怀疑这种商品化不会发生。
We’re going to make it more efficient, of course. The way that you framed the question is my mental model of our company. The input is electrons, the output is tokens. In the middle is Nvidia. Our job is to do as much as necessary and as little as possible to enable that transformation to be done at incredible capabilities. What I mean by “as little as possible,” whatever I don’t need to do, I partner with somebody and make it part of my ecosystem.
当然，我们会让它变得更高效。你刚才提出问题的方式，其实就是我对我们公司的心智模型。输入是电子，输出是词元。中间就是 Nvidia。我们的工作，是为了让这种转化以惊人的能力完成，尽可能做必要的事，同时尽可能少做不必要的事。我说“尽可能少做”，意思是，凡是我不必亲自做的事情，我就和别人合作，把它变成我的生态系统的一部分。
If you look at Nvidia today, we probably have the largest ecosystem of partners, both in the supply chain upstream and downstream, all of the computer companies, application developers, and model makers. AI is a five-layer cake, if you will. We have ecosystems across the entire five layers. We try to do as little as possible, but the part that we have to do, as it turns out, is insanely hard. I don’t think that gets commoditized.
如果你看看今天的 Nvidia，我们大概拥有最大的合作伙伴生态系统，包括供应链上游和下游、所有计算机公司、应用开发者，以及模型开发者。你可以把 AI 看成一块五层蛋糕。我们在整个五层中都有生态系统。我们努力尽可能少做，但事实证明，我们必须做的那一部分极其困难。我不认为那会被商品化。
In fact, I also don’t think the enterprise software companies, the tools makers… Most software companies today are tool makers. Some of them are not. Some of them are workflow codification systems. But for a lot of companies, they’re tool makers. For example, Excel is a tool, PowerPoint is a tool, Cadence makes tools, Synopsys makes tools. I actually see the opposite of what people see. I think the number of agents is going to grow exponentially, and the number of tool users is going to grow exponentially. It’s very likely that the number of instances of all these tools is going to skyrocket.
事实上，我也不认为企业软件公司、工具制造商会……今天大多数软件公司都是工具制造商。当然有些不是，有些是工作流编码系统。但对很多公司来说，它们就是工具制造商。比如，Excel 是工具，PowerPoint 是工具，Cadence 做工具，Synopsys 也做工具。我看到的其实和很多人看到的正好相反。我认为智能体的数量会指数级增长，工具使用者的数量也会指数级增长。所有这些工具的实例数量，很可能会急剧飙升。
It’s very likely that the number of instances of Synopsys Design Compiler is going to skyrocket, along with the number of agents using the floor planners, our layout tools, and our design rule checkers. Today we’re limited by the number of engineers. Tomorrow, those engineers are going to be supported by a bunch of agents. We’re going to be exploring the design space like you’ve never seen before, and we’re going to use the tools that we use today.
Synopsys Design Compiler 的实例数量很可能会急剧飙升，与此同时，使用布局规划工具、我们的版图工具以及设计规则检查工具的智能体数量也会大幅增加。今天，我们受限于工程师的数量。明天，这些工程师会得到一大批智能体的支持。我们将以前所未见的方式探索设计空间，而且我们会使用今天正在使用的这些工具。
I think tool use is going to cause the software companies to skyrocket. The reason why it hasn’t happened yet is because the agents aren’t good enough at using their tools yet. Either these companies are going to build the agents themselves, or agents are going to get good enough to be able to use those tools. I think it’s going to be a combination of both.
我认为，工具使用会推动软件公司急剧增长。之所以现在还没有发生，是因为智能体还不够擅长使用这些工具。要么这些公司会自己构建智能体，要么智能体会变得足够好，能够使用这些工具。我认为最后会是两者的结合。
Dwarkesh Patel
I think in your latest filings, you had almost a $100 billion in purchase commitments with foundries, memory, and packaging. SemiAnalysis has reported that you will have $250 billion of these kinds of purchase commitments. One interpretation is that Nvidia’s moat is really that you’ve locked up many years of these scarce components. Somebody else might have an accelerator, but can they actually get the memory to build it? Can they actually get the logic to build it? Is this really Nvidia’s big moat for the next few years?
我记得在你们最新的申报文件中，你们对晶圆代工、存储器和封装的采购承诺接近 1000 亿美元。SemiAnalysis 报道称，你们会有 2500 亿美元这类采购承诺。有一种解读是，Nvidia 真正的护城河在于，你们已经锁定了未来多年的这些稀缺组件。别人也许能做出加速器，但他们真的能拿到制造它所需的存储器吗？他们真的能拿到制造它所需的逻辑芯片产能吗？这是不是 Nvidia 未来几年真正的大护城河？
Jensen Huang
It’s one of the things that we can do that is hard for someone else to do. We’ve made enormous commitments upstream. Some of it is explicit, these commitments that you mentioned. Some of it is implicit. For example, a lot of the investments that are upstream are made by our supply chain because I said to the CEOs, “Let me tell you how big this industry is going to be, let me explain to you why, let me reason through it with you, and let me show you what I see.”
这是我们能做到、而别人很难做到的事情之一。我们已经对上游作出了巨大的承诺。其中一部分是明确的，也就是你刚才提到的那些承诺；另一部分是隐性的。比如，上游很多投资是我们的供应链做出的，因为我对那些首席执行官说：“让我告诉你这个行业将会变得多大，让我向你解释为什么，让我和你一起推演一遍，并且让我把我所看到的东西展示给你。”
As a result of that process of informing, inspiring, and aligning with CEOs of all different industries upstream, they’re willing to make the investments. Why are they willing to make the investments for me and not someone else? The reason for that is because they know that I have the capacity to buy their supply and sell it through my downstream. The fact is that Nvidia’s downstream supply chain and our downstream demand is so large, they’re willing to make the investment upstream.
正是通过这个向上游各个不同行业的首席执行官传递信息、激发信心并达成一致的过程，他们才愿意进行这些投资。为什么他们愿意为我投资，而不是为别人投资？原因在于，他们知道我有能力买下他们的供应，并通过我的下游把它卖出去。事实是，Nvidia 的下游供应链和我们的下游需求规模如此之大，所以他们愿意在上游进行投资。
If you look at GTC, people are marveled by the scale of it and the people that go. It’s a full 360 degrees, the entire universe of AI all in one place. They’re all in one place because they need to see each other. I bring them together so that the downstream can see the upstream, the upstream can see the downstream, and all of them can see the advances in AI. Very importantly, they can all meet the AI natives, all the AI startups being built, and all the amazing things happening so they can see firsthand all the things that I tell them. I spend a lot of my time informing, directly or indirectly, our supply chain, partners, and ecosystem about the opportunity in front of us.
如果你看看 GTC，人们会惊叹于它的规模，以及到场的人群。它是一个完整的 360 度场景，整个 AI 宇宙都聚在一个地方。他们之所以聚在同一个地方，是因为他们需要彼此看见。我把他们聚到一起，让下游能够看见上游，让上游能够看见下游，也让所有人都能看见 AI 的进展。非常重要的是，他们都可以见到 AI 原住民，见到所有正在创建的 AI 初创公司，见到所有正在发生的惊人事情，这样他们就能亲眼看到我告诉他们的一切。我花了大量时间，直接或间接地向我们的供应链、合作伙伴和生态系统说明我们面前的机会。
Some people always say, “Jensen, in most keynotes, it’s one announcement after another.” With our keynotes, there’s always a part of it that’s a little torturous in the sense that it almost comes across like education. In fact, that’s exactly on my mind. I need to make sure the entire supply chain, upstream and downstream, the ecosystem, understands what is coming at us, why it’s coming, when it’s coming, how big it’s going to be, and is able to reason about it systematically, just like I reason about it.
有些人总是说：“Jensen，大多数主题演讲都是一个公告接着一个公告。”但我们的主题演讲里，总有一部分有点折磨人，因为它听起来几乎像是在教育。事实上，我脑子里想的正是这个。我必须确保整个供应链，包括上游和下游，以及整个生态系统，都理解正在向我们走来的是什么，为什么会来，什么时候会来，规模会有多大，并且能够像我一样对它进行系统性推理。
Regarding the moat as you describe it, we’re able to build for a future. If our next several years are a trillion dollars in scale, we have the supply chain to do it. Without our reach, the velocity of our business… Just as there’s cash flow, there’s supply chain flow, there’s churns. Nobody is going to build a supply chain for an architecture if the business churns are low. Our ability to sustain the scale is only because our downstream demand is so great. And they see it, they hear about it, they see it all coming. That allows us to do the things we’re able to do at the scale we do them.
至于你所描述的护城河，我们能够为未来建设。如果未来几年我们的业务规模达到一万亿美元，我们有供应链去实现它。如果没有我们的触达能力，没有我们业务的速度……就像有现金流一样，也有供应链流动，也有周转。没有人会为一种架构建设供应链，如果这个业务的周转很低的话。我们之所以能够维持这种规模，只是因为我们的下游需求非常巨大。他们看得到，听得到，也看见这一切正在到来。这让我们能够以现在这样的规模去做我们能够做的事。
从技术上的优势走向供应链的优势，这是“实现企业全部潜力的能力”。
Dwarkesh Patel
I do want to understand more concretely whether the upstream can keep up. For many years now, you guys have been 2x-ing revenue year over year. You’ve been more than tripling the amount of flops you’re providing to the world year over year.
我确实想更具体地理解，上游到底能不能跟上。多年来，你们的收入一直在逐年翻倍。你们每年向世界提供的浮点运算量增长超过三倍。
Jensen Huang
And 2x-ing at this scale now is really incredible.
而在现在这个规模上继续翻倍，确实非常惊人。
Dwarkesh Patel
Exactly. But then you look at logic. You’re the biggest customer on TSMC’s N3 node, and you’re one of the biggest on N2. AI as a whole this year is going to be sixty percent of N3. It’s going to be 86% next year, according to SemiAnalysis. How do you double if you’re the majority? And how do you do that year over year? Are we in a regime now where the growth rate in AI compute has to slow because of upstream? Do you see a way to get around this? How do we build 2x more fabs year over year, ultimately?
正是如此。但你再看逻辑芯片。你们是 TSMC N3 节点最大的客户之一，也是 N2 节点最大的客户之一。根据 SemiAnalysis 的数据，今年 AI 整体将占 N3 产能的 60%；明年会达到 86%。如果你已经是多数需求方，怎么还能翻倍？又怎么能年复一年地做到这一点？我们现在是不是已经进入了这样一种状态：由于上游限制，AI 计算能力的增长率必须放缓？你是否看到了绕开这个限制的办法？归根到底，我们怎么可能每年多建出两倍的晶圆厂？
Jensen Huang
At some level, the instantaneous demand is greater than the supply upstream and downstream in the world. At any instant, we could be limited by the number of plumbers, which actually happens.
在某种层面上，全球上游和下游的即时需求都大于供给。在任何一个具体时点，我们都可能受限于水管工的数量，而这实际上确实会发生。
Dwarkesh Patel
The plumbers are invited to next year’s GTC.
水管工会被邀请参加明年的 GTC。
Jensen Huang
By the way, great idea. But that’s a good condition. You want an industry where the instantaneous demand is greater than the total supply of the industry. The opposite is obviously less good. If we’re too far apart, if one particular component is too far away, the industry swarms it. For example, notice people aren’t talking very much about CoWoS anymore.
顺便说一句，这是个好主意。但这是一种好的状态。你希望一个行业里的即时需求大于整个行业的总供给。反过来显然就没那么好。如果供需差得太远，如果某一个特定组件严重落后，整个行业就会蜂拥而上去解决它。比如，你会注意到，现在人们已经不怎么谈 CoWoS 了。
The reason for that is because for two years we swarmed the living daylights out of it. We doubled, doubled, doubled on several doubles. Now I think we’re in fairly good shape. TSMC now knows that CoWoS supply has to keep up with the rest of the logic demand and the memory demand. They’re scaling CoWoS and future packaging technologies at the same level as they scale logic. This is terrific, because for a long time, CoWoS and HBM memory were rather specialty. But they’re not specialties anymore. People now realize they’re mainstream computing technology.
原因是，过去两年里，我们几乎把它围攻到极限。我们一倍又一倍地扩产，连续翻了好几倍。现在我认为情况已经相当不错。TSMC 现在知道，CoWoS 供应必须跟上其余逻辑芯片需求和存储器需求。他们正在以扩展逻辑芯片产能同样的层级，扩展 CoWoS 和未来封装技术。这非常好，因为在很长一段时间里，CoWoS 和 HBM 存储器都更像是专门化技术。但它们现在不再是专门化技术了。人们现在意识到，它们是主流计算技术。
Of course, we’re now much more able to influence a larger scope of our supply chain. At the beginning of the AI revolution, all the things that I say now, I was saying five years ago. Some people believed in it and invested in it, for example, Sanjay and the Micron team. I still remember the meeting really well where I was clear about exactly what was going to happen, why it was going to happen, and the predictions of today. They really doubled down on it. We partnered with them across LPDDR and HBM memories, and they really invested in it. It obviously has been tremendous for the company. Some people came a little bit later, but now they’re all here.
当然，我们现在更有能力影响供应链中更大范围的环节。在 AI 革命刚开始的时候，我现在说的所有这些话，五年前我就在说了。有些人相信了，并且进行了投资，比如 Sanjay 和 Micron 团队。我仍然清楚记得那次会议，当时我非常明确地说明了到底会发生什么、为什么会发生，以及对今天的预测。他们确实加倍下注了。我们和他们在 LPDDR 与 HBM 存储器上进行了合作，他们也真正投入了资源。显然，这对这家公司来说非常巨大。有些人来得稍晚一些，但现在他们都已经到场了。
Each one of these bottlenecks gets a great deal of attention. Now we’re prefetching the bottlenecks years in advance. For example, the investments that we’ve done with Lumentum, Coherent, and the silicon photonics ecosystem over the last several years really reshaped the supply chain. We built up an entire supply chain around TSMC. We partnered with them on COUPE, invented a whole bunch of technology, and licensed those patents to the supply chain to keep it nice and open.
这些瓶颈中的每一个都会得到大量关注。现在，我们会提前几年预取这些瓶颈。比如，过去几年里，我们与 Lumentum、Coherent 以及硅光子生态系统所做的投资，真正重塑了供应链。我们围绕 TSMC 建立了一整套供应链。我们与他们在 COUPE 上合作，发明了一大批技术，并把这些专利授权给供应链，以保持它足够开放。
We’re preparing the supply chain through the invention of new technologies, new workflows, new testing equipment like double-sided probing, investing in companies, and helping them scale up their capacity. You can see that we’re trying to shape the ecosystem so that the supply chain is ready to support the scale.
我们正在通过发明新技术、新工作流、新测试设备，比如双面探针测试，通过投资公司，并帮助它们扩大产能，来为供应链做准备。你可以看到，我们正在努力塑造这个生态系统，使供应链准备好支撑这种规模。
Dwarkesh Patel
It seems like some bottlenecks are easier than others. Scaling up CoWoS versus scaling up—
看起来有些瓶颈比另一些更容易解决。扩大 CoWoS 产能，相比于扩大——
Jensen Huang
I went to the hardest one, by the way.
顺便说一句，我刚才说的是最难的那个。
Dwarkesh Patel
Which is?
哪一个？
Jensen Huang
Plumbers. Plumbers and electricians. This is one of the concerns that I have about the doomers describing the end of work and killing of jobs. If we discourage people from being software engineers, we’re going to run out of software engineers. The same prediction happened ten years ago. Some of the doomers were telling people, “Whatever you do, don’t be a radiologist.” You might hear some of those videos still on the web saying radiology is going to be the first career to go and the world is not going to need any more radiologists. Guess what we’re short of? Radiologists.
水管工。水管工和电工。这也是我对那些末日论者所说的“工作终结”和“岗位被消灭”感到担忧的原因之一。如果我们劝阻人们不要成为软件工程师，我们就会缺少软件工程师。十年前也发生过同样的预测。有些末日论者告诉人们：“无论你做什么，千万不要当放射科医生。”你现在可能还能在网上看到一些那类视频，说放射科会是第一个消失的职业，世界不再需要更多放射科医生。猜猜我们现在缺什么？放射科医生。
Dwarkesh Patel
Going back to this point about how some things you can scale, and other things… How do you actually manufacture 2x the amount of logic a year? Ultimately, memory and logic are bottlenecked by EUV. How do you get to 2x as many EUV machines year over year?
回到刚才关于有些东西可以扩张，而另一些东西……你们到底怎么每年制造两倍数量的逻辑芯片？归根到底，存储器和逻辑芯片都受限于 EUV。你怎么做到 EUV 机器数量逐年翻倍？
Jensen Huang
None of that is impossible to scale quickly. All of that is easy to do within two or three years. You just need a demand signal. Once you can build one, you can build ten, and once you can build ten, you can build a million. These things are not hard to replicate.
这些东西没有一个是不可能快速扩张的。所有这些在两三年内都很容易做到。你只需要一个需求信号。一旦你能造出一台，你就能造出十台；一旦你能造出十台，你就能造出一百万台。这些东西并不难复制。
Dwarkesh Patel
How far down the supply chain do you go? Do you go to ASML and say, “Hey, if I look out three years from now, for Nvidia to be generating two trillion a year in revenue, we need way more EUV machines”?
你们会深入供应链到什么程度？你会不会去找 ASML，说：“嘿，如果我往后三年看，为了让 Nvidia 每年创造两万亿美元收入，我们需要多得多的 EUV 机器”？
Jensen Huang
Some of them I have to directly, some of them indirectly, and some of them… If I can convince TSMC, ASML will be convinced. We have to think about the critical pinch points. But if TSMC is convinced, you’ll have plenty of EUV machines in a few years.
有些我必须直接去做，有些可以间接去做，还有一些……如果我能说服 TSMC，ASML 就会被说服。我们必须思考关键的卡点在哪里。但如果 TSMC 被说服了，几年后你就会有足够多的 EUV 机器。
My point is that none of the bottlenecks last longer than a couple of years, two, three years, none of them. Meanwhile, we’re improving computing efficiency by 10x 20x, and in the case of Hopper to Blackwell, 30x to 50x. We’re coming up with new algorithms because CUDA is so flexible. We’re developing all kinds of new techniques so that we drive efficiency in addition to increasing capacity. None of those things worry me. It’s the stuff that’s downstream from us. Energy policies that prevent energy from… You can’t create an industry without energy. You can’t create a whole new manufacturing industry without energy.
我的意思是，没有任何瓶颈会持续超过几年，两年、三年，一个都不会。与此同时，我们正在把计算效率提高 10 倍、20 倍；从 Hopper 到 Blackwell 的情况下，是 30 倍到 50 倍。因为 CUDA 非常灵活，我们正在提出新算法。我们正在开发各种新技术，所以除了增加产能之外，还能提高效率。这些事情都不让我担心。让我担心的是我们下游的东西。那些阻止能源发展的能源政策……没有能源，你不可能创造一个产业。没有能源，你不可能创造一个全新的制造业。
We want to reindustrialize the United States. We want to bring back chip manufacturing, computer manufacturing, and packaging. We want to build new things like EVs and robots. We want to build AI factories. You can’t build any of these things without energy, and those things take a long time. More chip capacity, that’s a 2-3 year problem. More CoWoS capacity, 2-3 year problem.
我们希望让美国重新工业化。我们希望把芯片制造、计算机制造和封装带回来。我们希望制造电动车和机器人这类新东西。我们希望建设 AI 工厂。没有能源，这些事情一个都建不起来，而这些事情都需要很长时间。更多芯片产能，这是一个两到三年的问题。更多 CoWoS 产能，也是两到三年的问题。
Dwarkesh Patel
Interesting. I feel like I have guests tell me the exact opposite thing sometimes. In this case, I just don’t have the technical knowledge to adjudicate.
有意思。我感觉有时候我的嘉宾会告诉我完全相反的事情。在这个问题上，我确实没有足够的技术知识来裁判谁对谁错。
Jensen Huang
The beautiful thing is you’re talking to the expert.
好处在于，你现在正在和专家谈。
00:16:25 – Will TPUs break Nvidia’s hold on AI compute?
00:16:25 – TPU 会打破 Nvidia 对 AI 计算的掌控吗？
Dwarkesh Patel
True. I want to ask about your competitors. If you look at the TPU, arguably two out of the top three models in the world, Claude and Gemini, were trained on TPU. What does that mean for Nvidia going forward?
确实。我想问问你们的竞争对手。如果你看 TPU，可以说世界前三大模型中有两个，Claude 和 Gemini，都是在 TPU 上训练出来的。这对 Nvidia 未来意味着什么？
Jensen Huang
We build a very different thing. What Nvidia built is accelerated computing, not a tensor processing unit. Accelerated computing is used for all kinds of things: molecular dynamics, quantum chromodynamics, data processing, data frames, structured data, and unstructured data. It’s also used for fluid dynamics and particle physics. In addition, we use it for AI.
我们构建的是非常不同的东西。Nvidia 构建的是加速计算，而不是张量处理单元。加速计算可以用于各种事情：分子动力学、量子色动力学、数据处理、数据框架、结构化数据和非结构化数据。它也用于流体动力学和粒子物理。此外，我们还把它用于 AI。
Accelerated computing is much more diverse. Although AI is the conversation today and is obviously very important and impactful, computing is much broader than that. Nvidia has reinvented the way computing is done, moving from general-purpose computing to accelerated computing. Our market reach is far greater than any TPU or ASIC can possibly have. If you look at our position, we’re the only company that accelerates applications of all kinds. We have a gigantic ecosystem. So all kinds of frameworks and algorithms run on Nvidia.
加速计算的用途要多样得多。虽然今天大家谈论的是 AI，而且 AI 显然非常重要、影响巨大，但计算本身比 AI 宽广得多。Nvidia 重新发明了计算的方式，把计算从通用计算推进到加速计算。我们的市场触达范围远远超过任何 TPU 或 ASIC 可能达到的范围。如果你看我们的定位，我们是唯一一家能够加速各种应用的公司。我们拥有一个巨大的生态系统。因此，各种框架和算法都运行在 Nvidia 上。
Because our computers are designed to be operated by other people, anyone who’s an operator can buy our systems. With most of these home-built systems, you have to be your own operator because they were never designed to be flexible enough for others to operate. Because anybody can operate our systems, we’re in every cloud, including Google, Amazon, Azure, and OCI.
因为我们的计算机从设计之初就是为了让别人来运营，所以任何运营方都可以买我们的系统。对于大多数自建系统来说，你必须自己当运营方，因为它们从来没有被设计成足够灵活、能让别人运营的系统。正因为任何人都可以运营我们的系统，所以我们进入了每一个云，包括 Google、Amazon、Azure 和 OCI。
If you want to operate it to rent, you better have a large ecosystem of customers in many industries to be the offtakers. If you want to operate it for yourself, we obviously have the ability to help you operate it yourself, like we did for Elon with xAI. And because we can enable operators in any company and any industry, you could use it to build a supercomputer for scientific research and drug discovery at Lilly. We can help them operate their own supercomputer and use it for the entire diversity of drug discovery and biological sciences that we accelerate.
如果你想运营它并出租，你最好拥有一个横跨许多行业的大型客户生态系统，作为承购方。如果你想自己运营，我们显然也有能力帮助你自己运营，就像我们为 Elon 的 xAI 所做的那样。并且，因为我们能够赋能任何公司、任何行业的运营方，所以你可以用它在 Lilly 建设一台用于科学研究和药物发现的超级计算机。我们可以帮助他们运营自己的超级计算机，并把它用于我们所加速的药物发现和生物科学中的各种不同任务。
There are just a whole bunch of applications that we can address that you can’t do with TPUs. Nvidia built CUDA to be a fantastic tensor processing unit as well, but it also handles every life cycle of data processing, computing, AI, and so on. Our market opportunity is just a lot larger, and our reach is a lot greater. Because we support every application in the world now, you can build Nvidia systems anywhere and know that there will be customers for it. It’s a very different thing.
有一大批应用是我们能够处理、而你无法用 TPU 处理的。Nvidia 构建 CUDA，使它也能成为一个出色的张量处理单元，但它同时还处理数据处理、计算、AI 等等的每一个生命周期。我们的市场机会要大得多，我们的触达范围也要大得多。因为我们现在支持世界上每一种应用，你可以在任何地方建设 Nvidia 系统，并且知道一定会有客户需要它。这是一件非常不同的事情。
技术层面的问题不好理解，但是专业化很大可能上会战胜多元化，Google在这个业务上显然是不够专注的，苹果能做成封闭式的系统，Google想做不等于能做到。
Dwarkesh Patel
This is going to be a long question. You have spectacular revenue, and you’re not making $60 billion a quarter from pharma and quantum. You’re making it because AI is an unprecedented technology that is growing unprecedentedly fast.
这个问题会比较长。你们的收入非常惊人，而你们每个季度赚到 600 亿美元，并不是靠制药和量子计算。你们之所以能做到，是因为 AI 是一种前所未有的技术，而且正在以前所未有的速度增长。
The question then is what is best for AI specifically. I’m not in the details, but I talk to my AI researcher friends and they say, “Look, when I use a TPU, it’s this big systolic array that’s perfect for doing matrix multiplies, whereas a GPU is very flexible. It’s great when you have lots of branching or irregular memory access.”
那么问题就是，具体对 AI 来说，什么才是最好的。我不了解细节，但我和做 AI 研究的朋友聊过，他们会说：“你看，当我使用 TPU 时，它是一个巨大的脉动阵列，非常适合做矩阵乘法；而 GPU 非常灵活，当你有大量分支或不规则内存访问时，它很强。”
But what is AI? It’s just these very predictable matrix multiplies again and again and again. You don’t have to give up any die area for warp schedulers or switches between threads and memory banks. And the TPU is really optimized for the bulk of this growth in revenue and use case for compute that is coming online right now. I wonder how you react to that.
但 AI 到底是什么？它不就是这些高度可预测的矩阵乘法，一遍又一遍、反复执行吗？你不需要为线程束调度器，或者线程与内存库之间的切换，牺牲任何裸片面积。而 TPU 确实是为当前正在上线的这部分计算收入增长和使用场景中的主体需求而优化的。我想知道你怎么看这个说法。
Jensen Huang
Matrix multiplies are an important part of AI, but they’re not the only part. If you want to come up with a new attention mechanism, disaggregate in a different way, or invent a whole new type of architecture altogether—like a hybrid SSM—you want an architecture that’s generally programmable. If you want to create a model that fuses diffusion and autoregressive techniques, you want an architecture that’s just generally programmable. We run everything you can imagine. That’s the advantage. It allows for the invention of new algorithms a lot more easily, because it’s a programmable system.
矩阵乘法是 AI 的重要组成部分，但它不是全部。如果你想提出一种新的注意力机制，想用不同方式进行解耦，或者发明一种全新的架构——比如混合 SSM——你就会想要一种通用可编程的架构。如果你想创造一个融合扩散技术和自回归技术的模型，你也会想要一种通用可编程的架构。你能想到的所有东西，我们都能运行。这就是优势。因为它是一个可编程系统，所以它让新算法的发明变得容易得多。
The ability to invent new algorithms is really what makes AI advance so quickly. TPUs, like anything else, are impacted by Moore’s Law, which we know is increasing by about 25% per year. The only way to really get 10x or 100x leaps is to fundamentally change the algorithm and how it’s computed every single year.
发明新算法的能力，才是真正让 AI 进步如此之快的东西。TPU 和其他任何东西一样，也受到 Moore’s Law 的影响，而我们知道，它每年大约提升 25%。真正实现 10 倍或 100 倍跃迁的唯一方法，是每一年都从根本上改变算法，以及改变它的计算方式。
That’s Nvidia’s fundamental advantage. The only reason we were able to make Blackwell to Hopper 50x… When I first announced Blackwell was going to be 35x more energy efficient than Hopper, nobody believed it. Then Dylan wrote an article saying I sandbagged, and it’s actually fifty times. You can’t reasonably do that with just Moore’s Law. The way we solve that problem is with new models, like MoEs, that are parallelized, disaggregated, and distributed across a computing system. Without the ability to really get down and come up with new kernels with CUDA, it’s really hard to do.
这就是 Nvidia 的根本优势。我们之所以能够让 Blackwell 相比 Hopper 提升 50 倍……当我最初宣布 Blackwell 的能效将比 Hopper 高 35 倍时，没有人相信。后来 Dylan 写了一篇文章，说我其实是保守说法，实际是 50 倍。仅靠 Moore’s Law，你不可能合理地做到这一点。我们解决这个问题的方式，是使用新的模型，比如 MoE，把它们并行化、解耦，并分布到一个计算系统中。没有真正深入下去、用 CUDA 写出新内核的能力，这件事很难做到。
It’s the combination of the programmability of our architecture and the fact that Nvidia is an extreme co-design company. We can even offload some of the computation into the fabric itself, like NVLink, or into the network with Spectrum-X. We could affect change across the processors, the system, the fabric, the libraries, and the algorithm simultaneously. Without CUDA to do that, I wouldn’t even know where to start.
这是我们架构的可编程性，与 Nvidia 作为一家极致协同设计公司的结合。我们甚至可以把部分计算卸载到互连结构本身，比如 NVLink，或者通过 Spectrum-X 卸载到网络中。我们能够同时在处理器、系统、互连结构、库和算法层面推动变化。没有 CUDA 来做这些，我甚至不知道该从哪里开始。
Dwarkesh Patel
This gets at an interesting question about Nvidia’s clientele. 60% of your revenue is coming from these big five hyperscalers. In a different era with different customers—let’s say professors running experiments—they need CUDA. They can’t use another accelerator. They just needed to run PyTorch with CUDA and have everything optimized.
这就引出了一个关于 Nvidia 客户群的有趣问题。你们 60% 的收入来自这五大超大规模云服务商。在另一个时代，客户不同——比如教授们在做实验——他们确实需要 CUDA。他们不能使用其他加速器。他们只是需要用 CUDA 跑 PyTorch，并让一切都被优化好。
But these hyperscalers have the resources to write their own kernels. In fact, they have to in order to get that last 5% of performance they need for their specific architecture. Anthropic and Google are mostly running their own accelerators or running TPUs and Trainium. But even OpenAI, using GPUs, has Triton because they need their own kernels. Down to CUDA C++, instead of using cuBLAS and NCCL, they’ve got their own stack which compiles to other accelerators as well. If most of your customers can and do make replacements for CUDA, to what extent is CUDA really the thing that is going to make frontier AI happen on Nvidia?
但这些超大规模云服务商有资源编写自己的内核。事实上，为了在其特定架构上获得最后 5% 的性能，他们也必须这样做。Anthropic 和 Google 大多运行自己的加速器，或者运行 TPU 和 Trainium。但即使是使用 GPU 的 OpenAI，也有 Triton，因为他们需要自己的内核。一直下探到 CUDA C++ 层面，他们不是使用 cuBLAS 和 NCCL，而是拥有自己的软件栈，并且这个软件栈也可以编译到其他加速器上。如果你们的大多数客户能够、也确实在做 CUDA 的替代品，那么 CUDA 到底在多大程度上，仍然是让前沿 AI 发生在 Nvidia 上的关键？
Jensen Huang
CUDA is a rich ecosystem. If you want to build on any computer first, building on CUDA first is incredibly smart. Because the ecosystem is so rich, we support every framework. If you want to create custom kernels… For example, we contribute enormously to Triton. So the back end of Triton has huge amounts of Nvidia technology.
CUDA 是一个丰富的生态系统。如果你想先在任何一种计算机上构建东西，先在 CUDA 上构建是非常聪明的。因为这个生态系统非常丰富，我们支持每一种框架。如果你想创建自定义内核……比如，我们对 Triton 做出了巨大贡献。所以 Triton 的后端包含大量 Nvidia 技术。
We’re delighted to help every framework become as great as it can be. There are lots and lots of frameworks. There’s Triton, vLLM, SGLang, and more. Now there’s a whole bunch of new reinforcement learning frameworks coming out, like verl and NeMo RL. With post-training and reinforcement learning, that entire area is just exploding. So if you want to build on an architecture, building on CUDA makes the most sense because you know the ecosystem is great.
我们很乐意帮助每一个框架变得尽可能好。框架非常多。有 Triton、vLLM、SGLang，还有更多。现在又出现了一大批新的强化学习框架，比如 verl 和 NeMo RL。随着后训练和强化学习的发展，整个领域正在爆发。所以，如果你想在某种架构上构建东西，在 CUDA 上构建最有道理，因为你知道这个生态系统很强大。
You know that if something happens, it’s more likely in your code and not in the mountain of code underneath. Don’t forget the amount of code you’re dealing with when building these systems. When something doesn’t work, was it you or was it the computer? You would like it to always be you and to be able to trust the computer. Obviously, we still have lots of bugs ourselves, but our system is so well wrung out that you can at least build on top of the foundation. That’s number one: the richness, programmability, and capability of the ecosystem.
你知道，如果出了问题，更可能是在你自己的代码里，而不是在底层那座代码大山里。不要忘了，在构建这些系统时，你要处理的代码量有多大。当某件事不能正常工作时，是你的问题，还是计算机的问题？你会希望它永远是你的问题，并且能够信任计算机。显然，我们自己仍然有很多缺陷，但我们的系统已经经过了充分锤炼，至少你可以在这个基础之上构建。这是第一点：生态系统的丰富性、可编程性和能力。
The second thing is, if you’re a developer building anything at all, the single most important thing you want is an install base. You want the software you write to run on a whole bunch of other computers. You’re not building software just for yourself. You’re building it for your fleet or everybody else’s fleet because you’re a framework builder. Nvidia’s CUDA ecosystem is ultimately its great treasure.
第二点是，如果你是一个开发者，不管你在构建什么，你最想要的一件事就是装机基础。你希望自己写的软件能够在大量其他计算机上运行。你不是只为自己构建软件。你是在为自己的机群，或者为其他所有人的机群构建软件，因为你是框架构建者。Nvidia 的 CUDA 生态系统，归根到底就是它巨大的宝藏。
We have several hundred million GPUs out there now. Every cloud has it. It goes back to the A10, A100, H100, H200, the L series, the P series. There’s a whole bunch of them. They’re in all kinds of sizes and shapes. If you’re a robotics company, you want that CUDA stack to actually run in the robot itself. We’re literally everywhere. The install base means that once you develop the software or the model, it’s going to be useful everywhere. That is just incredibly valuable.
现在外面已经有数亿块 GPU。每一个云都有它。从 A10、A100、H100、H200，到 L 系列、P 系列，都包括在内。数量非常多，形态和规模也各不相同。如果你是一家机器人公司，你会希望 CUDA 软件栈真的能在机器人本体上运行。我们几乎无处不在。装机基础意味着，一旦你开发出软件或模型，它就会在各处都有用。这一点极其有价值。
Lastly, the fact that we’re in every single cloud makes us genuinely unique. If you’re an AI company or developer, you’re not exactly sure which cloud service provider you’re going to partner with or where you’d like to run it. We run everywhere, including on-prem for you if you like. The combination of the richness of the ecosystem, the expansiveness of the install base, and the versatility of where we are makes CUDA invaluable.
最后，我们进入了每一个云，这让我们真正独特。如果你是一家 AI 公司或一个开发者，你不一定确定自己会和哪家云服务商合作，或者希望在哪里运行它。我们到处都能运行，如果你愿意，也可以在你的本地部署中运行。生态系统的丰富性、装机基础的广泛性，以及我们所在位置的多样性，三者结合起来，让 CUDA 具有不可估量的价值。
Dwarkesh Patel
That makes a lot of sense. I guess the thing I’m curious about is whether those advantages matter a lot to your main customers. There’s many people for whom they might matter. The kind of person who can actually build their own software stack makes up most of your revenue. Especially if you go to a world where AI is getting especially good at the things which have tight verification loops where you can RL on them…. This question of how do you write a kernel that does attention or MLP the most efficiently across a scale up? It’s a very verifiable sort of feedback loop.
这很有道理。我想我好奇的是，这些优势对你们的主要客户来说到底有多重要。确实有很多人会在意这些优势。但真正能够构建自己软件栈的那类人，构成了你们收入的大部分。尤其是如果我们进入一个世界，在那里 AI 尤其擅长那些拥有紧密验证循环、可以在其上进行强化学习的任务……比如，如何写一个内核，让它在扩展系统上最高效地执行注意力机制或 MLP？这是一个反馈循环非常容易验证的问题。
Can all the hyperscalers write these custom kernels for themselves? Nvidia still has great price performance, so they might still prefer to use Nvidia. But then the question is, does it just become a question of who is offering the best specs, the best flops and memory bandwidth for a given dollar. Whereas historically Nvidia has just had, and still has, the best margins in all of AI across hardware and software, +70%, because of this CUDA moat. And the question is, can you sustain those margins if for most of your customers, they can actually afford to build, instead of the CUDA moat?
所有这些超大规模云服务商能不能都为自己编写这些自定义内核？Nvidia 仍然拥有很好的性价比，所以他们可能仍然更愿意使用 Nvidia。但接下来的问题是，这会不会最终变成一个问题：谁能在每一美元价格下提供最好的规格、最好的浮点运算能力和内存带宽？而历史上，Nvidia 在整个 AI 的硬件和软件领域一直拥有、现在也仍然拥有最高的利润率，超过 70%，原因就是 CUDA 护城河。问题是，如果对大多数客户来说，他们实际上有能力自己构建，从而绕开 CUDA 护城河，你们还能不能维持这些利润率？
Jensen Huang
The number of engineers we have assigned to these AI labs is insane, working with them, optimizing their stack. The reason for that is because nobody knows our architecture better than we do. These architectures are not as general purpose as a CPU. A CPU is kind of like a Cadillac. It’s a nice cruiser. It never goes too fast. Everybody drives it pretty well. It’s got cruise control, and everything’s easy. But in a lot of ways, Nvidia’s GPUs, accelerators, are like F1 racers. I could imagine everybody’s able to drive it at a hundred miles an hour, but it takes quite a bit of expertise to be able to push it to the limit. We use a ton of AI to create the kernels that we have.
我们派给这些 AI 实验室的工程师数量非常惊人，他们和这些实验室一起工作，优化它们的软件栈。原因在于，没有人比我们更了解自己的架构。这些架构不像 CPU 那样通用。CPU 有点像 Cadillac，是一辆不错的巡航车，速度从来不会太快，大家都能开得很好，有定速巡航，一切都很容易。但在很多方面，Nvidia 的 GPU、加速器，更像是 F1 赛车。我可以想象，每个人都能把它开到每小时 100 英里，但要把它推到极限，就需要相当多的专业能力。我们使用大量 AI 来创建我们拥有的那些内核。
I’m pretty sure we’re going to still be needed for quite some time. Our expertise helps our AI lab partners to get another 2x out of their stack easily oftentimes. It’s not unusual that by the time we’re done optimizing their stack or optimizing a particular kernel, their model sped up by 3x, 2x, 50%. That’s a huge number, especially when you’re talking about the install base of the fleet that they have, of all the Hoppers and Blackwells that they have. When you increase it by a factor of two, that doubles the revenues. That directly translates to revenues.
我很确定，在相当长一段时间里，他们仍然会需要我们。我们的专业能力常常能轻松帮助 AI 实验室合作伙伴从它们的软件栈里再挤出 2 倍性能。当我们完成对它们的软件栈，或者某个特定内核的优化后，它们的模型速度提升 3 倍、2 倍、50%，这并不少见。这是一个巨大的数字，尤其是当你谈到它们已有机群的装机基础，谈到它们拥有的所有 Hopper 和 Blackwell 时。当你把它提升一倍时，收入也会翻倍。这会直接转化为收入。
Nvidia’s computing stack is the best performance per TCO in the world, bar none. Nobody can demonstrate to me that any single platform in the world today has a better performance-TCO ratio. Not one company. In fact, the benchmarks that are out there. Dylan’s InferenceMAX is sitting out there for everybody to use, and not one… TPU won’t come, Trainium won’t come.
Nvidia 的计算软件栈，在总拥有成本下的性能表现是世界第一，没有例外。没有任何人能向我证明，今天世界上有任何一个平台拥有更好的性能与总拥有成本之比。一家公司都没有。事实上，外面已经有基准测试。Dylan 的 InferenceMAX 就摆在那里，所有人都可以用，但没有一个……TPU 不会来，Trainium 也不会来。
I encourage them to use InferenceMAX and demonstrate their incredible inference cost. It’s really hard. Nobody wants to show up. MLPerf. I would welcome Trainium to demonstrate their 40% that they claim all the time. I would love to hear them demonstrate the cost advantage of TPUs. It makes no sense in my mind. It makes absolutely zero sense. On first principles, it makes no sense.
我鼓励他们使用 InferenceMAX，展示他们所谓惊人的推理成本优势。这真的很难。没人愿意出现。还有 MLPerf。我欢迎 Trainium 来证明他们一直宣称的 40% 优势。我也很愿意听他们证明 TPU 的成本优势。在我看来，这说不通。完全说不通。从第一性原理出发，它就说不通。
So I think the reason why we’re so successful is simply because our TCO is so great. Secondly, you say 60% of our customers are the top five, but most of that business is external. For example, most of Nvidia in AWS is for external customers, not internal use. Most of our customers at Azure, obviously all of our customers are external. All of our customers at OCI are external, not internal use. The reason why they favor us is because our reach is so great. We can bring them all of the great customers in the world. They’re all built on Nvidia. And the reason why all these companies are built on Nvidia is because our reach and our versatility is so great.
所以我认为，我们之所以如此成功，原因很简单，就是我们的总拥有成本非常好。其次，你说我们 60% 的客户是前五大超大规模云服务商，但其中大部分业务是面向外部的。比如，AWS 上的大多数 Nvidia 资源是给外部客户用的，不是内部使用。我们在 Azure 的大多数客户，显然也都是外部客户。我们在 OCI 的所有客户都是外部客户，不是内部使用。他们偏好我们的原因，是我们的触达范围非常大。我们能把全世界所有优秀客户都带给他们。这些客户都建立在 Nvidia 之上。而这些公司之所以建立在 Nvidia 之上，是因为我们的触达范围和通用性非常强。
So I think the flywheel is really install base, the programmability of our architecture, the richness of our ecosystem, and the fact that there’s so many AI companies in the world. There’s tens of thousands of them now. If you were one of those AI startups, what architecture would you choose? You would choose an architecture that’s most abundant. We’re the most abundant in the world. You’d choose the one that has the largest installed base. We’re the largest install base. And you’d choose the one that has a rich ecosystem.
所以我认为，这个飞轮真正由几个部分构成：装机基础、我们架构的可编程性、我们生态系统的丰富性，以及世界上有如此多 AI 公司这一事实。现在这样的公司有数万家。如果你是这些 AI 初创公司之一，你会选择什么架构？你会选择最充足的架构。我们是世界上最充足的。你会选择装机基础最大的那个。我们的装机基础最大。你也会选择拥有丰富生态系统的那个。
So that’s the flywheel. That’s the reason why, between the combination of: one, our perf per dollar is so great that they have the lowest cost tokens. Second, our perf per watt is the highest in the world. So if one of these companies, if our partners, built a one gigawatt data center, that one gigawatt data center better deliver the maximum amount of revenues and number of tokens, which directly translates to revenues. You want it to generate as many tokens as possible, maximize the revenues for that data center. We are the highest tokens per watt architecture in the world. Lastly, if your goal is to rent the infrastructure, we have the most customers in the world. So that’s the reason why the flywheel works.
这就是飞轮。原因在于几个因素的结合：第一，我们每一美元对应的性能非常好，所以他们拥有最低成本的词元。第二，我们每瓦性能是世界最高的。所以，如果这些公司之一，如果我们的合作伙伴建设了一座一吉瓦的数据中心，那么这座一吉瓦的数据中心最好能够产生最大数量的收入和最大数量的词元，而词元会直接转化为收入。你希望它生成尽可能多的词元，最大化这座数据中心的收入。我们是世界上每瓦词元产出最高的架构。最后，如果你的目标是出租基础设施，我们拥有世界上最多的客户。所以这就是飞轮能够运转的原因。
Dwarkesh Patel
Interesting. I guess the question comes down to, what is the actual market structure here? Because even if there’s other companies… There could have been a world where there’s tens of thousands of AI companies that have roughly equal share of compute. But even through these five hyperscalers, really the people on Amazon using the compute are Anthropic, OpenAI, and these big foundation labs who can themselves afford and have the ability to make different accelerators work.
有意思。我想问题归结为，这里的实际市场结构到底是什么？因为即使还有其他公司……本来可能存在这样一个世界：有数万家 AI 公司，它们大致平均地分享计算资源。但即便通过这五大超大规模云服务商，真正使用 Amazon 计算资源的人也是 Anthropic、OpenAI，以及这些大型基础模型实验室，而它们本身有财力，也有能力让不同加速器运转起来。
Jensen Huang
No, I think your premise is wrong.
不，我认为你的前提是错的。
Dwarkesh Patel
Maybe. But let me ask you a slightly different question.
也许是。但让我换一个稍微不同的问题。
Jensen Huang
Come back and make me correct your premise.
待会儿记得让我回来纠正你的前提。
Dwarkesh Patel
Okay. Let me just ask you a different question.
好。那我先问一个不同的问题。
Jensen Huang
But still make sure to make me come back and fix because it’s just too important to AI. It’s too important to the future of science. It’s too important to the future of the industry. That premise… Look —
但还是一定要让我回来把它纠正过来，因为这对 AI 太重要了。对科学的未来太重要了。对这个行业的未来也太重要了。这个前提……你看——
Dwarkesh Patel
Let me just finish the question and then we can address it together.
让我先把问题问完，然后我们一起处理这个前提。
Jensen Huang
Yeah.
好的。
Dwarkesh Patel
If all these things are true about price, performance, and performance per watt, et cetera, are true, why do you think it is the case that, say, Anthropic for example, just announced a couple days ago they have a multi-gigawatt deal with Broadcom and Google for TPUs and majority of their compute?
如果这些关于价格、性能、每瓦性能等等的说法都是真的，那么你认为为什么会出现这样的情况：比如 Anthropic，几天前刚刚宣布与 Broadcom 和 Google 达成了一项多吉瓦级别的 TPU 合作协议，而且这将构成它们大部分计算资源？
Obviously for Google, TPU is a majority of compute. So if I look at these big AI companies, it seems like a lot of their compute… There was some point where it’s all Nvidia and now it’s not. So I’m curious how to square, if these things are true on paper, why are they going with other accelerators?
显然，对 Google 来说，TPU 是其计算资源的大部分。所以如果我看这些大型 AI 公司，它们大量计算资源……曾经有一个时点几乎全是 Nvidia，而现在不是了。所以我好奇的是，如果这些纸面上的优势都是真的，那为什么它们还会选择其他加速器？这该如何协调？
Jensen Huang
Anthropic is a unique instance, not a trend. Without Anthropic, why would there be any TPU growth at all? It’s 100% Anthropic. Without Anthropic, why would there be Trainium growth at all? It’s 100% Anthropic. I think that’s fairly well known and well understood. It’s not that there’s an abundance of ASIC opportunities. There’s only one Anthropic.
Anthropic 是一个独特案例，不是趋势。没有 Anthropic，TPU 为什么还会有任何增长？那 100% 是 Anthropic。没有 Anthropic，Trainium 为什么还会有任何增长？那 100% 是 Anthropic。我认为这一点相当广为人知，也相当容易理解。并不是说 ASIC 机会非常多。Anthropic 只有一个。
以下是前一个问题的回答。
Jensen Huang
None of that is impossible to scale quickly. All of that is easy to do within two or three years. You just need a demand signal. Once you can build one, you can build ten, and once you can build ten, you can build a million. These things are not hard to replicate.
这些东西没有一个是不可能快速扩张的。所有这些在两三年内都很容易做到。你只需要一个需求信号。一旦你能造出一台，你就能造出十台；一旦你能造出十台，你就能造出一百万台。这些东西并不难复制。
Dwarkesh Patel
But OpenAI’s deals with AMD… They’re building their own Titan accelerator.
但 OpenAI 和 AMD 的交易……他们也在构建自己的 Titan 加速器。
Jensen Huang
Yeah, but I think we could all acknowledge they’re vastly Nvidia. We’re going to still do a lot of work together. I’m not offended by other people using something else and trying things. If they don’t try these other things, how would they know how good ours is? Sometimes you’ve got to be reminded of it. We have to continuously earn the position that we’re in.
是的，但我认为我们都可以承认，他们绝大部分还是 Nvidia。我们仍然会一起做大量工作。别人使用其他东西、尝试其他东西，我并不觉得被冒犯。如果他们不尝试这些其他东西，他们怎么知道我们的东西有多好？有时候你必须被提醒一下。我们必须持续赢得自己现在所处的位置。
There are always big claims. Look at the number of ASICs that have been canceled. Just because you’re going to build an ASIC… You still have to build something better than Nvidia. It’s not that easy building something better than Nvidia. It’s not sensible, actually. Nvidia’s got to be missing something, seriously. Because of our scale, our velocity, we’re the only company in the world that’s cranking it out every single year. Big leaps, every single year.
总会有人提出很大的宣称。看看有多少 ASIC 项目已经被取消。仅仅因为你要做一个 ASIC……你仍然必须做出比 Nvidia 更好的东西。做出比 Nvidia 更好的东西，并不那么容易。实际上，这甚至不太合理。除非 Nvidia 真的漏掉了什么重要东西。因为我们的规模、我们的速度，我们是世界上唯一一家每年都能持续推出这种东西的公司。每一年，都是大幅跃迁。
Dwarkesh Patel
I guess their logic is, “Hey, it doesn’t need to be better. It just needs to be not more than 70% worse,” because they’re paying you 70% margins.
我猜他们的逻辑是：“嘿，它不需要更好。它只需要不要差超过 70% 就行。”因为他们付给你们的是 70% 的利润率。
Jensen Huang
No, don’t forget, even in ASICs margins are really quite high. Nvidia’s margin is 70%, let’s say. But ASIC margins are 65%. What are you really saving?
不，别忘了，即使是 ASIC，利润率也相当高。比如说，Nvidia 的利润率是 70%。但 ASIC 的利润率是 65%。你到底真正省下了什么？
Dwarkesh Patel
Oh, you mean from Broadcom or something like that?
哦，你是说来自 Broadcom 或者类似公司那边？
Jensen Huang
Yeah, sure. You’ve got to pay somebody. I think the ASIC margins are incredibly good, from what I can tell. They believe it too. They’re quite proud of their incredible ASIC margins.
是的，当然。你总得付钱给某个人。据我所知，ASIC 的利润率非常好。他们自己也相信这一点。他们也相当自豪于自己惊人的 ASIC 利润率。
So, you asked the question why. A long time ago, we just didn’t have the ability to do it. At the time, I didn’t deeply internalize how difficult it would be to build a foundation AI lab like OpenAI and Anthropic, and the fact that they needed huge investments from the supplier themselves. We just weren’t in a position to make the multi-billion dollar investment into Anthropic so that they could use our compute. But Google and AWS were. They put in huge investments in the beginning so that Anthropic, in return, used their compute. We just weren’t in a position to do that at the time.
所以，你问为什么。很久以前，我们确实没有能力做这件事。当时，我没有深刻意识到，建立像 OpenAI 和 Anthropic 这样的基础 AI 实验室会有多难，也没有深刻意识到，它们需要供应商自身提供巨额投资。我们当时并没有能力向 Anthropic 投入数十亿美元，让他们使用我们的计算资源。但 Google 和 AWS 有这个能力。它们一开始就进行了巨额投资，作为回报，Anthropic 使用了它们的计算资源。我们当时没有条件做这件事。
I would say my mistake is I didn’t deeply internalize that they really had no other options, that a VC would never put in $5-10 billion of investment into an AI lab with the hopes of it turning out to be Anthropic. So that was my miss. But even if I understood it, I don’t think we would’ve been in a position to do that at the time. But I’m not going to make that same mistake again.
我会说，我的错误在于，我没有深刻意识到它们其实真的没有其他选择；风险投资机构永远不会向一家 AI 实验室投入 50 亿到 100 亿美元，只是希望它最终变成 Anthropic。所以这是我的失误。但即使我当时理解了这一点，我也不认为我们那时有条件这么做。不过，我不会再犯同样的错误。
I’m delighted to invest in OpenAI, and I’m delighted to help them scale, and I believe it’s essential to do so. And then, when I was able to, when Anthropic came to us, I’m delighted to be an investor, delighted to help them scale. We just weren’t, at the time, able to do it. If I could rewind everything—and Nvidia could have been as big back then as we are now—I would’ve been more than happy to do it.
我很高兴投资 OpenAI，也很高兴帮助他们扩大规模，而且我相信这样做至关重要。后来，当我们有能力时，当 Anthropic 来找我们时，我也很高兴成为投资者，很高兴帮助他们扩大规模。只是当时，我们确实没有能力这么做。如果一切能够倒回去——而且 Nvidia 当时就能像现在这么大——我会非常乐意这么做。
00:41:06 – Why doesn’t Nvidia become a hyperscaler?
00:41:06 – 为什么 Nvidia 不自己成为一家超大规模云服务商？
Dwarkesh Patel
This is actually quite interesting. For many years Nvidia has been the company in AI making money, making lots of money. Now you’re investing it. It’s been reported that you’ve done up to $30 billion in OpenAI and $10 billion in Anthropic. But now their valuations have increased, and I’m sure they’ll continue to increase.
这实际上很有意思。多年来，Nvidia 一直是 AI 领域真正赚钱的公司，而且赚了很多钱。现在你们开始把这些钱投出去。据报道，你们已经向 OpenAI 投资最高 300 亿美元，向 Anthropic 投资 100 亿美元。但现在它们的估值已经上升，而且我相信还会继续上升。
So if over these many years you were giving them the compute, you saw where it was headed, and they were worth like one tenth what they’re worth now a couple years ago—or even a year ago in some cases and you had all this cash — there’s a world where either Nvidia themselves becomes a foundation lab, does a huge investment to make that possible, or has made the deals you’ve made now at current valuations much earlier on. And you had the cash to do it. So I am curious, actually, why not have done it earlier?
所以，如果过去这些年你们一直在给它们提供计算资源，你们也看到了方向，而几年前它们的估值可能只有现在的十分之一——有些情况下甚至一年前也是如此，并且你们手里有这么多现金——那么本来可能存在这样一个世界：要么 Nvidia 自己成为一家基础模型实验室，进行巨额投资使之成为可能；要么你们更早就以当时的估值完成今天才做的这些交易。而且你们有现金去做。所以我确实很好奇，为什么没有更早做？
Jensen Huang
We did it as soon as we could have. We did it as soon as we could have, and if I could have, I would’ve done it even earlier. At the time that Anthropic needed us to do it, we just weren’t in a position to do it. It wasn’t in our sensibility to do so.
我们是在能够做的时候就立刻做了。我们是在能够做的时候就立刻做了；如果我当时能做，我会更早做。在 Anthropic 需要我们这样做的时候，我们确实没有条件这么做。那并不在我们当时的认知和习惯范围之内。
Dwarkesh Patel
How so? Was it like a cash thing?
为什么？是现金问题吗？
Jensen Huang
Yeah, the level of investment. We had never invested outside the company at the time, and not that much. We didn’t realize we needed to. I always thought that they could just go raise from VCs, for God’s sakes, like all companies do. But what they were trying to do couldn’t have been done through VCs. What OpenAI wanted to do couldn’t have been done through VCs. I recognize that now. I didn’t know it then.
是的，投资规模的问题。当时我们从来没有在公司外部做过投资，更没有做过那么大的投资。我们没有意识到自己需要这么做。我一直以为它们完全可以去找风险投资融资，天哪，就像所有公司那样。但它们试图做的事情，不可能通过风险投资完成。OpenAI 想做的事情，也不可能通过风险投资完成。我现在认识到了这一点。当时我不知道。
But that’s their genius. That’s why they’re smart. They realized then that they had to do something like that. And I’m delighted that they did. Even though we caused Anthropic to have to go to somebody else, I’m still happy that it happened. Anthropic’s existence is great for the world. I’m delighted for it.
但这就是他们的天才之处。这也是他们聪明的地方。他们当时就意识到，必须做类似这样的事情。我很高兴他们做到了。即使是因为我们导致 Anthropic 不得不去找别人，我仍然很高兴这件事发生了。Anthropic 的存在对世界是好事。我为此感到高兴。
Dwarkesh Patel
I guess you still are making a ton of money, and you’re making way more money quarter after quarter.
我想，你们仍然赚了很多钱，而且一个季度接一个季度赚得越来越多。
Jensen Huang
It’s still okay to have regrets.
有遗憾仍然是可以的。
Dwarkesh Patel
So the question still arises. Okay, now that we’re here and you have all this money that you keep making, what should Nvidia be doing with it? There’s one answer which is that there’s this whole middleman ecosystem that has popped up for converting CapEx into OpEx for these labs so that they can rent compute. Because the chips are really expensive, they make a lot of money over their lifetime because the AI models are getting better. So the value that they generate, their tokens, is increasing, but they’re expensive to set up. Nvidia has the money to do the CapEx. In fact, it’s been reported, you are backstopping CoreWeave up to $6.3 billion and have invested $2 billion.
所以问题仍然存在。好，现在我们到了这个阶段，而你们不断赚到这么多钱，Nvidia 应该拿这些钱做什么？有一种答案是，已经出现了整个中间商生态系统，专门为这些实验室把资本支出转换成运营支出，让它们可以租用计算资源。因为这些芯片真的很贵，但在它们的生命周期里会赚很多钱，因为 AI 模型正在变得更好。所以它们所产生的价值，也就是它们的词元，正在增加，但前期建设成本很高。Nvidia 有钱去做资本支出。事实上，据报道，你们为 CoreWeave 提供了最高 63 亿美元的兜底支持，并且已经投资了 20 亿美元。
Why doesn’t Nvidia become a cloud themselves? Why doesn’t it become a hyperscaler themselves and rent this compute out? You have all this cash to do it.
为什么 Nvidia 不自己成为一家云服务商？为什么不自己成为一家超大规模云服务商，并把这些计算资源出租出去？你们有这么多现金可以做这件事。
Jensen Huang
This is a philosophy of the company, and I think it’s wise. We should do as much as needed, as little as possible. What that means is, the work that we do with building our computing platform, if we don’t do it, I genuinely believe it doesn’t get done. If we didn’t take the risk that we take—if we didn’t build NVLink the way we built it, if we didn’t build the whole stack, if we didn’t create the ecosystem the way we did, if we didn’t dedicate ourselves to 20 years of CUDA while losing money most of that time—if we didn’t do it, nobody else would have done it.
这是公司的一个哲学，我认为它是明智的。我们应该做足必要的事，同时尽可能少做不必要的事。这意味着，我们在构建计算平台上所做的工作，如果我们不做，我真心相信它就不会被完成。如果我们没有承担自己承担的风险——如果我们没有以这种方式构建 NVLink，如果我们没有构建整个软件栈，如果我们没有以这种方式创建生态系统，如果我们没有在大部分时间都亏钱的情况下坚持投入 CUDA 二十年——如果我们不做这些事，就没有其他人会做。
If we didn’t create all the CUDA-X libraries so that they’re all domain-specific… A decade and a half ago, we pushed into domain-specific libraries because we realized that if we didn’t create these domain-specific libraries, whether it’s for ray tracing or image generation or even the early works of AI, these models, if we didn’t create them, for data processing, structured data processing, or vector data processing, if we didn’t create them, nobody would. I am completely certain of that. We created a library for computational lithography called cuLitho. If we didn’t create it, nobody would have. So accelerated computing wouldn’t advance the way it has if we didn’t do what we did.
如果我们没有创建所有 CUDA-X 库，让它们都成为面向特定领域的库……十五年前，我们推进面向特定领域的库，因为我们意识到，如果我们不创建这些面向特定领域的库，无论是用于光线追踪、图像生成，还是 AI 早期工作中的那些模型，如果我们不为数据处理、结构化数据处理或向量数据处理创建这些库，就没有人会做。我对此完全确信。我们创建了一个用于计算光刻的库，叫 cuLitho。如果我们不创建它，就没有人会创建。所以，如果我们没有做我们所做的事，加速计算就不会以今天这样的方式前进。
So we should do that. We should dedicate our company, all of our might, wholeheartedly to go do that. However, the world has lots of clouds. If I didn’t do it, somebody would show up. So following the recipe, the philosophy, of doing as much as needed but as little as possible—as little as possible—that philosophy exists in our company today. Everything I do, I do it with that lens.
所以我们应该做这件事。我们应该让整个公司、全部力量，全心投入去做这件事。然而，世界上有很多云。如果我不做云，总会有人出现来做。所以，遵循这个方法、这个哲学：做足必要的事，但尽可能少做——尽可能少做——这种哲学今天存在于我们公司内部。我做每一件事，都会用这个视角来看。
In the case of clouds, if we didn’t support CoreWeave to exist, these neoclouds, these AI clouds, wouldn’t exist. If we didn’t help CoreWeave exist, they would not exist. If we didn’t support Nscale, they wouldn’t be where they are today. If we didn’t support Nebius, they wouldn’t be what they are today. Now they’re doing fantastically.
以云为例，如果我们没有支持 CoreWeave 的存在，这些新型云、这些 AI 云，就不会存在。如果我们没有帮助 CoreWeave 存在，它们就不会存在。如果我们没有支持 Nscale，它们不会有今天的位置。如果我们没有支持 Nebius，它们也不会成为今天这样。现在它们做得非常好。
Is that a business model [inaudible]? We should do as much as needed, as little as possible. So we invest in our ecosystem because I want our ecosystem to thrive. I want the architecture, and AI, to be able to connect with as many industries as possible, as many countries as possible, and make it possible for the planet to be built on AI and to be built on the American tech stack. That vision is exactly what we’re pursuing.
这是不是一种商业模式？我们应该做足必要的事，同时尽可能少做。所以我们投资自己的生态系统，因为我希望我们的生态系统繁荣。我希望这种架构和 AI 能够连接尽可能多的行业、尽可能多的国家，并且让这个星球有可能建立在 AI 之上、建立在美国技术栈之上。这正是我们正在追求的愿景。
Now, one of the things that you mentioned… There are so many great, amazing foundation model companies, and we try to invest in all of them. This is another thing that we do. We don’t pick winners. We need to support everyone. It’s part of our joy of doing so. It’s imperative to our business. But we also go out of our way not to pick winners. So when I invest in one of them, I invest in all of them.
现在，你刚才提到的一件事……有这么多优秀、惊人的基础模型公司，我们努力投资所有这些公司。这也是我们会做的另一件事。我们不挑选赢家。我们需要支持每一个人。这本身也是我们做这件事的乐趣之一。它对我们的业务也是必要的。但我们也会特意避免挑选赢家。所以当我投资其中一家时，我也会投资所有其他家。
这个想法不错，巴菲特在医药行业和日本商社都采用了这样的策略，这个领域很难挑出最终的赢家。
Dwarkesh Patel
Why do you go out of your way not to pick winners?
你们为什么要特意避免挑选赢家？
Jensen Huang
Because it’s not our job to, number one. Number two, when Nvidia first started, there were 60 3D graphics companies. We are the only one that survived. If you would have taken those 60 graphics companies and asked yourself which one was going to make it, Nvidia would be at the top of that list not to make it.
第一，因为这不是我们的工作。第二，当 Nvidia 刚起步时，有 60 家 3D 图形公司。我们是唯一活下来的那一家。如果你当时拿那 60 家图形公司来看，并问自己哪一家会成功，Nvidia 会排在最不可能成功的名单前列。
This is long before you, but Nvidia’s graphics architecture was precisely wrong. It’s not a little bit wrong. We created an architecture that was precisely wrong, and it was an impossible thing for developers to support. It was never going to make it. We reasoned about it from good first principles, but we ended up with the wrong solution. Everybody would have counted us out. And here we are.
这是你出生很久以前的事，但 Nvidia 当时的图形架构是精确地错了。不是有一点错。我们创造了一种精确错误的架构，对开发者来说几乎不可能支持。它本来绝不可能成功。我们是从好的第一性原理出发推理的，但最后得出了错误的解决方案。所有人都会把我们排除在外。而我们现在还在这里。
So I have enough humility to recognize that. Don’t pick winners. Either let them all take care of themselves, or take care of all of them.
所以我有足够的谦逊来认识到这一点。不要挑选赢家。要么让它们自己照顾自己，要么就照顾所有人。
Dwarkesh Patel
One thing I didn’t understand is you said, “Look, we’re not prioritizing these neoclouds just because they are neoclouds and we want to prop them up.” But you also listed a bunch of neoclouds and said they wouldn’t exist if it wasn’t for NVIDIA. How are those two things compatible?
有一点我不太理解。你说：“看，我们并不是因为这些新型云是新型云、并且我们想扶持它们，就优先照顾它们。”但你又列举了一批新型云，并且说如果没有 NVIDIA，它们就不会存在。这两件事如何兼容？
Jensen Huang
First of all, they need to want to exist, and they come to ask us for help. When they want to exist and they have a business plan, expertise, and the passion for it… They obviously have to have some capabilities themselves. But if, at the end of the day, they need some investment in order to get it off the ground, we would be there for them. But the sooner they get their flywheel going...
首先，它们自己需要想要存在，并且来找我们寻求帮助。当它们想要存在，并且有商业计划、有专业能力、有热情……它们显然也必须自己具备一些能力。但如果最终它们需要一些投资才能起步，我们会在那里支持它们。不过，它们越早让自己的飞轮转起来……
Your question was, “Do we want to be in the financing business?” The answer is no. There are people in the financing business, and we’d rather work with all the people in the financing business than be a financier ourselves. Our goal is to focus on what we do, keep our business model as simple as possible, and support our ecosystem.
你的问题是：“我们想不想进入融资业务？”答案是不想。有人专门做融资业务，而我们宁愿和所有从事融资业务的人合作，也不愿自己成为融资方。我们的目标是专注于自己做的事，让我们的商业模式尽可能简单，并支持我们的生态系统。
When someone like OpenAI needs an investment of a $30 billion scale because it’s still before their IPO, and we deeply believe in them and I deeply believe that they’re going to be an… Well, they’re an extraordinary company already today. They’re going to be an incredible company. The world needs them to exist. The world wants them to exist. I want them to exist. They have the wind at their back. Let’s support them and let them scale. Those investments we’ll do because they need us to do it. But we’re not trying to do as much as possible. We’re trying to do as little as possible.
当像 OpenAI 这样的公司因为还没有上市而需要 300 亿美元规模的投资，而我们又深信它们，我也深信它们会成为一家……嗯，它们今天已经是一家非凡的公司。它们将会成为一家不可思议的公司。世界需要它们存在。世界希望它们存在。我也希望它们存在。它们顺风而行。那我们就支持它们，让它们扩大规模。这类投资我们会做，因为它们需要我们这么做。但我们并不是试图做尽可能多的事。我们是在试图做尽可能少的事。
“模糊正确”的边界，在很难挑选赢家的地方努力、试图划出一条边界，这是非常好的想法。
Dwarkesh Patel
This may be an obvious question, but we’ve lived many years in this situation where there’s a shortage of GPUs, and it’s grown now because models are getting better.
这可能是一个显而易见的问题，但我们已经在 GPU 短缺的状态下生活了很多年，而现在这种短缺还在扩大，因为模型正在变得更好。
Jensen Huang
We have a shortage of GPUs.
我们确实存在 GPU 短缺。
Dwarkesh Patel
Yes. Nvidia is known for divvying up the scarce allocation, not just based on high bidder, but rather on, “Hey, we want to make sure that these neoclouds exist. Let’s give some to CoreWeave, let’s give some to Crusoe, let’s give some to Lambda.” Why is it good for Nvidia? First of all, would you agree with this characterization of fracturing the market?
是的。Nvidia 以分配稀缺配额而闻名，而且并不只是根据谁出价最高来分配，而是会考虑：“嘿，我们要确保这些新型云存在。给 CoreWeave 一些，给 Crusoe 一些，给 Lambda 一些。”这对 Nvidia 为什么是好事？首先，你是否同意这种“分散市场”的描述？
Jensen Huang
No. No. Your premise is just wrong. We’re sufficiently mindful about these things. We’re very mindful about these things. First of all, if you don’t place a PO, all the talking in the world won’t make a difference. Until we get a PO, what are we going to do? So the first thing is, we work really hard with everybody to get a forecast done, because these things take a long time to build, and the data centers take a long time to build. We align ourselves with demand and supply and things like that through forecasting. Okay? That’s job number one.
不。不。你的前提就是错的。我们对这些事情有足够的谨慎。我们非常谨慎。首先，如果你不下采购订单，全世界所有的口头讨论都没有意义。在我们拿到采购订单之前，我们能做什么？所以第一件事是，我们非常努力地和每个人一起完成预测，因为这些东西需要很长时间来建造，数据中心也需要很长时间来建设。我们通过预测，让自己与需求、供给以及类似事项保持一致。明白吗？这是第一项工作。
Number two, we’ve tried to forecast with as many people as possible, but in the final analysis, you still have to place an order. Maybe, for whatever reason, you didn’t place your order. What can I do? At some point, first in, first out. But beyond that, if you’re not ready because your data center’s not ready, or certain components aren’t ready to enable you to stand up a data center, we might decide to serve another customer first. That’s just maximizing the throughput of our own factory. We might do some adjustments there.
第二，我们已经尽可能和更多人一起做预测，但归根到底，你仍然必须下订单。也许因为某种原因，你没有下订单。那我能怎么办？到了某个阶段，就是先到先得。除此之外，如果你还没准备好，因为你的数据中心还没准备好，或者某些组件还没准备好，无法让你把数据中心建起来，我们可能会决定先服务另一个客户。这只是为了最大化我们自己工厂的吞吐量。我们可能会在那里做一些调整。
Aside from that, the prioritization is first in, first out. You’ve got to place a PO. If you don’t place a PO… Now, of course, there are stories about that. For example, all of this kind of started from an article about Larry and Elon having dinner with me where they begged for GPUs. That never happened. We absolutely had dinner. We absolutely had dinner, and it was a wonderful dinner. At no time did they beg for GPUs. They just had to place an order. Once they place an order, we do our best to get the capacity to them. We’re not complicated.
除此之外，优先顺序就是先到先得。你必须下采购订单。如果你不下采购订单……当然，关于这一点有一些故事。比如，所有这些说法差不多都源自一篇文章，说 Larry 和 Elon 跟我吃饭时求我给他们 GPU。那从来没有发生过。我们确实一起吃了饭。我们确实一起吃了饭，而且那是一顿很愉快的晚餐。但他们从来没有在任何时候求我给 GPU。他们只需要下订单。一旦他们下了订单，我们就尽最大努力把产能交给他们。我们并不复杂。
Dwarkesh Patel
Okay. So it sounds like there’s a queue, and then based on whether your data center is ready and when you place a purchase order, you get them at a certain time. But it still doesn’t sound like the highest bidder just gets it. Is there a reason to do it…?
好。所以听起来像是有一个队列，然后根据你的数据中心是否准备好，以及你什么时候下采购订单，你会在某个时间拿到产品。但这听起来仍然不像是出价最高的人就能拿到。这样做有什么原因吗……？
Jensen Huang
We never do that.
我们从来不这么做。
Dwarkesh Patel
Okay.
好。
Jensen Huang
We never do.
我们从来不这么做。
Dwarkesh Patel
Why not just do high bidder?
为什么不直接价高者得？
Jensen Huang
Because it’s a bad business practice. You set your price and then people decide to buy it or not. I understand that others in the chip industry change their prices when demand is higher, but we just don’t. That’s just never been a practice of ours. You can count on us. I prefer to be dependable, to be the foundation of the industry. You don’t need to second-guess. If I quoted you a price, we quoted you a price. That’s it. If demand goes through the roof, so be it.
因为这是一种糟糕的商业做法。你设定价格，然后让人们决定买还是不买。我理解芯片行业里有些公司会在需求更高时改变价格，但我们就是不这样做。这从来不是我们的做法。你可以信赖我们。我更愿意成为可靠的一方，成为这个行业的基础。你不需要反复猜测。如果我给你报了一个价格，那就是我们给你的价格。就这样。如果需求冲上天，那就让它冲上天。
Dwarkesh Patel
On the other end, that’s why you have a productive relationship with TSMC, right?
反过来，这也是你们和 TSMC 之间有一种富有成效的关系的原因，对吧？
Jensen Huang
Yeah, Nvidia’s been in business with them for, I guess, coming up on 30 years. Nvidia and TSMC don’t have a legal contract. There’s always some rough justice. Sometimes I’m right, sometimes I’m wrong. Sometimes I got a better deal, sometimes I got a worse deal. But overall, the relationship is incredible. I can completely trust them. I can completely depend on them.
是的，Nvidia 和他们合作，我想快 30 年了。Nvidia 和 TSMC 之间没有法律合同。这里总有某种粗略的公平。有时候我是对的，有时候我是错的。有时候我拿到更好的交易，有时候我拿到更差的交易。但总体而言，这段关系非常了不起。我完全信任他们。我完全可以依靠他们。
One of the things you can count on with Nvidia is that this year, Vera Rubin is going to be incredible. Next year, Vera Rubin Ultra will come. The year after that, Feynman will come. And the year after that, I haven’t introduced the name yet. Every single year you can count on us. You’re going to have to go find another ASIC team in the world—pick your ASIC team—where you can say, “I can bet the farm, I can bet my entire business that you will be here for me every single year. Your token cost will decrease by an order of magnitude every single year. I can count on it like I can count on the clock.”
你可以信赖 Nvidia 的一点是，今年 Vera Rubin 会非常惊人。明年，Vera Rubin Ultra 会到来。再下一年，Feynman 会到来。再下一年，我还没有公布名字。每一年你都可以信赖我们。你必须去世界上找另一个 ASIC 团队——随便选一个 ASIC 团队——你能对它说：“我可以把农场押上，我可以把整个业务都押上，相信你每一年都会在那里支持我。你的词元成本每一年都会下降一个数量级。我可以像信赖时钟一样信赖这一点。”
I just said something about TSMC. For no other foundry in history can you possibly say that. You can say that about Nvidia today. You can count on us every single year. If you would like to buy a billion dollars worth of AI factory compute, no problem. If you’d like to buy a hundred million dollars, no problem. You’d like to buy $10 million, or just one rack, not a problem. Or just one graphics card, okay, no problem. If you would like to place an order for a $100 billion of AI factory, no problem. We’re the only company in the world where you can say that today.
我刚才说了关于 TSMC 的一些话。在历史上，你不可能对任何其他晶圆代工厂这么说。今天，你可以对 Nvidia 这么说。你每一年都可以信赖我们。如果你想购买价值 10 亿美元的 AI 工厂计算资源，没问题。如果你想买 1 亿美元，没问题。你想买 1000 万美元，或者只买一个机柜，也没问题。或者只买一块显卡，也可以，没问题。如果你想下一个 1000 亿美元 AI 工厂的订单，也没问题。今天，世界上只有我们一家公司能让你这样说。
I can say that about TSMC as well. I want to buy one, buy 1 billion, no problem. We just have to go through the process of planning for it, and all the things that mature people do. So I think this ability for Nvidia to be the foundation of the world’s AI industry, this is a position that has taken us a couple of decades to arrive at. Enormous commitment, enormous dedication. The stability of our company, the consistency of our company, is really important.
我也可以这样评价 TSMC。我想买一个，或者买 10 亿个，都没问题。我们只需要经历规划流程，以及所有成熟的人会做的事情。所以我认为，Nvidia 能够成为全球 AI 行业的基础，这个位置是我们花了几十年才走到的。巨大的承诺，巨大的投入。我们公司的稳定性、我们公司的一致性，真的非常重要。
00:57:36 – Should we be selling AI chips to China?
00:57:36 – 我们应该向中国出售 AI 芯片吗？
Dwarkesh Patel
Okay. I want to ask about China. I actually don’t know what I think about whether it’s good to sell chips to China or not, but I like to play devil’s advocate against my guests. So when Dario was on, who supports export controls, I asked him, why can’t America and China both have a country of geniuses in the datacenter? But since you’re on the opposite side, I’ll ask you in the opposite way.
好。我想问问中国问题。实际上，我也不知道自己到底认为向中国出售芯片是好事还是坏事，但我喜欢站在嘉宾的对立面扮演反方。所以当支持出口管制的 Dario 来的时候，我问他，为什么美国和中国不能都在数据中心里拥有一个由天才组成的国家？但既然你站在相反一边，我就用相反的方式问你。
One way to think about it is, Anthropic actually announced a couple days ago Mythos Preview. This model Mythos, they’re not even releasing publicly because they say it has such cyber-offensive capabilities that we don’t think the world is ready until we make sure these zero-days are patched up. But they say it found thousands of high-severity vulnerabilities across every major operating system, every browser. It found one in OpenBSD, which is this operating system that’s been specifically designed to not have zero days. It found one that’s existed for 27 years.
一种思考方式是，Anthropic 几天前其实发布了 Mythos Preview。这个模型 Mythos，他们甚至没有公开发布，因为他们说它具有如此强的网络攻击能力，以至于在确保这些零日漏洞被修补之前，他们认为世界还没有准备好。但他们说，它在每一个主要操作系统、每一种浏览器中都发现了数千个高严重性漏洞。它还在 OpenBSD 中发现了一个漏洞，而 OpenBSD 是一个专门被设计成不应有零日漏洞的操作系统。它发现的那个漏洞已经存在了 27 年。
So if Chinese companies and Chinese labs and the Chinese government had access to the AI chips to train a model like Claude Mythos with these cyber-offensive capabilities and run millions of instances of it with more compute, the question is, is that a threat to American companies, to American national security?
所以，如果中国公司、中国实验室和中国政府能够获得 AI 芯片，训练出像 Claude Mythos 这样具有网络攻击能力的模型，并用更多计算资源运行数百万个实例，那么问题是，这对美国公司、对美国国家安全是不是一种威胁？
Jensen Huang
First of all, Mythos was trained on fairly mundane capacity, and a fairly mundane amount of it. By an extraordinary company. The amount of capacity and the type of compute it was trained on is abundantly available in China. So you just have to first realize that chips exist in China.
首先，Mythos 是在相当普通的计算能力上训练出来的，而且使用的数量也相当普通。当然，这是由一家非凡的公司完成的。它训练所用的那种计算能力和计算类型，在中国是大量存在的。所以你首先必须意识到，芯片在中国是存在的。
They manufacture 60% of the world’s mainstream chips, maybe more. It’s a very large industry for them. They have some of the world’s greatest computer scientists. As you know, most of the AI researchers in all of these AI labs are Chinese. They have 50% of the world’s AI researchers. So the question is, considering all the assets they already have—they have an abundance of energy, they have plenty of chips, they’ve got most of the AI researchers—if you’re worried about them, what is the best way to create a safe world?
他们制造了全球 60% 的主流芯片，也许更多。对他们来说，这是一个非常大的产业。他们拥有世界上一些最优秀的计算机科学家。正如你所知道的，这些 AI 实验室里的许多 AI 研究人员都是中国人。他们拥有全球 50% 的 AI 研究人员。所以问题是，考虑到他们已经拥有的所有资产——他们有充足的能源，有大量芯片，也拥有大量 AI 研究人员——如果你担心他们，那么创造一个安全世界的最好方式是什么？
Victimizing them, turning them into an enemy, likely isn’t the best answer. They are an adversary. We want the United States to win. But I think having a dialogue and having research dialogue is probably the safest thing to do. This is an area that is glaringly missing because of our current attitude about China as an adversary. It is essential that our AI researchers and their AI researchers are actually talking. It is essential that we try to both agree on what not to use the AI for.
把他们受害者化，把他们变成敌人，很可能不是最好的答案。他们是一个对手。我们希望美国赢。但我认为保持对话，保持研究层面的对话，可能是最安全的做法。由于我们目前把中国视为对手的态度，这正是一个明显缺失的领域。我们的 AI 研究人员和他们的 AI 研究人员真正进行交流，这一点至关重要。我们努力共同同意 AI 不应该被用于哪些事情，这一点也至关重要。
With respect to finding bugs in software, of course, that’s what AI is supposed to do. Is it going to find bugs in a lot of software? Of course. There are lots and lots of bugs. There are lots of bugs in the AI software. That’s what AI is supposed to do, and I’m delighted that AI has reached a level where it could help us be so much more productive.
至于在软件中寻找漏洞，当然，这正是 AI 应该做的事情。它会在大量软件中发现漏洞吗？当然会。有非常多的漏洞。AI 软件中也有很多漏洞。这就是 AI 应该做的事，而我很高兴 AI 已经达到这样的水平，能够帮助我们变得高效得多。
One of the things that is underemphasized is the richness of the ecosystem around cybersecurity, AI cybersecurity and AI security and AI privacy and AI safety. There’s a whole ecosystem of AI startups that are trying to create this future for us, where you have one AI agent that’s incredible, surrounded by thousands of AI agents, keeping it safe, keeping it secure. That future surely is going to happen.
有一件事被低估了，那就是围绕网络安全、AI 网络安全、AI 安全、AI 隐私和 AI 安全性所形成的生态系统非常丰富。有一整个 AI 初创公司生态系统，正在努力为我们创造这样一个未来：你有一个非常强大的 AI 智能体，周围有成千上万个 AI 智能体保护它的安全，确保它不受侵害。这个未来肯定会发生。
The idea that you’re going to have an AI agent running around with nobody watching after it is kind of insane. We know very well that this ecosystem needs to thrive. It turns out this ecosystem needs open source. This ecosystem needs open models. They need open stacks so that all of these AI researchers and all these great computer scientists can go build AI systems that are as formidable and can keep AI safe. So one of the things that we need to make sure that we do is we keep the open source ecosystem vibrant. That can’t be ignored. A lot of that is coming out of China. We ought to not suffocate that.
认为你会有一个 AI 智能体到处运行、却没有任何东西看管它，这个想法有点疯狂。我们非常清楚，这个生态系统需要繁荣。事实证明，这个生态系统需要开源。这个生态系统需要开放模型。它们需要开放技术栈，让所有这些 AI 研究人员和优秀的计算机科学家能够去构建同样强大、并且能够保障 AI 安全的 AI 系统。所以我们需要确保的一件事是，让开源生态系统保持活力。这不能被忽视。其中很多成果来自中国。我们不应该扼杀它。
With respect to China, of course we want the United States to have as much computing as possible. We’re limited by energy, but we’ve got a lot of people working on that. We’ve got to not make energy a bottleneck for our country. But what we also want is to make sure that all the AI developers in the world are developing on the American tech stack, and making the contributions, the advancements of AI—especially when it’s open source—available to the American ecosystem. It would be extremely foolish to create two ecosystems: the open source ecosystem, and it only runs on a foreign tech stack, and a closed ecosystem that runs on the American tech stack. I think that would be a horrible outcome for the United States.
至于中国，当然，我们希望美国拥有尽可能多的计算能力。我们受到能源限制，但有很多人在解决这个问题。我们不能让能源成为我们国家的瓶颈。但我们还希望确保全世界所有 AI 开发者都在美国技术栈上开发，并且让 AI 的贡献和进步——尤其是开源时——能够进入美国生态系统。创造两个生态系统会极其愚蠢：一个开源生态系统，但它只运行在外国技术栈上；另一个封闭生态系统，运行在美国技术栈上。我认为，这对美国来说会是一个糟糕的结果。
Dwarkesh Patel
Since there are a lot of things, let me just triage the response. I think the concern, going back to the flop difference in the hacking, is yes, they have compute, but there’s some estimates that because they’re at 7nm—they don’t have EUVs because of chip-making export controls—the amount of flops they’re able to actually produce, they have one tenth the amount of flops that the US has.
这里面有很多点，让我先把回应分一下层次。我认为担忧在于，回到黑客攻击能力中的浮点运算差距，是的，他们有计算能力，但有一些估计认为，因为他们还在 7 纳米——由于芯片制造出口管制，他们没有 EUV——他们实际能够生产出来的浮点运算能力，只有美国的十分之一。
So with that, could they eventually train a model like Mythos? Yes. But the question is, because we have more flops, American labs are able to get to these levels of capabilities first. Because Anthropic got to it first, they say, “Okay, we’re going to hold onto it for a month while all these American companies, we’ll give them access to it. They’re going to patch up all their vulnerabilities, and now we release it.”
因此，在这种情况下，他们最终能不能训练出像 Mythos 这样的模型？可以。但问题是，因为我们拥有更多浮点运算能力，美国实验室能够先达到这些能力水平。由于 Anthropic 先做到了，他们可以说：“好，我们会先保留它一个月，同时把访问权限给这些美国公司。它们会修补所有漏洞，然后我们再发布它。”
Furthermore, even if they train a model like this, the ability to deploy it at scale… If you had a cyber hacker, it’s much more dangerous if they have a million of them versus a thousand of them. So that inference compute really matters a lot. In fact, the fact that they have so many AI researchers who are so good is the thing that makes it so scary, because what is it that makes those engineer researchers more productive? It’s compute.
进一步说，即使他们训练出了这样的模型，能否大规模部署它……如果你有一个网络黑客，那么拥有一百万个这样的黑客，比只有一千个要危险得多。所以推理计算能力非常重要。事实上，他们拥有这么多优秀 AI 研究人员这一点，正是让人感到害怕的地方，因为是什么让这些工程师和研究人员更高效？是计算能力。
If you talk to any AI lab in America, they say the thing that’s bottlenecking them is compute. There are quotes from the DeepSeek founder, or Qwen leadership or whatever. They say the thing they’re bottlenecked on is compute. So then the question is, isn’t it better that we get American companies, because they have more compute, to get to the Mythos-level capabilities first, prepare our society for it, before China can get to it because, they have less compute?
如果你和美国任何一家 AI 实验室交流，他们都会说限制他们的是计算能力。DeepSeek 创始人，或者 Qwen 领导层之类的人，也有类似说法。他们说自己的瓶颈是计算能力。那么问题就是，既然美国公司拥有更多计算能力，让美国公司先达到 Mythos 级别的能力，在中国因为计算能力较少而达到之前，先让我们的社会做好准备，这不是更好吗？
Jensen Huang
We should always be first and we should always have more. But in order for that outcome you described to be true, you have to take it to the extremes. They have to have no compute. If they have some compute, the question is how much is needed?
我们应该永远第一，也应该永远拥有更多。但要让你描述的那个结果成立，你必须把它推到极端。他们必须没有任何计算能力。如果他们有一些计算能力，问题就是到底需要多少。
The amount of compute they have in China is enormous. You’re talking about the country that is the second largest computing market in the world. If they want to aggregate their compute, they’ve got plenty of compute to aggregate.
他们在中国拥有的计算能力是巨大的。你说的是世界第二大计算市场。如果他们想把计算能力聚合起来，他们有大量计算能力可以聚合。
Dwarkesh Patel
But is that true? People do these estimates and they’re like, “SMIC is actually behind on the process nodes.”
但这是真的吗？人们做这些估算时会说：“SMIC 在制程节点上其实是落后的。”
Jensen Huang
I’m about to tell you.
我正要告诉你。
Dwarkesh Patel
Okay.
好。
Jensen Huang
The amount of energy they have is incredible. Isn’t that right? AI is a parallel computing problem, isn’t it? Why can’t they just put 4x, 10x, as many chips together because energy’s free? They have so much energy. They have datacenters that are sitting completely empty, fully powered. You know they have ghost cities, they have ghost datacenters too. They have so much infrastructure capacity. If they wanted to, they just gang up more chips, even if they’re 7nm.
他们拥有的能源数量是惊人的。不是吗？AI 是一个并行计算问题，不是吗？既然能源几乎是免费的，他们为什么不能把 4 倍、10 倍数量的芯片连接在一起？他们有如此多的能源。他们有一些数据中心完全空置，但已经通电。你知道他们有鬼城，也有鬼数据中心。他们拥有如此多的基础设施容量。如果他们愿意，即使芯片是 7 纳米，他们也只需要把更多芯片组合在一起。
Their capacity of building chips is one of the largest in the world. The semiconductor industry knows that they monopolize mainstream chips. They have over-capacity, they have too much capacity. So the idea that China won’t be able to have AI chips is completely nonsense.
他们制造芯片的能力是世界上最大的之一。半导体行业知道，他们垄断了主流芯片。他们有过剩产能，有太多产能。所以，认为中国无法拥有 AI 芯片，这种想法完全是胡说。
Now, of course, if you ask me, would the United States be further ahead if the entire world had no compute at all? But that’s just not an outcome. That’s not a scenario that’s true. They have plenty of compute already. The amount of threshold they need for the concern you’re worried about, they’ve already reached that threshold and beyond.
当然，如果你问我，如果全世界完全没有计算能力，美国是不是会领先更多？但这根本不是一个可能的结果。那不是一个真实的情境。他们已经拥有大量计算能力。就你所担心的问题而言，他们所需要达到的门槛，已经达到，而且已经超过了。
So I think you misunderstand that AI is a five-layer cake, and at the lowest layer is energy. When you have an abundance of energy, it makes up for chips. If you have an abundance of chips, it makes up for energy. For example, the United States is scarce on energy, which is the reason why Nvidia has to keep advancing our architecture and do this extreme co-design so that with the few chips that we ship—with the few chips, because the amount of energy is so limited—our throughput per watt is off the charts.
所以我认为你误解了一点：AI 是一块五层蛋糕，而最底层是能源。当你拥有充足能源时，它可以弥补芯片不足。如果你拥有充足芯片，它也可以弥补能源不足。比如，美国能源稀缺，这就是为什么 Nvidia 必须不断推进我们的架构，并进行这种极致协同设计，以便在我们交付的少量芯片上——之所以是少量芯片，是因为能源数量如此有限——实现极高的每瓦吞吐量。
But if your amount of watts is completely abundant, it’s free, what do you care about performance per watt for? You get plenty. You can use old chips to do. So 7nm chips are essentially Hopper. The ability for Hopper… I’ve got to tell you, today’s models are largely trained on Hopper, Hopper generation. So 7nm chips are plenty good. The abundance of energy is their advantage.
但如果你的瓦特数完全充足，甚至几乎免费，那你为什么还要在意每瓦性能？你有的是能源。你可以用旧芯片来做。所以 7 纳米芯片本质上就是 Hopper。我必须告诉你，今天的模型很大程度上是在 Hopper 上训练的，是 Hopper 这一代。所以 7 纳米芯片已经足够好。能源充足是他们的优势。
Dwarkesh Patel
But then there’s a question of whether they can actually manufacture enough chips.
但接下来的问题是，他们是否真的能制造出足够多的芯片。
Jensen Huang
But they do. What’s the evidence? Huawei just had the largest single year in the history of their company.
但他们能。证据是什么？Huawei 刚刚经历了公司历史上规模最大的一年。
Dwarkesh Patel
How many chips did they ship?
他们出货了多少芯片？
Jensen Huang
A ton. Millions. Millions is way more than Anthropic has.
很多。数百万。数百万远远超过 Anthropic 拥有的数量。
Dwarkesh Patel
There’s a question of how much logic SMIC can chip, and there’s a question of how much memory—
这里还有一个问题，SMIC 能出货多少逻辑芯片，以及有多少存储器——
Jensen Huang
I’m telling you what it is. They have plenty of logic, and they have plenty of HBM2 memory.
我现在就在告诉你事实是什么。他们有足够的逻辑芯片，也有足够的 HBM2 存储器。
Dwarkesh Patel
Right. But as you know, the bottleneck often in training and doing inference on these models is the amount of bandwidth. So if you have HBM2… I don’t know the numbers offhand but versus the newest thing you have, there could be almost an order of magnitude difference in memory bandwidth, which is huge.
对。但你也知道，在这些模型的训练和推理中，瓶颈往往是带宽数量。所以如果你用的是 HBM2……我不能脱口说出具体数字，但与最新产品相比，内存带宽可能相差接近一个数量级，而这是巨大的差距。
Jensen Huang
Huawei is a networking company.
Huawei 是一家网络公司。
Dwarkesh Patel
But that doesn’t change the fact that you need EUV for the most advanced HBM.
但这并不改变一个事实：最先进的 HBM 需要 EUV。
Jensen Huang
Not true. Not at all true. You could gang them together, just like we gang them together with NVL72. They’ve already demonstrated silicon photonics, connecting all of this compute together into one giant supercomputer. Your premise is just wrong.
不对。完全不对。你可以把它们组合在一起，就像我们用 NVL72 把它们组合在一起一样。他们已经展示了硅光子技术，可以把所有这些计算资源连接成一台巨大的超级计算机。你的前提就是错的。
The fact of the matter is, their AI development is going just fine. The best AI researchers in the world, because they’re limited in compute, they also come up with extremely smart algorithms. Remember, I just said that Moore’s law is advancing about 25% per year. However, through great computer science, we could still improve algorithm performance by 10x. What I’m saying is that great computer science is where the lever is.
事实是，他们的 AI 发展进展得很好。世界上最优秀的 AI 研究人员，因为受到计算能力限制，也会提出极其聪明的算法。记住，我刚才说过，Moore’s Law 每年大约推进 25%。然而，通过优秀的计算机科学，我们仍然可以把算法性能提高 10 倍。我的意思是，真正的杠杆在优秀的计算机科学那里。
There is no question, MoE is a great invention. There’s no question, all the incredible attention mechanisms reduce the amount of compute. We have got to acknowledge that most of the advances in AI came out of algorithm advances, not just the raw hardware. Now, if most advances came from algorithms and computer science and programming, tell me that their army of AI researchers is not their fundamental advantage. We see it. DeepSeek is not an inconsequential advance. The day that DeepSeek comes out on Huawei first, that is a horrible outcome for our nation.
毫无疑问，MoE 是一项伟大发明。毫无疑问，所有那些令人惊叹的注意力机制都减少了所需计算量。我们必须承认，AI 的大多数进步来自算法进步，而不仅仅是原始硬件。既然大多数进步来自算法、计算机科学和编程，那么你告诉我，他们那支 AI 研究人员大军难道不是他们的根本优势吗？我们看得见这一点。DeepSeek 不是一个无关紧要的进步。如果有一天 DeepSeek 首先在 Huawei 上发布，那对我们国家来说将是一个糟糕的结果。
Dwarkesh Patel
Why is that? Because currently you can have a model like DeepSeek that can run on any accelerator, if it’s open source. Why would that stop being the case in the future?
为什么？因为现在如果一个像 DeepSeek 这样的模型是开源的，它可以运行在任何加速器上。为什么未来这一点会不再成立？
Jensen Huang
Suppose it doesn’t. Suppose it’s optimized for Huawei, suppose it’s optimized for their architecture. It would put ours at a disadvantage. You described a situation that I perceive to be good news. A company developed software, developed an AI model, and it runs best on the American tech stack. I saw that as good news. You set it up as a premise that it was bad news. I’m going to give you the bad news, that AI models around the world are developed and they run best on non-American hardware. That is bad news for us.
假设它不会继续如此。假设它被针对 Huawei 优化，假设它被针对他们的架构优化。那会让我们的架构处于劣势。你描述的是一种我认为属于好消息的情况。一家公司开发了软件，开发了一个 AI 模型，而它在美国技术栈上运行得最好。我把这看作好消息。你却把它设定成一个坏消息的前提。我来告诉你什么才是坏消息：全世界的 AI 模型被开发出来，并且它们在非美国硬件上运行得最好。那对我们来说才是坏消息。
Dwarkesh Patel
I guess I just don’t see the evidence that there’s these huge disparities that would prevent you from switching accelerators. American labs are running their models across all the clouds, across all the different accelerators—
我想我只是没有看到证据表明，存在这些巨大的差异，会阻止你切换加速器。美国实验室正在所有云上、在各种不同加速器上运行它们的模型——
Jensen Huang
I am the evidence. You take a model that’s optimized for Nvidia and you try to run it on something else.
我就是证据。你拿一个针对 Nvidia 优化的模型，试着把它运行在别的东西上。
Dwarkesh Patel
But American labs do that.
但美国实验室确实这么做。
Jensen Huang
And they don’t run better. Nvidia’s success is perfect evidence. The fact that AI models are created on our stack, run best on our stack, how is that illogical to understand?
而且它们并不会运行得更好。Nvidia 的成功就是完美证据。AI 模型是在我们的技术栈上创建的，在我们的技术栈上运行得最好，这一点有什么难以理解的？
一个真实的风险，不仅仅是Huawei和DeepSeek，还包括Google，最终需要迭代的足够好，但Huawei和DeepSeek有可能在一个封闭的系统内，脱离Nvidia、Google竞争的环境下完成迭代。
Dwarkesh Patel
Anthropic’s models are run on GPUs, they’re run on Trainium, they’re run on TPUs.
Anthropic 的模型运行在 GPU 上，也运行在 Trainium 上，也运行在 TPU 上。
Jensen Huang
A lot of work has to go into it to change. But go to the global south, go to the Middle East. Coming out of the box, if all of the AI models run best on somebody else’s tech stack, you’ve got to be arguing some ridiculous claim right now that that’s a good thing for the United States.
要做这种切换，需要投入大量工作。但你去全球南方看看，去中东看看。如果所有 AI 模型开箱即用时，都在别人的技术栈上运行得最好，那你现在就必须提出一个荒谬的论点，说这对美国是好事。
Dwarkesh Patel
But I guess I don’t understand the argument. Say Chinese companies get to the next Mythos first. They find all the security vulnerabilities in American software first, but they can do it on Nvidia hardware and they ship it to the global south. They do it on Nvidia hardware. How is that good? Okay, it runs on Nvidia hardware—
但我想我还是不理解这个论点。假设中国公司先达到下一个 Mythos 级别。它们先发现美国软件中的所有安全漏洞，但它们是在 Nvidia 硬件上做到的，并把它推向全球南方。它们是在 Nvidia 硬件上做到的。这怎么会是好事？好吧，它运行在 Nvidia 硬件上——
Jensen Huang
It’s not good. It’s not good.
这不是好事。这不是好事。
Dwarkesh Patel
Right.
对。
Jensen Huang
It’s not good. So let’s not let it happen.
这不是好事。所以我们不要让它发生。
Dwarkesh Patel
Why do you think it’s perfectly fungible, that if you didn’t ship them compute it would exactly be replaced by Huawei? They are behind, right? They have worse chips than you.
为什么你认为这完全可以替代，也就是说，如果你不向他们供应计算资源，它就会被 Huawei 精确替代？他们是落后的，对吧？他们的芯片比你们差。
Jensen Huang
It’s completely… There’s evidence right now. Their chip industry’s gigantic.
这完全……现在就有证据。他们的芯片产业非常巨大。
Dwarkesh Patel
You can just look at the flop or bandwidth or memory comparisons between the H200 and the Huawei 910C. It’s like half to a third.
你只要比较一下 H200 和 Huawei 910C 之间的浮点运算能力、带宽或内存，就会发现它大概只有一半到三分之一。
Jensen Huang
They use more of it. They use twice as many.
他们会用更多。他们会用两倍数量。
Dwarkesh Patel
It seems like your argument is they have all this energy that’s ready to go, right? And they need to fill it with chips.
听起来你的论点是，他们拥有所有这些随时可用的能源，对吧？而他们需要用芯片把这些能源填满。
Jensen Huang
And they’re good at manufacturing.
而且他们擅长制造。
Dwarkesh Patel
And I’m sure eventually they would be able to just out-manufacture everybody. But there are these few critical years.
而且我相信他们最终会能够仅凭制造能力超过所有人。但现在有这样几个关键年份。
Jensen Huang
What is the critical year you’re talking about?
你说的关键年份是什么？
Dwarkesh Patel
These next few years. We’ve got these models that are going to be able to do all the cyber attacks.
接下来几年。我们会拥有这些能够执行各种网络攻击的模型。
Jensen Huang
In that case, if the next years are critical, then we have to make sure that all of the world’s AI models are built on the American tech stack, in these critical years.
既然如此，如果接下来几年是关键时期，那么我们就必须确保，在这些关键年份里，全球所有 AI 模型都建立在美国技术栈之上。
Dwarkesh Patel
If they’re built on the American tech stack, how would that prevent them, if they have more advanced capabilities, from launching the Mythos-equivalent cyber attacks?
如果它们建立在美国技术栈上，那又如何防止它们在拥有更先进能力时，发动 Mythos 等级的网络攻击？
Jensen Huang
There’s no guarantee either way.
无论哪种方式，都没有保证。
Dwarkesh Patel
But if you have it early, we can prepare for it.
但如果我们更早拥有它，就可以为它做准备。
Jensen Huang
Listen, why are you causing one layer of the AI industry to lose an entire market so that you could benefit another layer of the AI industry? There are five layers and every single layer has to succeed. The layer that has to succeed most is actually the AI applications. Why are you so fixated on that AI model? That one company? For what reason?
听着，为什么你要让 AI 行业中的某一层失去整个市场，以便让 AI 行业中的另一层受益？这里有五层，每一层都必须成功。最需要成功的那一层，其实是 AI 应用。你为什么如此执着于那个 AI 模型？那一家公司？出于什么原因？
Dwarkesh Patel
Because those models make possible these incredibly offensive capabilities, and you need compute to run them.
因为那些模型使这些极强的攻击能力成为可能，而你需要计算资源来运行它们。
Jensen Huang
The energy, the chips, and the ecosystem of AI researchers make it possible.
能源、芯片，以及 AI 研究人员生态系统，才让它成为可能。
Dwarkesh Patel
Okay, stepping back, it has to be the case that China is able to build enough 7nm capacity. And remember, they’re still stuck on 7nm while you’ll move on to 3nm and then 2nm or 1.6nm with Feynman. So while you’re on 1.6nm, they’re still going to be on 7nm, and they have to produce enough of it to make up for the shortfall. They have so much energy that the more chips you give them, the more compute they’d have. So it comes out as a question of, ultimately they are getting more compute. Compute is an input to training and inference—
好，退一步说，这必须建立在一个前提上：中国能够建设足够多的 7 纳米产能。而且记住，他们仍然卡在 7 纳米，而你们会推进到 3 纳米，然后随着 Feynman 进入 2 纳米或者 1.6 纳米。所以当你们处在 1.6 纳米时，他们仍然会停留在 7 纳米，并且必须生产足够多的 7 纳米芯片来弥补差距。他们有如此多的能源，以至于你给他们越多芯片，他们就会拥有越多计算资源。所以最终问题变成：他们确实获得了更多计算资源。计算资源是训练和推理的输入——
Jensen Huang
Listen, I just think you speak in absolutes. I think the United States ought to be ahead. The amount of compute in the United States is 100x more than anywhere else in the world. The United States ought to be ahead. Okay. The United States is ahead.
听着，我只是觉得你说话太绝对了。我认为美国应该领先。美国的计算资源数量是世界其他任何地方的 100 倍。美国应该领先。好。美国现在就是领先的。
Nvidia builds the most advanced technologies. We make sure that the US labs are the first to hear about it and have the first chance to buy it. And if they don’t have enough money, we even invest in them. The United States ought to be ahead. We want to do everything we can to make sure the United States is ahead. Number one point, do you agree? We’re doing everything we can to do that.
Nvidia 构建最先进的技术。我们确保美国实验室最先听到这些技术，并且拥有最先购买的机会。如果它们没有足够资金，我们甚至会投资它们。美国应该领先。我们想尽一切办法确保美国领先。第一点，你同意吗？我们正在尽一切努力做到这一点。
Dwarkesh Patel
But how is shipping chips to China keeping the US ahead if they’re bottlenecked on compute?
但如果中国的瓶颈是计算资源，那么向中国出货芯片，如何能让美国保持领先？
Jensen Huang
No, no. We’ve got Vera Rubin for the United States. We have Vera Rubin for the United States. Now, am I in the United States? Do you consider me part of the United States?
不，不。我们为美国准备了 Vera Rubin。我们为美国准备了 Vera Rubin。现在，我在美国吗？你认为我是美国的一部分吗？
Dwarkesh Patel
Yes.
是的。
Jensen Huang
Nvidia. You consider Nvidia a United States company? Okay. Number one, why is it that we don’t come up with a regulation that’s more balanced so that Nvidia can win around the world instead of giving up the world? Why would you want the United States to give up the world?
Nvidia。你认为 Nvidia 是一家美国公司吗？好。第一，为什么我们不制定一种更平衡的监管，让 Nvidia 能在全世界赢，而不是放弃全世界？你为什么希望美国放弃全世界？
The chip industry is part of the American ecosystem. It’s part of American technology leadership. It’s part of the AI ecosystem. It’s part of AI leadership. Why is it that your policy, your philosophy, leads to the United States giving up a vast part of the world’s market?
芯片行业是美国生态系统的一部分。它是美国技术领导力的一部分。它是 AI 生态系统的一部分。它也是 AI 领导力的一部分。为什么你的政策、你的哲学，会导致美国放弃全球市场中的一大块？
Dwarkesh Patel
I guess the claim here is… Dario had this quote where he said that it’s like Boeing bragging that we’re selling North Korea nukes, but the missile casings are made by Boeing. And that’s somehow enabling the US technology stack. Fundamentally, you’re giving them this capability.
我想这里的主张是……Dario 有一句话，他说这就像 Boeing 吹嘘说，我们正在向朝鲜出售核武器，但导弹外壳是 Boeing 制造的。因此这在某种程度上是在赋能美国技术栈。根本上，你是在给他们这种能力。
Jensen Huang
Comparing AI to anything that you just mentioned is lunacy.
把 AI 和你刚才提到的那些东西相比，是疯狂的。
Dwarkesh Patel
But AI is similar to enriched uranium, right? It can have positive uses, it can have negative uses. We still don’t want to send enriched uranium to other countries.
但 AI 和浓缩铀类似，对吧？它可以有正面用途，也可以有负面用途。我们仍然不希望把浓缩铀送到其他国家。
Jensen Huang
Who’s sending enriched—
谁在送浓缩——
Dwarkesh Patel
The analogy is that enriched uranium is like compute.
这个类比是说，浓缩铀就像计算资源。
Jensen Huang
It’s a lousy analogy. It’s an illogical analogy.
这是一个糟糕的类比。它是一个不合逻辑的类比。
Dwarkesh Patel
But if that compute can run a model that can do zero-day exploits against all American software, how is that not a weapon?
但如果这些计算资源可以运行一个模型，而这个模型能够针对所有美国软件执行零日漏洞利用，那它怎么不是武器？
Jensen Huang
First of all, the way to solve that problem is to have dialogues with the researchers and dialogues with China, and dialogues with all the countries to make sure that people don’t use technology in that way. That’s a dialogue that has to happen. Okay? Number one.
首先，解决这个问题的方法，是和研究人员进行对话，和中国进行对话，和所有国家进行对话，以确保人们不会以那种方式使用技术。这是必须发生的对话。明白吗？这是第一点。
Number two, we also need to make sure that the United States is ahead, that Vera Rubin, Blackwell, is available in the United States in abundance, mountains of it. Obviously, our results would show it. Abundance, tons of it. The amount of computing we have is great. We have amazing AI researchers here. It’s great. We ought to stay ahead.
第二，我们也需要确保美国领先，确保 Vera Rubin、Blackwell 在美国大量供应，像山一样多。显然，我们的业绩会显示这一点。大量供应，非常多。我们拥有的计算资源数量很大。我们这里有很出色的 AI 研究人员。这很好。我们应该保持领先。
However, we also have to recognize that AI is not just a model. AI is a five-layer cake. The AI industry matters across every single layer, and we want the United States to win at every single layer, including the chip layer. Conceding the entire market is not going to allow the United States to win the technology race long-term in the chip layer, in the computing stack. That is just a fact.
然而，我们也必须认识到，AI 不只是一个模型。AI 是一块五层蛋糕。AI 行业的每一层都很重要，而我们希望美国在每一层都赢，包括芯片层。放弃整个市场，不会让美国在芯片层、在计算技术栈中长期赢得技术竞赛。这就是事实。
Dwarkesh Patel
I guess then the crux comes down to, how does selling them chips now help us win in the long term? Tesla sold extremely good electric vehicles to China for a long time. iPhones are sold in China, extremely good. They didn’t cause them lock-in. China will still make their version of EVs and they’re dominating. Their smartphones are dominating.
我想，问题的核心就变成：现在向他们出售芯片，如何帮助我们在长期获胜？Tesla 很长时间以来一直向中国销售非常好的电动车。iPhone 也在中国销售，而且非常好。但它们没有造成锁定。中国仍然会制造自己的电动车版本，而且它们正在占据主导。它们的智能手机也在占据主导。
Jensen Huang
When we started the conversation today, you acknowledged that Nvidia’s position is very different. You used words like moat. The single most important thing to our company is the richness of our ecosystem, which is about developers. 50% of the AI developers are in China. The United States should not give that up.
我们今天开始谈话时，你承认 Nvidia 的位置非常不同。你用了护城河这样的词。对我们公司来说，最重要的一件事是我们生态系统的丰富性，而这关乎开发者。50% 的 AI 开发者在中国。美国不应该放弃这一点。
Dwarkesh Patel
But we have a lot of Nvidia developers in the US, and that doesn’t prevent American labs from also being able to use other accelerators in the future. In fact, right now they’re using other accelerators as well, which is fine and great. I don’t see why that wouldn’t be the case in China as well, if you sell them Nvidia chips, just the same way that Google can use TPUs and Nvidia—
但我们在美国也有很多 Nvidia 开发者，而这并不会阻止美国实验室未来也能够使用其他加速器。事实上，它们现在也在使用其他加速器，这没问题，也很好。如果你们向中国出售 Nvidia 芯片，我看不出为什么中国不会出现同样情况，就像 Google 可以同时使用 TPU 和 Nvidia——
Jensen Huang
We have to keep innovating and, as you probably know, our share is growing, not decreasing. The premise that even if we competed in China, that we’re going to lose that market anyways… You’re not talking to somebody who woke up a loser. That loser attitude, that loser premise makes no sense to me.
我们必须持续创新，而且你大概也知道，我们的份额正在增长，不是在下降。那种认为即使我们在中国竞争，也无论如何会失去那个市场的前提……你现在谈话的对象，不是一个一觉醒来就认输的人。那种失败者态度，那种失败者前提，对我来说毫无意义。
We’re not a car. We are not a car. The fact that I can buy this car brand one day and use another car brand another day, easy. Computing is not like that. There’s a reason why the x86 deal exists. There’s a reason why ARM is so sticky. These ecosystems are hard to replace. It costs an enormous amount of time and energy, and most people don’t want to do it. So it’s our job to continue to nurture that ecosystem, to keep advancing the technology so that we can compete in the marketplace.
我们不是汽车。我们不是汽车。我今天可以买这个汽车品牌，明天用另一个汽车品牌，这很容易。计算不是这样。x86 交易之所以存在，是有原因的。ARM 如此有黏性，也是有原因的。这些生态系统很难被替代。替代它们需要消耗巨大时间和精力，而大多数人并不想这么做。所以，我们的工作就是继续培育这个生态系统，继续推进技术，从而让我们能够在市场中竞争。
Conceding a marketplace based on the premise you described, I simply can’t acknowledge that. It makes no sense. Because I don’t think the United States is a loser. Our industry is not a loser. That losing proposition, that losing mindset, makes no sense to me.
基于你所描述的前提而放弃一个市场，我完全不能接受。那说不通。因为我不认为美国是失败者。我们的行业也不是失败者。那种失败主张、那种失败心态，对我来说毫无意义。
从哪个角度看问题非常关键，现在真是一个混乱竞争的时期，Jensen Huang和Dario Amodei都取得了规模巨大的成功，看上去Jensen Huang更成熟一些。
Dwarkesh Patel
Okay. I’ll move on. I just want to make sure that—
好。我换个话题。我只是想确认——
Jensen Huang
You don’t have to move on. I’m enjoying it.
你不必换话题。我挺享受这个讨论。
Dwarkesh Patel
Okay, great. Then I won’t. I appreciate that. But I think maybe the crux… and thanks for walking around the circles with me, because I think it helps bring out what the crux here is.
好，那很好。那我就不换。谢谢你愿意和我绕着这个问题走几圈，因为我觉得这有助于把核心问题显露出来。
Jensen Huang
The crux is you’re going to extremes. Your argument starts from extremes. That if we give them any compute at all in this narrow moment, we will lose everything.
核心在于你走向了极端。你的论证是从极端出发的。也就是说，如果我们在这个狭窄时刻给他们任何一点计算资源，我们就会失去一切。
Dwarkesh Patel
No, I think what my argument is—
不，我认为我的论点是——
Jensen Huang
Those extremes, they’re childish.
那些极端说法很幼稚。
Dwarkesh Patel
Let me just make my argument for myself. The idea is not that there is some key threshold of compute. It’s that any marginal compute is helpful. So if you have more compute, you can train a better model.
让我自己把我的论点说清楚。我的意思并不是存在某个关键计算门槛。我的意思是，任何边际计算资源都有帮助。所以，如果你有更多计算资源，你就能训练出更好的模型。
Jensen Huang
And I just want you to acknowledge that any marginal sales for the American technology industry is beneficial.
而我只是希望你承认，对美国科技产业来说，任何边际销售都是有益的。
Dwarkesh Patel
I actually don’t… If the AI models that run on those chips are capable of cyber offensive capabilities, or the chips are training models with cyber capabilities and running more instances of those models, it is not a nuclear weapon, but it enables a weapon of a kind.
我其实不……如果运行在那些芯片上的 AI 模型具备网络攻击能力，或者那些芯片正在训练具备网络能力的模型，并运行更多这样的模型实例，它不是核武器，但它确实赋能了某种武器。
Jensen Huang
The logic that you use, you might as well say it to microprocessors and DRAMs. You might as well say it to electricity.
按照你使用的逻辑，你也完全可以把它用于微处理器和 DRAM。你甚至也可以把它用于电力。
Dwarkesh Patel
But in fact we do have export controls on the technology that is relevant to making the most advanced DRAM. We have all kinds of export controls on China for all kinds of chip-making stuff.
但事实上，我们确实对制造最先进 DRAM 所需的相关技术实施出口管制。我们对中国实施了各种各样涉及芯片制造的出口管制。
Jensen Huang
We sell a lot of DRAM and CPUs into China, and I think it’s right.
我们向中国销售大量 DRAM 和 CPU，我认为这是正确的。
Dwarkesh Patel
I guess this goes back to the fundamental question of, is AI different? If you have the kind of technology where they can find these zero-days in software, is that something where we want to minimize China’s ability to get there first, to deploy it widely?
我想这又回到了一个根本问题：AI 是否不同？如果你拥有一种技术，它们可以用来发现软件中的这些零日漏洞，那么我们是否希望尽量减少中国率先达到这种能力、并大规模部署它的可能性？
Jensen Huang
We want the United States to be ahead. We can control that.
我们希望美国领先。这一点我们可以控制。
Dwarkesh Patel
How do we control that if the chips are already there and they’re using them to train that model?
如果芯片已经在那里，而且他们正在用这些芯片训练那个模型，我们如何控制这一点？
Jensen Huang
We have tons of compute. We have tons of AI researchers. We’re racing as fast as we can.
我们拥有大量计算资源。我们拥有大量 AI 研究人员。我们正在尽可能快地竞赛。
Dwarkesh Patel
Again, we have more nuclear weapons than anybody else, but we don’t want to send enriched uranium anywhere.
再说一次，我们拥有比任何人都多的核武器，但我们并不想把浓缩铀送到任何地方。
Jensen Huang
We’re not enriched uranium. It’s a chip, and it’s a chip that they can make themselves.
我们不是浓缩铀。它是一块芯片，而且是一块他们自己能够制造的芯片。
Dwarkesh Patel
But there’s a reason they’re buying it from you. We have quotes from the founders of Chinese companies that say that they’re bottlenecked on compute.
但他们从你们这里购买是有原因的。我们有中国公司创始人的引述，他们说自己的瓶颈是计算资源。
Jensen Huang
Because our chips are better. On balance, our chips are better. There’s just no question about it. In the absence of our chip… Can you acknowledge that Huawei had a record year? Can you acknowledge that a whole bunch of chip companies have gone public? Can you acknowledge that?
因为我们的芯片更好。总体而言，我们的芯片更好。这一点毫无疑问。在没有我们芯片的情况下……你能承认 Huawei 创下了历史最好年份吗？你能承认一大批芯片公司已经上市了吗？你能承认这些吗？
Dwarkesh Patel
Yes.
能。
Jensen Huang
Can you also acknowledge that we used to have a very large share in that market, and we no longer have a large share in that market? We can also acknowledge that China is about 40% of the world’s technology industry. To concede that market for the United States technology industry is a disservice to our country. It is a disservice to our national security. It is a disservice to our technology leadership, all for the benefit of one company. It makes no sense to me.
你是否也能承认，我们曾经在那个市场拥有很大份额，而现在已经不再拥有很大份额？我们还可以承认，中国大约占全球科技产业的 40%。让美国科技产业放弃那个市场，是对我们国家的不利。是对我们国家安全的不利。是对我们技术领导力的不利，而这一切只是为了让一家公司受益。这对我来说说不通。
Dwarkesh Patel
I guess I’m confused. It feels like you’re making two different statements. One is that we’re going to win this competition with Huawei because our chips are going to be way better if we’re allowed to compete. Another is that they would be doing the same exact thing without us anyway. How can both of those things be true at the same time?
我想我有点困惑。听起来你在说两个不同的判断。一个是，如果允许我们竞争，我们会在与 Huawei 的竞争中获胜，因为我们的芯片会好得多。另一个是，即使没有我们，他们无论如何也会做完全同样的事情。这两件事怎么能同时为真？
Jensen Huang
It’s obviously true. In the absence of a better choice, you’ll take the only choice you have. How is that illogical? It’s so logical.
这显然是真的。在没有更好选择的情况下，你会选择你唯一拥有的选择。这有什么不合逻辑？这非常合逻辑。
Dwarkesh Patel
The reason they want Nvidia chips is that they’re better.
他们想要 Nvidia 芯片的原因，是它们更好。
Jensen Huang
Yeah.
是的。
Dwarkesh Patel
Better is more compute. More compute means you can train a better model.
更好就是更多计算资源。更多计算资源意味着你可以训练出更好的模型。
Jensen Huang
No, it’s just better. It’s better because it’s easier to program. We have a better ecosystem. But whatever the better is, whatever the better is… And of course we’re going to send them compute. So what? The fact of the matter is that we get to benefit. Don’t forget, we get the benefit of American technology leadership. We get the benefit of developers working on the American tech stack. We get the benefit, as those AI models diffuse out into the rest of the world, that the American tech stack is therefore the best for it. We can continue to advance and diffuse American technology. That, I believe, is a positive. It’s a very important part of American technology leadership.
不，它只是更好。它更好，是因为它更容易编程。我们有更好的生态系统。但不管这个“更好”具体是什么，不管它是什么……当然，我们会向他们提供计算资源。那又怎样？事实是，我们会从中受益。不要忘了，我们获得的是美国技术领导力的好处。我们获得的是开发者在美国技术栈上工作的好处。随着这些 AI 模型扩散到世界其他地方，我们获得的好处是，美国技术栈因此最适合运行它们。我们可以继续推进并扩散美国技术。我认为，这是正面的。它是美国技术领导力中非常重要的一部分。
Now, the policies that you’re advocating resulted in the American telecommunications industry being policied out of basically the world, to the point where we don’t control our own telecommunications anymore. I don’t see that as smart. It’s a little narrow-minded, and it led to unintended consequences that I’m describing to you right now that you seem to have a very hard time understanding.
现在，你所主张的那些政策，导致美国电信产业基本上被政策排挤出了世界市场，以至于我们现在不再控制自己的电信基础设施。我不认为这是聪明的。它有点狭隘，而且带来了我现在正在向你描述的这些意外后果，而你似乎很难理解。
Dwarkesh Patel
Okay, let’s just step back. It seems like the crux here is there’s a potential benefit and there’s a potential cost. What we’re trying to figure out is, is the benefit worth the cost? I guess I’m trying to get you to acknowledge the potential cost. Compute is an input to training powerful models. Powerful models do have powerful offensive capabilities, like cyber attacks. It is a good thing that American companies got to Mythos-level capabilities first, and then now they’re going to hold off on those capabilities so that the American companies and American government can make their software more protected before that level of capability was announced.
好，我们退一步看。这里的核心似乎是：存在一个潜在收益，也存在一个潜在成本。我们想弄清楚的是，这个收益是否值得付出这个成本。我想我是在试图让你承认这个潜在成本。计算资源是训练强大模型的一项输入。强大的模型确实具有强大的攻击能力，比如网络攻击能力。美国公司率先达到 Mythos 级别能力是一件好事，而且现在它们会暂缓释放这些能力，让美国公司和美国政府能够在这种能力水平被公开之前，把自己的软件保护得更好。
If China had had more compute or more crowd compute, if they could have made a Mythos-level model earlier and deployed it widely, that would have been very bad. One of the reasons that hasn’t happened is that we have more compute thanks to companies like Nvidia in America. That is a cost of sending it to China. So let’s leave the benefit aside for a second. Do you acknowledge that this is a potential cost?
如果中国拥有更多计算资源，或者更多众包计算资源，如果他们能够更早做出 Mythos 级别模型并广泛部署，那会非常糟糕。这件事没有发生的原因之一，是因为在美国有像 Nvidia 这样的公司，我们拥有更多计算资源。这就是把它发送给中国的一个成本。所以我们先暂时把收益放在一边。你是否承认这是一个潜在成本？
Jensen Huang
I’ll also tell you the potential cost is we allow one of the most important layers of the AI stack, the chip layer, to concede an entire market—the second largest market in the world—so that they could develop scale, so that they could develop their own ecosystem, so that future AI models are optimized in a very different way than the American tech stack. As AI diffuses out into the rest of the world, their standards, their tech stack, will become superior to ours, because their models are open.
我也要告诉你另一个潜在成本：我们允许 AI 技术栈中最重要的层之一，也就是芯片层，放弃整个市场——世界第二大市场——从而让他们发展规模，让他们发展自己的生态系统，让未来的 AI 模型以一种与美国技术栈非常不同的方式被优化。随着 AI 扩散到世界其他地方，他们的标准、他们的技术栈，将会变得优于我们的技术栈，因为他们的模型是开放的。
Dwarkesh Patel
I guess I just believe enough in Nvidia’s kernel engineers and CUDA engineers to think that they could optimize—
我想，我只是足够相信 Nvidia 的内核工程师和 CUDA 工程师，相信他们能够优化——
Jensen Huang
AI is more than kernel optimization, as you know.
AI 不只是内核优化，你也知道。
Dwarkesh Patel
Of course, but there are so many things you can do, from distilling to a model that’s well-fit for your chips.
当然，但你们可以做很多事情，从蒸馏到一个更适配你们芯片的模型，等等。
Jensen Huang
We’re going to do our best.
我们会尽最大努力。
Dwarkesh Patel
You have all the software. It’s just hard to imagine that there’s a long-term lock-in to the Chinese ecosystem, even if they have a slightly better open source model for a while.
你们拥有所有软件。即使他们在一段时间里拥有一个稍微更好的开源模型，也很难想象会出现长期锁定到中国生态系统的情况。
Jensen Huang
China is the largest contributor to open source software in the world. Fact. China’s the largest contributor to open models in the world. Fact. Today it’s built on the American tech stack, Nvidia’s. Fact.
中国是全球开源软件的最大贡献者。事实。中国是全球开放模型的最大贡献者。事实。今天，它建立在美国技术栈上，也就是 Nvidia 的技术栈上。事实。
All five layers of the tech stack for AI are important. The United States ought to go win all five of them. They’re all important. The one that is the most important, of course, is the AI application layer. The layer that diffuses into society, the one that uses it most will benefit from this industrial revolution most. But my point is that every layer has to succeed.
AI 技术栈的五层都很重要。美国应该去赢下全部五层。它们都重要。当然，最重要的一层是 AI 应用层。真正扩散到社会中的那一层，使用 AI 最多的那一层，会从这场工业革命中受益最多。但我的观点是，每一层都必须成功。
If we scare this country into thinking that AI is somehow a nuclear bomb, so that everybody hates AI and everybody’s afraid of AI, I don’t know how you’re helping the United States. You’re doing it a disservice. If we scare everybody out of doing software engineering jobs because it’s going to kill every software engineering job—and we don’t have any software engineers as a result of that—we’re doing a disservice to the United States.
如果我们吓唬这个国家，让它以为 AI 在某种意义上是核弹，以至于每个人都讨厌 AI、每个人都害怕 AI，我不知道你这是如何帮助美国。你是在伤害美国。如果我们把所有人都吓得不去做软件工程工作，因为 AI 会消灭每一个软件工程岗位——结果我们因此没有软件工程师——那我们就是在伤害美国。
If we scare everybody out of radiology so nobody wants to be a radiologist because computer vision is completely free and no AI is going to do a worse job than a radiologist, we misunderstand the difference between a job and a task. The job of a radiologist is patient care. The task is to read a scan. If we misunderstand that so profoundly and we scare everybody out of going to radiology school, we’re not going to have enough radiologists and good enough healthcare.
如果我们把所有人都吓得不去做放射科医生，因为计算机视觉完全免费，而且 AI 不会比放射科医生做得更差，那么我们就误解了工作和任务之间的区别。放射科医生的工作是照护患者。任务是阅读扫描影像。如果我们如此深刻地误解这一点，并把所有人都吓得不去上放射科医学院，我们就不会有足够的放射科医生，也不会有足够好的医疗服务。
So I’m making the case that when you make a premise that is so extreme, everything goes from zero or infinity, we end up scaring people in a way that’s just not true. Life is not like that. Do we want the United States to be first? Of course we do. Do we need to be a leader in every layer of that stack? Of course we do. Of course we do. Today you’re talking about Mythos because Mythos is important. Sure. That’s fantastic.
所以我要说明的是，当你提出一个如此极端的前提，让所有事情都变成零或者无穷大时，我们最终会以一种并不真实的方式吓唬人。生活不是这样的。我们希望美国第一吗？当然希望。我们需要在这个技术栈的每一层都成为领导者吗？当然需要。当然需要。今天你谈论 Mythos，因为 Mythos 很重要。没错。这很好。
But in a few years time, I’m making you the prediction that when we want the American tech stack, when we want American technology to be diffused around the world—out to India, out to the Middle East, out to Africa, out to Southeast Asia—when our country would like to export, because we would like to export our technology, we would like to export our standards, on that day, I want you and I to have that same conversation again. I will tell you exactly about today’s conversation, about how your policy and what you imagined literally caused the United States to concede the second largest market in the world for no good reason at all.
但几年之后，我向你作一个预测：当我们希望美国技术栈扩散，当我们希望美国技术扩散到全世界——扩散到印度、中东、非洲、东南亚——当我们的国家希望出口，因为我们希望出口我们的技术、出口我们的标准，到那一天，我希望你和我再进行同样的对话。我会准确地告诉你今天这场对话，告诉你你的政策和你的想象，如何真的导致美国毫无正当理由地放弃了世界第二大市场。
We shouldn’t concede it. If we lose it, we lose it. But why do we concede it? Now nobody is advocating an all or nothing. Nobody’s advocating all or nothing, meaning we ship everything to China at all times. Nobody’s advocating that. We should always have the best technology here. We should always have the most technology here, and the first. But we should also try to compete and win around the world. Both of those things can simultaneously happen. It requires some amount of nuance, some amount of maturity instead of absolutes. The world is just not absolutes.
我们不应该主动放弃它。如果我们输了，那就是输了。但为什么要主动放弃？现在没有人在主张全有或全无。没有人在主张全有或全无，也就是说，不是说我们任何时候都把一切运往中国。没有人在主张这一点。我们应该始终把最好的技术留在这里。我们应该始终让这里拥有最多的技术，并且最先拥有。但我们也应该努力在全世界竞争并获胜。这两件事可以同时发生。它需要一定程度的细致权衡，一定程度的成熟，而不是绝对化。世界并不是绝对化的。
Dwarkesh Patel
Okay. The argument hinges on this. They’ve built models that are specified for the best chips that they make in a few years. Those chips get exported around the world. That sets the standard. Because of EUV export controls, as we said, you’re going to move on to 1.6nm. They’re still going to be on 7nm, even after a few years from now.
好。这个论证取决于这一点。他们会构建一些模型，专门适配他们几年后所能制造的最好芯片。这些芯片会出口到全世界。于是这就设定了标准。由于 EUV 出口管制，正如我们刚才说的，你们会推进到 1.6 纳米。而他们即使在几年之后，仍然会停留在 7 纳米。
It may make sense that domestically they would prefer, “Hey, we’ve got so much energy, we can manufacture at scale. We’ll still keep using 7nm.” But on the exporting thing, their 7nm chips have to be competitive against your 1.6nm chips. Their models have to be so far optimized for the 7nm that it’s better to run their models on 7nm than to run their models on your 1.6nm.
在国内，他们也许会倾向于说：“嘿，我们有这么多能源，我们可以大规模制造。我们仍然继续使用 7 纳米。”这可能说得通。但在出口这件事上，他们的 7 纳米芯片必须能和你们的 1.6 纳米芯片竞争。他们的模型必须被针对 7 纳米优化到这样一种程度：运行在 7 纳米上，反而比运行在你们的 1.6 纳米上更好。
Jensen Huang
Can we just look at the facts then? Is Blackwell 50 times more advanced lithography than Hopper? Is it 50 times? Not even close. I just kept saying it over and over again. Moore’s Law is dead. Between Hopper and Blackwell, from the transistors themselves, call it 75%. It was three years apart, 75%. Blackwell is 50 times Hopper.
那我们能不能只看事实？Blackwell 的光刻技术比 Hopper 先进 50 倍吗？有 50 倍吗？远远没有。我一直在反复说这一点。Moore’s Law 已经死了。从晶体管本身来看，Hopper 到 Blackwell 之间，大概提升 75%。它们相隔三年，提升 75%。但 Blackwell 是 Hopper 的 50 倍。
My point is, architecture matters. Computer science matters. Semiconductor physics matters as well, but computer science matters. The impact of AI largely comes from the computing stack, which is the reason why CUDA is so effective, which is the reason why CUDA is so beloved. It’s an ecosystem, a computing architecture that allows for so much flexibility that if you wanted to change an architecture completely—create something like MoE, create something like diffusion, create something that’s disaggregated—you could do so. It’s easy to do.
我的观点是，架构很重要。计算机科学很重要。半导体物理也重要，但计算机科学很重要。AI 的影响很大程度上来自计算技术栈，这就是 CUDA 如此有效的原因，也是 CUDA 如此受欢迎的原因。它是一个生态系统，是一种计算架构，提供了非常大的灵活性，使得如果你想完全改变一种架构——创造类似 MoE 的东西，创造类似扩散模型的东西，创造某种解耦的东西——你都可以做到。而且很容易做到。
So the fact of the matter is, AI is about the stack above as much as it is about the architecture below. To the extent that we have architectures and software stacks that are optimized for our stack, for our ecosystem, it is obviously good, because we started the conversation today about how Nvidia’s ecosystem is so rich. Why do people always love programming CUDA first? They do. They do. So do the researchers in China.
所以事实是，AI 既关乎上层技术栈，也关乎下层架构。只要我们拥有针对我们的技术栈、我们的生态系统优化的架构和软件栈，这显然是好事，因为我们今天一开始就谈到 Nvidia 的生态系统有多丰富。为什么人们总是喜欢先为 CUDA 编程？他们确实如此。他们确实如此。中国的研究人员也是如此。
But if we are forced to leave China, if we’re forced to leave China, first of all, it’s a policy mistake. Obviously it has backlash. It has turned out badly for the United States. It enabled, it accelerated their chip industry. It forced all of their AI ecosystem to focus on their internal architectures. It’s not too late, but nonetheless it has already happened.
但如果我们被迫离开中国，如果我们被迫离开中国，首先，这就是一个政策错误。显然，它会产生反作用。结果已经对美国不利。它赋能了、加速了他们的芯片产业。它迫使他们整个 AI 生态系统都聚焦于自己的内部架构。现在还不算太晚，但无论如何，这已经发生了。
You’re going to see in the future, they’re not stuck at 7nm, obviously. They’re good at manufacturing. They will continue to advance from 7nm and beyond. Now, is there a 10x difference between 5nm and 7nm? The answer is no. Architecture matters. Networking matters. That’s why Nvidia bought Mellanox. Networking matters. Energy matters. So all of that stuff matters. It’s not simplistic, like the way you’re trying to distill it.
未来你会看到，他们显然不会被困在 7 纳米。他们擅长制造。他们会继续从 7 纳米向前推进。现在，5 纳米和 7 纳米之间有 10 倍差距吗？答案是没有。架构很重要。网络很重要。这就是 Nvidia 收购 Mellanox 的原因。网络很重要。能源很重要。所以所有这些东西都重要。它并不像你试图提炼的那样简单。
01:35:06 – Why doesn’t Nvidia make multiple different chip architectures?
01:35:06 – 为什么 Nvidia 不制造多种不同的芯片架构？
Dwarkesh Patel
We can move on from China, but that actually raises an interesting question. We were discussing earlier these bottlenecks at TSMC and memory and so forth.
我们可以从中国问题转开，但这实际上引出了一个有趣的问题。我们之前讨论了 TSMC、存储器等等方面的瓶颈。
So if we’re in this world where you’re already the majority of N3—and at some point you’ll be N2 and you’ll be a majority of that—do you see that you could go back to N7, the spare capacity at an older process node, and say, “Hey, the demand for AI is so great and our capacity to expand the leading edge is not meeting it, so we’re going to make a Hopper or Ampere, but with everything we know about numerics today and all the other improvements you described”? Do you see that world happening before 2030?
所以，如果我们处在这样一个世界里：你们已经占据 N3 的大部分产能，而某个时候你们会转向 N2，并且也会占据那里的大部分产能——你是否认为你们可以回到 N7，利用较老制程节点上的闲置产能，然后说：“嘿，AI 需求太大，而我们扩展前沿制程产能的能力跟不上，所以我们要制造一个 Hopper 或 Ampere，但把我们今天对数值表示所知道的一切，以及你描述的其他所有改进都加入进去”？你认为 2030 年之前会出现这样的世界吗？
Jensen Huang
It’s not necessary to. The reason for that is because with every generation, the architecture is more than just the transistor scale. You’re doing so much engineering and packaging and stacking, and the numerics and the system architecture.
没有必要这样做。原因在于，每一代架构都不只是晶体管尺度的问题。你还在做大量工程、封装、堆叠、数值表示和系统架构。
When you run out of capacity, to easily go back to another node… That’s a level of R&D that no one could afford. We could afford to lean forward. I don’t think we could afford to go back. Now, if the world simply says… If on that day, let’s do the thought experiment, on that day we go, “Listen, we’re just never going to have more capacity ever again.” Would I go back and use 7nm? In a heartbeat, of course I would.
当你产能耗尽时，想轻易回到另一个节点……那需要的研发投入，没有人负担得起。我们有能力向前押注。我不认为我们有能力往回走。现在，如果世界只是说……我们做一个思想实验，如果到了那一天，我们说：“听着，我们以后再也不会有更多产能了。”我会不会回去使用 7 纳米？我会立刻这样做，当然会。
Dwarkesh Patel
One question somebody I was talking to had is, why doesn’t Nvidia run multiple different chip projects at the same time with totally different architecture? So you could do something like a Cerebras-style wafer scale. You could do a Dojo-style huge package. You could do one without CUDA. You have the resources and the engineering talent to do all of these in parallel. So why put all the eggs in one basket, given who knows where AI might go and architectures might go?
我和别人交流时，有人提出一个问题：为什么 Nvidia 不同时运行多个采用完全不同架构的芯片项目？比如，你们可以做类似 Cerebras 风格的晶圆级芯片。可以做类似 Dojo 风格的巨大封装。也可以做一个不使用 CUDA 的架构。你们有资源，也有工程人才，可以并行做所有这些事情。那么，既然没人知道 AI 会走向哪里、架构会走向哪里，为什么要把所有鸡蛋放在一个篮子里？
Jensen Huang
Oh, we could. It’s just that we don’t have a better idea. We could do all of those things. It’s just not better. We simulate it all in our simulator, proveably worse. So we wouldn’t do it. We’re working on exactly the projects that we want to work on. If the workload were to change dramatically—and I don’t mean the algorithms, I actually mean the workload, and that depends on the shape of the market—we may decide to add other accelerators.
哦，我们可以。只是我们没有更好的想法。我们可以做所有那些事情。只是它们并不更好。我们会在自己的模拟器里模拟所有这些方案，结果可证明地更差。所以我们不会做。我们正在做的，正是我们想做的项目。如果工作负载发生巨大变化——我说的不是算法，我真正指的是工作负载，而这取决于市场形态——我们可能会决定增加其他加速器。
For example, recently we added Groq, and we’re going to fold Groq into our CUDA ecosystem. We’re doing that now because the value of tokens has gone up so high that you could have different pricing of tokens. Back in the old days, just a couple years ago, tokens were either free or barely expensive. But now you can have different customers, and those customers want different answers. Because the customers make so much money—for example, our software engineers—if I can give them much more responsive tokens so that they’re even more productive than they are today, I would pay for it.
比如，最近我们加入了 Groq，并且会把 Groq 纳入我们的 CUDA 生态系统。我们现在这么做，是因为词元的价值已经上升得非常高，以至于你可以对词元进行不同定价。过去，也就是几年前，词元要么免费，要么几乎不贵。但现在你可以拥有不同类型的客户，而这些客户想要不同类型的答案。因为客户能从中赚很多钱——比如我们的软件工程师——如果我能给他们响应速度快得多的词元，让他们比今天更有生产力，我愿意为此付费。
But that market has only recently emerged. So I think we now have the ability to have the same model, based on the response time, have different segments. That’s the reason why we decided to expand the Pareto frontier and create a segment of inference that is faster response time, even though it’s lower throughput. Until now, higher throughput is always better. We think there could be a world where there could be very high ASP tokens, and even though the throughput is lower in the factory, the ASPs make up for it.
但这个市场只是最近才出现。所以我认为，我们现在有能力让同一个模型，根据响应时间，服务于不同细分市场。这就是为什么我们决定扩展 Pareto 前沿，并创造一个推理细分市场：响应时间更快，即使吞吐量更低。到目前为止，更高吞吐量一直被认为更好。我们认为，可能会有这样一个世界：存在平均售价非常高的词元，即使工厂里的吞吐量更低，平均售价也足以弥补这一点。
That’s the reason why we did it. But otherwise, from an architecture perspective, if I had more money, I would put more behind Nvidia’s architecture.
这就是我们这样做的原因。但除此之外，从架构角度看，如果我有更多钱，我会把更多资源投入到 Nvidia 的架构后面。
Dwarkesh Patel
I think this idea of extremely premium tokens and just the disaggregation of the inference market is a very interesting.
我认为这种极高端词元，以及推理市场解耦的想法，非常有意思。
Jensen Huang
The segmentation of it.
是它的细分。
Dwarkesh Patel
Yeah. Alright, final question. Suppose the deep learning revolution didn’t happen. What would Nvidia be doing? Obviously games, but given—
是的。好，最后一个问题。假设深度学习革命没有发生，Nvidia 会在做什么？显然有游戏业务，但考虑到——
Jensen Huang
Accelerated computing, the same thing we’ve been doing all along. The premise of our company is that Moore’s law is going to… General purpose computing is good for a lot of things, but for a lot of computation it’s not ideal.
加速计算，也就是我们一直以来都在做的同一件事。我们公司的前提是，Moore’s Law 将会……通用计算适合很多事情，但对很多计算来说，它并不理想。
So we combined an architecture called a GPU, CUDA, to a CPU, so that we can accelerate the workload of the CPU. Different kernels of code or algorithms could be offloaded onto our GPU. As a result, you speed up an application by 100x, 200x. Where can you use that? Obviously engineering and science and physics, data processing, computer graphics, image generation, all kinds of things. Even if AI doesn’t exist today, Nvidia would be very, very large.
所以我们把一种叫 GPU 的架构、CUDA，与 CPU 结合起来，这样我们就能加速 CPU 的工作负载。不同的代码内核或算法可以被卸载到我们的 GPU 上。结果是，你可以把一个应用加速 100 倍、200 倍。你能在哪里使用它？显然包括工程、科学、物理、数据处理、计算机图形、图像生成，以及各种各样的事情。即使今天 AI 不存在，Nvidia 也会非常、非常大。
The reason for that is fairly fundamental, which is that the ability for general purpose computing to continue to scale has largely run its course. And the only way… Not the only way, but the way to do that is through domain-specific acceleration. One of the domains that we started with was computer graphics, but there are many other domains. There’s all kinds. Particle physics and fluids, structured data processing, all kinds of different types of algorithms that benefit from CUDA.
原因相当根本：通用计算继续扩展的能力，基本上已经走完了自己的道路。而唯一的方法……不是唯一的方法，但一种方法，是通过面向特定领域的加速来实现。我们最初切入的领域之一是计算机图形，但还有很多其他领域。各种各样的领域都有。粒子物理和流体、结构化数据处理，以及各种不同类型的算法，都能从 CUDA 中受益。
Our mission was really to bring accelerated computing to the world and advance the type of applications that general purpose computing can’t do, and scale to the level of capability that helps break through certain fields of science. Some of the early applications were molecular dynamics, seismic processing for energy discovery, image processing of course, all of those kinds of fields where general purpose computing is just simply too inefficient to do so.
我们的使命实际上是把加速计算带给世界，并推动那些通用计算无法完成的应用，把能力扩展到足以帮助某些科学领域实现突破的水平。一些早期应用包括分子动力学、能源勘探中的地震数据处理，当然还有图像处理，以及所有这些通用计算做起来实在太低效的领域。
If there were no AI, I would be very sad. But because of the advances that we made in computing, we democratized deep learning. We made it possible for any researcher, any scientist, anywhere, any student, to be able to access a PC or a GeForce add-in card and do amazing science. That fundamental promise hasn’t changed, not even a little bit.
如果没有 AI，我会非常难过。但正因为我们在计算方面取得了进步，我们让深度学习变得民主化。我们让任何地方的任何研究人员、任何科学家、任何学生，都能够使用一台个人电脑或一块 GeForce 扩展卡，做出惊人的科学研究。这个根本承诺没有改变，一点都没有改变。
If you watch GTC, there’s the whole beginning part of it. None of it’s AI. That whole part of it with computational lithography or our quantum chemistry work, data processing work, all of that stuff is unrelated to AI. And it’s still very important. I know that AI is very interesting and quite exciting, but there’s a lot of people doing a lot of very important work that’s not AI related, and tensors are not the only way that you compute it. We want to help everybody.
如果你看 GTC，开头有一整部分内容都不是 AI。那一整部分内容涉及计算光刻、我们的量子化学工作、数据处理工作，所有这些都和 AI 无关。而它们仍然非常重要。我知道 AI 非常有意思，也相当令人兴奋，但有很多人在做大量非常重要、但与 AI 无关的工作，而且张量并不是你进行计算的唯一方式。我们希望帮助所有人。
Dwarkesh Patel
Jensen, thank you so much.
Jensen，非常感谢。
Jensen Huang
You’re welcome. I enjoyed it.
不客气。我很享受这次对话。
Dwarkesh Patel
Me too.
我也是。
热门主题
Recent Articles
2026-04-04 Andrej Karpathy.LLM Wiki
Refer To：《2026-04-04 Andrej Karpathy.LLM Wiki》。 LLM Wiki A pattern for building personal knowledge bases using LLMs. 一种使用 LLM 构建个人知识库的模式。 This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, ...2026-04-28 潘乱.从红果到AI短剧：谁在革谁的命？
Refer To：《从红果到AI短剧：谁在革谁的命？》。 红果短剧的快速崛起与用户增长逻辑 红果短剧在三年内实现日活过亿的爆发式增长，主要得益于其免费模式和对非长视频用户的有效触达。与优爱腾等长视频平台偏向正剧的定位不同，短剧更接近于电影的消费体验，但通过广告变现降低了消费门槛。AI 漫剧作为新兴品类，在去年下半年开始崭露头角，虽然与传统大制作动漫路径不同，但其生产效率和题材丰富度正在迅速提升，成为行业新的增长点。 王小书: (00:04) Hmm. 潘乱: (00:04) ...2020-12-10 王宁.潮流玩具风靡背后的心理学
Refer To：《泡泡玛特王宁：潮流玩具风靡背后的心理学》。 于近年来以Molly、Pucky、Dimoo等各类IP受到Z世代消费者欢迎的泡泡玛特，其实已经有十年历史。 “我从自己刷墙，开第一家实体店，做零售业，是在2008年5月13号，到这周末就是整整11年了。我们是创业老兵了，单泡泡玛特这个品牌就有9年。” ...2022-01-08 王宁.不做「你死我活」的生意
Refer To：《泡泡玛特王宁：不做「你死我活」的生意》。 今年全球最火的玩具，非Labubu莫属。 6月11日，一只稀有款薄荷色Labubu以人民币108万元成交价在二级市场拍出。就是下面这只—— 图片 6月14日，因为韩国地区线下销售太火爆，恐引发安全问题，泡泡玛特发公告暂停Labubu全系列销售。 Labubu全球爆火直接拉动泡泡玛特股价飙涨，今年以来，其股价涨幅超过200%，市值超过3500亿元，创始人王宁也因此取代牧原股份秦英林，成为新晋河南首富。 ...2026-05-13 Alex Wang.Meta's AI Chief On AI Beef, New Models And Life With Zuck
Refer To：《Meta's AI Chief On AI Beef, New Models And Life With Zuck》。 Meta Superintelligence Labs Structure and Strategic Compute Advantage Meta Superintelligence Labs 的组织结构与战略算力优势 Meta Superintelligence Labs (MSL) operates through a specialized ...