# Jensen Huang – TPU competition, why we should sell chips to China, & Nvidia’s supply chain moat
- Transcript ID: `fb65385aa939ab62`
- Source tier: `official_authored_text`
- Original language: `en_or_source_language_unlabeled`
- Translation status: `agent_translated_reviewed` (488/488 chunks)
- Original source: https://www.dwarkesh.com/p/jensen-huang
- Raw evidence: `data/transcripts/web/dwarkesh-jensen-huang-2026-04-15.json`
> 中文是派生翻译，英文是留存的源文本。中文未复核时，以英文和原始音视频为准。
## paragraph 1
**中文**
我向黄仁勋询问了 TPU 的竞争、英伟达对先进芯片所需且愈发成为瓶颈的供应链的掌控、我们是否应该向中国出售 AI 芯片、英伟达为什么不干脆成为一家超大规模云服务商、它如何开展投资，以及更多话题。祝大家看得愉快！
**English / Original**
I asked Jensen about TPU competition, Nvidia’s lock on the ever more bottlenecked supply chain needed to make advanced chips, whether we should be selling AI chips to China, why Nvidia doesn’t just become a hyperscaler, how it makes its investments, and much more. Enjoy!
## paragraph 2
**中文**
可在 YouTube 观看；也可在 Apple Podcasts 或 Spotify 收听。
**English / Original**
Watch on YouTube; listen on Apple Podcasts or Spotify.
## paragraph 3
**中文**
赞助商
**English / Original**
Sponsors
## paragraph 4
**中文**
Crusoe 的云采用最先进的 Blackwell GPU，并计划于今年晚些时候部署 Vera Rubin。但硬件只是故事的一部分——在推理方面，Crusoe 的 MemoryAlloy 技术实现了集群级 KV 缓存，首个词元生成时间（TTFT）最高可提速 10 倍，吞吐量较 vLLM 提升 5 倍。详情请访问 crusoe.ai/dwarkesh。
**English / Original**
Crusoe’s cloud runs on state-of-the-art Blackwell GPUs, with Vera Rubin deployment scheduled for later this year. But hardware is only part of the story—for inference, Crusoe’s MemoryAlloy tech implements a cluster-wide KV cache, delivering up to 10x faster TTFT and 5x better throughput than vLLM. Learn more at crusoe.ai/dwarkesh
## paragraph 5
**中文**
Crusoe 的云采用最先进的 Blackwell GPU，并计划于今年晚些时候部署 Vera Rubin。但硬件只是故事的一部分——在推理方面，Crusoe 的 MemoryAlloy 技术实现了集群级 KV 缓存，首个词元生成时间（TTFT）最高可提速 10 倍，吞吐量较 vLLM 提升 5 倍。详情请访问 crusoe.ai/dwarkesh。
**English / Original**
Crusoe’s cloud runs on state-of-the-art Blackwell GPUs, with Vera Rubin deployment scheduled for later this year. But hardware is only part of the story—for inference, Crusoe’s MemoryAlloy tech implements a cluster-wide KV cache, delivering up to 10x faster TTFT and 5x better throughput than vLLM. Learn more at crusoe.ai/dwarkesh
## paragraph 6
**中文**
Cursor 帮我用一个周末构建了一位 AI 协作研究员。现在，我有了一个 AI 智能体，可以通过 Google Docs 的行内评论串与它协作！其他智能体编程工具常让人感觉完全是个黑箱，而 Cursor 让我始终掌握整个实现过程。你可以在这里试用我的协作研究员，也可以今天就访问 cursor.com/dwarkesh，开始自己的 Cursor 项目。
**English / Original**
Cursor helped me build an AI co-researcher over the course of a weekend. Now I have an AI agent that I can collaborate with in Google Docs via inline comment threads! And while other agentic coding tools feel like a total black-box, Cursor let me stay on top of the full implementation. You can try my co-researcher out here, or get started on your own Cursor project today at cursor.com/dwarkesh
## paragraph 7
**中文**
Cursor 帮我用一个周末构建了一位 AI 协作研究员。现在，我有了一个 AI 智能体，可以通过 Google Docs 的行内评论串与它协作！其他智能体编程工具常让人感觉完全是个黑箱，而 Cursor 让我始终掌握整个实现过程。你可以在这里试用我的协作研究员，也可以今天就访问 cursor.com/dwarkesh，开始自己的 Cursor 项目。
**English / Original**
Cursor helped me build an AI co-researcher over the course of a weekend. Now I have an AI agent that I can collaborate with in Google Docs via inline comment threads! And while other agentic coding tools feel like a total black-box, Cursor let me stay on top of the full implementation. You can try my co-researcher out here, or get started on your own Cursor project today at cursor.com/dwarkesh
## paragraph 8
**中文**
Jane Street 花费约 2 万个 GPU 小时，在三个不同的语言模型中训练植入后门，随后向我的观众发起挑战，让他们找出触发条件。他们收到了一些很巧妙的解法——比如比较基础版本和微调版本，并外推所有差异来揭示隐藏后门——但没有人能把三个全部破解。如果这类开放问题令你兴奋，Jane Street 正在招聘。详情请访问 janestreet.com/dwarkesh。
**English / Original**
Jane Street spent ~20,000 GPU hours training backdoors into 3 different language models, then challenged my audience to find the triggers. They received some clever solutions—like comparing the base and fine-tuned versions and extrapolating any differences to reveal the hidden backdoor—but no one was able to solve all 3. So if open problems like this excite you, Jane Street is hiring. Learn more at janestreet.com/dwarkesh
## paragraph 9
**中文**
Jane Street 花费约 2 万个 GPU 小时，在三个不同的语言模型中训练植入后门，随后向我的观众发起挑战，让他们找出触发条件。他们收到了一些很巧妙的解法——比如比较基础版本和微调版本，并外推所有差异来揭示隐藏后门——但没有人能把三个全部破解。如果这类开放问题令你兴奋，Jane Street 正在招聘。详情请访问 janestreet.com/dwarkesh。
**English / Original**
Jane Street spent ~20,000 GPU hours training backdoors into 3 different language models, then challenged my audience to find the triggers. They received some clever solutions—like comparing the base and fine-tuned versions and extrapolating any differences to reveal the hidden backdoor—but no one was able to solve all 3. So if open problems like this excite you, Jane Street is hiring. Learn more at janestreet.com/dwarkesh
## paragraph 10
**中文**
时间戳
**English / Original**
Timestamps
## paragraph 11
**中文**
（00:00:00）——英伟达最大的护城河，是它对稀缺供应链的掌控吗？
**English / Original**
(00:00:00) – Is Nvidia’s biggest moat its grip on scarce supply chains?
## paragraph 12
**中文**
（00:16:25）——TPU 会打破英伟达对 AI 算力的掌控吗？
**English / Original**
(00:16:25) – Will TPUs break Nvidia’s hold on AI compute?
## paragraph 13
**中文**
（00:41:06）——英伟达为什么不成为超大规模云服务商？
**English / Original**
(00:41:06) – Why doesn’t Nvidia become a hyperscaler?
## paragraph 14
**中文**
（00:57:36）——我们应该向中国出售 AI 芯片吗？
**English / Original**
(00:57:36) – Should we be selling AI chips to China?
## paragraph 15
**中文**
（01:35:06）——英伟达为什么不打造多种不同的芯片架构？
**English / Original**
(01:35:06) – Why doesn’t Nvidia make multiple different chip architectures?
## paragraph 16
**中文**
访谈实录 00:00:00——英伟达最大的护城河，是它对稀缺供应链的掌控吗？
**English / Original**
Transcript 00:00:00 – Is Nvidia’s biggest moat its grip on scarce supply chains?
## paragraph 18
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 19
**中文**
我们已经看到许多软件公司的估值暴跌，因为人们预计 AI 会让软件商品化。有一种或许过于天真的思路是：你看，英伟达把一份 GDS2 文件发给台积电，台积电制造逻辑裸片和交换芯片，再把它们与 SK 海力士、美光和三星生产的 HBM 封装在一起，然后送到台湾的 ODM 厂商那里组装成机架。从根本上说，英伟达做的是由别人负责制造的软件；如果软件会商品化，英伟达也会被商品化吗？
**English / Original**
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?
## paragraph 20
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 21
**中文**
归根结底，总得有某种东西把电子转化为词元。把电子转化为词元，并让这些词元随时间变得更有价值，很难被彻底商品化。从电子到词元的转化是一段不可思议的旅程。创造一个词元，就像让一种分子比另一种分子更有价值，也就是让一个词元比另一个更有价值。为了让那个词元变得有价值，需要投入大量艺术性、工程、科学和发明；显然，我们正在实时见证这一切。人们远未深入理解其中的转化、制造以及全部科学，这段旅程也远未结束。我不认为它会被商品化。
**English / Original**
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.
## paragraph 22
**中文**
当然，我们会让它变得更高效。你刚才设问的方式，正是我理解我们公司的心智模型：输入是电子，输出是词元，中间是英伟达。我们的工作，是为实现这种能力惊人的转化，做一切必要之事，同时尽可能少做。所谓“尽可能少做”，就是凡非我必须亲自做的，我都会找人合作，让它成为我的生态系统的一部分。
**English / Original**
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.
## paragraph 23
**中文**
看看今天的英伟达，我们或许拥有规模最大的合作伙伴生态：既包括供应链上下游，也包括所有计算机公司、应用开发者和模型开发者。可以说，AI 是一块五层蛋糕，而我们的生态遍及这五层。我们努力尽可能少做，但事实证明，必须由我们来做的那部分难得惊人。我不认为它会被商品化。
**English / Original**
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.
## paragraph 24
**中文**
事实上，我也不认为企业软件公司和工具厂商会被商品化……今天大多数软件公司都是工具厂商，有些不是，其中一些是把工作流编码为系统的公司。但很多公司做的就是工具。例如 Excel 是工具，PowerPoint 是工具，Cadence 做工具，Synopsys 也做工具。我的看法其实与大家相反。我认为智能体数量将呈指数级增长，工具用户数量也将呈指数级增长。这些工具的运行实例数很可能会暴增。
**English / Original**
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.
## paragraph 25
**中文**
Synopsys Design Compiler 的运行实例数很可能会暴增，使用布局规划器、我们的版图工具和设计规则检查器的智能体数量也会一同猛增。如今我们受限于工程师人数。将来，这些工程师会得到一大批智能体的支持。我们会以前所未见的方式探索设计空间，而且还会使用今天就在用的这些工具。
**English / Original**
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.
## paragraph 26
**中文**
我认为，工具使用会推动软件公司飞速增长。之所以尚未发生，是因为智能体还不够擅长使用工具。要么这些公司自己构建智能体，要么智能体会进步到足以使用那些工具的程度。我认为最终会是两者的结合。
**English / Original**
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.
## paragraph 27
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 28
**中文**
我记得你们最近的申报文件显示，你们对晶圆代工、存储和封装的采购承诺接近 1000 亿美元。SemiAnalysis 报道称，这类采购承诺将达到 2500 亿美元。一种解读是，英伟达真正的护城河在于，你们已经锁定了未来多年的这些稀缺部件。别人也许有加速器，但他们真能拿到制造它所需的存储芯片吗？真能拿到制造它所需的逻辑芯片吗？这真的是英伟达未来几年最大的护城河吗？
**English / Original**
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?
## paragraph 29
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 30
**中文**
这是我们能做到、而别人很难做到的事情之一。我们对上游作出了巨额承诺，其中一部分是明确的，也就是你提到的那些承诺；还有一部分是隐性的。例如，上游很多投资是供应链在听取我的意见后作出的，因为我会对那些 CEO 说：“让我告诉你这个行业将变得多么庞大；让我解释原因，和你一起把逻辑推演清楚，再让你看看我所看到的未来。”
**English / Original**
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.”
## paragraph 31
**中文**
通过向各个上游行业的 CEO 传递信息、激发信心并协调方向，他们愿意作出这些投资。为什么他们愿意为我投资，而不是为别人？因为他们知道，我有能力买下他们的供应，并通过下游把产品卖出去。事实上，英伟达的下游供应链和下游需求规模如此之大，他们才愿意在上游投资。
**English / Original**
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.
## paragraph 32
**中文**
看看 GTC，人们会惊叹于它的规模和到场人群。那里汇聚了完整的 360 度全景，整个 AI 宇宙尽在一处。大家齐聚一堂，是因为他们需要彼此看见。我把他们聚在一起，让下游看见上游、上游看见下游，也让所有人看见 AI 的进展。更重要的是，他们都能见到 AI 原生公司、正在创建的所有 AI 初创企业和正在发生的种种精彩进展，从而亲眼看到我所讲述的一切。我花很多时间，直接或间接地向供应链、合作伙伴和生态系统说明我们眼前的机遇。
**English / Original**
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.
## paragraph 33
**中文**
总有人说：“Jensen，大多数主题演讲都是一个发布接着一个发布。”而我们的主题演讲总有一部分让人略感煎熬，因为听起来几乎像在上课。事实上，这正是我的用意。我必须确保整个供应链的上下游和生态系统都明白：什么正向我们走来、为什么会来、何时到来、规模会有多大，并且能像我一样，对此进行系统性推演。
**English / Original**
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.
## paragraph 34
**中文**
至于你所说的护城河，我们有能力为未来而建设。如果未来几年我们的规模达到一万亿美元，我们已经具备支撑它的供应链。没有我们的覆盖范围和业务速度……正如有现金流，也有供应链流转，也有周转次数。如果某种架构的业务周转率很低，就不会有人为它建立供应链。我们之所以能维持这种规模，只因为下游需求极其庞大。而他们看得见、听得到，也看得见这一切正在到来。这使我们能够以如今的规模做成这些事。
**English / Original**
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.
## paragraph 35
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 36
**中文**
我确实想更具体地了解上游能否跟上。多年来，你们的营收一直同比翻番，向全世界提供的浮点运算能力每年更是增长到三倍以上。
**English / Original**
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.
## paragraph 37
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 38
**中文**
而如今在这样的规模上再翻一番，着实不可思议。
**English / Original**
And 2x-ing at this scale now is really incredible.
## paragraph 39
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 40
**中文**
没错。但再看逻辑芯片：你们是台积电 N3 制程的最大客户，也是 N2 制程最大的客户之一。今年整个 AI 行业将占 N3 产能的 60%；据 SemiAnalysis 预测，明年将达到 86%。既然你们已占多数，怎么还能翻番？又如何年复一年地翻番？我们现在是否进入了这样一种局面：AI 算力的增速必然因上游限制而放缓？你看到绕开这一限制的办法了吗？归根结底，我们要怎样才能每年多建一倍的晶圆厂？
**English / Original**
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?
## paragraph 41
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 42
**中文**
从某种意义上说，在任何一个时点，瞬时需求都超过全球上下游的供给。在任意时刻，限制我们的可能是水管工的数量——这种情况确实发生过。
**English / Original**
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.
## paragraph 43
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 44
**中文**
水管工已经获邀参加明年的 GTC 了。
**English / Original**
The plumbers are invited to next year’s GTC.
## paragraph 45
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 46
**中文**
顺便说一句，这主意真不错。但这其实是一种好局面：一个行业的瞬时需求高于全行业总供给，才是你希望看到的，反过来显然不那么好。如果供需相差太远，或某个特定部件的缺口太大，整个行业就会蜂拥而上解决它。例如，你会注意到人们如今已经不太谈 CoWoS 了。
**English / Original**
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.
## paragraph 47
**中文**
原因是过去两年，我们几乎倾尽全行业之力猛攻这个问题，产能一再翻倍、翻倍、再翻倍。现在我认为状况相当不错。台积电如今明白，CoWoS 的供应必须跟上其余逻辑芯片和存储芯片的需求。他们扩展 CoWoS 和未来封装技术的力度，与扩展逻辑制程的力度相当。这非常好，因为很长一段时间里，CoWoS 和 HBM 存储都颇为小众；但它们已不再小众。大家现在意识到，它们是主流计算技术。
**English / Original**
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.
## paragraph 48
**中文**
当然，我们现在更有能力影响供应链的更大范围。AI 革命刚开始时，我现在说的这些话，五年前就在说。有些人相信并作出了投资，比如 Sanjay 和美光团队。我至今清楚记得那次会议，我明确说明了将会发生什么、为什么会发生，也给出了我们今天正在见证的这些预测。他们真正在这件事上加倍下注。我们与他们在 LPDDR 和 HBM 存储上展开合作，他们也切实进行了投资。显然，这给公司带来了巨大回报。有些人来得稍晚一些，但现在大家都已到场。
**English / Original**
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.
## paragraph 49
**中文**
每一个这样的瓶颈都会得到高度关注。如今，我们会提前数年预判并锁定瓶颈。例如，过去几年我们与 Lumentum、Coherent 以及硅光子生态系统共同开展的投资，真正重塑了供应链。我们围绕台积电建立了完整的供应链，与他们合作开发 COUPE，发明了大量技术，并把这些专利授权给供应链，以保持它良好、开放。
**English / Original**
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.
## paragraph 50
**中文**
我们通过发明新技术、建立新工作流、开发双面探测等新测试设备、投资企业并帮助它们扩大产能，让供应链做好准备。可以看到，我们正在塑造生态系统，使供应链能够支撑这样的规模。
**English / Original**
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.
## paragraph 51
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 52
**中文**
有些瓶颈似乎比另一些更容易解决。扩大 CoWoS 产能，相较于扩大——
**English / Original**
It seems like some bottlenecks are easier than others. Scaling up CoWoS versus scaling up—
## paragraph 53
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 54
**中文**
顺便说一句，我刚才举的已经是最难的那个。
**English / Original**
I went to the hardest one, by the way.
## paragraph 55
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 56
**中文**
也就是？
**English / Original**
Which is?
## paragraph 57
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 58
**中文**
水管工。水管工和电工。这正是我对那些末日论者的一项担忧：他们描述工作的终结和岗位的消亡。如果我们劝阻人们去当软件工程师，软件工程师就会不够用。十年前也出现过同样的预测。一些末日论者当时对人们说：“无论做什么，都别当放射科医生。”网上或许还能找到那类视频，说放射科会成为最先消失的职业，世界将不再需要任何放射科医生。猜猜我们现在缺什么？放射科医生。
**English / Original**
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.
## paragraph 59
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 60
**中文**
回到刚才那一点：有些东西可以扩产，另一些却……你们实际上要怎样才能每年制造两倍的逻辑芯片？归根结底，存储和逻辑芯片都受制于 EUV。如何做到 EUV 光刻机数量每年翻番？
**English / Original**
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?
## paragraph 61
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 62
**中文**
这些东西没有一样不能迅速扩产，全部都很容易在两三年内做到。你只需要明确的需求信号。既然能造一台，就能造十台；既然能造十台，就能造一百万台。这些东西并不难复制。
**English / Original**
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.
## paragraph 63
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 64
**中文**
你会沿供应链追溯到多深？你会不会找到 ASML 说：“嘿，展望三年后，如果英伟达要实现每年两万亿美元营收，我们需要多得多的 EUV 光刻机”？
**English / Original**
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”?
## paragraph 65
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 66
**中文**
有些公司我必须直接沟通，有些可以间接沟通，还有些……只要我能说服台积电，ASML 自然也会信服。我们必须关注关键的卡点。但如果台积电信服了，几年后你就会有充足的 EUV 光刻机。
**English / Original**
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.
## paragraph 67
**中文**
我的意思是，所有瓶颈都不会持续超过两三年，无一例外。与此同时，我们正把计算效率提升 10 倍、20 倍；从 Hopper 到 Blackwell，更是提升 30 到 50 倍。CUDA 非常灵活，所以我们还能提出新算法、开发各种新技术，在扩充产能之外继续提高效率。这些都不令我担心。我担心的是我们下游的事情，例如阻碍能源供应的能源政策……没有能源，你就无法创造一个产业；没有能源，也无法建立一个全新的制造业。
**English / Original**
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.
## paragraph 68
**中文**
我们希望推动美国再工业化，希望让芯片制造、计算机制造和封装回流，还希望制造电动汽车和机器人等新事物，建设 AI 工厂。没有能源，这些一样也造不了，而能源建设需要很长时间。增加芯片产能，是两三年能解决的问题；增加 CoWoS 产能，也是两三年的问题。
**English / Original**
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.
## paragraph 69
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 70
**中文**
有意思。我有时会听到节目嘉宾提出完全相反的说法。在这件事上，我确实缺乏足够的技术知识来判断孰是孰非。
**English / Original**
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.
## paragraph 71
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 72
**中文**
好在你现在谈话的对象正是专家。
**English / Original**
The beautiful thing is you’re talking to the expert.
## paragraph 73
**中文**
00:16:25——TPU 会打破英伟达对 AI 算力的掌控吗？
**English / Original**
00:16:25 – Will TPUs break Nvidia’s hold on AI compute?
## paragraph 74
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 75
**中文**
确实。我想问问你对竞争对手的看法。看看 TPU，可以说全球排名前三的模型中有两个——Claude 和 Gemini——是在 TPU 上训练的。这对英伟达未来意味着什么？
**English / Original**
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?
## paragraph 76
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 77
**中文**
我们打造的是截然不同的东西。英伟达打造的是加速计算，而不是张量处理单元。加速计算用于各种任务：分子动力学、量子色动力学、数据处理、数据框、结构化数据和非结构化数据；也用于流体动力学和粒子物理。此外，我们还用它做 AI。
**English / Original**
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.
## paragraph 78
**中文**
加速计算的用途要多样得多。尽管如今大家谈论的是 AI，而且 AI 显然非常重要、影响深远，但计算的范畴远比它广。英伟达重塑了计算方式，推动行业从通用计算转向加速计算。我们的市场覆盖面远非任何 TPU 或 ASIC 所能企及。纵观我们的定位，英伟达是唯一能加速各种应用的公司。我们拥有庞大的生态系统，因此各种框架和算法都能在英伟达平台上运行。
**English / Original**
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.
## paragraph 79
**中文**
我们的计算机从设计之初就考虑了让其他人运营，所以任何运营商都可以买我们的系统。而大多数自研系统从未被设计成足够灵活、可供他人运营，因此你必须自己当运营商。正因为任何人都能运营我们的系统，我们进入了每一家云，包括 Google、Amazon、Azure 和 OCI。
**English / Original**
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.
## paragraph 80
**中文**
如果你想运营它并对外出租，最好在众多行业拥有庞大的客户生态来承接产能。如果你想自己运营，我们显然也能帮助你，就像我们为 Elon 的 xAI 所做的那样。由于我们能让任何公司、任何行业成为运营商，你还可以在礼来建一台用于科研和药物发现的超级计算机。我们能帮助他们运营自己的超算，并将其用于我们所加速的各种药物发现和生命科学任务。
**English / Original**
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.
## paragraph 81
**中文**
我们能处理的应用多得很，其中很多是 TPU 做不了的。英伟达打造的 CUDA 同样可以成为出色的张量处理平台，但它还能覆盖数据处理、计算、AI 等每个生命周期。我们的市场机会大得多，覆盖范围也广得多。如今我们支持全球所有应用，所以你可以在任何地方建设英伟达系统，并确信它会有客户。这是完全不同的事物。
**English / Original**
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.
## paragraph 82
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 83
**中文**
接下来这个问题会比较长。你们的营收非常亮眼，而每季度 600 亿美元的收入并不是来自制药和量子计算。你们能做到这一点，是因为 AI 是一项前所未有、且正以前所未有速度增长的技术。
**English / Original**
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.
## paragraph 84
**中文**
那么问题就是，究竟什么方案最适合 AI。我不了解所有细节，但我和从事 AI 研究的朋友聊过，他们说：“你看，我使用 TPU 时，它就是一个很大的脉动阵列，非常适合做矩阵乘法；GPU 则非常灵活，在大量分支或不规则内存访问的场景下很出色。”
**English / Original**
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.”
## paragraph 85
**中文**
可 AI 是什么？无非就是一遍又一遍、极其可预测的矩阵乘法。你不必为 warp 调度器或线程与存储体之间的切换牺牲任何裸片面积。TPU 确实针对眼下正在涌现、带动大部分营收和算力使用增长的任务做了优化。我想知道你如何回应这一点。
**English / Original**
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.
## paragraph 86
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 87
**中文**
矩阵乘法是 AI 的重要组成部分，但并非全部。如果你想提出新的注意力机制、用不同方式解耦，或干脆发明一种全新架构——比如混合 SSM——就会希望架构具备通用可编程性。如果你想创建一个融合扩散技术与自回归技术的模型，同样会希望架构具备通用可编程性。我们可以运行你能想到的一切，这就是优势。因为它是可编程系统，所以发明新算法要容易得多。
**English / Original**
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.
## paragraph 88
**中文**
发明新算法的能力，才真正让 AI 进步得如此之快。TPU 与其他任何技术一样，都受摩尔定律影响；我们知道，摩尔定律带来的提升约为每年 25%。真正实现 10 倍或 100 倍跃升的唯一办法，是每年从根本上改变算法及其计算方式。
**English / Original**
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.
## paragraph 89
**中文**
这就是英伟达的根本优势。我们能让 Blackwell 相比 Hopper 提升 50 倍，唯一的原因就是这一点……我最初宣布 Blackwell 的能效将是 Hopper 的 35 倍时，没人相信。后来 Dylan 写文章说我故意给了保守数字，实际是 50 倍。单靠摩尔定律不可能合理实现这一点。我们的解决方式，是采用 MoE 等新模型，把它们并行化、解耦并分布到整个计算系统。若不能真正深入底层、用 CUDA 编写新内核，这件事就很难做到。
**English / Original**
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.
## paragraph 90
**中文**
这来自我们架构的可编程性，也来自英伟达是一家实行极致协同设计的公司。我们甚至能把部分计算卸载到 NVLink 这样的互连结构中，或通过 Spectrum-X 卸载到网络中。我们可以同时改变处理器、系统、互连、软件库和算法。没有 CUDA，我甚至不知道该从哪里着手。
**English / Original**
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.
## paragraph 91
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 92
**中文**
这引出了一个有关英伟达客户群的有趣问题。你们 60% 的营收来自五家大型超大规模云服务商。换一个时代、面对不同客户——比如运行实验的教授——他们确实需要 CUDA，没法用别的加速器；他们只需要用 CUDA 跑 PyTorch，并让所有东西都得到优化。
**English / Original**
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.
## paragraph 93
**中文**
但这些超大规模云服务商有资源编写自己的内核。事实上，为了针对自身特定架构榨出最后 5% 的性能，他们必须这么做。Anthropic 和 Google 大多在运行自研加速器，或使用 TPU 和 Trainium。即便使用 GPU 的 OpenAI 也有 Triton，因为他们需要自己的内核。他们一路深入到 CUDA C++，不使用 cuBLAS 和 NCCL，而是拥有自己的一套技术栈，这套栈也能编译到其他加速器上。如果你们的大多数客户都能而且确实会构建 CUDA 的替代方案，那么 CUDA 在多大程度上仍会是让前沿 AI 选择英伟达的关键？
**English / Original**
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?
## paragraph 94
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 95
**中文**
CUDA 是一个丰富的生态系统。如果你想优先面向某种计算机进行开发，先在 CUDA 上构建是极其明智的选择。因为生态非常丰富，我们支持所有框架。如果你想创建自定义内核……例如，我们为 Triton 作出了大量贡献，所以 Triton 的后端包含大量英伟达技术。
**English / Original**
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.
## paragraph 96
**中文**
我们很乐意帮助每个框架做到最好。框架非常多，有 Triton、vLLM、SGLang 等。现在又涌现出一大批新的强化学习框架，例如 verl 和 NeMo RL。随着后训练和强化学习的发展，这整个领域正在爆发。因此，如果你要基于一种架构构建系统，选择 CUDA 最为合理，因为你知道它的生态十分出色。
**English / Original**
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.
## paragraph 97
**中文**
你知道，若出了问题，更可能是你自己的代码，而不是底下那座代码大山。别忘了，构建这些系统时要面对多么庞大的代码量。东西不能运行时，问题出在你，还是出在计算机？你会希望问题永远出在自己身上，这样就能信任计算机。显然，我们自己仍有很多缺陷，但我们的系统已经历极其充分的锤炼，至少让你能在这个基础上继续构建。这是第一点：生态系统的丰富性、可编程性和能力。
**English / Original**
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.
## paragraph 98
**中文**
第二，如果你是一名开发者，无论在构建什么，最想要的单一要素都是装机基础。你希望自己编写的软件能在许多其他计算机上运行。你不是只为自己开发软件；无论你是为自己的机群还是别人的机群开发，作为框架开发者，你的目标都不止自己。英伟达的 CUDA 生态归根结底是它的巨大财富。
**English / Original**
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.
## paragraph 99
**中文**
现在外面已有数亿块我们的 GPU，每一家云都有。从 A10、A100、H100、H200，到 L 系列、P 系列，型号非常多，尺寸和形态也多种多样。如果你是一家机器人公司，就会希望那套 CUDA 技术栈也能直接在机器人本体上运行。我们确实无处不在。装机基础意味着，你一旦开发出软件或模型，它便能在各处发挥作用。这具有难以置信的价值。
**English / Original**
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.
## paragraph 100
**中文**
最后，我们进入了每一家云，这一点让我们真正独一无二。如果你是一家 AI 公司或一名开发者，未必已经确定要与哪家云服务商合作，也未必确定希望在哪里运行。我们无处不在；如果你愿意，也可以在你的本地环境中运行。生态的丰富、装机基础的广泛，再加上部署位置的多样性，共同让 CUDA 具有无可替代的价值。
**English / Original**
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.
## paragraph 101
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 102
**中文**
这很有道理。我好奇的是，这些优势对你们的主要客户究竟有多重要。它们可能对很多人很重要，但真正有能力自建软件栈的那类客户，却贡献了你们大部分营收。尤其是假如未来 AI 特别擅长处理那些具备严密验证闭环、可以通过强化学习训练的任务……比如，如何编写一个内核，让注意力或 MLP 在纵向扩展系统中达到最高效率？这类反馈闭环非常容易验证。
**English / Original**
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.
## paragraph 103
**中文**
所有超大规模云服务商都能为自己编写这些定制内核吗？英伟达的性价比仍然很高，因此它们或许仍会偏爱英伟达。但问题就变成了：竞争是否只取决于在同样一美元下，谁能提供最佳规格、最多浮点运算能力和最高内存带宽？而在历史上，凭借 CUDA 护城河，英伟达在整个 AI 行业的软硬件领域一直拥有、现在也仍拥有最高的利润率，超过 70%。那么问题是：如果大多数客户都有能力自己构建技术栈，而不再依赖 CUDA 护城河，你们还能维持这种利润率吗？
**English / Original**
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?
## paragraph 104
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 105
**中文**
我们派驻这些 AI 实验室、与他们合作并优化其技术栈的工程师多得惊人。原因是没有人比我们更了解自己的架构。这些架构不像 CPU 那么通用。CPU 有点像凯迪拉克，是一辆舒适的巡航车，速度从不会太快，人人都能开得不错；它有定速巡航，一切都很轻松。但从很多方面看，英伟达的 GPU 和加速器更像 F1 赛车。我可以想象，人人都能把它开到每小时一百英里，但要把它推到极限，需要相当深的专业知识。我们大量使用 AI 来创建自己的内核。
**English / Original**
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.
## paragraph 106
**中文**
我相当确信，在未来很长一段时间里，大家仍然会需要我们。我们的专业能力往往能轻松帮助 AI 实验室合作伙伴把技术栈性能再提升一倍。等我们优化完他们的技术栈或某个特定内核，模型提速三倍、两倍或 50% 都很常见。这是极其可观的数字，尤其考虑到他们机群的装机规模，以及拥有的全部 Hopper 和 Blackwell。性能翻倍，就等于收入翻倍，会直接转化为营收。
**English / Original**
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.
## paragraph 107
**中文**
英伟达计算技术栈的单位总拥有成本（TCO）性能举世最佳，没有例外。世界上今天没有任何一家公司能向我证明，某个平台的性能与 TCO 之比优于我们，一家也没有。其实公开基准就在那里，Dylan 的 InferenceMAX 人人都能用，却没有一家……TPU 不来，Trainium 也不来。
**English / Original**
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.
## paragraph 108
**中文**
我鼓励他们使用 InferenceMAX，展示其号称惊人的推理成本优势。可这很难，没人愿意参赛。还有 MLPerf。我非常欢迎 Trainium 来证明他们总在宣称的 40% 优势，也很想听他们展示 TPU 的成本优势。在我看来，这根本说不通，完全说不通，从第一性原理出发也说不通。
**English / Original**
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.
## paragraph 109
**中文**
所以，我认为我们如此成功，原因很简单：我们的 TCO 极具优势。第二，你说我们 60% 的客户是前五大公司，但这部分业务大多服务于外部客户。例如，AWS 上的大多数英伟达算力是供外部客户使用，而非 AWS 内部使用。Azure 上显然也都是外部客户，OCI 上同样全是外部客户，不是内部用途。他们青睐我们，是因为我们的触达范围极广。我们能为他们带来全世界所有优秀客户，因为这些客户都基于英伟达构建。而这些公司选择英伟达，正因为我们的覆盖面和通用性极强。
**English / Original**
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.
## paragraph 110
**中文**
因此，我认为这个飞轮真正由装机基础、架构的可编程性、生态的丰富性，以及全球数量众多的 AI 公司共同驱动。现在已有数万家 AI 公司。如果你是其中一家 AI 初创企业，会选择什么架构？你会选供应最充足的，而我们是全球供应最充足的；你会选装机基础最大的，而我们的装机基础最大；你还会选生态丰富的。
**English / Original**
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.
## paragraph 111
**中文**
这就是飞轮。它之所以运转，是因为几项因素相互结合：第一，我们每美元性能非常出色，使客户获得最低成本的词元；第二，我们的每瓦性能全球最高。如果合作伙伴建设一座一吉瓦的数据中心，它最好能产生最大收入和最多词元，而词元会直接转化为收入。你会希望它尽可能多地生成词元，把数据中心收入最大化。我们是全球每瓦生成词元最多的架构。最后，如果你的目标是出租基础设施，我们拥有全球最多的客户。这就是飞轮能够运转的原因。
**English / Original**
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.
## paragraph 112
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 113
**中文**
有意思。我想问题归结为：这里真实的市场结构究竟怎样？即使存在其他公司……原本可能出现一种格局，让数万家 AI 公司各自占有大致相等的算力份额。但即使通过这五家超大规模云服务商，真正在 Amazon 上使用算力的也是 Anthropic、OpenAI 和这些大型基础模型实验室；它们本身有财力也有能力让不同加速器工作。
**English / Original**
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.
## paragraph 114
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 115
**中文**
不，我认为你的前提错了。
**English / Original**
No, I think your premise is wrong.
## paragraph 116
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 117
**中文**
也许吧。但让我换个角度问。
**English / Original**
Maybe. But let me ask you a slightly different question.
## paragraph 118
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 119
**中文**
待会儿一定要绕回来，让我纠正你的前提。
**English / Original**
Come back and make me correct your premise.
## paragraph 120
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 121
**中文**
好吧。让我先问另一个问题。
**English / Original**
Okay. Let me just ask you a different question.
## paragraph 122
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 123
**中文**
但一定记得让我回头纠正它，因为这对 AI 太重要了，对科学的未来太重要了，对行业的未来也太重要了。那个前提……你看——
**English / Original**
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 —
## paragraph 124
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 125
**中文**
让我先把问题问完，然后我们一起讨论它。
**English / Original**
Let me just finish the question and then we can address it together.
## paragraph 126
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 127
**中文**
好。
**English / Original**
Yeah.
## paragraph 128
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 129
**中文**
如果价格、性能、每瓦性能等说法都是真的，那你认为为什么会出现这种情况：比如 Anthropic 几天前刚宣布，与 Broadcom 和 Google 达成了数吉瓦规模的 TPU 合作，而其大部分算力也将来自那里？
**English / Original**
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?
## paragraph 130
**中文**
对 Google 而言，TPU 显然占其算力的大多数。所以观察这些大型 AI 公司时，似乎它们的很大一部分算力……曾几何时全部来自英伟达，现在却不是了。我很好奇，既然纸面上的这些优势都成立，它们为什么还会采用其他加速器？这两者该如何解释？
**English / Original**
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?
## paragraph 131
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 132
**中文**
Anthropic 是一个特殊个案，不是一种趋势。没有 Anthropic，TPU 凭什么增长？TPU 的增长 100% 来自 Anthropic。没有 Anthropic，Trainium 又凭什么增长？其增长也 100% 来自 Anthropic。我认为这一点相当广为人知，也为大家所理解。并不存在大量 ASIC 机会，因为 Anthropic 只有一家。
**English / Original**
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.
## paragraph 133
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 134
**中文**
但 OpenAI 和 AMD 达成了合作……而且他们正在打造自己的 Titan 加速器。
**English / Original**
But OpenAI’s deals with AMD… They’re building their own Titan accelerator.
## paragraph 135
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 136
**中文**
是，但我想大家都承认，他们绝大多数算力仍是英伟达。我们还会继续开展大量合作。别人使用其他产品、尝试不同方案，我并不介意。若不试试别的，他们又怎么知道我们的东西有多好？有时确实需要重新体会一下。我们必须持续赢得自己今天的地位。
**English / Original**
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.
## paragraph 137
**中文**
总会有各种宏大宣称。看看已经取消了多少 ASIC 项目。仅仅因为你打算打造一款 ASIC……你仍然必须做出比英伟达更好的产品。打造比英伟达更好的产品没那么容易。事实上，这并不合算；除非英伟达严重漏掉了什么。凭借我们的规模和速度，我们是全球唯一一家每年都能推出新一代产品的公司，而且每年都有巨大飞跃。
**English / Original**
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.
## paragraph 138
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 139
**中文**
我猜他们的逻辑是：“嘿，它不需要更好，只要别差到超过 70% 就行”，因为他们向你们支付的利润率是 70%。
**English / Original**
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.
## paragraph 140
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 141
**中文**
不，别忘了，即使 ASIC 的利润率也相当高。假设英伟达的利润率是 70%，ASIC 的利润率也有 65%。你究竟省下了多少？
**English / Original**
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?
## paragraph 142
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 143
**中文**
哦，你是说从 Broadcom 之类的公司采购？
**English / Original**
Oh, you mean from Broadcom or something like that?
## paragraph 144
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 145
**中文**
对，当然。你总得付钱给某个人。据我所知，ASIC 的利润率高得惊人。他们自己也这么认为，而且对其惊人的 ASIC 利润率颇为自豪。
**English / Original**
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.
## paragraph 146
**中文**
你问为什么。很久以前，我们根本没有能力这么做。当时我没有深刻认识到，建立 OpenAI、Anthropic 这样的基础 AI 实验室会有多困难，也没有意识到它们需要供应商亲自提供巨额投资。我们当时没有条件向 Anthropic 投入数十亿美元，让他们使用我们的算力；但 Google 和 AWS 有。他们早期投入巨资，换取 Anthropic 使用其算力。我们当时确实没有能力这样做。
**English / Original**
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.
## paragraph 147
**中文**
要说我的错误，就是没有深刻认识到他们真的别无选择：风险投资机构绝不会抱着它最终可能成为 Anthropic 的希望，向一家 AI 实验室投入 50 亿到 100 亿美元。这是我的疏漏。但即便当时理解了，我也不认为我们有能力这么做。不过，我不会再犯同样的错误。
**English / Original**
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.
## paragraph 148
**中文**
我很高兴投资 OpenAI，也很高兴帮助他们扩大规模，而且我认为这样做至关重要。后来，在我有能力之后，Anthropic 找到我们时，我也很高兴成为投资者，并帮助他们扩张。只是当时我们做不到。如果一切可以重来——如果那时的英伟达就能像今天这样强大——我会非常乐意这么做。
**English / Original**
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.
## paragraph 149
**中文**
00:41:06——英伟达为什么不成为超大规模云服务商？
**English / Original**
00:41:06 – Why doesn’t Nvidia become a hyperscaler?
## paragraph 150
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 151
**中文**
这其实很有意思。多年来，英伟达一直是 AI 领域真正赚钱、而且赚很多钱的公司。现在你们正把这些钱投出去。据报道，你们对 OpenAI 的投资最高达 300 亿美元，对 Anthropic 达 100 亿美元。而现在它们的估值已经上涨，我相信还会继续上涨。
**English / Original**
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.
## paragraph 152
**中文**
所以，这么多年来，你们一直向它们提供算力，也看到了趋势所向；就在几年前——有时甚至只是一年前——它们的估值还只有现在的十分之一，而你们手里又有这么多现金。原本可能出现另一种世界：要么英伟达自己成为一家基础模型实验室，投入巨资实现这一点；要么把你们如今按当前估值达成的投资交易，早得多就完成。你们当时也有现金。所以我确实很好奇，为什么没有更早行动？
**English / Original**
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?
## paragraph 153
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 154
**中文**
我们一有能力就做了，确实是在具备条件的第一时间做的。如果可以，我还会更早行动。在 Anthropic 需要我们出手时，我们确实没有条件这么做，那也不符合我们当时的思维方式。
**English / Original**
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.
## paragraph 155
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 156
**中文**
为什么？是现金方面的问题吗？
**English / Original**
How so? Was it like a cash thing?
## paragraph 157
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 158
**中文**
对，是投资规模的问题。我们当时从未对公司外部进行投资，更别说那么大的金额。我们没意识到有这个必要。我一直以为，看在老天的份上，他们完全可以像所有公司一样去找风投融资。但他们想做的事，不可能靠风险投资完成；OpenAI 想做的事也不可能靠风险投资完成。我现在明白了，当时却不知道。
**English / Original**
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.
## paragraph 159
**中文**
但这正是他们的天才之处，也是他们聪明的原因。他们那时就意识到必须采取那样的办法。我很高兴他们做到了。即便是因为我们无法出手，才迫使 Anthropic 转而寻找别人，我仍然庆幸这件事发生了。Anthropic 的存在对世界大有裨益，我为此感到高兴。
**English / Original**
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.
## paragraph 160
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 161
**中文**
不过，你们仍在赚很多钱，而且一个季度比一个季度赚得更多。
**English / Original**
I guess you still are making a ton of money, and you’re making way more money quarter after quarter.
## paragraph 162
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 163
**中文**
即使如此，有遗憾也没关系。
**English / Original**
It’s still okay to have regrets.
## paragraph 164
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 165
**中文**
所以问题依然存在。好，现在已经走到这一步，而且你们还在源源不断赚这么多钱，英伟达应该拿这些钱做什么？一种答案是：为了让这些实验室租用算力，已经出现了一个庞大的中间商生态，专门把资本支出转化为运营支出。芯片确实很贵，但随着 AI 模型不断进步，它们在整个寿命周期里能赚很多钱；它们创造的价值，也就是词元，正在增长，只是初始建设成本很高。英伟达有钱承担资本支出。事实上，据报道，你们为 CoreWeave 提供了最高 63 亿美元的兜底承诺，并已投资 20 亿美元。
**English / Original**
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.
## paragraph 166
**中文**
英伟达为什么不亲自做云？为什么不自己成为超大规模云服务商，把这些算力出租出去？你们有这么多现金，完全能做。
**English / Original**
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.
## paragraph 167
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 168
**中文**
这是公司的经营哲学，我认为它很明智：做一切必要之事，同时尽可能少做。意思是，对于我们打造计算平台所做的工作，我真心相信，如果我们不做，就不会有人做。如果我们不承担那些风险——不以这样的方式打造 NVLink，不构建完整技术栈，不以这种方式创建生态系统，不在大部分时间都亏钱的情况下仍为 CUDA 奉献二十年——那就不会有别人去做。
**English / Original**
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.
## paragraph 169
**中文**
如果我们不创建所有具备领域专用性的 CUDA-X 库……十五年前，我们开始大力开发领域专用库，因为意识到，如果我们不为光线追踪、图像生成乃至早期 AI 工作创建这些专用库，不为数据处理、结构化数据处理或向量数据处理创建这些模型和库，就不会有人来做。我对此完全确信。我们还为计算光刻创建了名为 cuLitho 的库；如果我们不做，也不会有别人做。所以，若没有我们所做的一切，加速计算不可能像今天这样发展。
**English / Original**
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.
## paragraph 170
**中文**
因此，这些事我们应该做，应该让全公司倾尽全力、全心投入。但世界上已经有许多云；如果我不做云，总会有人出现。所以，我们遵循“做一切必要之事，同时尽可能少做”的原则——尽可能少做。这一哲学如今存在于公司内部，我做任何事都会透过这层视角来判断。
**English / Original**
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.
## paragraph 171
**中文**
在云服务方面，如果我们没有支持 CoreWeave 生存下来，这些新型云和 AI 云就不会存在。没有我们的帮助，CoreWeave 不会存在；没有我们的支持，Nscale 不会有今天的成就；Nebius 也不会成为今天的样子。如今它们都发展得非常出色。
**English / Original**
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.
## paragraph 172
**中文**
这算是一种商业模式吗？[听不清] 我们应该做一切必要之事，同时尽可能少做。因此，我们投资自己的生态系统，因为我希望它繁荣。我希望我们的架构和 AI 能连接尽可能多的行业、尽可能多的国家，让整个地球都能基于 AI、基于美国技术栈来建设。这正是我们正在追求的愿景。
**English / Original**
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.
## paragraph 173
**中文**
你刚才还提到一点……优秀、惊人的基础模型公司如此之多，我们努力投资所有这些公司。这也是我们的做法：不挑选赢家，而是需要支持每一家。这样做本身让我们快乐，对业务也至关重要。我们还会特意避免押注某个赢家。因此，只要投资其中一家，我就会投资所有家。
**English / Original**
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.
## paragraph 174
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 175
**中文**
你为什么要特意避免挑选赢家？
**English / Original**
Why do you go out of your way not to pick winners?
## paragraph 176
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 177
**中文**
第一，因为那不是我们的职责。第二，英伟达刚起步时，市场上有 60 家 3D 图形公司，只有我们一家活了下来。如果当时把这 60 家公司摆在面前，让你判断谁能成功，英伟达会高居“最不可能成功”的名单榜首。
**English / Original**
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.
## paragraph 178
**中文**
那是你出生前很久的事，但英伟达当时的图形架构恰恰是错的，不是略有偏差，而是精准地错了。我们创建了一种开发者根本不可能支持的架构，注定无法成功。我们从合理的第一性原理出发进行推演，却得出了错误答案。所有人都会把我们排除在外，但我们今天仍站在这里。
**English / Original**
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.
## paragraph 179
**中文**
因此，我有足够的谦逊来认识这一点：不要挑选赢家。要么让他们全都自力更生，要么就照顾所有人。
**English / Original**
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.
## paragraph 180
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 181
**中文**
有一点我没明白。你说：“我们不会仅仅因为这些公司是新型云、因为想扶持它们，就给予优先待遇。”但你又列举了许多新型云，并说若没有英伟达，它们根本不会存在。这两种说法如何相容？
**English / Original**
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?
## paragraph 182
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 183
**中文**
首先，它们自己必须想要存在，并前来向我们求助。当它们确实想做这件事，拥有商业计划、专业能力和热情时……显然，它们本身也必须具备一定能力。但如果到最后，项目要启动确实还需要一些投资，我们会伸出援手。不过，它们越早让自己的飞轮转起来越好……
**English / Original**
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...
## paragraph 184
**中文**
你的问题是：“我们想进入融资业务吗？”答案是不想。市场上有人专门从事融资业务；与其自己成为出资方，我们更愿意与所有融资机构合作。我们的目标是专注于自己擅长的事，尽可能保持商业模式简单，并支持生态系统。
**English / Original**
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.
## paragraph 185
**中文**
当 OpenAI 这样的公司因为尚未上市，需要 300 亿美元规模的投资，而我们又深信它——我也深信它将成为一家……其实它今天已经是一家非凡的公司，将来会更了不起。世界需要它存在，世界希望它存在，我也希望它存在。它正处于顺风之中，那就支持它、让它扩大规模。我们会做这类投资，因为他们需要我们这样做。但我们的目标不是尽可能多做，而是尽可能少做。
**English / Original**
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.
## paragraph 186
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 187
**中文**
这个问题也许很显然，但多年来我们一直处于 GPU 短缺的局面；现在模型越来越好，短缺也随之加剧。
**English / Original**
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.
## paragraph 188
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 189
**中文**
我们的 GPU 确实供不应求。
**English / Original**
We have a shortage of GPUs.
## paragraph 190
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 191
**中文**
是。众所周知，英伟达分配稀缺供货时，不只是价高者得，而会考虑：“嘿，我们要确保这些新型云能生存。给 CoreWeave 一些，给 Crusoe 一些，再给 Lambda 一些。”这对英伟达有什么好处？首先，你同意这种主动拆分市场的描述吗？
**English / Original**
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?
## paragraph 192
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 193
**中文**
不，不。你的前提就是错的。我们对这些事情足够谨慎，也非常上心。首先，如果你不下采购订单，说再多也没用。我们收到采购订单之前，能做什么？所以第一件事，是努力与每一家客户共同完成需求预测，因为这些产品需要很长时间才能制造，数据中心也需要很长时间才能建成。我们通过预测来协调需求、供应等各方面，好吗？这是头等工作。
**English / Original**
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.
## paragraph 194
**中文**
第二，我们努力和尽可能多的客户一起做预测，但归根结底，你仍然必须下单。也许你出于某种原因没有下订单，那我能怎么办？到了某个节点，就按先入先出处理。但除此之外，如果你还没准备好——比如数据中心尚未就绪，或支持数据中心上线的某些部件没准备好——我们可能决定先服务另一位客户。这只是为了最大化我们自家工厂的吞吐量；我们可能会因此做些调整。
**English / Original**
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.
## paragraph 195
**中文**
除此之外，优先次序就是先入先出。你必须下采购订单；如果不下单……当然，外面有一些传闻。例如，这类说法都源于一篇文章，声称 Larry 和 Elon 与我共进晚餐时恳求我给他们 GPU。这件事从未发生。我们确实共进了晚餐，而且那顿饭非常愉快；但他们从未在任何时刻恳求 GPU。他们只需要下单。订单一下，我们就会尽最大努力把产能交付给他们。我们的做法并不复杂。
**English / Original**
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.
## paragraph 196
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 197
**中文**
好。听起来是有一条队列，然后根据数据中心是否就绪，以及何时提交采购订单，在特定时间拿到货。但听起来仍不是单纯价高者得。这样做有什么原因吗……？
**English / Original**
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…?
## paragraph 198
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 199
**中文**
我们从来不那么做。
**English / Original**
We never do that.
## paragraph 200
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 201
**中文**
好。
**English / Original**
Okay.
## paragraph 202
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 203
**中文**
从来不会。
**English / Original**
We never do.
## paragraph 204
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 205
**中文**
为什么不干脆价高者得？
**English / Original**
Why not just do high bidder?
## paragraph 206
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 207
**中文**
因为那是不良的商业惯例。你定好价格，然后由客户决定买不买。我知道芯片行业里有些公司会在需求升高时涨价，但我们不会；那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企业，成为行业的基石，让客户不必反复猜测。我报给你什么价格，就是什么价格，仅此而已。即使需求冲上天，也就随它去。
**English / Original**
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.
## paragraph 208
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 209
**中文**
从供应链另一端看，这也是你们能与台积电保持高效合作关系的原因，对吧？
**English / Original**
On the other end, that’s why you have a productive relationship with TSMC, right?
## paragraph 210
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 211
**中文**
对，英伟达与他们合作大概快 30 年了。英伟达和台积电之间没有法律合同，总是凭一种大致的公平来相处。有时我判断对，有时判断错；有时拿到更好的交易，有时条件差一些。但总体而言，这段关系非常了不起。我可以完全信任他们，也可以完全依赖他们。
**English / Original**
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.
## paragraph 212
**中文**
英伟达有一点值得信赖：今年的 Vera Rubin 会非常惊艳；明年 Vera Rubin Ultra 会到来；后年 Feynman 会到来；再下一年的产品名我还没公布。你每一年都可以依靠我们。你得去世界上另找一支 ASIC 团队——随便挑一支——然后问自己能否说：“我敢押上全部家当、押上整个企业，相信你年年都会为我交付；我的词元成本每年都会下降一个数量级，而且这像时钟一样可靠。”
**English / Original**
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.”
## paragraph 213
**中文**
我刚才对台积电说了一番话。历史上没有哪家其他晶圆代工厂能让你这样评价；而今天，你也能这样评价英伟达，年年都可以依赖我们。如果你想买价值 10 亿美元的 AI 工厂算力，没问题；想买 1 亿美元，没问题；想买 1000 万美元，或只买一个机架，也没问题；哪怕只买一张显卡，也可以，没问题。你若想下单建设价值 1000 亿美元的 AI 工厂，同样没问题。今天全球只有我们一家能做到这一点。
**English / Original**
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.
## paragraph 214
**中文**
我也可以这样评价台积电。想买一个，或买十亿个，都没问题；我们只需完成规划流程，以及成熟企业该做的一切。我认为，英伟达成为全球 AI 行业基石的能力，是我们花费二十多年才赢得的地位，背后是巨大的承诺与投入。公司的稳定性和一贯性至关重要。
**English / Original**
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.
## paragraph 215
**中文**
00:57:36——我们应该向中国出售 AI 芯片吗？
**English / Original**
00:57:36 – Should we be selling AI chips to China?
## paragraph 216
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 217
**中文**
好。我想问问中国。我其实不确定向中国出售芯片究竟是好是坏，但我喜欢站在嘉宾观点的对立面唱反调。支持出口管制的 Dario 来做客时，我问他：为什么美国和中国不能都在数据中心里拥有一个“天才之国”？既然你站在相反一边，我就反过来问你。
**English / Original**
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.
## paragraph 218
**中文**
一种思考方式是：Anthropic 几天前刚公布 Mythos Preview。他们甚至不公开发布 Mythos 模型，因为据称它的网络攻击能力太强；在确保相关零日漏洞得到修补前，他们认为世界尚未准备好。但他们说，这个模型在所有主要操作系统和浏览器中发现了数千个高危漏洞。它还在 OpenBSD 中发现了一个漏洞——这个操作系统是专门以杜绝零日漏洞为目标设计的，而该漏洞已存在 27 年。
**English / Original**
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.
## paragraph 219
**中文**
所以，如果中国公司、中国实验室和中国政府能获得 AI 芯片，用来训练具备这类网络攻击能力的 Claude Mythos，并凭借更多算力运行数百万个实例，那么问题是：这会不会威胁美国公司和美国国家安全？
**English / Original**
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?
## paragraph 220
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 221
**中文**
首先，Mythos 是用相当普通的算力类型和相当普通的算力规模训练出来的，只不过训练它的是一家非凡公司。中国完全不缺这种规模和类型的算力。所以首先必须认识到，中国是有芯片的。
**English / Original**
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.
## paragraph 222
**中文**
他们制造全球 60% 的主流芯片，或许还更多；这是他们规模极大的产业。他们拥有世界上一些最优秀的计算机科学家。你也知道，这些 AI 实验室里的 AI 研究人员很多都是华人；全球 50% 的 AI 研究人员都在那里。所以，考虑到他们已有的一切资产——能源充裕、芯片充足、拥有最多的 AI 研究人员——如果你担心他们，那么创造一个安全世界的最佳办法是什么？
**English / Original**
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?
## paragraph 223
**中文**
把他们当作受害对象、逼成敌人，可能不是最佳答案。他们是竞争对手，我们希望美国获胜。但我认为，开展对话和研究交流或许才是最安全的做法。由于当前把中国视为对手的态度，这种交流明显缺失。我们的 AI 研究人员与他们的 AI 研究人员保持沟通至关重要；双方努力就 AI 绝不能用于哪些事情达成共识，也至关重要。
**English / Original**
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.
## paragraph 224
**中文**
至于发现软件缺陷，这当然就是 AI 应该做的事。它会在大量软件中找到缺陷吗？当然会，软件里有数不清的缺陷，AI 软件本身也有很多缺陷。这正是 AI 的用途；AI 已发展到能帮助我们大幅提高生产力的程度，我对此很高兴。
**English / Original**
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.
## paragraph 225
**中文**
有一点常被低估，就是网络安全、AI 网络安全、AI 安全、AI 隐私和 AI 可靠性周边生态的丰富程度。一整个 AI 初创企业生态正在努力为我们创造这样的未来：一个能力惊人的 AI 智能体，周围有数千个 AI 智能体守护它的安全与稳健。这样的未来一定会到来。
**English / Original**
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.
## paragraph 226
**中文**
让一个 AI 智能体四处行动，却无人看管，这种想法近乎疯狂。我们非常清楚，这个生态系统必须繁荣。而事实证明，它需要开源软件、开放模型和开放技术栈，才能让所有 AI 研究人员和优秀计算机科学家构建同样强大的 AI 系统，保障 AI 安全。因此，我们必须确保开源生态保持活力，不能忽视这一点。其中许多成果来自中国，我们不应该扼杀它。
**English / Original**
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.
## paragraph 227
**中文**
谈到中国，我们当然希望美国拥有尽可能多的算力。我们受能源限制，但已经有很多人在解决这个问题；不能让能源成为国家的瓶颈。同时，我们也希望全世界所有 AI 开发者都在美国技术栈上开发，并让 AI 的贡献和进步——尤其是开源成果——惠及美国生态。若人为造出两个生态：开源生态只能运行在外国技术栈上，而运行在美国技术栈上的却是封闭生态，那将愚蠢至极。我认为这会给美国带来极其糟糕的结果。
**English / Original**
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.
## paragraph 228
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 229
**中文**
你谈到了很多方面，我来梳理一下回应。回到网络攻击所需浮点算力的差距，担忧在于：没错，他们有算力，但有些估算认为，由于只能使用 7 纳米制程——受芯片制造出口管制影响，他们没有 EUV——他们实际能提供的浮点算力只有美国的十分之一。
**English / Original**
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.
## paragraph 230
**中文**
因此，他们最终能训练出 Mythos 这样的模型吗？能。但问题在于，美国实验室算力更多，所以能率先达到这种能力水平。正因为 Anthropic 先实现了，它可以说：“好，我们先保留一个月，向这些美国公司开放，让它们修补所有漏洞，之后再发布。”
**English / Original**
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.”
## paragraph 231
**中文**
此外，即使他们训练出这样的模型，大规模部署能力也很重要。如果你有一个网络黑客，拥有一百万个这样的实例显然比只有一千个危险得多。因此，推理算力非常关键。事实上，他们拥有这么多优秀 AI 研究人员，恰恰让问题更令人担忧；而什么能提高这些工程师和研究人员的生产力？就是算力。
**English / Original**
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.
## paragraph 232
**中文**
你去问美国任何一家 AI 实验室，他们都会说瓶颈是算力。DeepSeek 创始人、Qwen 负责人或其他人也有类似表态，说自己的瓶颈就是算力。那么问题是：既然美国公司算力更多，难道不该让它们先达到 Mythos 级能力，让我们的社会先做好准备，然后中国才因算力较少而随后达到吗？
**English / Original**
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?
## paragraph 233
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 234
**中文**
我们应当永远率先达到，也应当永远拥有更多。但要让你描述的结果成立，就必须把条件推到极端：他们必须完全没有算力。只要还有一定算力，问题就变成究竟需要多少。
**English / Original**
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?
## paragraph 235
**中文**
中国拥有的算力规模非常庞大。你谈论的是全球第二大计算市场；如果他们想汇集算力，能汇集的算力多得很。
**English / Original**
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.
## paragraph 236
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 237
**中文**
但真是这样吗？有人做过估算，认为“中芯国际的制程节点其实落后”。
**English / Original**
But is that true? People do these estimates and they’re like, “SMIC is actually behind on the process nodes.”
## paragraph 238
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 239
**中文**
我正要告诉你答案。
**English / Original**
I’m about to tell you.
## paragraph 240
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 241
**中文**
好。
**English / Original**
Okay.
## paragraph 242
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 243
**中文**
他们拥有的能源多得惊人，不是吗？AI 是一个并行计算问题，对吧？既然能源近乎免费，为什么不能把四倍、十倍的芯片连接起来？他们能源太多了，甚至有完全空置却电力设施齐全的数据中心。你知道他们有“鬼城”，同样也有“幽灵数据中心”。他们拥有如此庞大的基础设施容量；只要愿意，即使芯片是 7 纳米的，也可以把更多芯片集群起来。
**English / Original**
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.
## paragraph 244
**中文**
他们的芯片制造能力位居世界前列。半导体行业都知道，他们在主流芯片领域占据垄断地位，产能过剩，甚至产能太多。因此，说中国无法拥有 AI 芯片完全是无稽之谈。
**English / Original**
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.
## paragraph 245
**中文**
当然，如果你问：假如全世界完全没有算力，美国是否会领先得更多？但那根本不可能发生，也不是现实情景。他们已经拥有充足算力；对于你担忧的风险来说，他们早已达到甚至超过所需门槛。
**English / Original**
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.
## paragraph 246
**中文**
所以，我认为你没有理解 AI 是一块五层蛋糕，最底层是能源。能源充裕可以弥补芯片不足，芯片充裕也可以弥补能源不足。例如，美国能源稀缺，这正是英伟达必须不断推进架构、实行这种极致协同设计的原因。我们出货的芯片数量有限——因为能源供应非常有限——因此必须让每瓦吞吐量高得惊人。
**English / Original**
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.
## paragraph 247
**中文**
但如果你拥有极其充裕、近乎免费的电力，又何必在意每瓦性能？只要数量够多就行，也可以用旧芯片来做。7 纳米芯片本质上相当于 Hopper。我必须告诉你，今天的模型大多就是在 Hopper 这一代上训练的，所以 7 纳米芯片完全够用。能源充裕是他们的优势。
**English / Original**
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.
## paragraph 248
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 249
**中文**
但接下来还有一个问题：他们实际上能不能制造足够多的芯片？
**English / Original**
But then there’s a question of whether they can actually manufacture enough chips.
## paragraph 250
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 251
**中文**
他们能。证据是什么？华为刚刚创下公司历史上最高的单年度业绩。
**English / Original**
But they do. What’s the evidence? Huawei just had the largest single year in the history of their company.
## paragraph 252
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 253
**中文**
他们出货了多少芯片？
**English / Original**
How many chips did they ship?
## paragraph 254
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 255
**中文**
非常多，数以百万计。数百万远多于 Anthropic 拥有的数量。
**English / Original**
A ton. Millions. Millions is way more than Anthropic has.
## paragraph 256
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 257
**中文**
一方面要看中芯国际能生产多少逻辑芯片，另一方面还要看有多少存储——
**English / Original**
There’s a question of how much logic SMIC can chip, and there’s a question of how much memory—
## paragraph 258
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 259
**中文**
我告诉你的就是现实：他们有充足的逻辑芯片，也有充足的 HBM2 存储。
**English / Original**
I’m telling you what it is. They have plenty of logic, and they have plenty of HBM2 memory.
## paragraph 260
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 261
**中文**
对。但你也知道，训练这些模型和用它们推理时，瓶颈往往是带宽。假如使用 HBM2……我一时记不起具体数字，但它和你们最新产品的内存带宽可能相差近一个数量级，这个差距非常大。
**English / Original**
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.
## paragraph 262
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 263
**中文**
华为是一家网络设备公司。
**English / Original**
Huawei is a networking company.
## paragraph 264
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 265
**中文**
但这改变不了一个事实：最先进的 HBM 仍需要 EUV。
**English / Original**
But that doesn’t change the fact that you need EUV for the most advanced HBM.
## paragraph 266
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 267
**中文**
不对，完全不对。你可以把它们连接起来，就像我们用 NVL72 把芯片连在一起。他们已经展示了硅光子技术，可以把所有这些算力连接成一台巨型超级计算机。你的前提就是错的。
**English / Original**
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.
## paragraph 268
**中文**
事实是，他们的 AI 发展得很好。世界上最优秀的 AI 研究人员因为算力受限，反而会提出极其聪明的算法。记住，我刚说过摩尔定律每年大约推进 25%；但凭借优秀的计算机科学，我们仍能把算法性能提高 10 倍。我的意思是，优秀的计算机科学才是真正的杠杆。
**English / Original**
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.
## paragraph 269
**中文**
毫无疑问，MoE 是伟大的发明；那些出色的注意力机制也确实减少了计算量。我们必须承认，AI 的大部分进步来自算法进展，而不只是原始硬件。既然多数进步来自算法、计算机科学和编程，你怎么能说他们庞大的 AI 研究人员队伍不是根本优势？我们已经看到了：DeepSeek 绝不是无关紧要的进展。哪一天 DeepSeek 首先在华为平台上发布，对我们国家来说就是糟糕的结果。
**English / Original**
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.
## paragraph 270
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 271
**中文**
为什么？像 DeepSeek 这样的模型如果开源，如今可以在任何加速器上运行。为什么未来会不再如此？
**English / Original**
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?
## paragraph 272
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 273
**中文**
假设事实并非如此。假设它针对华为、针对他们的架构做了优化，那会让我们的架构处于劣势。你描述的是一种在我看来属于好消息的情形：一家公司开发了软件和 AI 模型，而且在美国技术栈上运行得最好。我把它视为好消息，你却把它设定成坏消息的前提。真正的坏消息是：全世界开发出的 AI 模型都在非美国硬件上运行得最好。那才对我们不利。
**English / Original**
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.
## paragraph 274
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 275
**中文**
我只是没有看到存在如此巨大差距的证据，足以阻止人们切换加速器。美国实验室会让模型横跨各种云、各种不同加速器运行——
**English / Original**
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—
## paragraph 276
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 277
**中文**
我就是证据。你把针对英伟达优化的模型拿去别的平台上运行试试。
**English / Original**
I am the evidence. You take a model that’s optimized for Nvidia and you try to run it on something else.
## paragraph 278
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 279
**中文**
但美国实验室确实这么做。
**English / Original**
But American labs do that.
## paragraph 280
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 281
**中文**
但在别的平台上运行得并不好。英伟达的成功就是最充分的证据。AI 模型在我们的技术栈上创建，也在我们的技术栈上运行得最好，这一点有什么难以理解的？
**English / Original**
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?
## paragraph 282
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 283
**中文**
Anthropic 的模型既在 GPU 上运行，也在 Trainium 和 TPU 上运行。
**English / Original**
Anthropic’s models are run on GPUs, they’re run on Trainium, they’re run on TPUs.
## paragraph 284
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 285
**中文**
要切换过去，需要投入大量工作。但你去看看全球南方，看看中东。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，而你现在却要提出某种荒唐论断，声称这对美国是好事，那就太离谱了。
**English / Original**
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.
## paragraph 286
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 287
**中文**
但我想我还是不理解这个论点。假设中国公司率先做出下一代 Mythos，率先发现美国软件的所有安全漏洞，却是在英伟达硬件上完成，并把它推向全球南方。他们是在英伟达硬件上做的，这有什么好处？好吧，它运行在英伟达硬件上——
**English / Original**
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—
## paragraph 288
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 289
**中文**
这不是好事，不是好事。
**English / Original**
It’s not good. It’s not good.
## paragraph 290
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 291
**中文**
对。
**English / Original**
Right.
## paragraph 292
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 293
**中文**
这不是好事。所以别让它发生。
**English / Original**
It’s not good. So let’s not let it happen.
## paragraph 294
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 295
**中文**
你为什么认为算力完全可以相互替代，仿佛你们不向他们出货，就会由华为一比一补上？他们确实落后，对吧？他们的芯片不如你们。
**English / Original**
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.
## paragraph 296
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 297
**中文**
完全可以……眼下就有证据。他们的芯片产业规模极其庞大。
**English / Original**
It’s completely… There’s evidence right now. Their chip industry’s gigantic.
## paragraph 298
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 299
**中文**
只要比较 H200 和华为 910C 的浮点算力、带宽或内存就能看出来；后者大约只有前者的二分之一到三分之一。
**English / Original**
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.
## paragraph 300
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 301
**中文**
那他们就多用一些，用两倍数量。
**English / Original**
They use more of it. They use twice as many.
## paragraph 302
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 303
**中文**
你的论点似乎是，他们有大量现成能源，对吧？现在需要用芯片把这些容量填满。
**English / Original**
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.
## paragraph 304
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 305
**中文**
而且他们擅长制造。
**English / Original**
And they’re good at manufacturing.
## paragraph 306
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 307
**中文**
我也相信他们最终会凭制造规模超过所有人。但眼前这几年至关重要。
**English / Original**
And I’m sure eventually they would be able to just out-manufacture everybody. But there are these few critical years.
## paragraph 308
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 309
**中文**
你所说的“关键年份”究竟是哪一年？
**English / Original**
What is the critical year you’re talking about?
## paragraph 310
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 311
**中文**
就是接下来这几年。我们将拥有能够实施各种网络攻击的模型。
**English / Original**
These next few years. We’ve got these models that are going to be able to do all the cyber attacks.
## paragraph 312
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 313
**中文**
既然如此，如果未来几年至关重要，那我们就必须确保在这段关键时期，全球所有 AI 模型都建立在美国技术栈之上。
**English / Original**
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.
## paragraph 314
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 315
**中文**
如果它们建立在美国技术栈之上，但拥有更先进的能力，这如何阻止它们发动相当于 Mythos 的网络攻击？
**English / Original**
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?
## paragraph 316
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 317
**中文**
无论哪种情况都没有保证。
**English / Original**
There’s no guarantee either way.
## paragraph 318
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 319
**中文**
但如果我们率先拥有它，就能提前做好准备。
**English / Original**
But if you have it early, we can prepare for it.
## paragraph 320
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 321
**中文**
听着，你为什么要让 AI 产业中的一层失去整个市场，只为了让另一层受益？AI 有五层，每一层都必须成功。最需要成功的其实是 AI 应用层。你为什么如此执着于那个 AI 模型、那一家公司？理由是什么？
**English / Original**
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?
## paragraph 322
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 323
**中文**
因为这些模型让极其强大的攻击能力成为可能，而运行它们需要算力。
**English / Original**
Because those models make possible these incredibly offensive capabilities, and you need compute to run them.
## paragraph 324
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 325
**中文**
能源、芯片和 AI 研究人员生态共同让它成为可能。
**English / Original**
The energy, the chips, and the ecosystem of AI researchers make it possible.
## paragraph 326
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 327
**中文**
好，退一步看，前提必须是中国能建设足够多的 7 纳米产能。记住，他们仍受困于 7 纳米，而你们会前进到 3 纳米，再到 Feynman 的 2 纳米或 1.6 纳米。等你们用上 1.6 纳米时，他们可能仍停留在 7 纳米，因此必须生产足够数量来弥补差距。他们能源如此充裕，你给他们的芯片越多，他们拥有的算力就越多。所以归根结底，结论是他们会获得更多算力；而算力是训练和推理的一项投入——
**English / Original**
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—
## paragraph 328
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 329
**中文**
听着，我只是觉得你总用绝对化的方式说话。我认为美国应该领先。美国拥有的算力是世界其他任何地方的 100 倍，美国理应领先。好，美国现在确实领先。
**English / Original**
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.
## paragraph 330
**中文**
英伟达打造最先进的技术。我们确保美国实验室率先获知，并最先获得购买机会；如果它们资金不足，我们甚至会投资。美国应该领先，我们想尽一切办法确保美国领先。这是第一点，你同意吗？我们正竭尽所能实现这一点。
**English / Original**
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.
## paragraph 331
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 332
**中文**
但如果他们的瓶颈是算力，向中国运送芯片怎么能让美国保持领先？
**English / Original**
But how is shipping chips to China keeping the US ahead if they’re bottlenecked on compute?
## paragraph 333
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 334
**中文**
不，不。我们把 Vera Rubin 提供给美国，我们有供美国使用的 Vera Rubin。现在，我算美国的一部分吗？你认为我是美国的一部分吗？
**English / Original**
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?
## paragraph 335
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 336
**中文**
是。
**English / Original**
Yes.
## paragraph 337
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 338
**中文**
那英伟达呢？你认为英伟达是一家美国公司吗？好。第一，为什么我们不能制定更均衡的法规，让英伟达赢得全球市场，而不是放弃全世界？你为什么希望美国放弃全世界？
**English / Original**
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?
## paragraph 339
**中文**
芯片产业是美国生态的一部分，是美国技术领导力的一部分，也是 AI 生态和 AI 领导力的一部分。为什么你的政策、你的理念会导致美国放弃全球市场的很大一部分？
**English / Original**
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?
## paragraph 340
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 341
**中文**
我想这里的主张是……Dario 有句比喻：这就像 Boeing 炫耀我们向朝鲜出售核武器，但导弹外壳由 Boeing 制造，因此不知怎的就促进了美国技术栈。归根结底，你是在赋予他们这种能力。
**English / Original**
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.
## paragraph 342
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 343
**中文**
把 AI 与你刚才提到的任何东西相比，都是疯狂的。
**English / Original**
Comparing AI to anything that you just mentioned is lunacy.
## paragraph 344
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 345
**中文**
但 AI 与浓缩铀相似，对吧？它既可以有正面用途，也可以有负面用途。我们仍不愿把浓缩铀送给其他国家。
**English / Original**
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.
## paragraph 346
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 347
**中文**
谁在输送浓缩——
**English / Original**
Who’s sending enriched—
## paragraph 348
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 349
**中文**
这个类比是说，浓缩铀就像算力。
**English / Original**
The analogy is that enriched uranium is like compute.
## paragraph 350
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 351
**中文**
这是个糟糕的类比，也不合逻辑。
**English / Original**
It’s a lousy analogy. It’s an illogical analogy.
## paragraph 352
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 353
**中文**
但如果这种算力能运行一个对所有美国软件发动零日漏洞攻击的模型，它怎么会不是武器？
**English / Original**
But if that compute can run a model that can do zero-day exploits against all American software, how is that not a weapon?
## paragraph 354
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 355
**中文**
首先，解决这个问题的方式，是与研究人员、中国以及所有国家展开对话，确保人们不以那种方式使用技术。这种对话必须发生，好吗？这是第一点。
**English / Original**
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.
## paragraph 356
**中文**
第二，我们还必须确保美国领先，确保 Vera Rubin、Blackwell 在美国大量供应，堆积如山。我们的业绩显然能证明这一点：供应非常充裕，数量极多。我们拥有强大的算力，也有出色的 AI 研究人员，一切都很好。我们应该保持领先。
**English / Original**
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.
## paragraph 357
**中文**
然而，我们也必须认识到 AI 不只是一个模型，而是一块五层蛋糕。AI 产业的每一层都很重要，我们希望美国在每一层都获胜，包括芯片层。如果放弃整个市场，美国就无法在芯片层和计算技术栈的长期技术竞赛中获胜，这是事实。
**English / Original**
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.
## paragraph 358
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 359
**中文**
那么核心问题似乎是：现在向他们出售芯片，如何帮助我们长期获胜？Tesla 长期向中国销售极其优秀的电动汽车，iPhone 也在中国销售，产品都非常出色，却没有形成锁定。中国仍会制造自己的电动汽车，而且正在占据主导；他们的智能手机也在占据主导。
**English / Original**
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.
## paragraph 360
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 361
**中文**
我们今天开始谈话时，你承认英伟达的地位非常不同，还用了“护城河”这样的词。对我们公司最重要的一点，就是由开发者构成的丰富生态。全球 50% 的 AI 开发者在中国，美国不应该放弃他们。
**English / Original**
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.
## paragraph 362
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 363
**中文**
但美国有很多英伟达开发者，这并不妨碍美国实验室未来也使用其他加速器。事实上，他们现在也在使用其他加速器，这完全没问题，而且很好。我不明白为什么如果你们向中国出售英伟达芯片，中国不会出现同样情况，就像 Google 可以同时使用 TPU 和英伟达——
**English / Original**
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—
## paragraph 364
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 365
**中文**
我们必须持续创新；你大概也知道，我们的份额正在增长，而非下降。认为即便我们在中国竞争，最终也必然会失去那个市场……你现在面对的不是一个一觉醒来就认输的人。那种失败者心态、失败者前提，对我来说毫无道理。
**English / Original**
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.
## paragraph 366
**中文**
我们不是汽车，不是汽车。今天买这个品牌的车，明天换另一个品牌，很容易；计算并非如此。x86 生态之所以存在、ARM 之所以黏性极强，都是有原因的。这些生态很难替代，迁移要耗费巨量时间和精力，大多数人不愿这么做。因此，我们的职责是持续培育生态、不断推进技术，从而在市场中竞争。
**English / Original**
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.
## paragraph 367
**中文**
基于你描述的前提就放弃一个市场，我实在无法认同。这毫无道理，因为我不认为美国是失败者，我们的产业也不是失败者。那种注定失败的主张和心态，对我而言毫无道理。
**English / Original**
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.
## paragraph 368
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 369
**中文**
好，我换个话题。我只是想确认——
**English / Original**
Okay. I’ll move on. I just want to make sure that—
## paragraph 370
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 371
**中文**
你不必换话题，我聊得很享受。
**English / Original**
You don’t have to move on. I’m enjoying it.
## paragraph 372
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 373
**中文**
好，太好了，那我就不换。我很感谢。不过我想核心也许在于……也谢谢你陪我绕着这个问题反复讨论，因为这有助于揭示真正的分歧所在。
**English / Original**
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.
## paragraph 374
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 375
**中文**
核心在于你走向了极端。你的论证从极端条件出发：仿佛在这个短暂窗口里，只要给他们任何一点算力，我们就会失去一切。
**English / Original**
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.
## paragraph 376
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 377
**中文**
不，我的论点其实是——
**English / Original**
No, I think what my argument is—
## paragraph 378
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 379
**中文**
那些极端说法很幼稚。
**English / Original**
Those extremes, they’re childish.
## paragraph 380
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 381
**中文**
让我自己把论点讲清楚。我的意思并不是存在某个关键算力阈值，而是任何增量算力都有帮助。算力越多，就能训练出更好的模型。
**English / Original**
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.
## paragraph 382
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 383
**中文**
而我只希望你承认，美国科技产业每增加一笔边际销售都有益处。
**English / Original**
And I just want you to acknowledge that any marginal sales for the American technology industry is beneficial.
## paragraph 384
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 385
**中文**
我其实不承认……如果运行在这些芯片上的 AI 模型具备网络攻击能力，或者这些芯片被用于训练具备网络能力的模型并运行更多模型实例，那虽不是核武器，却会赋能某种武器。
**English / Original**
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.
## paragraph 386
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 387
**中文**
照你这套逻辑，同样的话也可以用于微处理器和 DRAM，甚至可以用于电力。
**English / Original**
The logic that you use, you might as well say it to microprocessors and DRAMs. You might as well say it to electricity.
## paragraph 388
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 389
**中文**
但事实上，我们确实对制造最先进 DRAM 所需的技术实施出口管制，也针对中国、针对各种芯片制造设备采取了各类出口管制。
**English / Original**
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.
## paragraph 390
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 391
**中文**
我们向中国出售大量 DRAM 和 CPU，我认为这是正确的。
**English / Original**
We sell a lot of DRAM and CPUs into China, and I think it’s right.
## paragraph 392
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 393
**中文**
我想这又回到根本问题：AI 是否有所不同？如果有一种技术能让他们在软件中找到这些零日漏洞，我们是否应该尽量降低中国率先实现并广泛部署它的能力？
**English / Original**
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?
## paragraph 394
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 395
**中文**
我们希望美国领先，这是我们能够掌控的。
**English / Original**
We want the United States to be ahead. We can control that.
## paragraph 396
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 397
**中文**
如果芯片已经在那里，他们正用来训练那种模型，我们要怎么掌控？
**English / Original**
How do we control that if the chips are already there and they’re using them to train that model?
## paragraph 398
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 399
**中文**
我们有大量算力，也有大量 AI 研究人员，正在以最快速度竞赛。
**English / Original**
We have tons of compute. We have tons of AI researchers. We’re racing as fast as we can.
## paragraph 400
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 401
**中文**
还是那个比喻：我们的核武器比任何国家都多，但我们不会因此就愿意把浓缩铀送往各地。
**English / Original**
Again, we have more nuclear weapons than anybody else, but we don’t want to send enriched uranium anywhere.
## paragraph 402
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 403
**中文**
我们不是浓缩铀。这是一枚芯片，而且是他们自己也能制造的芯片。
**English / Original**
We’re not enriched uranium. It’s a chip, and it’s a chip that they can make themselves.
## paragraph 404
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 405
**中文**
但他们从你们这里购买是有原因的。中国公司的创始人公开说过，他们受制于算力瓶颈。
**English / Original**
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.
## paragraph 406
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 407
**中文**
因为我们的芯片更好。综合而言，我们的芯片确实更好，这毫无疑问。但如果没有我们的芯片……你能承认华为刚创下纪录年度吗？能承认一大批芯片公司已经上市吗？能承认这些事实吗？
**English / Original**
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?
## paragraph 408
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 409
**中文**
能。
**English / Original**
Yes.
## paragraph 410
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 411
**中文**
你还能承认，我们过去在那个市场占有很大份额，如今却不再如此吗？我们也可以承认，中国约占全球科技产业的 40%。让美国科技产业放弃这个市场，是对国家的伤害，是对国家安全的伤害，也是对技术领导力的伤害，而这一切只为一家公司的利益。我觉得这毫无道理。
**English / Original**
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.
## paragraph 412
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 413
**中文**
我有些困惑。你似乎同时提出了两个不同论断：一是，只要允许竞争，我们会凭远胜华为的芯片赢得这场竞争；二是，即使没有我们，他们也会做完全相同的事情。这两点怎么能同时成立？
**English / Original**
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?
## paragraph 414
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 415
**中文**
显然能成立。没有更好的选择时，你就会采用唯一可用的选择。这哪里不合逻辑？逻辑再清楚不过。
**English / Original**
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.
## paragraph 416
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 417
**中文**
他们想要英伟达芯片，是因为英伟达的产品更好。
**English / Original**
The reason they want Nvidia chips is that they’re better.
## paragraph 418
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 419
**中文**
对。
**English / Original**
Yeah.
## paragraph 420
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 421
**中文**
更好就意味着更多算力，更多算力意味着可以训练出更好的模型。
**English / Original**
Better is more compute. More compute means you can train a better model.
## paragraph 422
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 423
**中文**
不，只是综合体验更好。它更好，是因为更容易编程，也因为我们的生态更优秀。但无论“更好”具体指什么……我们当然会向他们提供算力，那又怎样？事实是，我们也会从中受益：别忘了，我们会受益于美国的技术领导地位，受益于开发者在美国技术栈上工作；随着那些 AI 模型扩散到世界其他地方，美国技术栈也会因此成为最适合它们的平台。我们可以继续推进并传播美国技术。我认为这是积极的，也是美国技术领导力极其重要的一部分。
**English / Original**
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.
## paragraph 424
**中文**
而你主张的那些政策，实际上通过政策手段把美国电信产业基本逐出了全球市场，以至于我们如今甚至无法掌控自己的电信产业。我不认为这明智。那种思路有些狭隘，并导致了我现在向你描述的意外后果，而你似乎很难理解。
**English / Original**
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.
## paragraph 425
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 426
**中文**
好，我们退一步看。这里的核心似乎是既有潜在收益，也有潜在成本；我们要弄清的是收益是否值得付出成本。我想让你承认这种潜在成本：算力是训练强大模型的投入，而强大模型确实具有强大的攻击能力，例如网络攻击。美国公司率先达到 Mythos 级能力是件好事；现在他们会暂缓释放这种能力，让美国公司和美国政府有时间先加固软件，然后再公开这一能力水平。
**English / Original**
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.
## paragraph 427
**中文**
如果中国拥有更多算力或更多汇集算力，能更早造出 Mythos 级模型并广泛部署，那将非常糟糕。之所以尚未发生，原因之一就是在英伟达等美国公司的推动下，我们拥有更多算力。向中国提供算力会带来这种成本。先把收益放到一边，你是否承认这是潜在成本？
**English / Original**
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?
## paragraph 428
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 429
**中文**
我也来告诉你另一种潜在成本：我们让 AI 技术栈中最重要的层之一——芯片层——放弃全球第二大市场，让他们得以发展规模、建立自己的生态，使未来 AI 模型按照与美国技术栈截然不同的方式优化。随着 AI 向世界其他地区扩散，由于他们的模型是开放的，他们的标准和技术栈将变得优于我们。
**English / Original**
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.
## paragraph 430
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 431
**中文**
我想，我对英伟达的内核工程师和 CUDA 工程师足够有信心，相信他们可以完成优化——
**English / Original**
I guess I just believe enough in Nvidia’s kernel engineers and CUDA engineers to think that they could optimize—
## paragraph 432
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 433
**中文**
你也知道，AI 不只是内核优化。
**English / Original**
AI is more than kernel optimization, as you know.
## paragraph 434
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 435
**中文**
当然，但你们能做的事情很多，从蒸馏到得到一个适配自家芯片的模型。
**English / Original**
Of course, but there are so many things you can do, from distilling to a model that’s well-fit for your chips.
## paragraph 436
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 437
**中文**
我们会尽最大努力。
**English / Original**
We’re going to do our best.
## paragraph 438
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 439
**中文**
你们掌握所有软件。即使他们暂时拥有稍好一点的开源模型，也很难想象中国生态会形成长期锁定。
**English / Original**
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.
## paragraph 440
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 441
**中文**
中国是全球开源软件最大的贡献者，这是事实；中国也是全球开放模型最大的贡献者，这是事实；而今天这些成果建立在美国技术栈，也就是英伟达平台之上，这同样是事实。
**English / Original**
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.
## paragraph 442
**中文**
AI 技术栈的五层都很重要，美国应该赢下全部五层，每一层都重要。当然，最重要的是 AI 应用层；它会渗透进社会，使用最多的一方也将从这场工业革命中获益最多。但我的意思是，每一层都必须成功。
**English / Original**
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.
## paragraph 443
**中文**
如果我们把这个国家吓得以为 AI 就是某种核弹，让人人憎恨 AI、惧怕 AI，我不知道这怎么能帮助美国；这反而是在伤害美国。如果我们吓得所有人都不敢从事软件工程，因为 AI 会消灭每一个软件工程岗位，最终导致我们没有软件工程师，那也是在伤害美国。
**English / Original**
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.
## paragraph 444
**中文**
如果我们把所有人都吓得不敢学放射学，认为计算机视觉完全免费、AI 绝不会比放射科医生做得差，于是没人愿意成为放射科医生，那就是混淆了职业与任务。放射科医生的职业是照护患者，读取影像只是其中一项任务。如果我们如此严重地误解这一点，吓得所有人都不去学放射学，最后就不会有足够的放射科医生，也不会有足够好的医疗服务。
**English / Original**
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.
## paragraph 445
**中文**
所以我要说的是，当你的前提极端到一切非零即无穷时，最终会用不真实的东西吓到人们。生活不是那样。我们希望美国率先吗？当然。我们需要在技术栈每一层都保持领导吗？当然，当然需要。今天你谈 Mythos，是因为 Mythos 很重要，没错，这很好。
**English / Original**
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.
## paragraph 446
**中文**
但我预测，几年后，当我们希望美国技术栈和美国技术传播到全球——印度、中东、非洲、东南亚——当我国希望出口技术和标准时，我希望你我再次进行同样的对话。届时我会原原本本提醒你今天这番讨论：你的政策和设想如何毫无正当理由地导致美国亲手放弃全球第二大市场。
**English / Original**
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.
## paragraph 447
**中文**
我们不应该主动放弃。若竞争失败，那是失败；可为什么要不战而弃？现在没有人主张全有或全无，也没有人主张时时刻刻向中国运送所有产品。我们应该始终让最好的技术、最多的技术首先留在这里；但也应努力参与全球竞争并取胜。这两件事可以同时发生。它需要一些细致权衡和成熟判断，而不是绝对化。世界绝不是非黑即白。
**English / Original**
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.
## paragraph 448
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 449
**中文**
好。这个论点取决于一点：他们会针对几年后自家能制造的最佳芯片开发模型，再把这些芯片出口全球，从而建立标准。但正如我们所说，由于 EUV 出口管制，你们将前进到 1.6 纳米，而几年后他们仍会停留在 7 纳米。
**English / Original**
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.
## paragraph 450
**中文**
在国内，他们也许会想：“嘿，我们能源这么多，又能大规模制造，就继续用 7 纳米。”这或许合理。但在出口市场上，他们的 7 纳米芯片必须与英伟达的 1.6 纳米芯片竞争；他们的模型必须针对 7 纳米优化到极致，以至于在 7 纳米上运行反而胜过在你们 1.6 纳米芯片上运行。
**English / Original**
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.
## paragraph 451
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 452
**中文**
那我们就看看事实。Blackwell 的光刻技术比 Hopper 先进 50 倍吗？是 50 倍吗？差得远。我已经反复强调：摩尔定律已死。从晶体管本身看，Hopper 到 Blackwell 相隔三年，提升姑且算 75%；但 Blackwell 的实际性能是 Hopper 的 50 倍。
**English / Original**
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.
## paragraph 453
**中文**
我的意思是，架构很重要，计算机科学很重要。半导体物理当然也重要，但计算机科学同样关键。AI 的影响很大程度来自计算技术栈，这正是 CUDA 如此有效、如此受喜爱的原因。它既是生态，也是计算架构，提供了极大的灵活性：如果想彻底改变架构——创造 MoE、扩散模型或解耦式架构——都可以做到，而且很容易。
**English / Original**
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.
## paragraph 454
**中文**
因此，事实是 AI 既关乎底层架构，也同样关乎上层技术栈。只要架构和软件栈是针对我们的平台和生态优化的，显然就是好事；毕竟我们今天一开场就谈到英伟达生态多么丰富。为什么人们总喜欢优先为 CUDA 编程？他们确实如此，中国研究人员也一样。
**English / Original**
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.
## paragraph 455
**中文**
但如果我们被迫离开中国，首先，这是政策错误；它显然引发了反作用，对美国造成了不良后果。它赋能并加速了中国芯片产业，还迫使其整个 AI 生态转而专注于本土架构。现在还不算太晚，但这些后果已经发生。
**English / Original**
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.
## paragraph 456
**中文**
未来你会看到，他们显然不会永远困在 7 纳米。他们擅长制造，会继续从 7 纳米向更先进制程发展。那么 5 纳米和 7 纳米之间有十倍差距吗？没有。架构重要，网络重要——这就是英伟达收购 Mellanox 的原因——能源也重要。这一切都重要，问题并不像你试图概括的那么简单。
**English / Original**
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.
## paragraph 457
**中文**
01:35:06——英伟达为什么不打造多种不同的芯片架构？
**English / Original**
01:35:06 – Why doesn’t Nvidia make multiple different chip architectures?
## paragraph 458
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 459
**中文**
我们可以结束中国这个话题，不过它确实引出了一个有趣问题。之前我们讨论了台积电、存储等方面的瓶颈。
**English / Original**
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.
## paragraph 460
**中文**
如果我们进入这样一个世界：你们已经占 N3 产能的大多数，某个时点又会进入 N2 并同样占据多数；那你是否可能退回 N7，利用旧制程节点的闲置产能，说：“AI 需求太大，而最先进制程的扩产跟不上，所以我们要制造一款 Hopper 或 Ampere，但融入今天对数值格式的全部认识，以及你刚才提到的所有改进”？你认为这种情况会在 2030 年前发生吗？
**English / Original**
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?
## paragraph 461
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 462
**中文**
没有必要。因为每一代架构都远不只是晶体管尺度变化；其中还包含大量工程、封装、堆叠、数值格式和系统架构工作。
**English / Original**
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.
## paragraph 463
**中文**
产能用尽后，再轻易退回另一个制程节点……那需要的研发投入没有谁承受得起。我们有能力向前推进，却未必承担得起向后退。当然，假如世界有一天真的宣告——我们做个思想实验——“听着，我们永远不会再有新增产能了”，那我会不会回头使用 7 纳米？会毫不犹豫，当然会。
**English / Original**
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.
## paragraph 464
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 465
**中文**
有人向我提出过一个问题：英伟达为什么不同时运行多个架构完全不同的芯片项目？比如做 Cerebras 风格的晶圆级芯片、Dojo 风格的巨型封装，也可以做一个不使用 CUDA 的架构。你们有资源和工程人才并行推进所有这些项目。既然没人知道 AI 和架构将走向哪里，为什么把所有鸡蛋放在一个篮子里？
**English / Original**
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?
## paragraph 466
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 467
**中文**
哦，我们当然能做，只是没有更好的方案。这些东西都能做，只不过不会更好。我们在模拟器里把它们全模拟过，结果可以证明更差，所以不会去做。我们正在做的，正是自己想做的项目。如果工作负载发生剧烈变化——我说的不是算法变化，而是真正的工作负载变化，这取决于市场形态——我们也许会决定增加其他加速器。
**English / Original**
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.
## paragraph 468
**中文**
例如，我们最近加入了 Groq，并将把 Groq 纳入 CUDA 生态。现在这样做，是因为词元价值已变得非常高，因而可以形成不同的词元定价。过去，也就是短短几年前，词元不是免费，就是成本微不足道。现在则有不同类型的客户，他们需要不同类型的答案。由于这些客户能创造巨大价值——比如我们的软件工程师——如果我能提供响应快得多的词元，让他们比今天更高效，我愿意为此付费。
**English / Original**
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.
## paragraph 469
**中文**
但这个市场只是最近才出现。我认为，同一个模型如今可以根据响应时间划分不同细分市场。正因如此，我们决定拓展帕累托前沿，创建一个响应更快、但吞吐量较低的推理细分市场。此前一直是吞吐量越高越好；而我们认为，未来可能出现单价非常高的词元，即使工厂吞吐量较低，也能由更高的平均销售价格（ASP）弥补。
**English / Original**
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.
## paragraph 470
**中文**
这就是我们这样做的原因。但除此之外，从架构角度看，如果我有更多资金，仍会把更多资源投入英伟达的架构。
**English / Original**
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.
## paragraph 471
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 472
**中文**
我认为，这种超高溢价词元的概念，以及推理市场的解耦，非常有意思。
**English / Original**
I think this idea of extremely premium tokens and just the disaggregation of the inference market is a very interesting.
## paragraph 473
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 474
**中文**
也就是市场细分。
**English / Original**
The segmentation of it.
## paragraph 475
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 476
**中文**
对。好，最后一个问题。假设深度学习革命从未发生，英伟达会在做什么？游戏当然会做，但考虑到——
**English / Original**
Yeah. Alright, final question. Suppose the deep learning revolution didn’t happen. What would Nvidia be doing? Obviously games, but given—
## paragraph 477
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 478
**中文**
加速计算，还是我们一直以来在做的事。公司的出发点是摩尔定律将会……通用计算适合很多事情，但对许多计算任务并不理想。
**English / Original**
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.
## paragraph 479
**中文**
所以，我们把一种名为 GPU 的架构、CUDA 与 CPU 结合起来，加速 CPU 的工作负载。不同的代码内核或算法可以卸载到 GPU 上，从而让应用提速 100 倍、200 倍。它能用于哪里？显然包括工程、科学、物理、数据处理、计算机图形、图像生成等各种领域。即使今天不存在 AI，英伟达仍会是一家非常、非常庞大的公司。
**English / Original**
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.
## paragraph 480
**中文**
原因相当根本：通用计算持续扩展的能力基本已走到尽头。而继续扩展的唯一方式……不算唯一方式，但重要方式，就是领域专用加速。我们最先切入的领域之一是计算机图形，但还有许多其他领域，包括粒子物理、流体、结构化数据处理，以及各种能从 CUDA 获益的算法。
**English / Original**
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.
## paragraph 481
**中文**
我们的使命，真正说来是把加速计算带给全世界，推进通用计算无法处理的应用，并把能力扩展到足以帮助某些科学领域取得突破的水平。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；在所有这些领域，通用计算都效率太低，根本无法胜任。
**English / Original**
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.
## paragraph 482
**中文**
如果没有 AI，我会非常难过。但也正因为我们在计算领域取得的进步，深度学习才得以普及。我们让世界任何地方的研究人员、科学家和学生，都能使用一台 PC 或一张 GeForce 扩展卡开展了不起的科学研究。这项根本承诺没有丝毫改变。
**English / Original**
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.
## paragraph 483
**中文**
如果你看 GTC，会发现开头整整一大段都与 AI 无关。其中的计算光刻、量子化学和数据处理工作，全都不属于 AI，却依然非常重要。我知道 AI 很有趣、很令人兴奋，但许多人也在从事大量与 AI 无关的重要工作，而张量并不是完成计算的唯一方式。我们希望帮助所有人。
**English / Original**
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.
## paragraph 484
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 485
**中文**
Jensen，非常感谢你。
**English / Original**
Jensen, thank you so much.
## paragraph 486
**中文**
黄仁勋
**English / Original**
Jensen Huang
## paragraph 487
**中文**
不客气，我聊得很愉快。
**English / Original**
You’re welcome. I enjoyed it.
## paragraph 488
**中文**
德瓦克什·帕特尔
**English / Original**
Dwarkesh Patel
## paragraph 489
**中文**
我也是。
**English / Original**
Me too.