Jensen Huang – TPU competition, why we should sell chips to China, & Nvidia’s supply chain moat

频道: Dwarkesh Podcast
视频: https://www.dwarkesh.com/p/jensen-huang
原文语言: en
统计: 共 119 轮 · Dwarkesh Patel 38 · Jensen Huang 81

[0:00] Dwarkesh Patel

We've seen the valuations of a bunch of software companies crash because people are expecting AI to commoditize software.And there's a potentially naive way of thinking about things, which is like, 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 and Micron and Samsung make.Then it sends it to an ODM in Taiwan where they assemble the racks.And so NVIDIA is fundamentally making software that other people are manufacturing.And if software gets commoditized, does NVIDIA get commoditized?

我们已经看到一批软件公司的估值崩掉，因为大家预期 AI 会把软件商品化（commoditize，即变成谁都能做、卖不上价的大路货）。有一种可能很天真的想法是这样：你看，NVIDIA 把一个 GDS2 文件（芯片版图的最终交付格式）发给台积电（TSMC），台积电做出逻辑裸片（logic die）、做出交换芯片，再把它们和 SK 海力士、美光（Micron）、三星做的 HBM（高带宽内存）封装在一起，然后送到台湾的 ODM 厂去组装机柜。所以说到底，NVIDIA 做的其实是软件，制造是别人干的。那如果软件被商品化了，NVIDIA 会不会也被商品化？

[0:32] Jensen Huang

Well, in the end, something has to transform electrons to tokens.That transformation, there's no, the transformation of electrons to tokens and making those tokens more valuable over time.I don't, I think that that's hard to, hard to completely commoditize.The transformation from electrons to tokens is such an incredible journey.And making that token, you know, it's like making one molecule more valuable than another molecule, making one token more valuable than another.The amount of artistry, engineering, science, invention that goes into making that token valuable, obviously we're watching it happening in real time.And so the transformation, the manufacturing, all of the science that goes in there is far from deeply understood.And it's far from, the journey is far from over.And so I doubt that it will happen.We're going to make it more efficient, of course.I mean, the whole thing about NVIDIA, in fact, the way that you framed the question is my mental model of our company.The input is electron.The output is tokens.That is in the middle, NVIDIA.And our job is to do as much as necessary, as little as possible, to enable that transformation to be done at incredible capabilities.And what I mean by as little as possible, whatever I don't need to do, I partner with somebody and I make it part of my ecosystem to do.

说到底，总得有东西把电子转换成 token。这个转换过程——把电子变成 token，并且让这些 token 随时间越来越有价值——我认为很难被彻底商品化。从电子到 token 的这段旅程本身就极其了不起。让一个 token 比另一个 token 更有价值，就像让一个分子比另一个分子更有价值一样；要让 token 变得有价值，背后投入的技艺、工程、科学、发明，我们正在实时地看着它发生。所以这个转换、这套制造、里面所有的科学，远远谈不上被彻底理解，这段旅程也远远没有结束。所以我怀疑（商品化）会发生。我们当然会让它更高效。其实 NVIDIA 这家公司整件事，恰恰就是你提问的那个框架，就是我对我们公司的心智模型：输入是电子，输出是 token，中间那一块，就是 NVIDIA。我们的工作是「做到必要的那么多、同时尽可能少做」，好让这个转换以惊人的能力完成。所谓「尽可能少做」是指：凡是我不必自己做的，我就找伙伴来做，把它变成我生态的一部分。

[2:16] Jensen Huang

And if you look at NVIDIA today, we probably have the largest ecosystem of partners, both in supply chain upstream, supply chain downstream, all of the computers, computer companies, and all the application developers, and all the model makers, and all the, you know, AI is a five-layer cake, if you will.And we have ecosystems across the entire five layers.And so we try to do as little as possible.But the part that we have to do, as it turns out, is insanely hard.And I don't think that that gets commoditized.In fact, I also don't think that the enterprise software companies, the tools makers, you know, most of the software companies today are tools makers.Some of them are not, but some of them are workflow codification, you know, systems.But for a lot of companies, they're tool makers.For example, you know, Excel is a tool.PowerPoint's a tool.Cadence makes tools.Synopsys makes tools.I actually see the opposite of what people see.I think the number of agents are going to grow exponentially.The number of tool users are going to grow exponentially.And it's very likely that the number of instances of all these tools are going to skyrocket.It is very likely the number of instances of Synopsys design compiler is going to skyrocket.

今天你看 NVIDIA，我们大概拥有最大的伙伴生态：上游供应链、下游供应链、所有的计算机厂商、所有的应用开发者、所有的模型厂商——AI 是一块「五层蛋糕」，我们在五层里都有生态。所以我们尽量少做，但我们必须做的那部分，事实证明难得离谱，我不认为它会被商品化。事实上，我也不认为那些企业软件公司、工具厂商会被商品化。今天大多数软件公司是工具制造商，有些不是，有些是把工作流固化成系统。但对很多公司来说，它们是做工具的。比如 Excel 是工具，PowerPoint 是工具，Cadence 做工具，Synopsys 做工具。我看到的结论恰恰跟大家相反：我认为 agent（智能体）的数量会指数级增长，工具使用者的数量会指数级增长，而所有这些工具的实例数很可能会暴涨。Synopsys 的 Design Compiler（芯片综合工具）的实例数很可能暴涨。

[3:45] Jensen Huang

And the number of agents that are going to be using the floor planners and all of our layout tools and our design rule checkers, the number of agents that are today were limited by the number of engineers.Tomorrow, those engineers are going to be supported by a bunch of agents.And we're going to be exploring the design space like you've never seen explored before and want to use the tools that we use today.And so I think tool use is going to cause these 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.And so 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.And I think it's going to be a combination of both.I think in your latest filings, you had almost $100 billion in purchase commitments with people, foundries, memory, packaging.And then Semi-Analysis has reported that you will have $250 billion of these kinds of purchase commitments.And so one interpretation is NVIDIA's mode is really that you've locked up many years of these scarce components that are, you know, somebody else might have an accelerator, but can they actually get the memory to build it?

还有那些会去用布局规划器（floor planner）、各种版图工具、设计规则检查器的 agent 数量——今天 agent 的数量受限于工程师的数量，明天这些工程师背后会有一大堆 agent 支持。我们会以前所未见的方式去探索设计空间，而且会想用我们今天用的这些工具。所以我认为工具调用会让这些软件公司起飞。之所以还没发生，是因为 agent 还不够会用它们的工具。所以要么这些公司自己去做 agent，要么 agent 变得足够好、能用上这些工具——我觉得两者都会发生。

〔Dwarkesh〕 你们最新的财报文件里，跟代工厂、内存厂、封装厂的采购承诺接近 1,000 亿美元。而 SemiAnalysis 报道说你们这类采购承诺会达到 2,500 亿美元。所以一种解读是：NVIDIA 真正的护城河（moat），是你们已经把这些稀缺元器件锁定了好多年——别人也许能做出加速器，但他们真的拿得到内存吗？

[4:55] Dwarkesh Patel

Can they actually get the logic to build it?And this is really NVIDIA's big mode for the next few years.Well, it's one of the things that we can do that is hard for someone else to do.The reason why we could, 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 and let me explain to you why.And let me reason through it with you.And let me show you what I see.And so as a result of that process of informing, inspiring, aligning with CEOs of all different industries upstream,they're willing to make the investments.Now, why are they willing to make the investments for me and not someone else?

他们真的拿得到逻辑芯片产能吗？这才是 NVIDIA 未来几年真正的大护城河。

〔Jensen〕 这确实是我们能做、而别人很难做的事情之一。我们之所以能，是因为我们在上游做了巨大的承诺。有些是显性的，就是你提到的那些承诺；有些是隐性的。比如上游的很多投资是我们的供应链自己掏钱做的，因为我跟那些 CEO 说：让我告诉你这个行业将会有多大，让我解释给你听为什么，让我跟你一起把这个逻辑推一遍，让我把我看到的东西给你看。正是通过这个「告知、激励、对齐」的过程，跟上游各行各业的 CEO 对齐，他们才愿意做这些投资。那么问题来了，他们为什么愿意为我投、而不为别人投？

[5:50] Jensen Huang

And the reason for that is because they know that I have the capacity to buy it, buy their supply, and sell it through my downstream.The fact that NVIDIA's downstream supply chain and our downstream demand is so large,they're willing to make the investment upstream.And so if you look at GTC, and people are marveled by the scale of GTC and the people that go,it's a 360 degree.It's the entire universe of AI all in one place.And they're all in one place because they need to see each other.I bring them together so that the downstream could see the upstream, the upstream could see the downstream,and all of them could see all the advances in AI.And very importantly, they can all meet the AI natives and all the AI startups that are all being builtand all the amazing things that are happening so that they could see firsthand all the things that I tell them.And so I spend a lot of my time informing directly or indirectly our supply chain and our partners and our ecosystemabout the opportunity that's in front of us.Most of my keynotes, some people always say,Jensen, in most keynotes, it's like one announcement after another announcement after another announcement after another announcement.Our keynotes are, there's always a part of it that's a little torturous in the sense that it almost comes across like education.

原因是他们知道我有能力把他们的供给买下来，并通过我的下游卖出去。NVIDIA 的下游供应链和下游需求如此之大，所以他们愿意在上游投资。你看 GTC（NVIDIA 的年度技术大会），大家惊叹于 GTC 的规模和到场的人——它是 360 度的，整个 AI 宇宙都在同一个地方。他们都在那儿，是因为他们需要看见彼此。我把他们聚到一起，让下游看见上游、上游看见下游，让所有人都看到 AI 的全部进展。非常重要的一点是，他们还能见到所有 AI 原生公司和 AI 创业公司、见到正在发生的所有神奇的事，这样他们就能亲眼确认我跟他们讲的那些事。所以我花大量时间，直接或间接地向我们的供应链、伙伴、生态传达摆在我们面前的机会。我大部分主题演讲——有人总说，Jensen，你的 keynote 就是一个发布接着一个发布。其实我们的 keynote 里总有一部分有点「折磨人」，因为它几乎像在上课。

[7:21] Jensen Huang

And in fact, that's exactly on my mind.I need to make sure that 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 is it going to be, and be able to reason about it systematically,just like I reason about it.And so I think the mode as you describe it, we're able to, of course, build for a future.If our next several years is a trillion dollars in scale, we have the supply chain to do it.Without our reach, the velocity of our business, you know, just as there's cash flow, there's supply chain flow, there turns.Nobody's going to build a supply chain for an architecture if the architecture, the business turns as low.And so our ability to sustain the scale is only because our downstream demand is so great and they see it and they all hear about it.They see it all coming.And so that allows us to do the things that we're able to do at the scale we're able to do.I do want to understand more concretely whether the upstream can keep up.For many years now, you guys have been 2Xing revenue year over year.You guys have been more than tripling the amount of flops you're providing to the world year over year.

事实上我心里想的正是这个。我得确保整条上下游供应链、整个生态都明白：什么正朝我们扑来、为什么来、什么时候来、会有多大，并且能像我一样系统性地推理它。所以你说的那个护城河，我们当然有能力为未来做建设。如果我们未来几年的规模是一万亿美元，我们有能撑起来的供应链。没有我们的触达面、没有我们业务的周转速度——就像有现金流一样，供应链也有流动、也有周转次数。如果一个架构的生意周转很低，没人会为它建供应链。我们之所以能维持这个规模，唯一原因就是我们的下游需求太大了，而且他们看得见、听得到，他们看到这一切正在到来。这才让我们能以现在这个规模做我们做的事。

〔Dwarkesh〕 我想更具体地弄清楚上游到底跟不跟得上。这么多年你们的收入一直在同比翻倍，向世界提供的算力（flops）一直在同比三倍以上地增长。

[8:44] Dwarkesh Patel

And 2Xing at the scale now is really incredible.Exactly.Yeah.So then you look at logic, say.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 60% of N3.It's going to be 86% next year according to some analysis.How do you 2X if you're the majority?

在现在这个体量上还能翻倍，真的很惊人。〔Jensen〕 没错。〔Dwarkesh〕 那我们看逻辑芯片。你们是台积电 N3 制程最大的客户，也是 N2 上最大的客户之一。有分析说 AI 今年整体会占掉 N3 产能的 60%，明年会到 86%。当你已经占大头的时候，怎么再翻倍？

[9:07] Dwarkesh Patel

And how do you do that year over year?So are we in a regime now where the growth rate in the AI compute has to slow because of upstream?Do you see a way to get around these, you know, how do we build 2X more fabs year over year ultimately?

而且要年复一年地翻倍？所以我们是不是已经进入了一个「因为上游限制、AI 算力增速必须放慢」的阶段？你有没有看到绕过这些约束的办法——说到底，我们怎么可能一年建出两倍多的晶圆厂？

[9:21] Jensen Huang

Yeah.At some level, the instantaneous demand is greater than the supply upstream and downstream in the world.And it could be at any instance, we could be limited by the number of plumbers, which actually happens.The plumbers are admitted to next year's GTC.Yeah.You know, by the way, great idea.But that's a good condition.You want a market, 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 item, one particular component is too far away, obviously the industry swarms it.So, for example, notice people aren't talking very much about co-ops anymore.And the reason for that is because for two years we swarmed a living daylights out of it.And we double, double, double on several doubles.And now I think we're in a fairly good shape.And TSMC now knows that co-ops supply has to keep up with the rest of the logic demand and the memory demand.And so they're scaling co-ops and they're scaling future packaging technologies at the same level as they scale logic, which is terrific.Because for a long time, co-ops was rather specialty.And HBM memory was rather specialty.But they're not specialties anymore.

是这样。在某种程度上，全世界的瞬时需求就是大于上下游的供给。任何时刻我们都可能被某个环节卡住，比如管道工（plumber）的数量——这是真的发生过的。〔Dwarkesh〕 那明年 GTC 得请管道工入场了。〔Jensen〕 顺便说，这主意不错。但这是个好状况。你要的就是一个瞬时需求大于全行业总供给的市场和行业，反过来显然就没那么好。如果差得太远、如果某个特定环节、某个特定元器件差得太远，行业显然会一拥而上去解决它。举例来说，注意到大家现在不太谈 CoWoS（台积电的先进封装技术）了吧？原因就是这两年我们把它往死里猛攻，翻倍、翻倍、再翻倍，翻了好几轮，现在我觉得状况相当不错了。而且台积电现在明白，CoWoS 的供给必须跟上逻辑需求和内存需求。所以他们在扩 CoWoS，也在以跟扩逻辑产能同样的力度扩未来的封装技术，这非常好。因为很长一段时间里，CoWoS 属于比较特殊的小众品类，HBM 内存也是小众品类，但它们现在不再是小众了。

[10:55] Jensen Huang

People now realize they're mainstream computing technology.And then, of course, we're now much more able to influence a larger scope of our supply chain.In the past, in the beginning of the AI revolution, all the things that I say now, I was saying five years ago.And 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's going to happen and why it's going to happen and the predictions of today.And they really doubled down on it.And we partnered with them across LPDDR, across HBM memories.They really invested in it.And it obviously has been tremendous for the company.Some people came a little bit later.But now they're all here.And so I think each one of these bottlenecks gets a great deal of attention.And now we're prefetching the bottlenecks years in advance.So, for example, the investments that we've done with Lumentum and Coherent and all of the silicon photonics ecosystem,the last several years, we really reshaped the ecosystem and the supply chain, silicon photonics.We built up an entire supply chain around TSMC.We partnered with them on Coop, invented a whole bunch of technology.

大家现在意识到它们是主流计算技术。当然，我们现在也更有能力去影响供应链里更大范围的环节。在 AI 革命刚开始的时候，我现在说的这些话，五年前我就在说了。有些人信了、也投了。比如美光的 Sanjay（Sanjay Mehrotra，美光 CEO）和他的团队，我到现在都清楚记得那场会——我很明确地讲了到底会发生什么、为什么会发生，以及今天这些预测。他们真的重仓押了下去。我们跟他们在 LPDDR、在 HBM 内存上都合作了，他们真的投了，这对那家公司显然是巨大的回报。有些人来得晚一点，但现在他们都来了。所以我认为每一个瓶颈都会得到大量关注，而现在我们是提前好几年就去「预取」（prefetch）这些瓶颈。比如我们跟 Lumentum、Coherent 以及整个硅光子（silicon photonics）生态做的那些投资，过去几年我们真的重塑了硅光子的生态和供应链。我们围绕台积电搭起了一整条供应链，跟他们一起做 CoWoS，发明了一大堆技术。

[12:31] Jensen Huang

We licensed those patents to the supply chain, keep it nice and open.And so we're preparing the supply chain through invention of new technologies, new workflows, new testing equipment, double-sided probing,investing in companies, helping them scale up their capacity.And so you could see that we're trying to shape the ecosystem so that it's ready, the supply chain, so that it's ready to support the scale.It seems like some bottlenecks are easier than others.And so scaling up Coop versus scaling up...I went to the hardest one, by the way.Which is?

我们把这些专利授权给供应链，保持它足够开放。所以我们是在通过发明新技术、新工艺流程、新测试设备（比如双面探针测试）、投资这些公司、帮他们扩产能，来把供应链准备好。你能看到我们在努力塑造这个生态和供应链，让它做好准备去支撑这个规模。

〔Dwarkesh〕 看起来有些瓶颈比另一些好解。比如扩 CoWoS 相比扩……

〔Jensen〕 顺便说，我可是挑了最难的那个。

〔Dwarkesh〕 哪个？

[13:05] Jensen Huang

Plumbers.Yeah.That's true.Yeah, yeah.I actually went to the hardest one.Yeah.Yeah, plumbers and electricians.And the reason for that is because...And this is one of the concerns that I have about the doomers, describing the end of work and killing of jobs.And one of the things that if we discourage people from being software engineers, we're going to run out of software engineers.And the same prediction 10 years ago, some of the doomers were saying that...We're telling people, whatever you do, don't be a radiologist.And you might hear some of those videos are still on the web.You know, radiology is going to be the first career to go.The world's not going to need any more radiologists.Guess what we're short of?

管道工。〔Dwarkesh〕 是啊，确实。〔Jensen〕 我真的挑了最难的那个：管道工和电工。原因在于——这也是我对那些「末日论者」（doomer）的担忧之一，他们描述工作的终结、岗位被杀死。如果我们劝阻人们不要去当软件工程师，我们就真的会没有软件工程师。十年前也有同样的预测，那些末日论者说：不管你干什么，千万别去当放射科医生。你现在还能在网上找到那些视频。他们说放射科会是第一个消失的职业，世界不再需要放射科医生了。结果猜猜我们现在缺什么？

[13:53] Jensen Huang

Radiologists.Oh, but okay.So going back to this point about, well, some things you scale, other things like...How do you actually get...How do you actually manufacture 2x the amount of logic a year?Ultimately, that's bottlenecked by...Memory and logic are bottlenecked by EUV.How do you get to 2x as many EUV machines a year?

放射科医生。

〔Dwarkesh〕 好，那回到刚才那个点：有些东西你能扩，有些东西……你到底怎么才能一年真的制造出两倍的逻辑芯片？说到底这被什么卡住？内存和逻辑都被 EUV（极紫外光刻）卡住。你怎么让 EUV 光刻机一年出货量翻倍？

[14:09] Jensen Huang

Yeah.Year over year.None of that's impossible to scale quickly.You just need to...You could do...All of that is easy to do within two or three years.You just need a demand signal.It's not...Once you can build one, you can build 10.And once you can build 10, you can build a million.And so these things are not hard to replicate.How far down the supply chain do you go?

是啊，年复一年。这些都不是快速扩产做不到的事。你只需要……这些事在两三年内都很容易做到，你只需要一个需求信号。一旦你能造出一台，你就能造出十台；一旦你能造十台，你就能造一百万台。所以这些东西不难复制。

〔Dwarkesh〕 你会顺着供应链往下走多远？

[14:32] Dwarkesh Patel

Do you go to ASML and say, hey, if I look out three years from now, for me to...For NVIDIA to be generating 2 trillion in a year in revenue, we need way more AUV machines?And...Some of them I have to directly...Some of them I indirectly.And some of them...If I can convince TSMC, ASML will be convinced.And so that's...You know, we have to think about the critical pinch points.And...But if TSMC is convinced, you'll have plenty of EUV machines in a few years.And so none of that...My point is that none of the bottlenecks last longer than a couple, two, three years.None of them.And meanwhile, we're improving computing efficiency by 10x, 20x.In the case of Hopper to Blackwell, some 30, 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.And so...So those are things that none of that worry me.It's the stuff that's downstream from us.Energy policies that prevent energy from...You know, you can't grow...You can't create an industry without energy.You can't create a whole new manufacturing industry without energy.We want to re-industrialize the United States.We want to bring back chip manufacturing and computer manufacturing and packaging.

你会去找 ASML 说，嘿，如果我看三年后，NVIDIA 要做到一年 2 万亿美元收入，我们需要多得多的 EUV 光刻机吗？

〔Jensen〕 有些我得直接去谈，有些我间接影响。有些是这样：如果我能说服台积电，ASML 自然就被说服了。所以我们要想清楚关键的卡点在哪。但只要台积电被说服了，几年后你就会有充足的 EUV 光刻机。我的意思是，没有哪个瓶颈能撑过两三年，一个都没有。与此同时我们在把计算效率提升 10 倍、20 倍——Hopper 到 Blackwell 这一代甚至是 30 倍、50 倍。因为 CUDA 足够灵活，我们不断想出新算法，开发各种新技术，在扩容之外还能提效。所以这些事都不让我担心。真正让我担心的是我们下游的东西：那些阻碍能源供应的能源政策——没有能源你没法造出一个产业，没有能源你没法建起一个全新的制造业。我们想让美国重新工业化，我们想把芯片制造、计算机制造和封装带回来。

[15:57] Jensen Huang

And we want to build new things like EVs and robots.And we want to build AI factories.And you can't build any of these things without energy.And those things take a long time.But more chip capacity, that's a two, three-year problem.More COAS capacity, two, three-year problem.Interesting.I feel like I have guests tell me the exact opposite thing sometimes.And I don't...In this case, I just don't have the technical knowledge to adjudicate, but...Well, the beautiful thing is you're talking to the expert.Yeah.True, true.Okay, I want to ask about your competitors.

我们还想造电动车和机器人这些新东西，我们想建 AI 工厂。没有能源你什么都建不了，而这些（能源）要花很长时间。但更多芯片产能，那是个两三年的问题；更多 CoWoS 产能，两三年的问题。

〔Dwarkesh〕 有意思。我感觉有些嘉宾跟我讲的完全相反。这件事上我没有足够的技术知识来裁决，但……

〔Jensen〕 好在你正在跟专家说话。

〔Dwarkesh〕 对对，没错。好，我想问问你的竞争对手。

[16:28] Dwarkesh Patel

Yeah.So, if you look at 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?Well, we have a very different...We build a very different thing.You know, what NVIDIA built is accelerated computing, not a tensor processing unit.And accelerated computing is used for all kinds of things.You know, molecular dynamics and quantum chromodynamics.And it's used for data processing, data frames, structured data, unstructured data.It's used for fluid dynamics, particle physics.You know, and in addition, we use it for AI.And so, accelerated computing is much more diverse.And although AI is the conversation today is obviously very important and impactful,computing is much broader than that.And what NVIDIA has done is reinvented the way computing is done from general purpose computingto accelerated computing.Our market reach is far greater than any TPU, any ASIC can possibly have.And so, if you look at our position, we're the only company that accelerates applications of all kinds.We have a gigantic ecosystem.And so, all kinds of frameworks and algorithms all run on NVIDIA.And because our computers are designed to be operated by other people,

你看 TPU（Google 的张量处理单元），可以说全世界前三的模型里有两个——Claude 和 Gemini——是在 TPU 上训练的。这对 NVIDIA 的未来意味着什么？

〔Jensen〕 我们做的是很不一样的东西。NVIDIA 造的是加速计算（accelerated computing），不是张量处理单元。加速计算被用在各种各样的地方：分子动力学、量子色动力学，用于数据处理、数据帧（data frame）、结构化数据、非结构化数据，用于流体力学、粒子物理。此外我们才用它做 AI。所以加速计算要多样得多。虽然今天的话题 AI 显然非常重要、影响巨大，但计算的范围要宽得多。NVIDIA 做的事，是把计算的方式从通用计算重新发明成了加速计算。我们的市场触达面，比任何 TPU、任何 ASIC（专用集成电路）都要大得多。所以看我们的位置：我们是全世界唯一一家能加速各种应用的公司。我们有一个巨大的生态，各种框架、各种算法都跑在 NVIDIA 上。而且因为我们的计算机被设计成可以被别人运营……

[18:08] Jensen Huang

anyone who's an operator could buy our systems.Most of these home-built systems, you have to be your own operatorbecause it was never designed to be flexible enough for other people to operate.And so, as a result of the fact that anybody can operate our systems,we're in every cloud, including Google and Amazon and, you know, Azure and OCI, right?

……任何一个运营商都能买我们的系统。大多数自研自建的系统，你必须自己当运营商，因为它从来就没被设计得足够灵活、能让别人来运营。正因为任何人都能运营我们的系统，我们出现在每一朵云上，包括 Google、Amazon、Azure、OCI（甲骨文云），对吧？

[18:32] Jensen Huang

And so, whether you want to operate it to rent or operate it,if you want to operate it to rent, you better have a large ecosystem of customersin many industries that be the off-takers.If you're operating it, if you want to operate it for yourself,we obviously have the ability to help you operate yourself,like, for example, for Elon with XAI.And because we could enable operators in any company, in any industry,you could use it to build a supercomputer for scientific research and drug discovery at Lilly.And so, we can help them operate their own supercomputerand use it for the entire diversity of drug discovery and biological sciences that we accelerate.And so, there are just, you know, a whole bunch of applications that we can addressthat you can't do so with TPUs.Because NVIDIA has built CUDA as a fantastic tensor processing unit as well,but it does, you know, it does every life cycle of data processing and computing and AI and so on and so forth.And so, our market opportunity is just a lot larger.Our reach is a lot greater.And because we have such a large, we basically support every application in the world now,you could build NVIDIA systems anywhere and know that there will be customers for it.

所以不管你是想运营它来出租，还是自己用。如果你要运营它来出租，你最好在很多行业都有一大批客户来接货（off-taker）。如果你是自己运营自己用，我们显然也有能力帮你——比如帮 Elon 的 xAI。因为我们能让任何公司、任何行业的运营方跑起来，你可以用它建一台做科研和药物发现的超级计算机，比如礼来（Lilly）。我们能帮他们运营自己的超算，用于我们所加速的整个药物发现和生命科学的多样场景。所以有一大堆应用是我们能覆盖、而 TPU 做不到的。因为 NVIDIA 把 CUDA 也做成了一个极好的张量处理单元，但它同时还覆盖数据处理、计算、AI 等等的整个生命周期。所以我们的市场机会大得多，触达面广得多。而且因为我们基本上支撑了当今世界上的每一个应用，你可以在任何地方建 NVIDIA 系统，并且确信会有客户。

[19:58] Jensen Huang

And so, it's a very different thing.This is going to be sort of a long question, but, you know, you have spectacular revenue.And this revenue is mostly, you're not making $60 billion a quarter from pharma and quantum.You're making it because AI is unprecedented technology that is going unprecedentedly fast.And so, then the question is, what is best for AI specifically?

所以这是很不一样的东西。

〔Dwarkesh〕 这可能是个比较长的问题。你们的收入非常惊人，而这些收入主要不是来自制药和量子——你一个季度赚 600 亿美元，不是靠这些。你赚这些钱，是因为 AI 是一项以前所未有的速度推进的、前所未有的技术。所以问题就变成：具体到 AI，什么才是最好的？

[20:18] Dwarkesh Patel

And 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, when you have irregular memory access.But these, you know, what is AI?

我不懂具体细节，但我跟做 AI 研究的朋友聊，他们说：你看，我用 TPU 的时候，它是一个巨大的脉动阵列（systolic array），做矩阵乘法非常完美；而 GPU 非常灵活，当你有大量分支、有不规则内存访问的时候它很棒。但是，AI 到底是什么？

[20:37] Dwarkesh Patel

Just like these very predictable matrix multiplies again and again and again.And you don't have to give up any die area for warp schedulers, for, you know,switches between threads and memory banks.And so, the TPU is really optimized for the majority, the bulk of this growth in revenueand use case for a compute that is coming online right now.Yeah, I wonder how you react to that.Matrix multiplies is an important part of AI, but it's not the only part of AI.

不就是这些非常可预测的矩阵乘法，一遍又一遍地做吗？而且你不用为线程束调度器（warp scheduler）、为线程和内存 bank 之间的切换让出任何芯片面积。所以 TPU 其实是为当下这波算力需求和收入增长的大头做了优化。我想听听你怎么回应。

〔Jensen〕 矩阵乘法是 AI 的重要一部分，但不是 AI 的全部。

[21:05] Jensen Huang

And if you want to come up with a new attention mechanism, or if you want to disaggregate in a different way,if you want to come up with a whole new type of architecture altogether, for example, you know, a hybrid SSM,if you want to use a, you want to create a model that fuses diffusion and autoregressive somehow,you want an architecture that's just generally programmable.And we run everything you can imagine.And so, that's the advantage.It allows for invention of new algorithms a lot more easily.And so, because it's a programmable system.And the ability to invent new algorithms is really what makes AI advance so quickly.You know, TPUs, like anything else, is impacted by Moore's Law.And we know that Moore's Law is increasing about 25% per year.And so, the only way to really get 10x leaps, 100x leaps, is to fundamentally change the algorithm and how it's computed every single year.And that's NVIDIA's fundamental advantage.The only reason why we were able to make Blackwell to Hopper 50 times, you know, I said it was 35 times.And when I first announced Blackwell was going to be 35 times more energy efficient than Hopper, nobody believed it.And then Dylan wrote an article.He said, in fact, I sandbagged it's actually 50 times.

如果你想搞出一种新的注意力机制，或者你想用不同的方式做分离式部署（disaggregate），或者你想搞一种全新的架构——比如混合 SSM（状态空间模型），比如你想做一个把扩散模型和自回归融合起来的模型——你就需要一个通用可编程的架构。而我们什么都能跑。这就是优势所在：它让发明新算法容易得多，因为它是可编程的系统。而发明新算法的能力，恰恰是 AI 进步这么快的真正原因。TPU 跟别的东西一样，受摩尔定律限制。而我们知道摩尔定律现在每年大概只带来 25% 的提升。所以要拿到 10 倍、100 倍的跃迁，唯一办法就是每一年都从根本上改变算法以及计算它的方式。这才是 NVIDIA 的根本优势。我们之所以能把 Blackwell 相对 Hopper 做到 50 倍——我当时说的是 35 倍。我第一次宣布 Blackwell 会比 Hopper 能效高 35 倍时，没人信。后来 Dylan（Dylan Patel，SemiAnalysis 创始人）写了篇文章，说我其实是保守了，实际上是 50 倍。

[22:47] Jensen Huang

And you can't reasonably do that with just Moore's Law.And so, the way that we solve that problem is new models, MOEs, paralyzed and disaggregated and distributed across a computing system.And without the ability to really get down and come up with new kernels with CUDA, it's really hard to do.And so, the combination of the programmability of our architecture, the fact that NVIDIA is an extreme co-design company where we could even offload some of the computation into the fabric itself, NVLink, for example, into the network Spectrum X.And that we could affect change across the processors, the system, the fabric, the libraries, the algorithm.All of that was done simultaneously.Without CUDA to do that, I wouldn't even know where to start.My sponsor, Crusoe, was among the first clouds to offer NVIDIA's Blackwell and Blackwell Ultra platforms.And they just announced their NVIDIA Vera Rumen deployment, scheduled for later this year.But access to state-of-the-art hardware is only part of the story.For example, most inference engines already do KV caching for a single user's forward passes.But Crusoe does it across users and GPUs.So, if a thousand agents are running on the same system prompt, Crusoe only has to compute the KV cache once for it to become available to every single GPU in the cluster.

光靠摩尔定律你不可能合理地做到这一点。我们解决这个问题的方式是新模型：MoE（混合专家）、并行化、分离式部署、分布到整个计算系统上。而如果没有能力沉到底层、用 CUDA 写出新的 kernel（核函数），这非常难做到。所以是我们架构的可编程性，加上 NVIDIA 是一家极致协同设计（co-design）的公司——我们甚至能把一部分计算卸载到互联结构本身，比如 NVLink，比如网络里的 Spectrum-X——再加上我们能同时在处理器、系统、互联、库、算法各层做出改变，这一切是同时发生的。没有 CUDA 来做这件事，我都不知道该从哪儿下手。

〔广告〕 我的赞助商 Crusoe 是最早提供 NVIDIA Blackwell 和 Blackwell Ultra 平台的云之一，他们刚刚宣布了将于今年晚些时候上线的 NVIDIA Vera Rubin 部署。但拿到最先进的硬件只是故事的一部分。比如，大多数推理引擎已经会为单个用户的前向传播做 KV 缓存，而 Crusoe 是跨用户、跨 GPU 做缓存。所以如果有一千个 agent 跑在同一个系统提示词上，Crusoe 只需要算一次 KV 缓存，集群里每一块 GPU 都能用上。

[24:19] Dwarkesh Patel

This is especially important as systems get more authentic and require much longer prefixes in order to use tools and access files.In a recent benchmark, Crusoe was able to deliver up to 10 times faster time-to-first token and up to 5 times better throughput than VLLM.This is just one among many reasons that you should run your inference workload with Crusoe.And if you need GPUs for training, you don't need to switch clouds.Crusoe's got you covered there too.Go to crusoe.ai.thor cache to learn more.So, this gets at an interesting question about NVIDIA's clientele, where if 60% of your revenue is coming from these big five hyperscalers,you know, in a different era with different customers, let's say it's professors who are running experiments,and they are helped a bunch by they need CUDA.They can't use another accelerator.They need to just run PyTorch with CUDA and have everything optimized.But if you've got these hyperscalers, they have the resources to write their own kernels.In fact, they have to to get that extra last 5% that they need for their specific architecture.Anthropic, Google are mostly running their own accelerators or running TPUs and Tranium.But even OpenAI using GPUs has Triton, which they're like, we need our own kernels.

〔广告〕 随着系统越来越 agent 化、需要更长的前缀来调用工具和访问文件，这一点尤其重要。在最近一次基准测试中，Crusoe 相比 vLLM 做到了最快 10 倍的首 token 时延、最高 5 倍的吞吐。这只是你该把推理负载放在 Crusoe 上的众多理由之一。如果你需要 GPU 做训练，也不用换云，Crusoe 一样能覆盖。去 crusoe.ai/dwarkesh 了解更多。

〔Dwarkesh〕 这引出一个关于 NVIDIA 客户结构的有趣问题：如果你 60% 的收入来自那五大超大规模云厂商（hyperscaler）……在另一个时代、面对不同客户，比如是做实验的教授，他们非常需要 CUDA 的帮助，他们没法用别的加速器，他们就是要跑 PyTorch 加 CUDA、并且一切都被优化好。但如果是这些超大规模厂商，他们有资源自己写 kernel；实际上他们必须写，才能为自己的特定架构榨出最后那 5%。Anthropic、Google 大多在跑自研加速器或者 TPU 和 Trainium。就算是用 GPU 的 OpenAI 也有 Triton，他们的态度是：我们需要自己的 kernel。

[25:36] Dwarkesh Patel

So, down to CUDA C++, instead of using Kublaz and Nickel and everything,they've got their own stack, which compiles to other accelerators as well.And so, 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?

往下到 CUDA C++ 这一层，他们不用 cuBLAS、NCCL 那一套，而是有自己的技术栈，并且能编译到别的加速器上。所以如果你的大多数客户既有能力、也确实在做 CUDA 的替代品，那 CUDA 到底在多大程度上是「让前沿 AI 必须发生在 NVIDIA 上」的那个东西？

[25:58] Jensen Huang

Yeah. CUDA is a rich ecosystem.And so, if you want to build on any computer first, building on CUDA first is incredibly smart.And because the ecosystem is so rich, we support every framework.If you want to create custom kernels, if you need, for example, we contribute enormously to Triton.And so, the back end of Triton, huge amounts of NVIDIA technology.We're delighted to help every framework become as great as it can be.And there's lots and lots of frameworks.There's Triton, there's VLM, there's SGLang, and there's more, right?

CUDA 是一个丰富的生态。所以如果你想在任何一台计算机上做开发，先在 CUDA 上做是极其聪明的选择。因为生态如此丰富，我们支持每一个框架。如果你想写自定义 kernel——比如 Triton，我们对它的贡献极大，Triton 的后端里有海量 NVIDIA 的技术。我们很乐意帮每一个框架变得尽可能好。框架非常非常多，有 Triton、有 vLLM、有 SGLang，还有更多，对吧？

[26:38] Jensen Huang

And now there's a whole bunch of new reinforcement learning frameworks coming out.You know, you got Vero, you got Nemo, RL, you got a whole bunch of new.And then now with post-training and reinforcement learning, that entire area is just exploding, right?

现在还冒出一大堆新的强化学习框架，有 verl、有 NeMo RL，还有一堆新的。现在有了后训练（post-training）和强化学习，整个这块正在爆发式增长，对吧？

[26:53] Jensen Huang

And so, if you want to build on an architecture, building on a CUDA makes the most sense.Because you know that the ecosystem is great.You know that if something happens, it's more likely in your code and not in the mountain of code underneath.You know, don't forget the amount of code that you're dealing with when you're building these systems.When something doesn't work, was it you or was it the computer?

所以如果你要基于某个架构做开发，基于 CUDA 是最合理的。因为你知道生态很好；你知道一旦出问题，问题更可能在你自己的代码里，而不是在你底下那座代码大山里。别忘了，你在搭这些系统时要面对的代码量有多大。当有东西跑不通时，到底是你的问题，还是计算机的问题？

[27:15] Jensen Huang

You would like it always to be you and to be able to trust the computer.And obviously, we still have lots and lots of bugs ourselves.But our system is so well wrung out that you could at least build on top of the foundation.So, that's number one, is that the richness of the ecosystem, the programmability of it, the capability of it.The second thing is, if you were a developer and you were building anything at all,the single most important thing you want more than anything is install base.You want the software that you run to run on a whole bunch of other computers.You don't want to build a software.You're not building software just for yourself.You're building software for your fleet or for everybody else's fleet because you're a framework builder.And NVIDIA's CUDA ecosystem is ultimately its great treasure.We are now, I don't know how many, several hundred million GPUs.Every cloud has it.It goes back to A10, A100, H100, H200, you know, the L series, the P series.I mean, there's a whole bunch of them.And they're in all kinds of sizes and shapes.And if you're a robotics company, you want that CUDA stack to actually run in the robot itself.We're literally everywhere.And so, the install base says that once you develop the software, once you develop the model, it's going to be useful everywhere.

你当然希望永远是「你的问题」，希望能信任这台计算机。显然我们自己也还有非常非常多的 bug，但我们的系统被磨合得足够好，你至少可以在这个地基上往上搭。所以第一点是生态的丰富度、可编程性和能力本身。第二点是：如果你是开发者，不管你在做什么，你最想要的东西胜过一切，就是装机量（install base）。你希望你写的软件能跑在一大堆别的计算机上。你不是只为自己写软件，你是为你自己的机队、或者为别人的机队写软件，因为你是个框架作者。而 NVIDIA 的 CUDA 生态归根到底是它最大的宝藏。我们现在有——我不知道具体数——好几亿块 GPU，每一朵云上都有。往回追溯到 A10、A100、H100、H200，还有 L 系列、P 系列，一大堆，各种尺寸各种形态都有。如果你是家机器人公司，你希望那套 CUDA 栈能真的跑在机器人本体里。我们真的无处不在。所以装机量意味着：一旦你开发出软件、开发出模型，它在任何地方都有用。

[28:37] Jensen Huang

And so, the install base is just too incredibly valuable.And then lastly, the fact that we're in every single cloud makes us genuinely unique.Because, you know, you're an AI company and you're an AI developer, you're not exactly sure which CSP you're going to partner with and where you would like to run it.And we run it everywhere, including on-prem for you if you like.And so, I think that the richness of the ecosystem, the expansiveness of the install base, and the versatility of where we are, that combination makes CUDA invaluable.So, 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.Like, there's many people who they might matter for, the kind of person who can actually build their own software stack, who will make 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.And then 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.And so, oh, can everybody, can all the hyperscalers write these custom kernels for themselves?

所以装机量的价值实在太高了。最后一点，我们出现在每一朵云上，这让我们真正独一无二。因为你是一家 AI 公司、是个 AI 开发者，你不太确定自己会跟哪家云服务商（CSP）合作、想在哪儿跑。而我们哪儿都能跑，你要愿意，本地私有部署我们也能。所以生态的丰富、装机量的广度、以及我们所在位置的多样性，这个组合让 CUDA 无可替代。

〔Dwarkesh〕 这很有道理。我好奇的是，这些优势对你的主力客户是不是也同样重要？有很多人确实需要这些优势，但那些真正能自己搭全栈的人，才是你收入的大头。尤其如果我们走向一个世界：AI 在那些「验证回路很紧」、可以直接做强化学习的任务上变得特别强。那么「怎么在一个 scale-up（单机柜纵向扩展）域里最高效地写一个注意力或 MLP 的 kernel」，这是一个非常可验证的反馈回路。所以，是不是所有超大规模厂商都能自己写这些定制 kernel？

[29:56] Dwarkesh Patel

And they might still, NVIDIA still has great PICE 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 and memory bandwidth for a given dollar?

即便如此，NVIDIA 的性价比仍然很好，他们可能还是更愿意用 NVIDIA。但接下来的问题是：这是不是就退化成了「谁在单位美元下给出最好的规格——最好的算力、内存和内存带宽」？

[30:12] Dwarkesh Patel

Where historically NVIDIA has just had and still has, you know, the best margins in all of AI across hardware and software, 70% plus, because of this CUDA mode.And the question is, oh, can you sustain those margins if for most of your customers they can actually afford to build instead of the CUDA mode?

历史上 NVIDIA 一直有、现在也仍然有全 AI 领域最高的毛利率，软硬件加起来 70% 以上，靠的就是 CUDA 这条护城河。问题是：如果你的大多数客户其实有能力自建、绕开 CUDA 这条护城河，你还能维持这个毛利率吗？

[30:32] Jensen Huang

The number of engineers we have assigned to these AI labs is insane, working with them, optimizing their stack.And the reason for that is because nobody knows our architecture better than we do.And these architectures are not as general purpose as a CPU.The reason why a CPU is so, you know, a CPU is kind of like a Cadillac, you know.It's just always, you know, it's a nice cruiser.It never goes too fast.Everybody drives it pretty well, you know.It's got cruise control, you know, and everything is easy.But in a lot of ways, NVIDIA's GPUs are, accelerators are kind of like F1 racers.And, yeah, I could imagine everybody's able to drive it at 100 miles an hour.But it takes quite a bit of expertise to be able to push it to the limit.And we use a ton of AI to create the kernels that we have.And I'm pretty sure we're going to still be needed for quite some time.And so our expertise helps our AI labs partners get another 2x out of their stack easily, oftentimes.It's not unusual that we, you know, by the time that 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,

我们派驻到这些 AI 实验室、跟他们一起优化技术栈的工程师数量多到离谱。原因是没有人比我们更懂我们自己的架构。而且这些架构不像 CPU 那样通用。CPU 有点像凯迪拉克，永远是那种舒服的巡航车，从来不开太快，谁都能开得挺好，有定速巡航，一切都很轻松。但在很多方面，NVIDIA 的 GPU、加速器更像 F1 赛车。是的，我能想象每个人都能把它开到 100 英里/小时，但要把它推到极限需要相当多的专业能力。而且我们大量使用 AI 来写我们的 kernel。我很确定我们还会被需要相当长一段时间。我们的专业能力常常能帮 AI 实验室伙伴从他们的栈里再榨出 2 倍性能，这很常见。等我们优化完他们的栈、或者优化完某个特定 kernel，他们的模型提速 3 倍、2 倍、50% 都不稀奇。那是个巨大的数字，尤其当你面对的是他们手里整个机队的装机量时。

[32:07] Jensen Huang

of all the hoppers and blackwalls that they have.When you increase it by a factor of 2, that doubles the revenues.That directly translates to revenues.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 better performance TCO ratio.Not one company.And in fact, the benchmarks are out there.Dylan's, right, inference max is sitting out there for everybody to use.And not one TPU won't come.Tranium won't come.I encourage them to use inference max and demonstrate their incredible inference cost.It's really, really hard.Nobody wants to show up.MLperf, I would welcome Tranium 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.And so I think the reason why we're so successful is simply because our TCO is so great.There's a second, you say 60% of our customers are the top five, but most of that business is external.For example, most of AWS is, most of NVIDIA and AWS is for external customers, not internal use.Most of our customers at Azure, obviously, all of our customers are external.

他们手里所有的 Hopper 和 Blackwell，当你把它提升 2 倍，那就是收入翻倍，直接转化成收入。NVIDIA 的计算栈是全世界每一美元总拥有成本（TCO）下性能最好的，没有之一。没有人能向我证明今天世界上有任何一个平台的性能/TCO 比更好，一家都没有。而且基准测试就摆在那里，Dylan 的 InferenceMAX 就在那儿，谁都能用。没有一家 TPU 会来，Trainium 也不会来。我鼓励他们用 InferenceMAX 来展示他们惊人的推理成本，这真的、真的很难，没人愿意露面。MLPerf 上，我也欢迎 Trainium 来展示他们成天宣称的那 40%。我很想听他们证明 TPU 的成本优势。在我看来这说不通，完全说不通，从第一性原理上就说不通。所以我认为我们之所以这么成功，原因很简单，就是我们的 TCO 太好了。还有第二点，你说我们 60% 的客户是前五大厂商，但那部分生意大多是对外的。比如 AWS 上大部分 NVIDIA 的量是给外部客户的，不是内部自用；Azure 上我们的客户显然也全都是外部的。

[33:46] Jensen Huang

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.And 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.And if you were one of those AI startups, what architecture would you choose?

OCI 上我们的客户全是外部的，不是内部自用。他们之所以偏爱我们，是因为我们的触达面太广了。我们能把全世界最好的客户带给他们，因为这些客户全都建在 NVIDIA 上。而所有这些公司之所以建在 NVIDIA 上，是因为我们的触达面和通用性太强了。所以我认为这个飞轮真正的构成是：装机量、我们架构的可编程性、生态的丰富度，以及世界上有这么多 AI 公司——现在有数万家。如果你是其中一家 AI 创业公司，你会选哪个架构？

[34:26] Jensen Huang

You would choose an architecture that's the most abundant, or the most abundant in the world.The one has the largest install base, or the most largest install base, and one that has a rich ecosystem.And 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.And 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 and maximize the revenues for that data center.We have the highest tokens per watt architecture in the world.And then lastly, if your goal is to rent the infrastructure, we have the most customers in the world.And so that's the reason why the flywheel works.Interesting.I guess the question comes down to what is the actual market structure here?

你会选世界上最普及的那个架构，装机量最大的那个，生态最丰富的那个。这就是飞轮。这就是为什么——第一，我们的每美元性能太好，所以他们的 token 成本最低；第二，我们的每瓦性能全世界最高。如果我们的伙伴建了一座 1 吉瓦的数据中心，那座 1 吉瓦数据中心最好能产出最大化的收入和 token 数量，而 token 数量直接转化成收入。你希望它尽可能多产 token，把那座数据中心的收入最大化。我们有全世界每瓦 token 数最高的架构。最后，如果你的目标是把基础设施租出去，我们手上有全世界最多的客户。这就是飞轮转起来的原因。

〔Dwarkesh〕 有意思。我想问题归根到底是：这里真实的市场结构是什么？

[35:30] Dwarkesh Patel

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 if even through these five hyperscalers, really the people on Amazon using the computer, Anthropic, OpenAI, and these big foundation labs who can themselves afford and have the ability to make different accelerators work.No, I think your assumption is premise is wrong.Maybe.Yeah.But let me ask you a slightly different question.Come back and make me correct your premise.Okay.Let me just ask you a different question, which is, okay.But still make sure that 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, the premise, look.Let me just finish the question and then you can address it together.Yeah.So what do you think, if all these things are true about price performance and performance for what, 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.

因为就算有别的公司，本来也可能存在一个世界：有上万家 AI 公司，算力份额大致均分。但如果实际上是通过这五家超大规模厂商——真正在 Amazon 上用算力的是 Anthropic、OpenAI 这些大基础模型实验室，而他们自己既负担得起、也有能力把不同的加速器跑起来……

〔Jensen〕 不，我认为你的假设、你的前提是错的。

〔Dwarkesh〕 也许吧。那我换个稍微不同的问题问。

〔Jensen〕 你回头得让我把你的前提纠正过来。

〔Dwarkesh〕 好。那我问个不同的问题——

〔Jensen〕 但一定要让我回来纠正，因为这对 AI 太重要了，对科学的未来太重要了，对整个行业的未来太重要了。那个前提，那个前提，你看——

〔Dwarkesh〕 让我先把问题问完，然后你可以一起回答。如果关于性价比、每瓦性能等等这些说法都成立，那你觉得为什么会出现这种情况：比如 Anthropic 几天前刚宣布跟博通（Broadcom）和 Google 达成了多吉瓦级的 TPU 协议，占了他们大部分算力。

[36:45] Dwarkesh Patel

Obviously, for Google, TPUs are majority of the compute.So if I look at these big AI companies, it seems like a lot of their companies, there was some point where it was all NVIDIA and now it's not.And so I'm curious how to square, if these things are true on paper, why are they going with other accelerators?

而 Google 这边，TPU 显然占了大部分算力。所以我看这些大 AI 公司，感觉很多公司都曾经是全 NVIDIA，现在不是了。所以我很好奇怎么解释：如果纸面上这些都成立，他们为什么要转向别的加速器？

[37:03] Jensen Huang

Yeah.Anthropic is a unique instance and not a trend.Without Anthropic, why would there be any TPU growth at all?It's 100% Anthropic.Without Anthropic, why would there be any Tranium 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.But OpenAI deals with AMD.They're building their own Titan accelerator.Yeah, but they're mostly, I think we could all acknowledge, they're vastly NVIDIA.And we're going to still do a lot of work together.Yeah.Yeah.And we're not, 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?

Anthropic 是一个独特的个案，不是趋势。没有 Anthropic，TPU 的增长从哪来？100% 来自 Anthropic。没有 Anthropic，Trainium 的增长从哪来？100% 来自 Anthropic。我觉得这点大家都相当清楚、也都理解。并不是有一大堆 ASIC 的机会，只有一个 Anthropic。

〔Dwarkesh〕 但 OpenAI 也跟 AMD 有合作，他们也在造自己的 Titan 加速器。

〔Jensen〕 是，但我想我们都能承认，他们绝大部分还是 NVIDIA。我们还会一起做很多事。而且我不会因为别人用别的东西、去试试别的而觉得被冒犯。如果他们不去试试别的，怎么知道我们的有多好？

[37:54] Jensen Huang

You know, and sometimes you've got to be reminded of it.And we have to continuously earn the position that we're in.There are always big claims.And 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.And it's not that easy building something better than NVIDIA.It's not sensible, actually.You know, NVIDIA has got to be missing something seriously.You know, because 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.I guess their logic is that, 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.No, no, no.Don't forget.But even in ASICs, margin's really quite high.NVIDIA's margin's 70%, let's say.But in ASIC margin, 65%.What are you really saving?

有时候你是需要被提醒一下的。而我们必须持续赢得我们现在的位置。总是有各种大话。而且看看有多少 ASIC 项目被取消了。你要造一个 ASIC，你还是得造出比 NVIDIA 更好的东西，而造出比 NVIDIA 更好的东西并不容易，实际上也不划算。NVIDIA 得是漏了什么大招才行。因为我们的规模、我们的速度——我们是全世界唯一一家每一年都往外出货、每一年都有大跃进的公司。

〔Dwarkesh〕 我猜他们的逻辑是：不需要更好，只要别差过 70% 就行，因为他们本来就在为你的 70% 毛利买单。

〔Jensen〕 不不不，别忘了，就算是 ASIC，毛利也相当高。假设 NVIDIA 毛利是 70%，ASIC 的毛利是 65%，你到底省下了什么？

[38:51] Dwarkesh Patel

Oh, you mean from Broadcom or something like that?Yeah, sure.You got to pay somebody.Yeah.And so I think the ASIC margins are incredibly good from what I can tell.And they believe it so too.And so they're quite proud of their incredible ASIC margins.And so you asked the question why.A long time ago, we just didn't have the ability to do it.And at the time, I didn't deeply internalize how difficult it would be to build a foundation AI lab like OpenAI and Anthropic.

你是说博通这类公司的毛利？

〔Jensen〕 对啊，你总得付钱给某个人。所以据我所知，ASIC 的毛利好得不得了，他们自己也这么认为，他们对自己那惊人的 ASIC 毛利相当自豪。所以你问为什么。很久以前，我们就是没有那个能力去做。而且当时我没有深刻地内化「建一个像 OpenAI、Anthropic 这样的基础模型实验室有多难」。

[39:32] Jensen Huang

And the fact that they needed huge investments from the supplier themselves.We just weren't in a position to make the multibillion dollar investment into Anthropic so that they could use our compute.But Google and AWS were.And they put in huge investments in the beginning so that Anthropic, in return, use their compute.We just weren't in a position to do so at the time.Nor did I, 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 five, $10 billion of investment into an AI lab with the hopes of it turning out to be Anthropic.And so that was my miss.But even if I understood it, I don't think we would have been in a position to do that at the time.But I'm not going to make that same mistake again.And 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.But we just weren't at the time able to do so.If I could rewind everything, NVIDIA could have been as big back then as we are now.I would have been more than happy to do it.This is actually quite interesting, which is for many years, NVIDIA has been this, the company in AI making money, making lots of money.

以及他们需要供应商本身给出巨额投资这件事。我们当时没有能力向 Anthropic 投几十亿美元、好让他们用我们的算力，但 Google 和 AWS 有。他们在早期投了巨额资金，作为交换 Anthropic 用他们的算力。我们当时就是做不到。而且我要说，我的失误在于我没有深刻内化他们真的别无选择——VC 是绝不会往一个 AI 实验室里投 50 亿、100 亿美元、指望它长成 Anthropic 的。这是我看漏的地方。但就算我当时懂了，我觉得我们当时也没有能力那么做。不过我不会再犯同样的错了。我很高兴能投资 OpenAI，很高兴能帮他们扩张，我相信这是必要的。后来当我有能力了、Anthropic 来找我们时，我也很高兴成为投资人、高兴帮他们扩张。只是我们当时做不到。如果一切能倒带，如果 NVIDIA 当年就有今天这么大，我会非常乐意去做。

〔Dwarkesh〕 这其实挺有意思的：很多年来，NVIDIA 一直是 AI 里那家赚钱的公司，赚了很多钱。

[41:17] Dwarkesh Patel

And 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.And so over these many years, you were giving them the compute, you saw where I was headed, and then they were worth like one-tenth what they are 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.

现在你在把它投出去。有报道说你已经投了 OpenAI 多达 300 亿美元、Anthropic 100 亿美元。但他们的估值已经涨上去了，而且我相信还会继续涨。所以这么多年里，你把算力给了他们，你看到了方向，而他们一两年前的估值只有现在的十分之一，有些情况下甚至就是一年前。而你手上有这么多现金。本来存在一个世界：要么 NVIDIA 自己成为一家基础模型实验室、做一笔巨额投资把它做成；要么用当年的估值、而不是现在的估值，早早地做成你今天做的这些交易。

[41:57] Dwarkesh Patel

And you had the cash to do it.So I am curious, actually, why not have done it earlier?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 have 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.How so? Like a cash thing or just?

而且你有钱去做。所以我确实好奇，为什么没有更早做？

〔Jensen〕 我们一有能力就做了。我们一有能力就做了。如果我能，我会更早做。在 Anthropic 需要我们做的那个时点，我们就是没有能力做，那不在我们当时的认知里。

〔Dwarkesh〕 怎么讲？是现金的问题还是？

[42:23] Jensen Huang

Yeah, the level of investment.We never invested outside the company at the time, and not that much.And we didn't realize we needed to.I always thought that they could just go raise 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.And I recognize that now.I didn't know it then.But that's their genius.That's why they're smart.And so they realized that then that they had to do something like that.And I'm delighted that they did.And 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.I guess you still are making a ton of money, and you're making way more money quarter after quarter.It's still okay to have regrets.So the question still arises, okay, well, now that we're here, and you have all this money that you keep making, what should NVIDIA be doing with it?

是投资的量级问题。我们当时从来不在公司外部做投资，也没有那么大的量。而且我们没意识到我们需要那么做。我一直以为他们可以去融 VC 啊，天哪，所有公司不都这样吗。但他们想做的事没法通过 VC 完成，OpenAI 想做的事没法通过 VC 完成。我现在认识到了，当时不知道。但那正是他们的天才之处，那正是他们聪明的地方。他们当时就意识到必须做那样的安排。我很高兴他们那么做了。即便我们害得 Anthropic 不得不去找别人，我依然为这件事的发生感到高兴。Anthropic 的存在对世界是好事，我为此高兴。

〔Dwarkesh〕 我猜你还是赚了一大笔钱，而且每个季度赚得越来越多。不过有遗憾也没关系。所以问题还是在：既然我们已经走到这儿，你还在不停地赚钱，NVIDIA 该拿这些钱做什么？

[43:36] Dwarkesh Patel

And there's one answer which says, look, 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 AML is getting better.The value that they generate through tokens is increasing, but they're expensive to set up.NVIDIA has the money due to the CapEx.And in fact, it's been reported you're backstopping core.We have up to $6.3 billion and have invested 2B.But yeah, why doesn't NVIDIA become a cloud themselves?

有一个答案是：现在冒出了一整个「中间商生态」，把资本开支（capex）转换成运营开支（opex），好让这些实验室能租算力。因为芯片非常贵，它们在生命周期里能赚很多钱——因为经济寿命在变好、它们通过 token 产生的价值在增加——但把它们架起来很贵。而 NVIDIA 有这笔钱能覆盖资本开支。事实上有报道说你在给 CoreWeave 兜底。〔Jensen〕 我们承诺了最高 63 亿美元，已经投了 20 亿。〔Dwarkesh〕 那为什么 NVIDIA 不干脆自己做云呢？

[44:11] Dwarkesh Patel

Why doesn't it become a hyperscaler themselves and rent this compute out?You have all this cash to do it.This is a philosophy of the company, and I think it's wise.We should do as much as needed, as little as possible.And 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 MVLink the way we built, if we didn't build the whole stack, if we didn't create the ecosystem the way we did it, 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.If we didn't create all the CUDAX libraries so that they're all domain-specific, you know, this is several, 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, structure data processing or vector data processing, if we didn't create them, nobody would.And I am completely certain of that.We created a library for computational lithography called Coolitho.

为什么不自己成为一家超大规模云厂商、把这些算力租出去？你有这么多现金可以做。

〔Jensen〕 这是公司的一条哲学，我认为它是明智的：我们应该「做到必要的那么多，同时尽可能少做」。意思是，我们在搭建计算平台上做的工作，如果我们不做，我真心相信它就不会有人做。如果我们不冒我们冒的那些险，如果我们不像那样去造 NVLink，如果我们不去做整个栈，如果我们不像那样去建生态，如果我们不在大部分时间亏钱的情况下把 CUDA 坚持二十年，如果我们不做，没有别人会做。如果我们不做出所有那些 CUDA-X 领域专用库——这是十几年前的事，我们推进领域专用库，是因为我们意识到如果我们不去创造这些领域专用库，无论是给光线追踪、图像生成，还是 AI 的早期工作、这些模型，如果我们不为数据处理、结构化数据处理、向量数据处理做出来，就没人会做。我对此完全确信。我们做过一个叫 cuLitho 的计算光刻库。

[45:26] Jensen Huang

If we didn't create it, nobody would have.And so accelerated computing wouldn't advance the way it has if we didn't do what we did.And 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'd show up.And 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.And everything I do, I do it with that lens.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 where they are today.Now they are, they're doing fantastically.Is that a business model where, no, we should do as much as needed, as little as possible.And so we invest in our ecosystem because I want our ecosystem to thrive.And I want the architecture and I want 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.

如果我们不做，没人会做。所以如果我们不做我们做过的这些事，加速计算不会像今天这样推进。所以这些事我们应该做，我们应该把公司、把全部力量、全心全意投进去做这些。然而，世界上有的是云。如果我不做，总会有人冒出来。所以按照「做到必要的那么多、但尽可能少做」这条配方和哲学——尽可能少做——这条哲学今天就存在于我们公司里，我做的每件事都用这个镜头去看。就云而言，如果我们不支持 CoreWeave 存在，这些新云（NeoCloud）、这些 AI 云就不会存在。如果我们不帮 CoreWeave，它们不会存在；如果我们不支持 Nscale，他们不会有今天；如果我们不支持 Nebius，他们不会有今天。现在他们都做得很棒。这是不是一种商业模式？不，我们应该做到必要的那么多、尽可能少做。所以我们投资我们的生态，因为我想让我们的生态繁荣。我想让这个架构、让 AI 能连接到尽可能多的行业、尽可能多的国家，让这个星球有可能建立在 AI 之上、建立在美国技术栈之上。

[46:54] Jensen Huang

And so that vision, I think, is exactly what we're pursuing.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.And this is another thing that we do.We don't pick winners.And we like, we need to support everyone.And it's part of our joy of doing so.It's an imperative to our business.But we also go out of our way not to pick winners.And so when I invest in one of them, I invest in all of them.Why do you go out of your way to not to pick winners?

所以那个愿景，我认为正是我们在追求的。你提到的一点是：现在有这么多了不起的基础模型公司，我们尽量都投。这是我们做的另一件事：我们不挑赢家。我们需要支持每一个人，这是我们做这件事的乐趣之一，也是我们业务的必然要求。但我们也刻意避免挑赢家。所以当我投资其中一家时，我就投资所有家。

〔Dwarkesh〕 你为什么要刻意不去挑赢家？

[47:29] Jensen Huang

Because it's not our job to, number one.Number two, when NVIDIA first started, there were 60 graphics companies, 60 3D graphics companies.We are the only one that survived.If you would have taken those 60 companies, 60 graphics companies, and asked yourself which one was going to make it, NVIDIA would be the top of that list not to make it.You know, 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 in the wrong solution.And everybody would have counted us out.And here we are.And so I have enough humility to recognize that, you know, don't pick winners.Either let them all take care of themselves or take care of all of them.One thing I didn't understand is you said, look, we're not prioritizing these neoclouts just because there are neoclouts and we want to prop them up.But you also said, you listed a bunch of neoclouts and you said they wouldn't exist if it wasn't for NVIDIA.And so how are those two things compatible?

第一，因为这不是我们的工作。第二，NVIDIA 刚起步的时候，有 60 家图形公司、60 家 3D 图形公司，只有我们活下来了。如果你当年拿那 60 家公司来问自己「哪家能活下来」，NVIDIA 会排在「最不可能活下来」的名单最前面。这远在你出生之前——NVIDIA 的图形架构是精确地错的，不是错一点点，我们造了一个精确错误的架构，对开发者来说根本没法支持，它绝无可能成功。我们从很好的第一性原理去推理，但最后落到了错误的解上。所有人都会把我们排除掉。然而我们就在这里。所以我有足够的谦卑去认识到：别挑赢家。要么让他们各自照顾自己，要么照顾所有人。

〔Dwarkesh〕 有一点我没听懂：你说我们不是因为新云是新云、想把他们扶起来才优先给他们。但你又列了一堆新云，说要不是因为 NVIDIA 他们不会存在。这两件事怎么兼容？

[48:50] Jensen Huang

First of all, they need to want to exist.And they come to ask us for help.And when they want to exist and they have a business plan and, you know, they have expertise and, you know, they have the passion for it,they obviously have to have some capabilities themselves.But if at the end of the day they need some investment and we're to get it off the ground, we would be there for them.But the sooner they get their flywheel going, you know, your question was, do we want to be in the financing business?

首先，他们得自己想存在。他们是来找我们求助的。当他们想存在、有商业计划、有专长、有热情，他们显然自己也得有些能力。但如果到最后他们需要一笔投资才能起步，我们会在那儿。不过他们的飞轮越早转起来越好。你的问题是，我们想不想做金融生意？

[49:21] Jensen Huang

The answer is no.Yeah.We don't want to be – we want to – because there are people in the financing business.And we'd rather work with all of the people who are in the financing business than to be a financier ourselves.And so I think our goal is to focus on what we do, keep our business model as simple as possible, support our ecosystem.When someone like OpenAI needs an investment of $30 billion scale because it's still before their IPO and we deeply believe in them.We deeply believe that – I deeply believe that they're going to be – 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.And they have everything – they have the wind at their back.Let's support them and let them scale.And so those investments will 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.I spend way too much time copy-pasting text back and forth from Google Docs to chatbots.And so I built what's basically a cursor for writing, which operates the way I think an AI co-researcher should operate.I can tag it and it can talk with me through inline comment threads and help me dig deeper and brainstorm.

答案是不想。因为已经有人在做金融生意了。我们宁可跟所有做金融的人合作，也不想自己去当金融家。所以我认为我们的目标是聚焦我们做的事，把商业模式保持得尽可能简单，支持我们的生态。当像 OpenAI 这样的公司需要 300 亿美元量级的投资、因为他们还没上市而我们又深深相信他们——我深深相信他们会成为——好吧，他们今天已经是一家非凡的公司了，他们会成为一家了不起的公司。世界需要他们存在，世界希望他们存在，我希望他们存在。他们万事俱备、顺风顺水。那就支持他们，让他们规模化。所以那些投资我们会做，因为他们需要我们做。但我们不是在试图做尽可能多，我们是在试图做尽可能少。

〔广告〕 我花了太多时间在 Google Docs 和聊天机器人之间来回复制粘贴文本，所以我做了一个基本上是「写作版 Cursor」的东西，它按我认为 AI 协作研究员该有的方式工作。我可以 @ 它，它能通过行内评论线程跟我对话，帮我挖得更深、一起头脑风暴。

[50:39] Dwarkesh Patel

I built this entire thing over the weekend with Cursor and their new Composer 2 model.With a lot of agentic coding tools, I feel like I have no idea what's going on under the surface.I just have to relinquish control and hope for the best.But Cursor, let me try a bunch of different ideas while staying on top of the implementation.I did most of my brainstorming in the agents window.And after I got some basic files in place, I used a diff window to track changes.The few times that I needed to make a quick tweak by hand, I just used the editor.If you want to try my AI co-researcher yourself, I've linked the GitHub repo in the description.And if you have a tool that you've been wanting to build, you should make it happen.Go to cursor.com slash thwarkash to get started.This might be sort of 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.We have a shortage of GPUs.Yes.Yeah.And NVIDIA is known for divvying up the scarce allocation, not just based on highest bidder, but rather on, hey, we want to make sure that these Neo clouds exist.Let's give some to CoreWeave.Let's give some to Crusoe.

〔广告〕 我用 Cursor 和他们的新模型 Composer 2，一个周末就把这整个东西做出来了。用很多 agent 编程工具时，我感觉完全不知道底下在发生什么，只能交出控制权、祈祷结果还行。但 Cursor 让我可以试一堆不同的想法，同时又对实现保持掌控。我大部分头脑风暴是在 agents 窗口里做的，等基本文件到位后，我用 diff 窗口跟踪改动。少数几次需要手动快速改一下，我就直接用编辑器。如果你想自己试试我这个 AI 协作研究员，我把 GitHub 仓库链接放在简介里了。如果你有一直想做的工具，就动手做出来吧。去 cursor.com/dwarkesh 开始。

〔Dwarkesh〕 这可能是个显而易见的问题：我们已经在 GPU 短缺的状况里过了很多年，而且因为模型越来越好，短缺变得更严重了。我们缺 GPU。〔Jensen〕 是的。〔Dwarkesh〕 而 NVIDIA 以「分配稀缺产能时不是单纯按出价高低，而是要确保这些新云能存在」著称——分一些给 CoreWeave，分一些给 Crusoe。

[51:41] Dwarkesh Patel

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?No.No.Yeah.Your premise is just wrong.Yeah.Yeah.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.And so until we get a PO, what are we going to do?

分一些给 Lambda。这对 NVIDIA 为什么是好事？首先，你同意「把市场打散」这个描述吗？

〔Jensen〕 不同意。

〔Dwarkesh〕 不同意。

〔Jensen〕 你的前提就是错的。我们对这些事足够审慎，我们非常审慎。首先，如果你不下采购订单（PO），说再多也没用。所以在拿到 PO 之前，我们能做什么？

[52:12] Jensen Huang

And 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.And so we align ourselves with demand and supply and things like that through forecasting.Okay.That's job number one.Number two, everybody who, you know, we've tried to forecast with as many people as possible, but in the final analysis, you still have to place an order.And maybe for whatever reason you didn't place your order, what can I do?

所以第一件事是，我们非常努力地跟所有人一起把预测做出来，因为这些东西建起来要很长时间，数据中心建起来也要很长时间。所以我们通过预测把自己和需求、供给对齐。好，这是第一号工作。第二，我们尽可能跟更多人一起做预测，但最终你还是得下订单。如果因为某种原因你没有下单，我能怎么办？

[52:46] Jensen Huang

And so at some point, first in, first out.But beyond that, if you're not ready because your data center is 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.And so we might do some adjustments there.Aside from that, the prioritization is first in, first out.Yeah.You got to place a PO.If you don't place a PO.Now, of course, there are stories about that.You know, like for example, all of this kind of started from, it was an article about Larry and Elon having dinner with me where they begged for GPUs.That never happened.And we absolutely had dinner.We absolutely had dinner.And it was a wonderful dinner.In no time did they beg for GPUs.And so they just had to place an order.And once they placed an order, we do our best to get the capacity to them.We're not complicated.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 can have a certain time.But it still doesn't sound like High Spitter just gets it.Is there a reason to do it?

所以在某个点上，就是先到先得。除此之外，如果你还没准备好——你的数据中心没建好、或者某些配套没到位让你立不起数据中心——我们可能会决定先服务另一个客户，那只是在最大化我们自己工厂的产出。所以我们可能会在那儿做些调整。除此之外，优先级就是先到先得。你得下 PO。如果你不下 PO……当然，关于这个有各种故事。比如这一切最初是源于一篇文章，说 Larry（拉里·埃里森）和 Elon 跟我吃饭时求我给 GPU。那从来没发生过。我们确实一起吃了饭，那是一顿很愉快的晚餐，但他们从来没求过 GPU。他们只需要下个订单。一旦下了订单，我们会尽全力把产能给到他们。我们不复杂。

〔Dwarkesh〕 好，所以听起来是有一个排队队列，然后根据你的数据中心是否就绪、以及你什么时候下单来定时间。但听起来还是不像「出价最高的人拿走」。有没有理由这么做？

[54:13] Jensen Huang

We never do that.Okay.We never do that.Why not just do High Spitter?Because it's a bad business practice.You set your price.You set your price.And then people decide to buy it or not.And I understand that others in the chip industry change their prices when demand is higher.But we just don't.We just don't.That's just never been a practice of ours.You can count on us.You know, I prefer to be dependable, to be the foundation of the industry.And you don't need to second guess.You know, if I quoted you a price, we quoted you a price.That's it.And if demand goes through the roof, so be it.And on the other end, that's why you have a productive relationship with TSMC, right?

我们从不那么做。〔Dwarkesh〕 好。〔Jensen〕 我们从不那么做。〔Dwarkesh〕 为什么不价高者得？

〔Jensen〕 因为那是糟糕的商业习惯。你定你的价，你定了价，然后别人决定买不买。我知道芯片行业里别人会在需求高的时候改价，但我们就是不。我们就是不，那从来不是我们的做法。你可以指望我们。我更愿意做那个可靠的、做行业地基的角色，让你不需要反复猜疑。如果我给你报了价，那就是报了价，就这样。如果需求爆表，那就爆表吧。

〔Dwarkesh〕 另一头，这也是你跟台积电关系顺畅的原因，对吧？

[55:04] Jensen Huang

Yeah.Yeah, yeah.NVIDIA has been in business.We've been doing business with them for, I guess, coming up on 30 years.And NVIDIA and TSMC don't have a legal contract.There is always some rough justice.And sometimes I'm right.Sometimes I'm wrong.Sometimes I got a better deal.Sometimes I got a worse deal.But overall and the whole, the relationship is incredible.And I can completely trust them.I can completely depend on them.And one of the things that you can count on with NVIDIA is that next year, this year, Verrubin is going to be incredible.Next year, Verrubin Ultra will come.The year after that, Feynman will come.And the year after that, I haven't introduced the name yet.And so every single year you can count on us.And this is an, 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 of, I can bet my entire business that you will be here for me every single year.Your cost, your token cost will decrease by an order of magnitude every single year.I can count on it, but I can count on the clock.Well, I just said something about TSMC.No other foundry in history can you possibly say that.You can say that about NVIDIA today.

对，对对。NVIDIA 跟他们做生意快 30 年了。而 NVIDIA 和台积电之间没有法律合同。总会有一种「粗糙的公道」：有时我占理，有时我不占理；有时我拿到更好的条件，有时更差。但总体而言，这段关系好得不得了，我可以完全信任他们、完全依赖他们。而 NVIDIA 身上你可以指望的一件事是：今年 Vera Rubin 会很棒，明年 Vera Rubin Ultra 会来，再下一年 Feynman 会来，再下一年的名字我还没公布。所以每一年你都可以指望我们。你去全世界找一个 ASIC 团队看看——挑一个你敢说「我可以把整个农场押上、可以把我整个生意押上，你每一年都会在这儿等着我，我的 token 成本每年会下降一个数量级」的 ASIC 团队。我可以指望它，我可以像指望时钟一样指望它。刚才我也这么说过台积电。历史上没有任何别的晶圆厂能让你这么说，但今天你可以对 NVIDIA 这么说。

[56:30] Jensen Huang

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.If 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 hundred billion dollar AI factory, no problem.We're the only company in the world where you can say that today.I can say that about TSMC as well.I want to buy one, buy one billion, no problem.We just got to go through the process of planning for it and, you know, all the things that mature people do.You know?

每一年你都可以指望我们。你想买 10 亿美元的 AI 工厂算力，没问题；想买 1 亿美元，没问题；想买 1,000 万美元、或者就一个机柜，没问题；或者就一块显卡，好的，没问题。你想下一张 1,000 亿美元 AI 工厂的订单，没问题。我们是当今世界上唯一一家你能这么说的公司。台积电我也能这么说：我想买一个、买十亿个，没问题。我们只需要走一遍规划的流程，做所有成熟的人会做的事，对吧？

[57:11] Jensen Huang

And so, so I, I think the, the, this ability for NVIDIA to be the foundation of the world's AI industry, this is a, this is a position that has taken us decade, several dec, couple of decades to arrive at.So, so I think that's an enormous commitment, enormous dedication.And, um, the stability of our company, the consistency of our company is really, really important.Okay.I want to ask about China.Yeah.And I always like to take, uh, I don't, I actually don't know what I think about whether it's good to sell ships to China or not, but I like played devil's advocate against my guests.So when Dario was on, who supports export control, I asked him, well, why can't America and China both have country of geniuses in a data center?

所以我认为，NVIDIA 能成为世界 AI 产业地基的这个能力，是我们花了几十年才到达的位置。所以我认为这是巨大的承诺、巨大的投入。而我们公司的稳定性、一致性真的非常非常重要。

〔Dwarkesh〕 好，我想问问中国。〔Jensen〕 好。〔Dwarkesh〕 我总喜欢——其实我自己也不确定卖芯片给中国到底好不好，但我喜欢跟嘉宾唱反调。所以 Dario（Anthropic CEO 达里奥·阿莫代伊）来的时候，他支持出口管制，我就问他：为什么美国和中国不能都拥有「数据中心里的天才之国」？

[57:52] Dwarkesh Patel

But since, um, you're on the opposite side, I'll ask you in the opposite way.Um, and look, one way to think about it is, uh, Anthropic actually announced a couple of days ago, mythos preview, this model mythos of not even releasing publicly because they say it has such cyber offensive capabilities that we don't think the world is ready until we get, 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 open BSD, which is this operating system that has been specifically designed to not have zero days and it found one, uh, for 27 years it's existed.Um, and 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, oh, is that a threat to American companies to American national security?

而既然你站在相反的一边，我就反过来问你。有一种看法是这样：Anthropic 几天前刚宣布了 Mythos 预览版，这个模型他们甚至没有公开发布，因为他们说它的网络攻击能力太强，在确保那些零日漏洞被补上之前，他们认为世界还没准备好。他们说它在每一个主流操作系统、每一个浏览器里都找到了数以千计的高危漏洞。它在 OpenBSD 里找到了一个——那是个专门为「没有零日漏洞」而设计的操作系统，存在了 27 年，它找到了一个。所以如果中国公司、中国的实验室和中国政府拿到了能训练出 Claude Mythos 这种网络攻击能力模型的 AI 芯片，还能用更多算力跑上百万个实例，那问题就是：这对美国公司、对美国国家安全是不是威胁？

[58:46] Jensen Huang

Uh, first of all, um, mythos was, was, uh, trained on fairly mundane capacity and a fairly mundane amount of it, um, by an extraordinary company.Uh, and so the amount of capacity and the type of compute that's, it was trained on is abundantly available in China.And so you just have to first realize that chips exist in China.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, most of them are Chinese.They have 50% of the world's AI researchers.And so the question is, if you're concerned about them, what is the, considering all the assets they already have?

首先，Mythos 是用相当平常的产能、相当平常的量训出来的，只不过是被一家非凡的公司训出来的。所以它训练所用的产能规模和算力类型，在中国是大量存在的。你首先得认识到：芯片在中国是存在的。他们制造了全世界 60% 甚至更多的主流芯片，这对他们是个非常大的产业。他们有一些全世界最顶尖的计算机科学家。你也知道，这些 AI 实验室里大多数 AI 研究员都是中国人，他们拥有全世界 50% 的 AI 研究员。所以问题是，如果你担心他们，考虑到他们已经拥有的全部资产，那什么才是……

[59:47] Jensen Huang

They have an abundance of energy.They have plenty of chips.They got most of the AI researchers.If you're worried about them, what is the best way to create a safe world?Well, victimizing them, um, uh, turning them into an enemy, uh, likely isn't the best answer.They are an adversary.We want United States to win.Um, but I think having a, having a dialogue and having research dialogue is probably the safest thing to do.This is an area that, 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 how to, what not to use the AI for.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?

他们有充沛的能源，他们有大量的芯片，他们有大多数 AI 研究员。如果你担心他们，创造一个安全世界的最好方式是什么？把他们受害者化、把他们变成敌人，很可能不是最好的答案。他们是对手，我们希望美国赢。但我认为，保持对话、保持研究层面的对话，很可能是最安全的做法。而正因为我们当下把中国当对手的态度，这块恰恰是明显缺失的。我们的 AI 研究员和他们的 AI 研究员真正对上话，这是必要的；我们双方就「不该把 AI 用于什么」达成一致，这是必要的。至于在软件里找 bug——当然，那本来就是 AI 该干的事。它会在很多软件里找到 bug 吗？

[1:00:54] Jensen Huang

Of course.There's lots and lots of bugs.There are lots of bugs in the AI software.And so, um, that's what AI is supposed to do.And I'm delighted that, that, uh, uh, AI has reached a level where it could help us be so much more productive.Um, one of the things that, that, um, is, is, uh, under, under emphasized is the richness of ecosystem around cybersecurity, AI cybersecurity and AI security and AI privacy and AI safety.That whole ecosystem of AI startups that are trying to create this future for us, where, 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.And the idea that you're going to have an AI agent running around with nobody watching after it is kind of insane.And so, uh, 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, um, AI safe.And, uh, and, and, and so one of the things that we need to make sure that we do is we keep the, the open source ecosystem vibrant.

当然会。bug 多得是。AI 软件里也有很多 bug。所以那就是 AI 该做的事，我很高兴 AI 达到了能让我们生产力提升这么多的水平。有一件被低估的事，是围绕网络安全的生态有多丰富——AI 网络安全、AI 安全、AI 隐私、AI 安全性。整个那一批 AI 创业公司正在为我们创造这样一个未来：你有一个非常强的 AI agent，周围有上千个 AI agent 在保护它、保障它的安全。那个未来一定会到来。而「让一个 AI agent 到处跑却没人看着它」这种想法，是相当疯狂的。所以我们非常清楚，这个生态需要繁荣。而事实证明，这个生态需要开源，这个生态需要开放模型，他们需要开放的技术栈，好让所有这些 AI 研究员、所有这些优秀的计算机科学家去构建同样强悍的 AI 系统、去保障 AI 的安全。所以我们必须确保的一件事，就是让开源生态保持活力。

[1:02:28] Jensen Huang

And, um, and that can't be ignored.That can't be ignored.And, and, and a lot of that is coming out of China.Um, uh, we, we, we added, we added not suffocate that, you know, with respect to China, we want to have, of course, we want United States to have as much computing as possible.Uh, we're, we're limited by energy.Um, but you know, we got a lot of people working on that and we, we had to not make energy a, a bottleneck for our country.Um, but what we also want is we want 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.And it would be extremely foolish to create two ecosystems, the open source ecosystem.And it only runs on the Chinese tech tech, a foreign tech stack and a closed ecosystem.And that runs on the American tech stack.I think that that would be, that would be a horrible outcome for the United States.Since there are a lot of things, let me just triage the, um, response.I mean, 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 seven nanometer, uh, they don't have UVs because of chip making expert controls.

这一点不能被忽视，不能被忽视。而其中很大一部分是从中国出来的。就中国而言，我们当然希望美国拥有尽可能多的算力，我们受限于能源，但我们有很多人在解决这个问题，我们绝不能让能源成为国家的瓶颈。但我们同样想要的是：确保世界上所有 AI 开发者都在美国技术栈上开发，并把 AI 的进步——尤其是开源的那部分——贡献回美国生态。搞出两个生态是极其愚蠢的：开源生态只跑在中国技术栈、外国技术栈上，闭源生态跑在美国技术栈上。我认为那对美国来说是可怕的结果。

〔Dwarkesh〕 有很多点，让我把回应分个类。我觉得回到黑客攻击那个算力差距的问题：是的，他们有算力，但有估算说，因为他们卡在 7 纳米、因为芯片制造的出口管制他们没有 EUV……

[1:03:55] Dwarkesh Patel

The amount of flops they're about to actually produce, they have like one 10th, the amount of flops that the U S has.And so with that, could they train eventually a model like Mythos?Yes.But the question is because we have more flops, uh, American labs are able to get to these level of capabilities first.And 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 give them access to it.They're going to patch up all their vulnerabilities.And now we release it.Furthermore, if they, even if they train a model like this, the ability to deploy that scale, you know, 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.And in fact, the fact that they have so many AI researchers are so good is the thing that makes it so scary because what is it that makes those engineer researchers more productive is compute.Um, if you talk to any AI lab in America, they say the thing that's bottleneck and they miss compute.So, and there are quotes from DeepSeq founder or, uh, Quen leadership or whatever.They say like the thing we're bottlenecked on is compute.

他们实际能产出的算力大概只有美国的十分之一。基于这一点，他们最终能训出像 Mythos 那样的模型吗？能。但问题是，因为我们有更多算力，美国的实验室能先到达这些能力水平。而因为 Anthropic 先到了，他们说，好，我们先压一个月，让所有这些美国公司先拿到访问权、把他们的漏洞都补上，然后我们再发布。再者，就算他们训出了这样的模型，部署的规模也是问题——如果你有一个网络黑客，他有一百万个实例比有一千个实例危险得多。所以推理算力真的非常重要。而且，正因为他们有这么多这么优秀的 AI 研究员，这才更可怕，因为让这些工程师和研究员更高效的东西正是算力。你去问美国任何一家 AI 实验室，他们都说卡住他们的是算力不够。而且我们有 DeepSeek 创始人或者通义千问（Qwen）负责人的引述，他们说我们的瓶颈就是算力。

[1:04:53] Dwarkesh Patel

Um, so then the question is, isn't it better that we get to get American companies because they have more compute, get to, get, get to the level of spot or mythos level capabilities.And so, um, uh, uh, and if they have some compute, the question is how much is needed?

所以问题是：让美国公司——因为他们有更多算力——先到达 Mythos 那种能力水平，是不是更好？以及，如果他们有一些算力，问题是需要多少算力？

[1:05:03] Jensen Huang

The amount of compute they have in China is enormous.it's it it itmean, you're talking about the country. It's the second largest computing market in the world.If they want to deploy, aggregate their compute, they got plenty of compute to aggregate.But is that true? I mean, there's like people do these estimates and they're like, well,this make is actually behind on the process notes. I'm about to tell you. The amount ofenergy they have is incredible. Isn't that right? AI is a parallel computing problem, isn't it?

中国拥有的算力总量是巨大的。你说的是一个国家啊，它是世界第二大计算市场。如果他们想部署、想把算力聚合起来，他们有大把算力可聚合。

〔Dwarkesh〕 但这是真的吗？有人做过估算，说他们在制程节点上其实是落后的……

〔Jensen〕 我这就告诉你。他们拥有的能源量是惊人的，对不对？AI 是个并行计算问题，不是吗？

[1:05:58] Jensen Huang

Why can't they just put four, 10 times as much chips together because energy is free. They haveso much energy. They have data centers that are sitting completely empty, fully powered.They have ghost cities, they have ghost data centers. They have so much capacity of infrastructure.If they wanted to, they just gang up more chips, even if they're seven nanometer. And their capacityof building chips is one of the largest in the world. The semiconductor industry knowsthat they monopolize mainstream chips. They overcapacity, they have too much capacity.And so the idea that China won't be able to have AI chips is completely nonsense. Now, of course,if you ask me, would the United States be further ahead if the entire world had no compute at all?

既然能源是免费的、他们有那么多能源，他们为什么不能把四倍、十倍的芯片堆在一起？他们有完全空置、电力却已经到位的数据中心。他们有鬼城，他们有鬼数据中心，他们有那么多基础设施产能。只要他们想，他们就多堆一些芯片，哪怕是 7 纳米的。而他们造芯片的产能是世界上最大的之一。半导体行业都知道他们垄断了主流芯片，他们产能过剩，他们产能太多了。所以「中国拿不到 AI 芯片」这个想法完全是胡说。当然，如果你问我，假如全世界完全没有算力，美国会不会领先更多？

[1:06:51] Jensen Huang

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 worriedabout, they've already reached that threshold and beyond. And so I think you misunderstand that AI isa five-layer cake. And at the lowest layer is energy. When you have abundance of energy, it makes up forchips. If you have abundance of chips, it makes up for energy. For example, 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 thatwith the few chips that we ship, okay, with the few chips, because the amount of energy is so limited,our throughput per watt is off the charts. But if your amount of watts is completely abundant, it's free.What do you care about performance for watt for? You can use old chips to do so. So seven nanometer chips areessentially hopper. The ability for hopper, I got to tell you, today's models are largely trained on hopper.Yeah. Hopper generation. And so, so hopper, seven nanometer chips are plenty good. The abundance ofenergy is their advantage. But then there's a question of, okay, well, can they actually

但那不是一个会发生的结果，那不是一个成立的情景。他们已经有大量算力了。你担心的那件事所需要的门槛，他们早就达到、并且超过了。所以我认为你误解了：AI 是一块五层蛋糕，最底层是能源。当你有充沛的能源，它可以补上芯片的不足；当你有充沛的芯片，它可以补上能源的不足。比如美国的能源是稀缺的，这正是 NVIDIA 必须不断推进架构、做极致协同设计的原因——用我们出货的那么少的芯片，因为能源太有限，我们的每瓦吞吐必须高到爆表。但如果你的瓦数完全充沛、几乎免费，你还在乎每瓦性能干什么？你可以用老芯片来做。所以 7 纳米芯片本质上就是 Hopper。Hopper 的能力——我得告诉你，今天的模型大部分是在 Hopper 上训练的。〔Dwarkesh〕 是的。〔Jensen〕 Hopper 这一代。所以 Hopper、7 纳米芯片完全够用。能源的充沛才是他们的优势。〔Dwarkesh〕 但接下来的问题是，他们真的能……

[1:08:14] Dwarkesh Patel

manufacture enough chips given their... But they do. What's, what's the evidence? Huawei just had thelargest single year in the history of the company. How many chips did they ship? A ton. Millions.Millions. Millions is way more, way more than Anthropic has.So there's a question of how much logic, Smick and Chef. Then there's a question of how muchmemory... I'm telling you what it is. They have plenty of, they have plenty of logic and theyhave plenty of HBM2 memory. Right. But as you know, the bottleneck often in training and doinginference on these models is the amount of bandwidth. So if you, HBM2, I don't know the numbers offhand,but like versus the newest thing you have, you know, you can be almost an order of magnitudedifference of memory bandwidth, which is huge. Huawei's a networking company. Huawei's a networkingcompany. But that doesn't change the fact that you need an EUV for the most advanced HBM.Not true. Not at all true. You could gang them together just like we gang them together withNBLink72. They've already demonstrated Silicon Photonics connecting all of these compute togetherinto one giant supercomputer. Your premise is just wrong. The fact of the matter is their AI

……造出足够多的芯片吗，考虑到他们的……

〔Jensen〕 但他们做到了。

〔Dwarkesh〕 证据是什么？

〔Jensen〕 华为刚刚经历了公司历史上最大的一年。

〔Dwarkesh〕 他们出了多少芯片？

〔Jensen〕 一大堆。数以百万计。数以百万计。数以百万计，那比 Anthropic 手上的多得多、多得多。

〔Dwarkesh〕 那还有个问题是逻辑芯片有多少，中芯国际（SMIC）……然后还有内存有多少……

〔Jensen〕 我这就告诉你是什么情况。他们有大量的逻辑芯片，他们有大量的 HBM2 内存。

〔Dwarkesh〕 对。但你也知道，训练和推理这些模型的瓶颈常常是带宽。HBM2 我不记得具体数字，但跟你们最新的东西比，内存带宽可能差将近一个数量级，那是巨大的差距。

〔Jensen〕 华为是家网络公司。华为是家网络公司。

〔Dwarkesh〕 但这不改变一个事实：最先进的 HBM 需要 EUV。

〔Jensen〕 不对，完全不对。你可以把它们串联起来，就像我们用 NVLink 72 串起来一样。他们已经演示过用硅光子把所有这些算力连成一台巨型超级计算机。你的前提就是错的。事实是他们的 AI 发展进行得很好。

[1:09:26] Jensen Huang

development is going just fine. And the best AI researchers in the world,because they are limited in compute, they also come up with extremely smart algorithms. Rememberwhat I said. I said that Moore's law is advancing about 25% per year. However, through great computerscience, we could still improve algorithm performance by 10x. What I'm saying is great computer science.is where the lever is. There is no question. MOE is a great invention. There's no question. All theincredible attention mechanisms reduce the amount of compute. We have got to acknowledge that most of theadvances in AI came out of algorithm advances, not just the raw hardware. Now, if most advances came fromalgorithms and computer science and programming tell me that their army of AI researchers is not theirfundamental advantage, and we see it. DeepSeq is not inconsequential advance. And the day that DeepSeqcomes out on Huawei first, that is a horrible outcome for our nation.Why? Is that because, I mean, currently you can have a model like DeepSeq that can run on anyaccelerator if it's open source. Why would that stop being the case in the future?

而且全世界最好的 AI 研究员，正因为他们在算力上受限，他们才会想出极其聪明的算法。记住我说的：我说摩尔定律每年大约推进 25%，然而通过出色的计算机科学，我们仍然能把算法性能提升 10 倍。我要说的是，出色的计算机科学才是杠杆所在，这毫无疑问。MoE 是个伟大的发明，毫无疑问。所有那些了不起的注意力机制都在减少所需的算力。我们必须承认，AI 的大部分进步来自算法进步，而不只是原始硬件。那么，如果大部分进步来自算法、计算机科学和编程，你倒是告诉我，他们那支 AI 研究员大军怎么就不是他们的根本优势？而且我们看到了。DeepSeek 不是无关紧要的进展。而 DeepSeek 首发在华为上的那一天，对我们国家是个可怕的结果。

〔Dwarkesh〕 为什么？我是说，现在你有一个像 DeepSeek 这样的模型，如果它是开源的，它可以跑在任何加速器上。未来为什么会不再是这样？

[1:10:50] Jensen Huang

Well, suppose it doesn't. Suppose it's optimized for Huawei. Suppose it's optimized for theirarchitecture. It would put others at a disadvantage. You described the situation that I perceived to begood news that a company developed software, developed an AI model, and it runs best on theAmerican 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 beston not American hardware. That is bad news for us.I guess I just don't see the evidence that there's these huge disparities that would prevent you fromswitching accelerators. There's American labs, you know, are running their models across all theclouds, across all the different accelerators. I am the evidence. You take a model that'soptimized for NVIDIA and you try to run on something else.But the American labs do that.And they don't run better. NVIDIA's success is perfect evidence. The fact that AI models arecreated on our stack runs best on our stack. How is that illogical to understand?

那假设它不是呢？假设它是为华为优化的，假设它是为他们的架构优化的，那就会让别人处于劣势。你刚才描述的那个情形——一家公司开发了软件、开发了一个 AI 模型，而它在美国技术栈上跑得最好——我看到的是好消息，你却把它设成了坏消息的前提。我告诉你什么才是坏消息：全世界开发的 AI 模型，在非美国的硬件上跑得最好，那对我们才是坏消息。

〔Dwarkesh〕 我只是没看到有什么巨大的差距会阻止你切换加速器的证据。美国的实验室在所有的云、所有不同的加速器上跑他们的模型。

〔Jensen〕 我就是证据。你拿一个为 NVIDIA 优化的模型，试着在别的东西上跑。

〔Dwarkesh〕 但美国的实验室就是这么做的。

〔Jensen〕 而且跑得没那么好。NVIDIA 的成功就是完美的证据。AI 模型在我们的栈上被创造出来、在我们的栈上跑得最好，这有什么不合逻辑的？

[1:11:57] Dwarkesh Patel

I'm just looking, look, Anthropics models are run on GPUs, they're run on Tranium,they're run on TPUs.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 our UnitedStates.But I guess I don't understand arguments. So like, if, say, Chinese companies get to thenext Mythos first, they find that all the security runner really is an American software first,but they can do it on NVIDIA hardware and they ship it to the global South that does it onNVIDIA hardware. Like, how is that good? I mean, I just, okay, they've run on NVIDIA hardware.It's not good.It's not good.Right.It's not good. So let's not let it happen.Why do you think it's perfectly fungible that if you didn't ship them to compute,it would exactly be replaced by Huawei? They are behind, right? They have worse chips than you.It's completely, there's evidence right now. Their chip industry is gigantic.You can just look at the flop or bandwidth or memory comparisons between the H200 andthe Huawei 910C. It's like half, half a third.

我只是看到，Anthropic 的模型在 GPU 上跑、在 Trainium 上跑、在 TPU 上跑。

〔Jensen〕 要换是要投入大量工作的。但你去看全球南方、去看中东，如果开箱即用，所有 AI 模型都在别人的技术栈上跑得最好，你现在得给出一个多离谱的论断，才能说那对美国是好事。

〔Dwarkesh〕 但我不太理解这个论证。比如说中国公司先做出下一个 Mythos，他们发现所有那些安全漏洞其实主要在美国软件里，但他们是在 NVIDIA 硬件上做的，然后他们把它输出到全球南方、也跑在 NVIDIA 硬件上——这怎么就是好事？我是说，好吧，它们跑在 NVIDIA 硬件上。

〔Jensen〕 那不好。

〔Dwarkesh〕 那不好。

〔Jensen〕 对，那不好。所以别让它发生。

〔Dwarkesh〕 你为什么认为这是完全可替代的——如果你不卖给他们算力，就会被华为一比一地补上？他们是落后的，对吧？他们的芯片比你的差。

〔Jensen〕 完全不是，现在就有证据，他们的芯片产业是巨大的。

〔Dwarkesh〕 你可以直接看 H200 和华为 910C 的算力、带宽或内存对比，大概是一半、三分之一。

[1:12:56] Jensen Huang

They use more of it. They use twice as many.I guess 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.And they're good at manufacturing.And I'm sure eventually they would be able to just out-manufacture everybody,but there's these few critical years.What is the critical year you're talking about?

那他们就用更多，他们用两倍的量。

〔Dwarkesh〕 我猜你的论证是：他们有大量现成的能源，他们需要用芯片把它填满。

〔Jensen〕 而且他们擅长制造。

〔Dwarkesh〕 我相信最终他们能在制造上超过所有人，但眼下有这么关键的几年。

〔Jensen〕 你说的关键的几年是哪几年？

[1:13:12] Dwarkesh Patel

These next few years, we've got these models that are going to be able to do all the cyber attacks.If the critical years, the next critical years is critical,then we have to make sure that all of the world's AI models are built on American tech stack.These critical years.Okay. How would that prevent, if they're built on American tech stack,how would that prevent them from, if they have more advanced capabilities fromlaunching the Mythos equivalent cyber attacks on their source?

就是接下来这几年，我们有了能做各种网络攻击的模型。

〔Jensen〕 如果这几年是关键的几年，那我们就必须确保全世界所有的 AI 模型都建在美国技术栈上。就是这关键的几年。

〔Dwarkesh〕 好。可如果他们建在美国技术栈上，这怎么能阻止他们——如果他们有了更先进的能力——发起 Mythos 那种级别的网络攻击？

[1:13:33] Jensen Huang

There's no guarantee either way.But if you have it early, we can prepare for it.Listen, why are you causing one layer of the AI industry to lose an entire marketso that you could benefit another layer of the AI industry?

两种情况都没有保证。但如果你更早拿到它，我们就能做准备。听着，你为什么要让 AI 产业的一层丢掉整整一个市场，好让 AI 产业的另一层受益？

[1:13:54] Jensen Huang

There's 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?Because those models make possible these incredibly offensive capabilities,and you need compute to run them.The energy, the chips, the ecosystem of AI researchers make it possible.A few months ago, Jane Street spent about 20,000 GPU hours trading backdoors into three different language models.Then, they challenged my audience to find the trigger phrases.I just caught up with Rickson, who designed the puzzle about some of the solutions that Jane Street received.If you think the base model was here and the backdoor model was here,you can kind of linearly interpolate the weights to, like, adjust the strength of the backdoor.But you can also extrapolate it to make the backdoor even stronger.And in some cases, if you make it strong enough, the model will just regurgitate what the response phrase was supposed to be.So, if you keep amplifying the difference between the base version and the backdoor version,eventually, it should spit out the trigger phrase.But this technique only worked on two out of the three models.

有五层，每一层都必须成功。最必须成功的那一层其实是 AI 应用。你为什么这么执着于那个 AI 模型？那一家公司？为了什么理由？

〔Dwarkesh〕 因为那些模型让这些极强的攻击能力成为可能，而你需要算力去跑它们。

〔Jensen〕 能源、芯片、AI 研究员的生态才让它成为可能。

〔广告〕 几个月前，Jane Street 花了大约 2 万 GPU 小时，在三个不同的语言模型里训练进了后门，然后他们向我的听众发起挑战：找出触发短语。我刚跟设计这道谜题的 Rickson 聊了 Jane Street 收到的一些解法。（Rickson：）如果你认为基础模型在这里、带后门的模型在那里，你可以对权重做线性插值来调节后门的强度，但你也可以往外做外推、让后门更强。有些情况下，如果你把它加强到足够强，模型会直接把本该输出的回应短语吐出来。所以如果你不断放大基础版本和后门版本之间的差异，最终它应该会把触发短语吐出来。但这个技巧只在三个模型里的两个上奏效。

[1:15:00] Dwarkesh Patel

Even Rickson isn't sure why it didn't work on the other.Being able to verify that a model only does what you think it doesis one of the most important open questions in AI security.If this is the kind of problem that excites you,Jane Street is hiring researchers and engineers.Go to jainestreet.com slash thorkash to learn more.Okay, stepping back, it has to be the case that China is able to build enough 7 nanometer capacity.And remember, they're still stuck on 7 nanometer.Well, you'll move on to 3 nanometer and then 2 nanometer or 1.6 nanometer with Feynman.So, while you're on 1.6 nanometer, they're still going to be on 7 nanometer.And they have to produce enough of it to make up for the shortfall.And they have so much energy that the more chips you give them, the more compute they'd have.Right?

〔广告〕 连 Rickson 自己也不确定为什么在第三个上没奏效。能够验证一个模型只做你以为它会做的事，是 AI 安全里最重要的开放问题之一。如果你被这类问题激起兴趣，Jane Street 正在招研究员和工程师，去 janestreet.com/dwarkesh 了解更多。

〔Dwarkesh〕 好，退一步说，这必须建立在中国能建出足够的 7 纳米产能的前提上。而且记住，他们还卡在 7 纳米，而你会推进到 3 纳米、然后 2 纳米、再到 Feynman 的 1.6 纳米。所以当你在 1.6 纳米时，他们还在 7 纳米。他们必须产出足够多来补上这个缺口。而他们有那么多能源，你给他们越多芯片，他们就有越多算力，对吧？

[1:15:38] Dwarkesh Patel

Like, so, it just, there's, it comes up to the question of,ultimately, they are getting more compute.Compute is right, input to training and inference.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 100 times more than anywhere else in the world.The United States ought to be ahead.Okay?

所以归根到底问题是，他们终究会拿到更多算力，而算力是训练和推理的输入。

〔Jensen〕 我只是觉得你在用绝对化的方式说话。我认为美国应该领先。美国的算力总量是世界上任何其他地方的 100 倍，美国应该领先，好吗？

[1:15:59] Jensen Huang

The United States is ahead.NVIDIA builds the most advanced technologies.We make sure that the U.S. labs are the first to hear about it and 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?

美国现在就是领先的。NVIDIA 造最先进的技术，我们确保美国的实验室最先听到、最先有机会买到。如果他们钱不够，我们甚至投资他们。美国应该领先，我们在尽一切努力确保美国领先。这是第一点。你同意吗？

[1:16:22] Jensen Huang

And we're doing everything we can to do that.But how is shipping ships to China, keeping the U.S. in there?No, no, no.If they're bottleneck on compute.We got Vera Rubin for United States.We got Vera Rubin for United States.Now, United States.Am I in United States?

而我们在尽一切努力做到这一点。

〔Dwarkesh〕 但把芯片卖给中国怎么能保持美国领先？如果他们的瓶颈是算力的话。

〔Jensen〕 不不不。我们有 Vera Rubin 给美国，我们有 Vera Rubin 给美国。现在，美国——我算不算美国的一部分？

[1:16:36] Jensen Huang

Do you consider me part of the United States?Yes.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?

你认为我算美国的一部分吗？

〔Dwarkesh〕 算。

〔Jensen〕 NVIDIA。你认为 NVIDIA 是一家美国公司。好，第一点。那为什么我们不能想出一个更平衡的监管规则，让 NVIDIA 能在全世界赢，而不是把世界拱手让人？

[1:16:56] Jensen Huang

Why would you want United States to give up the world?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 United States giving up a vast part of the world's market?

你为什么会希望美国放弃全世界？芯片产业是美国生态的一部分，它是美国技术领导力的一部分，它是 AI 生态的一部分，它是 AI 领导力的一部分。为什么你的政策、你的哲学会导致美国放弃世界市场中如此巨大的一块？

[1:17:20] Dwarkesh Patel

I guess the claim here is, Dario had this quote where he said, 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 U.S. technology stack.Like fundamentally, you're giving them this capability.Comparing AI to anything that you just mentioned is lunacy.But AI is similar to enriched uranium, right?

我想这里的主张是——Dario 有句话，他说这就像波音吹嘘「我们在卖核弹给朝鲜，但导弹外壳是波音造的」，这怎么就叫在赋能美国技术栈？根本上说，你是在把这个能力给他们。

〔Jensen〕 把 AI 拿来跟你刚才说的任何东西比，都是疯话。

〔Dwarkesh〕 但 AI 类似于浓缩铀，对吧？

[1:17:39] Dwarkesh Patel

And then it can have positive uses.It can have negative uses.We still don't want to send enriched uranium to other countries.Who's sending enriched uranium?The analogy is enriched uranium.Because it's a lousy analogy.It's an illogical analogy.But if that compute can run a model that can do zero-day exploits against all American software, how is that not a weapon?

它可以有正面用途，也可以有负面用途，但我们仍然不想把浓缩铀送去别的国家。

〔Jensen〕 谁在送浓缩铀？

〔Dwarkesh〕 这个类比是浓缩铀。

〔Jensen〕 因为那是个糟糕的类比，是个不合逻辑的类比。

〔Dwarkesh〕 但如果那些算力能跑一个可以对所有美国软件做零日攻击的模型，它怎么就不是武器？

[1:18:03] Jensen Huang

First of all, the way to solve that problem is to have dialogues with the researchers and dialogues with China and dialogues with all the countries to make sure that people don't use technology in that way.That's a dialogue that has to happen.Okay?

首先，解决那个问题的方式是跟研究员对话、跟中国对话、跟所有国家对话，确保人们不会那样使用技术。那个对话必须发生，好吗？

[1:18:16] Jensen Huang

Number one.Number two, we also need to make sure that United States is ahead.Everything that Rubin, Vera Rubin, Blackwell is available in United States in abundance.Tons of it, obviously, our results would show it.Abundance, tons of it.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.However, we also have to recognize that AI is not just a model.That AI is a five-layer cake.That AI industry matters across every single layer.And we want United States to win at every single layer, including the chip layer.And conceding the entire market is not going to allow United States to win the technology race long-term in the chip layer, in the computing stack.That is just a fact.I guess then the crux comes down to how does selling them chips now help us win in the long-term?

第一点。第二点，我们也必须确保美国领先。Rubin、Vera Rubin、Blackwell 在美国都是充裕供应的，量非常大，我们的业绩显然能证明这一点。充裕，量非常大，量非常大。我们拥有的算力非常好，我们这里有出色的 AI 研究员，很好，我们应该保持领先。然而，我们也必须认识到 AI 不只是一个模型，AI 是一块五层蛋糕，AI 产业在每一层上都重要，我们希望美国在每一层上都赢，包括芯片这一层。而把整个市场让出去，不会让美国在长期的技术竞赛里、在芯片层、在计算栈上获胜。这就是事实。

〔Dwarkesh〕 那我想症结就归结到：现在卖芯片给他们，怎么帮我们在长期获胜？

[1:19:16] Jensen Huang

Like 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 or smartphones are dominating.The way we started the conversation today, you would acknowledge and you acknowledged that NVIDIA's position is very different.You use words like moat.The single most important thing to our company is our richness of our ecosystem, which is about developers.50% of the AI developers are in China.We don't want to...We shouldn't...The United States should not give that up.But we have a lot of NVIDIA developers in the U.S.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...We have to keep innovating and, you know, 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.I don't...You're not talking to somebody who woke up a loser.

比如特斯拉长期以来一直在往中国卖极好的电动车，iPhone 也在中国卖，非常好，这并没有造成锁定。中国照样会做他们自己版本的电动车，而且他们正在主导；智能手机上他们也在主导。

〔Dwarkesh〕 按我们今天开场那样谈，你会承认、你也确实承认了 NVIDIA 的位置非常不同。你用了「护城河」这样的词。

〔Jensen〕 对我们公司最重要的一件事是我们生态的丰富度，而那是关于开发者的。50% 的 AI 开发者在中国。我们不想……我们不应该……美国不应该把那个放弃掉。

〔Dwarkesh〕 但我们在美国有很多 NVIDIA 开发者。而这并不妨碍美国的实验室将来也能用别的加速器。事实上现在他们就在用别的加速器，这没问题、也挺好。我不明白为什么在中国就不会是这样。如果你卖给他们 NVIDIA 芯片，就像 Google 可以同时用 TPU 和 NVIDIA 一样……

〔Jensen〕 我们必须持续创新。而且你大概也知道，我们的份额在涨，不是在跌。「就算我们在中国竞争，反正也会丢掉那个市场」这个前提——我……你不是在跟一个一觉醒来就当输家的人说话。

[1:20:28] Jensen Huang

And that loser attitude, that loser premise makes no sense to me.We are not...We're not a car.We are not a car.It...The fact that I can buy a car, 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 still exists.There's a reason why ARM is so sticky.These ecosystems, these ecosystems are hard to replace.It costs an enormous amount of time and energy and most people don't want to do it.And so it's our job to continue to nurture that ecosystem, to keep advancing the technologyso that we could compete in the marketplace.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.And that losing proposition, that losing mindset makes no sense to me.Okay.I'll move on.I just want to make sure that...You don't have to move on.I'm enjoying it.Okay.Great.Yeah, yeah.Then I will.I appreciate that.But I think that maybe the crux and thanks for walking around the circles with me becausethen I think it helps bring out what the crux here is.The crux is you're going to extremes.Your argument starts from extremes.

那种输家心态、那种输家前提，我完全无法理解。我们不是一辆车。我们不是一辆车。我今天买这个牌子的车、明天换另一个牌子，很容易。计算不是那样的。x86 之所以还存在是有原因的，ARM 之所以粘性这么强是有原因的。这些生态很难被替换，替换要花巨量的时间和精力，而大多数人不想干这事。所以我们的工作就是继续培育那个生态、继续推进技术，好让我们能在市场上竞争。基于你描述的那个前提去让出一个市场，我根本无法认同。它说不通，因为我不认为美国是输家，我们的产业不是输家。那种失败的主张、那种失败者心态，我完全无法理解。

〔Dwarkesh〕 好，我换个话题。我只是想确认……

〔Jensen〕 你不用换，我挺享受的。

〔Dwarkesh〕 好，太好了，那我就继续，我很感激。但我觉得也许这个症结——谢谢你陪我绕这些圈子，因为这样才能把症结逼出来。

〔Jensen〕 症结在于你在走极端，你的论证是从极端出发的。

[1:21:48] Jensen Huang

That if we give them any compute at all, in this narrow moment, we will lose everything.No, I think what my argument is...Those extremes, they're childish.Let me just make my argument for myself.They're childish.Yeah.The idea is not that there is some key threshold of compute.Yeah.It is that any marginal compute is helpful, right?

就是「如果我们给他们任何算力，在这个狭窄的时刻，我们就会失去一切」。

〔Dwarkesh〕 不，我的论证是——

〔Jensen〕 那些极端说法是幼稚的。

〔Dwarkesh〕 让我自己把我的论证说完。

〔Jensen〕 那是幼稚的。

〔Dwarkesh〕 我的意思不是存在某个算力的关键阈值，而是任何边际上的算力都是有帮助的，对吧？

[1:22:08] Dwarkesh Patel

So if you have more compute, you can train a better model.And I just want you to acknowledge that any marginal sales for American technology industryis beneficial.I actually don't...I mean, if the AI models that run on those chips...Yeah....are capable of cyber offensive capabilities,or training models are capable of cyber defensive, running more models of those instance,it is not a nuclear weapon, but it enables a weapon of a kind.The logic that you use, you might as well say it to microprocessors and DRAMs.You might as well say it to electricity.But in fact, we do have expert controls on the technology that is relevant to makingthe most advanced DRAM, right?

所以如果你有更多算力，你就能训出更好的模型。而我只是想让你承认，对美国科技产业来说任何边际销售都是有好处的。

〔Jensen〕 我其实不……

〔Dwarkesh〕 我是说，如果跑在那些芯片上的 AI 模型具备网络攻击能力，或者训练出的模型具备网络攻防能力，跑更多这样的实例——它不是核武器，但它使某种武器成为可能。

〔Jensen〕 你用的这套逻辑，你不如把它用到微处理器和 DRAM 上，你不如把它用到电力上。

〔Dwarkesh〕 但事实上，我们对制造最先进 DRAM 相关的技术确实有出口管制，对吧？

[1:22:42] Dwarkesh Patel

We have all kinds of expert controls on China for all kinds of chip making stuff.We sell a lot of DRAM and CPUs into China.And I think it's right.I guess this goes back to the fundamental question of, is AI different, right?

我们对中国在各种芯片制造相关的东西上都有各种出口管制。

〔Jensen〕 我们往中国卖了大量 DRAM 和 CPU，我认为这是对的。

〔Dwarkesh〕 我想这又回到了那个根本问题：AI 是不是不一样？

[1:22:54] Dwarkesh Patel

If you have the kind of technology that can find these zero days in software, is thatsomething where we want to minimize China's ability to get there first, to deploy it lightly?We want the United States to be ahead.We can control that.How do we control that if the chips are already there and they're using that to train thatmodel?

如果你有这种能在软件里找出零日漏洞的技术，这是不是我们想尽量降低中国率先拿到、率先轻易部署的能力的那种东西？

〔Jensen〕 我们希望美国领先，我们能控制这一点。

〔Dwarkesh〕 如果芯片已经在那儿了、他们正在用它训练那个模型，我们怎么控制？

[1:23:11] Jensen Huang

We have tons of compute.We have tons of AI researchers.We're racing as fast as we can.Again, we have more nuclear weapons than anybody else, but we don't want to send enricheduranium anywhere.We're not enriched uranium.It's a chip.And it's a chip that they can make themselves.But there's a reason they're buying it from you, right?

我们有大量算力，我们有大量 AI 研究员，我们在尽可能快地跑。

〔Dwarkesh〕 再说一次，我们的核武器比谁都多，但我们不想把浓缩铀送到任何地方。

〔Jensen〕 我们不是浓缩铀，这是芯片，而且是他们自己能造的芯片。

〔Dwarkesh〕 但他们从你这儿买是有原因的，对吧？

[1:23:30] Dwarkesh Patel

And we have quotes from the founders of Chinese companies that say that we're bottleneckedon computer.Because our chips are better.On balance, our chips are better.There's just no question about it.In the absence of our chip, in the absence of our chip, can you acknowledge that Huaweihad a record year?

而且我们有中国公司创始人的引述，说我们的瓶颈是算力。

〔Jensen〕 因为我们的芯片更好。总体上我们的芯片更好，这毫无疑问。在没有我们芯片的情况下——在没有我们芯片的情况下，你能不能承认华为创下了纪录年？

[1:23:42] Jensen Huang

Can you acknowledge that a whole bunch of chip companies have gone public?Can you acknowledge that?Yes.Can you acknowledge that?Can you can also acknowledge that the fact that we used to have a very large share in thatmarket and we no longer have the large share in that market?

你能不能承认一大堆芯片公司已经上市了？你能承认吗？

〔Dwarkesh〕 能。

〔Jensen〕 你能承认吗？你能不能也承认，我们曾经在那个市场有非常大的份额，而现在我们在那个市场已经没有大份额了？

[1:23:54] Jensen Huang

We can also acknowledge that China is about 40% of the world's technology industry.That market, to leave that market, concede that market for the United States technologyindustry 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.I guess I'm confused of, 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 tobe way better if we're allowed to compete.And another is that they would be doing the same exact thing without us anyways.Right?

我们还可以承认，中国大约占全世界科技产业的 40%。离开那个市场、为美国科技产业让出那个市场，是对我们国家的伤害，是对我们国家安全的伤害，是对我们技术领导力的伤害。而这一切只是为了一家公司的利益。这在我看来完全说不通。

〔Dwarkesh〕 我有点困惑，感觉你在做两个不同的陈述。一个是：如果允许我们竞争，我们会赢下跟华为的这场竞争，因为我们的芯片会好得多。另一个是：反正没有我们他们也会做一模一样的事。对吧？

[1:24:28] Dwarkesh Patel

How can those two things be at the same time?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.The reason they want NVIDIA chips is they're better.Better is more compute.More compute means you can train a better model.It's better because it's easier to program.We have a better ecosystem.Whatever the better is.Whatever the better is.And of course, we're going to send them compute.So what?

这两件事怎么能同时成立？

〔Jensen〕 这显然是成立的。在没有更好选择的情况下，你会选你唯一有的那个选择。这怎么不合逻辑？这太合逻辑了。他们想要 NVIDIA 芯片的原因是它更好。更好意味着更多算力，更多算力意味着你能训出更好的模型。更好也因为它更好编程，我们有更好的生态。不管「更好」指的是什么，不管「更好」指的是什么。而我们当然会送算力给他们。那又怎样？

[1:24:55] Jensen Huang

So what?The fact of the matter is, we get the 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.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.Now, the policy that you're advocating resulted in the American telecommunication industry beingpolicy out of basically the world to the point where we don't control our own telecommunicationsanymore.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 seemto have a very hard time understanding.Okay.Let's just step back.It seems like the crux here is there's a potential benefit and there's a potential cost.And we're trying to figure out, is the benefit worth the cost?

那又怎样？事实是，我们得到了好处。别忘了，我们得到了美国技术领导力的好处，我们得到了开发者在美国技术栈上工作的好处，我们得到了这些 AI 模型扩散到世界其他地方时的好处——因为美国技术栈是跑它们的最佳选择，我们可以继续推进并扩散美国技术。我相信这是正面的，这是美国技术领导力非常重要的一部分。而你主张的那种政策，结果就是让美国的电信产业基本被赶出了全世界，以至于我们连自己的电信都控制不了了。我不觉得那聪明，那有点狭隘，而且它导致了我现在正在跟你描述的、而你似乎很难理解的意外后果。

〔Dwarkesh〕 好，我们退一步。看起来症结是：有一个潜在收益，也有一个潜在成本，我们在试图弄清楚，收益是否值这个成本。

[1:25:58] Dwarkesh Patel

I guess I'm trying to get you to acknowledge the potential cost.The compute is an input to training powerful models.Powerful models do have powerful, you know, offensive capabilities like cyber attacks.It is a good thing that American companies got to claw at mythos-level capabilities first.And then now they're going to hold off on those capabilities so that the American companiesand American government can make their software more protected before this level of capabilitywas announced.If China had had more computer, had more power compute, had made a mythos-level model earlierand 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 companieslike NVIDIA in America.That is a cost of sending it to China.And so let's leave the benefit aside for a second.Do you acknowledge that this is a potential cost?

我想让你承认那个潜在成本。算力是训练强大模型的输入，强大模型确实具备强大的攻击能力，比如网络攻击。美国公司先摸到 Mythos 级别的能力是件好事，然后他们现在会先按住这些能力，好让美国公司和美国政府在这个能力等级被公布之前先把自己的软件保护起来。如果中国有更多算力、更强算力，更早做出 Mythos 级别的模型并且广泛部署，那会非常糟糕。这件事之所以没发生，原因之一就是我们有更多算力，得益于美国的 NVIDIA 这类公司。这就是把算力送去中国的成本。所以我们先把收益放一边，你承认这是一个潜在成本吗？

[1:26:44] Jensen Huang

I will also tell you the potential cost is we allow one of the most important layers ofthe AI stack, the chip layer, to concede an entire market, the second largest market inthe 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 becomesuperior to ours because their models are open.I guess I just believe enough in NVIDIA's kernel engineers and Kudo engineers to thinkthat they could optimize.AI is more than kernel optimization, as you know.Of course, but there's so many things you can do from distilling to a model that's wellfit for your chips.We're going to do our best.You have all the software.It's just hard to imagine that there's a long-term lock-in to Chinese ecosystem.Even if they have this slightly better open source model for a while.China is the largest contributor to open source software in the world.Fact.Right?

我也要告诉你另一个潜在成本：我们让 AI 栈里最重要的一层——芯片层——让出了整整一个市场、世界第二大市场，好让他们去发展规模、去发展自己的生态，好让未来的 AI 模型以一种跟美国技术栈非常不同的方式被优化。当 AI 扩散到世界其他地方，他们的标准、他们的技术栈会变得比我们的更强，因为他们的模型是开放的。

〔Dwarkesh〕 我只是足够相信 NVIDIA 的 kernel 工程师和 CUDA 工程师，觉得他们能优化过来。

〔Jensen〕 你也知道，AI 不只是 kernel 优化。

〔Dwarkesh〕 当然，但从蒸馏到做一个适配你芯片的模型，你能做的事太多了。

〔Jensen〕 我们会尽力。

〔Dwarkesh〕 你们有全部的软件。很难想象会有对中国生态的长期锁定，就算他们有一阵子拥有稍好一点的开源模型。

〔Jensen〕 中国是全世界开源软件最大的贡献者。事实。对吧？

[1:27:52] Jensen Huang

China is the largest contributor to open models in the world.Fact.Today, it's built on the American tech stack, NVIDIA's.Fact.All five layers of the tech stack for AI is important.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.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 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.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 disservice to the United States.And we misunderstand the difference between a job and the task.The task, the job of a radiologist, patient care, task, to read a scan.

中国是全世界开放模型最大的贡献者。事实。今天，它建立在美国技术栈上，NVIDIA 的。事实。AI 技术栈的五层全都重要，美国应该去把这五层全赢下来，它们都重要。当然，最重要的那一层是 AI 应用层，那是扩散进社会、被使用得最多的一层，也是从这场工业革命中受益最多的一层。但我的意思是，每一层都必须成功。如果我们把这个国家吓得以为 AI 就是核弹，让所有人都恨 AI、都怕 AI，我不知道你是在怎么帮美国，你是在帮倒忙。如果我们把所有人吓得不敢去做软件工程的工作，因为它会杀死每一个软件工程岗位，结果我们没有软件工程师了，我们就是在帮美国的倒忙。如果我们把所有人吓得不敢去做放射科，因为计算机视觉完全免费、没人想当放射科医生了，那也是在帮美国的倒忙。而我们混淆了「工作」和「任务」的区别：放射科医生的工作是照护病人，任务才是读片。

[1:29:18] Jensen Huang

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.And 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.Life is not like that.Do we want United States to be first?

如果我们把这一点误解得如此彻底，把所有人吓得不去读放射科，我们就不会有足够的放射科医生和足够好的医疗。所以我想说的是：当你提出一个如此极端的前提、一切非零即无穷，我们最后就把人们吓成了一个根本不真实的样子。生活不是那样的，生活不是那样的。我们想不想让美国第一？

[1:29:52] Jensen Huang

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.Is today you're talking about mythos because mythos is important?Sure, that's fantastic.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.We should always have the best technology here.

我们当然想。我们需不需要在这个栈的每一层都做领导者？当然需要，当然需要。今天你谈 Mythos，是因为 Mythos 重要？当然，那很好。但我给你一个预测：几年之后，当我们希望美国技术栈、希望美国技术扩散到世界各地，到印度、到中东、到非洲、到东南亚，当我们的国家想要出口——因为我们想出口我们的技术、想出口我们的标准——到那一天，我希望你我再进行一次同样的对话。我们应该永远在这里拥有最好的技术。

[1:31:11] Jensen Huang

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.Okay.The argument hinges on they've built models that are specified for their architect.They're the best chips that they make in a few years.And 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 1.6 nanometer.They're still going to be on 7 nanometer, even after a few years from now.And it may make sense that domestically they would prefer, hey, we've got so much energy.We can manufacture it at such scale.We'll still keep using 7 nanometer.But the exporting thing, their 7 nanometer chips have to be competitive against your 1.6 nanometer chips.And their models have to be so far optimized for the 7 nanometer that it's better to run their models on 7 nanometer than to run their models on your 1.6 nanometer.Can we just look at the facts then?

我们应该永远在这里拥有最多的技术、而且最先拥有，但我们也应该努力去全世界竞争并获胜。这两件事可以同时发生。这需要一些细微的分寸感、一些成熟，而不是绝对化。世界就不是绝对的。

〔Dwarkesh〕 好。这个论证的关键假设是：他们造出了为自己架构专门优化的模型，那是他们几年后能造的最好的芯片，然后那些芯片被出口到全世界，从而立下标准。但因为 EUV 出口管制，就像我们说的，你会推进到 1.6 纳米，而几年后他们还在 7 纳米。国内他们可能确实更愿意说「我们有这么多能源、能大规模制造，就继续用 7 纳米」，这说得通。但出口这件事上，他们的 7 纳米芯片必须能跟你的 1.6 纳米芯片竞争，而且他们的模型必须为 7 纳米优化到「在 7 纳米上跑他们的模型比在你的 1.6 纳米上跑更好」的程度。

〔Jensen〕 那我们能不能就看看事实？

[1:32:19] Jensen Huang

Okay.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.My point is, architecture matters.Computer science matters.Semiconductor physics matter as well.But computer science matters.AI, 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.And 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 people always love programming on CUDA first.

好。Blackwell 的光刻工艺比 Hopper 先进 50 倍吗？有 50 倍吗？差得远。我一遍又一遍地说：摩尔定律已经死了。从 Hopper 到 Blackwell，单看晶体管本身，算 75%（提升），中间隔了三年，75%。而 Blackwell 是 Hopper 的 50 倍。我的意思是，架构重要，计算机科学重要。半导体物理当然也重要，但计算机科学重要。AI 的影响很大程度上来自计算栈，这正是 CUDA 如此有效的原因，也是 CUDA 如此被喜爱的原因。它是一个生态、一个计算架构，它允许如此大的灵活性——如果你想彻底换一个架构，做出 MoE 这样的东西、做出扩散这样的东西、做出分离式的东西，你都能做到，很容易做到。所以事实是，AI 既关乎下面的架构，也同样关乎上面的栈。就我们拥有为自己的栈、自己的生态优化的架构和软件栈而言，这显然是好事。因为我们今天的对话就是从「NVIDIA 的生态如此丰富、为什么大家总是最爱在 CUDA 上编程」开始的。

[1:33:53] Jensen Huang

They do.They do.And so do the researchers in China.But if we are forced to leave China, if we're forced to leave China, it would be – well, first of all, it's a policy mistake.Obviously, it has backlash.Obviously, it has fired – has turned out badly for the United States.

他们确实如此，他们确实如此。中国的研究员也是如此。但如果我们被迫离开中国，如果我们被迫离开中国，那——首先，这是个政策错误。它显然会有反噬，它显然已经对美国产生了糟糕的结果。

[1:34:18] Jensen Huang

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.You're going to see in the future, they're not stuck at 7 nanometer, obviously.They're good at manufacturing.They will continue to advance from 7 and beyond.Now, is there 10x difference between 5 nanometer and 7 nanometer?

它让、它加速了他们的芯片产业，它迫使他们整个 AI 生态聚焦到自己的内部架构上。现在还不算太晚，但无论如何，这已经发生了。你以后会看到，他们显然不会卡在 7 纳米，他们擅长制造，他们会从 7 纳米继续往前推进。那么，5 纳米和 7 纳米之间有 10 倍差距吗？

[1:34:51] Jensen Huang

The answer is no.Architecture matters.Networking matters.That's why NVIDIA bought Mellanox.Networking matters.Energy matters.And so all of that stuff matters.It's not simplistic like the way you're trying to distill it.We can move on from China.But that actually raises an interesting question about – we were discussing earlier these bottlenecks at TSMC and memory and so forth.And so if we're in this world where you're already the majority of N3, at some point you'll be N2, 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 about everything we know about a numerix today and all the other improvements you described.Do you see that world happening before 2030?

答案是没有。架构重要，网络重要——这就是 NVIDIA 收购 Mellanox 的原因，网络重要——能源也重要。所有这些东西都重要，不像你试图把它简化成的那样简单。

〔Dwarkesh〕 中国这块我们可以翻篇了。但这其实引出一个有意思的问题：我们前面在讨论台积电、内存这些地方的瓶颈。如果我们处在这样一个世界里，你已经占了 N3 的大头、到某个时点你会占 N2 的大头，你会不会回头去用 N7、去用老制程节点上的闲置产能，说「AI 的需求太大了，而我们扩先进制程的能力跟不上，所以我们要用今天我们懂的所有数值格式（numerics）和你描述的其他改进，再造一个 Hopper 或者 Ampere」？你觉得这个世界会在 2030 年前发生吗？

[1:35:44] Jensen Huang

It's not necessary to.And the reason for that is because with every generation, the architecture is more than just the transistor scale.It also – you're doing so much engineering and packaging and stacking and the numerics and the system architecture.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.You know, 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 7 in a heartbeat?

没有必要那么做。原因是每一代的架构都不只是晶体管尺度，你还在封装、堆叠、数值格式和系统架构上做大量工程。当你没产能了、想轻松地退回到另一个节点，那是没有人负担得起的研发量。我们负担得起往前倾，我不认为我们负担得起往回走。当然，如果这个世界干脆说——我们做个思想实验——如果到那一天我们说，听着，我们永远不会再有更多产能了，那我会不会毫不犹豫地退回去用 7 纳米？

[1:36:39] Jensen Huang

You know, of course I would.Yeah.One question somebody I was talking to had is why NVIDIA doesn't run multiple different chip projects at the same time with totally different architectures?So you could do like a Cerebra-style wafer scale.You could do a Dojo-style huge package.You could do one without CUDA.You know, you have the resources and the engineering talent to do all of these in parallel.So why put all the eggs in one basket given who knows where AI might go and architectures might go?

我当然会。

〔Dwarkesh〕 是啊。我跟人聊时有人问，为什么 NVIDIA 不同时跑好几个架构完全不同的芯片项目？比如你可以做 Cerebras 那种晶圆级的，可以做 Dojo 那种超大封装的，可以做一个不带 CUDA 的。你有资源、有工程人才可以并行做这些。既然谁也不知道 AI 和架构会走向哪儿，为什么要把鸡蛋放在一个篮子里？

[1:37:05] Jensen Huang

Oh, we could.It's just that we don't have a better idea.Yeah.Yeah, yeah.We could do all of those things.it's just not better. And we simulate it all. They're in our simulator provably worse.And so we wouldn't do it. Yeah. We're working on exactly the projects that we want to work on.And if the workload were to change dramatically, and I don't mean the algorithms, I actually meanthe workload. And that depends on the shape of the market. We may decide to add other accelerators.Like for example, recently we added Grok. And we're going to fold Grok into our CUDA ecosystem.And we're doing that now because the value of tokens have gone up so high that you could havedifferent pricing of tokens. Back in the old days, just a couple of years ago, tokens are either freeor barely expensive. But now you can have different customers and those customers want differentanswers. And so because the customers make so much money, like for example, our software engineers,if I can give them much more responsive tokens so that they're even more productive than they aretoday, I would pay for it. But that market has only recently emerged. And so I think that we now havethe ability to have the same model based on the response time have different segments. And that's

哦，我们可以啊，只是我们没有更好的想法。我们可以做所有那些事，只是它们不更好。而且我们全都仿真过，在我们的仿真器里它们是可证明更差的，所以我们不会去做。我们正在做的恰恰就是我们想做的项目。而如果工作负载发生剧变——我说的不是算法，我指的是工作负载本身，那取决于市场的形态——我们可能会决定加上别的加速器。比如我们最近加上了 Groq，我们会把 Groq 折进我们的 CUDA 生态。我们现在做这件事，是因为 token 的价值已经涨得如此之高，以至于你可以对 token 做不同的定价。在过去，就在几年前，token 要么免费、要么便宜得几乎不要钱。但现在你可以有不同的客户，那些客户想要不同的答案。因为客户赚的钱太多了，比如我们的软件工程师，如果我能给他们响应快得多的 token、让他们比今天更高效，我愿意为此付钱。但那个市场只是最近才出现。所以我认为我们现在有能力让同一个模型按响应时间分出不同的细分市场。

[1:38:47] Jensen Huang

the reason why we decided to expand the Pareto frontier and create a segment of inference that is fasterresponse time, even though it's lower throughput. Until now, higher throughput is always better.We think that there could be a world where there could be very high ASP tokens. And even thoughthe throughput is lower in the factory, the ASPs make up for it. That's the reason why we did it.But otherwise, from an architecture perspective, I think NVIDIA's architectures, I would rather put,if I had more money, I put more behind the architecture.I think this idea of extremely premium tokens and just the disaggregation of the inference market is veryinteresting.The segmentation of it.Yeah. Final question. Suppose the deep learning revolution didn't happen. What would NVIDIA be doing?

这就是我们决定扩展帕累托前沿、造出一个「响应更快但吞吐更低」的推理细分市场的原因。在此之前，吞吐越高总是越好。我们认为可能存在一个世界，里面有非常高单价（ASP）的 token，即便工厂的吞吐更低，单价也能补回来。这就是我们这么做的原因。但除此之外，从架构角度讲，如果我有更多钱，我会把更多钱押在架构上。

〔Dwarkesh〕 我觉得「极高溢价 token」这个想法、以及推理市场的分离化非常有意思。

〔Jensen〕 是它的细分化。

〔Dwarkesh〕 对。最后一个问题：假设深度学习革命没有发生，NVIDIA 会在做什么？

[1:39:45] Dwarkesh Patel

Obviously games, but given...Accelerated computing.Accelerated computing. The same thing we've been doing all along. The premise of our company is thatMoore's law is going to... General force computing is good for a lot of things, but for a lot ofcomputation, it's not ideal. And so we combined an architecture called a GPU, CUDA, to a CPU so thatwe can accelerate the workload of the CPU. And so different kernels of code or algorithms could beoffloaded onto our GPU. And as a result, you speed up an application by 100x, 200x. And where can youuse that? Well, obviously engineering and science and physics and so on. So data processing,computer graphics, image generation, I mean, all kinds of things. Even if AI doesn't exist today,NVIDIA will be very, very large. Yeah. And so I think the reason for that is fairly fundamental,which is the ability for general purpose computing to continue to scale has largely run its course.And the only... Not the only way, but the way to do that is through domain-specific acceleration.And one of the domain that we started with was computer graphics. But there are many,many other domains. I mean, there's all kinds of scientific particle physics and fluids and

显然是游戏，但除此之外……

〔Jensen〕 加速计算。

〔Dwarkesh〕 加速计算。

〔Jensen〕 就是我们一直在做的那件事。我们公司的前提是：通用计算对很多事情是好的，但对很多计算它并不理想。所以我们把一个叫 GPU 的架构、加上 CUDA，跟 CPU 组合起来，以便加速 CPU 的工作负载。于是不同的代码核函数或算法可以被卸载到我们的 GPU 上，结果你把一个应用加速 100 倍、200 倍。这能用在哪儿？显然是工程、科学、物理等等。还有数据处理、计算机图形、图像生成——各种各样的东西。就算今天 AI 不存在，NVIDIA 也会非常非常大。原因相当根本：通用计算继续扩展的能力基本已经走到头了，而做到这件事的方式——不是唯一的方式，但那个方式——就是领域专用加速。我们起步的领域之一是计算机图形，但还有非常非常多别的领域。我是说，各种科学的粒子物理、流体……

[1:41:11] Jensen Huang

you know, and so structured data processing, all kinds of different types of algorithms that benefitfrom CUDA. And so our mission was really to bring accelerated computing to the world and advance thetype of applications that general purpose computing can do and scale to the level of capability thathelps break through certain fields of science. And so some of the early applications were moleculardynamics, seismic processing for energy discovery, image processing, of course. And so all of thosekinds of fields where general purpose computing is just simply too inefficient to do so. And so,if there's 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 studentto be able to access a PC or a GeForce adding card and do amazing science. And that fundamental promisehasn't changed, not even a little bit. And so 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 chemistrywork or all of that stuff, data processing work, all of that stuff is unrelated to AI. And it's stillvery important. I mean, I know that AI is very interesting and quite exciting, but there's a lot

……以及结构化数据处理，各种能从 CUDA 受益的算法。所以我们的使命真的是把加速计算带给世界，推进通用计算所能做的应用类型，并把能力扩展到能在某些科学领域取得突破的水平。所以早期的一些应用是分子动力学、用于能源勘探的地震数据处理，当然还有图像处理——所有那些通用计算实在太低效的领域。所以如果没有 AI，我会非常难过。但因为我们在计算上取得的进步，我们让深度学习民主化了，我们让任何研究者、任何科学家、任何地方的任何学生都能用一台 PC 或一块 GeForce 显卡去做了不起的科学。那个根本承诺一点都没有改变。所以如果你看 GTC，开头一整块都不是 AI——计算光刻、我们的量子化学工作、数据处理工作，那一整块都跟 AI 无关，但它仍然非常重要。我知道 AI 很有意思、很令人兴奋，但有很多人……

[1:42:57] Jensen Huang

of people doing a lot of very important work that's not AI related and tensors is not the only way thatyou compute with. And we want to help everybody.Jensen, thank you so much.You're welcome. I enjoyed it.Me too. Sweet.

……在做很多非常重要的、跟 AI 无关的工作，而张量也不是唯一的计算方式。我们想帮助每一个人。

〔Dwarkesh〕 Jensen，非常感谢你。

〔Jensen〕 不客气，我很享受。

〔Dwarkesh〕 我也是。太棒了。