NVIDIA: OpenAI, Future of Compute, and the American Dream | BG2 w/ Bill Gurley and Brad Gerstner

Interactive Transcript

spk_0 I think that OpenAI is likely going to be the next multi-trillion dollar

spk_0 hyper-scale company.

spk_0 Jensen, great to be back. Of course, with my partner Clark Tang,

spk_0 you know, I can't believe it's been-

spk_0 Welcome to MV.

spk_0 Oh, and nice glasses.

spk_0 There's actually a really good idea. The problem is now everybody's going to want you to

spk_0 wear them all the time. They're going to say, where are the red glasses?

spk_0 I can vouch for that. So it's been over a year since we did the last pod.

spk_0 Yeah. Over 40% of your revenue today is inference.

spk_0 But inference is about ready because of chain of reasoning.

spk_0 Yeah. Right. It's about to go up by a billion times.

spk_0 Right. By the way, a million x by a billion x.

spk_0 That's right. That's the part that most people have, you know,

spk_0 haven't completely internalized. This is that industry we were talking about,

spk_0 but this is the industrial revolution.

spk_0 Honestly, it's felt like you and I have had a continuation of the pod every day since then,

spk_0 you know, in AI time. It's been about a hundred years.

spk_0 I was rewatching the pod recently and the many things that we talked about that stood out.

spk_0 The most- the one that was probably most profound for me was you pounding the table

spk_0 that, you know, remember at the time there was kind of a slump in terms of pre-training?

spk_0 And people were like, oh my god.

spk_0 Pre-training. Right. The end of pre-training. We're over building.

spk_0 Yeah. This is about a year and a half ago.

spk_0 And you said inference isn't going to a hundred x, a thousand x.

spk_0 It's going to one billion x, which brings us to where we are today.

spk_0 You know, you announce this huge deal.

spk_0 We ought to start there. I underestimated.

spk_0 I mean, it's going record. I underestimated.

spk_0 We now have three scaling laws, right? We have pre-training scaling law.

spk_0 We have post-training scaling law. Post-training is basically like,

spk_0 AI practicing. Yes.

spk_0 Practicing a skill until it gets it right.

spk_0 And so it tries a whole bunch of different ways.

spk_0 And in order to do that, you've got to do inference.

spk_0 So now training and inference are now integrated in reinforcement learning.

spk_0 Really complicated. And so that's called post-training.

spk_0 And then the third is inference.

spk_0 The old way of doing inference was one shot.

spk_0 Right.

spk_0 But the new way of doing inference, which we appreciate, is thinking.

spk_0 So think before you answer.

spk_0 Yeah.

spk_0 And so now you have three scaling laws.

spk_0 The longer you think, the better the quality answer you get.

spk_0 While you're thinking you do research, you go check on some ground truth.

spk_0 And you learn some things.

spk_0 You think some more.

spk_0 You go learn some more.

spk_0 And then you generate an answer.

spk_0 Don't just generate right off the bat.

spk_0 And so thinking, post-training, pre-training.

spk_0 We now have three scaling laws, not one.

spk_0 You knew that last year.

spk_0 But is your level of confidence this year

spk_0 in the inference is going to 1 billion X.

spk_0 And where that will take the levels of intelligence?

spk_0 Is it higher?

spk_0 Are you more confident this year than you were a year ago?

spk_0 A more confident this year.

spk_0 And the reason for that is because look at the agentic systems now.

spk_0

spk_0 And AI is no longer a language model.

spk_0 And AI is a system of language models.

spk_0 And they're all running concurrently.

spk_0 Maybe using tools.

spk_0 Some of us using tools.

spk_0 Some of us using doing research.

spk_0 And yeah, there's a whole bunch of stuff.

spk_0 Okay.

spk_0 And it's all multi-modality.

spk_0 Look at all the video that's been generated.

spk_0 I mean, it's just too crazy stuff.

spk_0

spk_0 It really brings us to, you know, kind of the seminal moment this week

spk_0 that everybody's talking about.

spk_0 The massive deal you announced a couple of days ago.

spk_0 With OpenAI, Stargate, where you're going to be a preferred partner.

spk_0 Invest $100 billion in the company.

spk_0 Over a period of time, they're going to build 10 gigs.

spk_0 And if they used Nvidia for those 10 gigs,

spk_0 that could be upwards of 400 billion in revenue to Nvidia.

spk_0 So help us understand.

spk_0 And you know, just tell us a little bit about that partnership.

spk_0 What it means to you, right?

spk_0 And why that investment makes so much sense for Nvidia.

spk_0 So first of all, I'll answer that last question first.

spk_0 Okay.

spk_0 I'll come back and go into my way.

spk_0 Good.

spk_0 I think that OpenAI is likely going to be the next multi-trillion dollar

spk_0 hyperscale company.

spk_0 I think you want to know.

spk_0 Why do you call it a hyperscale company?

spk_0 Hyperscale like, like Metta's a hyperscale.

spk_0 Google's a hyperscale.

spk_0 They're going to have consumer and enterprise services.

spk_0 And they are very likely going to be the world's next multi-trillion dollar

spk_0 hyperscale company.

spk_0 And I think you would agree with that.

spk_0 I agree.

spk_0 If that's the case, the opportunity to invest

spk_0 before they get there, this is some of the smartest investments we can possibly imagine.

spk_0 And you have to invest in things you know.

spk_0 Right.

spk_0 And it turns out we happen to know this space.

spk_0 And so the opportunity to invest in that, the return on that money is going to be fantastic.

spk_0 So we love the opportunity to invest.

spk_0 We don't have to invest.

spk_0 Right.

spk_0 And it's not required for us to invest, but they're giving us the opportunity to invest.

spk_0 Fantastic thing.

spk_0 Now let me start from the beginning.

spk_0 So we're partnering with OpenAI in several projects.

spk_0 First, the first project is the build out of Microsoft Azure.

spk_0 We're going to continue to do that.

spk_0 And that partnership is going fantastically when we have several years of build out to do,

spk_0 hundreds of billions that always work just to do there.

spk_0

spk_0 The second is the OCI build out.

spk_0 And I think there's some 567 gigawatts that are about to be built out.

spk_0 And so working with OCI and OpenAI and Softbank to build that out.

spk_0 Those projects are contracted or working on it, lots of work to do.

spk_0 And then the third, the third is CoreWeave.

spk_0 Right.

spk_0 And so all of CoreWeave 4, I'm talking about OpenAI still.

spk_0 Yes.

spk_0 Okay, everything in the context of OpenAI.

spk_0 And so the question is what is this new partnership?

spk_0 This new partnership is about helping OpenAI work in

spk_0 partnering with OpenAI to build their own self-build AI infrastructure for the first time.

spk_0 Right.

spk_0 And so this is us working directly with OpenAI at the chip level, at the software level,

spk_0 at the systems level, at the AI factory level,

spk_0 to help them become a fully operated, hyper-scale company.

spk_0 That's right.

spk_0 I mean, this is going to go on for some time.

spk_0 It's going to supplement.

spk_0 It's going to supplement the amount of, you know,

spk_0 they're going through two exponentials, as you know.

spk_0 Right.

spk_0 The first exponential is the number of customers is growing exponentially.

spk_0 Right.

spk_0 And the reason for that is the AI is getting better,

spk_0 the use case is getting better.

spk_0 Just about every application is connected to OpenAI now.

spk_0 And so they're going through the usage exponential.

spk_0 The second exponential is the computational exponential ever-used.

spk_0 Yes.

spk_0 Right.

spk_0 Yes.

spk_0 Instead of just a one-shot inference, it's now thinking before it answers.

spk_0 And so these two exponentials compounding the compute requirements.

spk_0 And so we got to build out all these different projects.

spk_0 And so this last one is an additive on top of everything that they've already announced.

spk_0 All the things that we're already working on with them.

spk_0 It's additive on top of that.

spk_0 And it's going to support this incredible exponential.

spk_0 One of the things you said that's really interesting to me is kind of,

spk_0 you know, they're going to be high probability,

spk_0 multi-trillion dollar company in your mind.

spk_0 You think it's a great investment.

spk_0 At the same time, you know, they're self-building.

spk_0 You're helping them self-build their data centers.

spk_0 So here to four, they've been outsourcing to Microsoft to build the data center.

spk_0 Now they want to build full stack factories themselves.

spk_0 They want to do, they want to basically have a relationship with us,

spk_0 the way that Elon and Exhaust really.

spk_0 Correct.

spk_0 I mean, Elon and Exhaust build.

spk_0 Exactly.

spk_0 But I think that's, this is a very big deal.

spk_0 When you think about the advantage that Colossus had.

spk_0 They're building full stack.

spk_0 That is a hyper-scaler.

spk_0 Because if they don't use the capacity,

spk_0 they could sell it to somebody else.

spk_0 That's right.

spk_0 And the same way, Stargate, they're building monstrous capacity.

spk_0 They think they'll need to use most of it.

spk_0 But it puts them in a position to sell it to somebody else as well.

spk_0 Sounds very much like AWS or GCP or Azure.

spk_0 That's what you're saying.

spk_0 Yeah.

spk_0 I think they'll likely use it themselves.

spk_0 And just in the case of Exhaust, they'll likely use it themselves.

spk_0 But they would like to have the same direct relationship with us,

spk_0 direct working relationship and direct purchasing relationship.

spk_0 I'm Meta.

spk_0 Just as with Zuck and Meta has with us, it's exactly a direct.

spk_0 Our relationship with between us and Sundar and Google, direct.

spk_0 Our partnership with Satya and Azure Direct, is that right?

spk_0 And so they've gotten to a large enough scale.

spk_0 They believe it's time for them to start building these direct relationships.

spk_0 I'm delighted to support that.

spk_0 And Satya knows it and Larry knows it.

spk_0 And everybody's aware of what's going on.

spk_0 And everybody's very supportive of it.

spk_0 So one of the things I find mysterious, you just mentioned,

spk_0 Oracle 300 billion Colossus, what they're building.

spk_0 We know what the sovereigns are building.

spk_0 We know what the hyperscalers are building.

spk_0 You know, Sam's talking in terms of trillions.

spk_0 But of the 25 cell site analysts on Wall Street who cover your stock.

spk_0 If I look at the consensus estimate, it basically has your growth flatlining starting in 2027.

spk_0 8% growth 2027 through 2030.

spk_0 Okay. That is the 25 people and their only job, they get paid to forecast the growth rate for

spk_0 Nvidia. So clearly we're comfortable with that, but

spk_0 we're comfortable with that.

spk_0 Okay. We have no trouble beating the numbers on a regular basis.

spk_0 Right. No, I understand that.

spk_0 But there is this interesting disconnect.

spk_0 No.

spk_0 Right. I hear it every day on CNBC in Bloomberg.

spk_0 And I think it goes to some of these questions around, you know,

spk_0 shortages leading to a glut that they don't believe.

spk_0 They say, okay, we'll give you credit for 26, but 27, you know,

spk_0 maybe we'll have two months and you're not going to need that.

spk_0 But it is interesting to me.

spk_0 And I think it's important to point out that your consensus forecast is that this won't happen.

spk_0 Right.

spk_0 And we also put together forecast, you know, for the company,

spk_0 taking into account all of these numbers.

spk_0 And what it shows me is still, even though we're two and a half years into the age of AI,

spk_0 a massive divergence of belief between what we hear Sam Altman saying,

spk_0 you say and Sundar saying, such a saying, and what Wall Street still believes.

spk_0 And, you know, again, you're comfortable with that.

spk_0 I also don't think it's inconsistent.

spk_0 Okay. So explain that about that.

spk_0 So first of all, for the builders, we're supposed to be building for opportunity.

spk_0 Right.

spk_0 We're builders.

spk_0 Let me give you three points to think through.

spk_0 And these three points, it'll help you, hopefully, be more comfortable with Envidia in

spk_0 this future. So the first point, and this is the laws of physics point.

spk_0 This is the most important point that general purpose computing is over, and the future is

spk_0 accelerated computing and AI computing.

spk_0

spk_0 That's the first point.

spk_0 And so the way to think about that is there's how much, how many trillions of dollars of computing

spk_0 infrastructures in the world that has to be refreshed.

spk_0 Right.

spk_0 And when it gets refreshed, it's going to be accelerated.

spk_0 That's right.

spk_0 And so the first thing you have to realize is that general purpose computing, and nobody disputes

spk_0 that. Everybody goes, yeah, we completely agree with that.

spk_0 General purpose computing is over.

spk_0 Moore's Law is dead.

spk_0 People say these things.

spk_0 And so what does that mean?

spk_0 So general purpose computing is going to go to accelerated computing.

spk_0 Our partnership with Intel is recognizing that general purpose computing needs to be fused

spk_0 with accelerated computing to create opportunities for them.

spk_0 Is that right?

spk_0 And so one, general purpose computing is shifting to accelerated computing and AI, too.

spk_0 The first use case of AI is actually already everywhere.

spk_0

spk_0 It's in search, recommender engines.

spk_0 Isn't that right?

spk_0 In shopping, the basic hyperscale computing infrastructure used to be CPUs doing

spk_0 recommenders is now going to GPUs doing AI.

spk_0 Right.

spk_0 So you just take classical computing.

spk_0 It's going to accelerate computing AI.

spk_0 You take hyperscale computing is going from CPUs to accelerated computing and AI.

spk_0 And then now that's the second point.

spk_0 Just feeding the meta as the Googles, the bi-dances, the Amazon's,

spk_0 and take their classical traditional way of doing hyperscale and moving into AI,

spk_0 that's hundreds of billions of dollars.

spk_0 And because that may mean four billion people on the planet today, if you take TikTok, meta into

spk_0 account, Google into account, who are already demanding workloads that are driven by

spk_0 several of these things.

spk_0 That's exactly right.

spk_0 And so without even thinking about AI creating new opportunities,

spk_0 it's about AI shifting how you used to do something to the way new way of doing something.

spk_0 And then now let's talk about the future.

spk_0 I just so far have only spoken kind of largely about what is mundane stuff.

spk_0 Just mundane stuff.

spk_0 The old way is now wrong.

spk_0 You're no longer going to use fuel light lanterns.

spk_0 You're going to go to electricity.

spk_0 That's all.

spk_0

spk_0 And you no longer, you know, prop planes.

spk_0 You're going to go to jets.

spk_0 That's all.

spk_0 You know, that's so far.

spk_0 You know, that's all I've talked about.

spk_0 And then now that the incredible thing is, when you go to AI,

spk_0 when you go to accelerated computing, then what happens?

spk_0 What are the new applications that emerge as a result?

spk_0 And that's all the AI stuff that we're talking about.

spk_0 And that's the that opportunity.

spk_0 What is it?

spk_0 What does that look like?

spk_0 Well, the simple way of thinking about that is

spk_0 where motors replaced labor and physical activity.

spk_0 We now have AI, these AI supercomputers,

spk_0 these AI factories that I talk about.

spk_0 They're going to generate tokens to augment human intelligence, right?

spk_0 And human intelligence represents what?

spk_0 55, 65% of the world's GDP.

spk_0 Let's call it 50 trillion dollars.

spk_0 And that 50 trillion dollars is going to get augmented by something.

spk_0 And so let's just let's come back to a single person.

spk_0 Suppose I were to hire a $100,000 employee.

spk_0 And I augmented that $100,000 employee with a $10,000 AI.

spk_0 And that $10,000 AI as a result made the $100,000 employee twice more productive,

spk_0 three times more productive.

spk_0 Would I do it?

spk_0 Heartbeat.

spk_0 I'm doing it across every single person in our company right now, right?

spk_0 Every single coagents.

spk_0 That's right.

spk_0 Every single worker.

spk_0 Every single chip designer in our company already has AI's working with them.

spk_0 100% coverage.

spk_0 As a result, the number of chips we're building is better.

spk_0 The number is growing.

spk_0 The pace at which we're doing it is right.

spk_0 And so we're growing faster as a company.

spk_0 As a result, we're hiring more people.

spk_0 Our productivity is greater.

spk_0 Our top line is greater.

spk_0 Our profitability is greater.

spk_0 What's not to love about that?

spk_0 Now apply the NVIDIA story to the world's GDP.

spk_0 And so what's likely to happen?

spk_0 Is that that $50 trillion is augmented by, let's pick a number, $10 trillion.

spk_0 That $10 trillion needs to run on a machine.

spk_0 Now the reason that AI is different than IT in the past.

spk_0 In a way, software was written a priori and then it runs on a CPU.

spk_0 And it runs a person would operate it.

spk_0 In the future, of course, AI is generating tokens.

spk_0 But a machine has to generate the tokens and it's thinking.

spk_0 So that software is running all the time.

spk_0 Whereas in the past, the software is written once.

spk_0 Now the software is in fact writing all the time.

spk_0 It's thinking.

spk_0 In order for the AI to think, it needs a factory.

spk_0 And so let's say that that $10 trillion of token generated,

spk_0 50% gross margins,

spk_0 and 5 trillion of it needs a factory.

spk_0 It needs an AI infrastructure.

spk_0 So if you told me that on an annual basis,

spk_0 the cap ex of the world was about $5 trillion,

spk_0 I would say the math seems to make sense.

spk_0 And that's kind of the future, right?

spk_0 The going from general purpose computing,

spk_0 accelerated computing, replacing all the hyper-scales with AI,

spk_0 and then now augmenting human intelligence for the world's GDP.

spk_0 And today that market is about, our estimate is about 400 billion annually.

spk_0 So the TAM is a 4 to 5x increase.

spk_0 Over where it is today.

spk_0 Yeah.

spk_0 Eddie last night, Eddie, Elie Boba said,

spk_0 between now and the end of the year,

spk_0 and excuse me, now in the end of the decade,

spk_0 they're going to increase their data center power by 10x.

spk_0 Right.

spk_0 You just said how much?

spk_0 4x.

spk_0 There you go.

spk_0 They're going to increase power by 10x.

spk_0 And we correlate the power.

spk_0 And Vita's revenue is almost correlated to power.

spk_0 Yes.

spk_0 Yeah.

spk_0 That's right.

spk_0 Yeah. Because more of us.

spk_0 Another thing, what else did he say?

spk_0 He said, token generation is doubling every few months.

spk_0 Yeah.

spk_0 What's that saying?

spk_0 The Perf Per watt has to keep on going exponentially.

spk_0 That's why Vita is like cranking it out with Perf Per watt.

spk_0 And revenue Per watt is, you know,

spk_0 watt is basically revenues in this future.

spk_0 Embedded in this assumption, I find it very fascinating.

spk_0 Historical context.

spk_0 Right.

spk_0 For 2000 years, basically GDP did not grow.

spk_0 Okay.

spk_0 And then we get the industrial revolution.

spk_0 GDP accelerates.

spk_0 We get the digital revolution.

spk_0 GDP accelerates.

spk_0 And it basically what you're saying.

spk_0 It's got best in this set.

spk_0 He said, I think we're going to have 4% GDP growth next year.

spk_0 Basically, what you're saying is the world's GDP growth is going to accelerate.

spk_0 Because now we are giving the world billions of co-workers that will do work for us.

spk_0 And if GDP is an amount of output for a fixed amount of labor and capital,

spk_0 right, it has to accelerate.

spk_0 It has to.

spk_0 Right.

spk_0 It has to.

spk_0 Look at what's going on with AI.

spk_0 As a result of the technology of AI,

spk_0 and that technology of AI, let's just call it the large language models and all the AI agents.

spk_0 It's now creating a new industry of AI agents.

spk_0 There's no question about that.

spk_0 Okay.

spk_0 So that's opening eyes to fastest growing revenue company in history.

spk_0 Right.

spk_0 And they're growing exponentially.

spk_0 Right. And so AI itself is a fast growing industry.

spk_0 Because of AI needs a factory behind it.

spk_0 Right.

spk_0 An infrastructure behind it.

spk_0 There's this industry is growing.

spk_0 My industry is growing.

spk_0

spk_0 And because my industry is growing,

spk_0 the industry underneath it is growing.

spk_0 Energy is growing.

spk_0 Power.

spk_0 Shell.

spk_0 This is the end.

spk_0 This is like renaissance for the energy industry.

spk_0 Nuclear energy.

spk_0 Gas turbines.

spk_0 I mean, look at all of those companies in the infrastructure ecosystem.

spk_0 Underneath us,

spk_0 they're doing incredibly well.

spk_0 Everybody's growing.

spk_0 These numbers have everybody talking about a glut or a bubble.

spk_0 Right.

spk_0 Zuckerberg said last week on a podcast,

spk_0 you know, he said,

spk_0 listen, I think it's quite possible at some point

spk_0 that we will have an air pocket.

spk_0 And Meta may in fact overspend by $10 billion or whatever,

spk_0 but he said it doesn't matter.

spk_0 It's so existential to the future of his business

spk_0 that it's a risk that they have to take.

spk_0 But when you think about that,

spk_0 it sounds a little bit like prisoners dilemma, right?

spk_0 And walk us again through.

spk_0 These are very happy prisoners.

spk_0 Walk us again through, right?

spk_0 To date, our estimate is that we're going to have

spk_0 a hundred billion of AI revenue in 2026.

spk_0 Excluding Meta,

spk_0 excluding, you know, the GPUs running recommender engines.

spk_0 Yeah.

spk_0 Okay.

spk_0 So there's,

spk_0 or search or correct.

spk_0 So there's other stuff.

spk_0 But let's call it a hundred billion.

spk_0 What is that industry anyways?

spk_0 What is the industry already in the hyperscale?

spk_0 What is the hyperscale's, you know, between...

spk_0 Trillion.

spk_0 Yeah, exactly.

spk_0 By the way, that industry is going to AI.

spk_0 Before anybody starts at zero, you got to start there.

spk_0 But I think the skeptics would say,

spk_0 we need to go from a hundred billion of AI revenue in 26

spk_0 to at least a trillion of AI revenue in 2030.

spk_0 Okay.

spk_0 You just were talking a minute ago about five trillion

spk_0 when you look at kind of global GDP.

spk_0 If you do it, did a bottoms up,

spk_0 can you see your way to a trillion dollars of AI driven revenues

spk_0 from a hundred billion over the course of the next five years?

spk_0 Are we growing that fast?

spk_0 Yes.

spk_0 And I would also say we're already there.

spk_0 Okay.

spk_0 Explain that.

spk_0 Because the hyperscalers,

spk_0 they went from CPUs to AI.

spk_0 Okay.

spk_0 Their entire revenue base is all now AI driven.

spk_0 Correct.

spk_0 You can't do TikTok without AI.

spk_0 Correct.

spk_0 You can't do YouTube short without AI.

spk_0 You can't, you know, you can't do any of the stuff without AI.

spk_0 The amazing things that Meta is doing for,

spk_0 for, you know, customized content, personalized content,

spk_0 you can't do that without AI.

spk_0 It's all of that stuff used to be humans, you know, doing

spk_0 content a priori, creating four choices that are then

spk_0 selected by a recommended engine.

spk_0

spk_0 And now it's infinite number of choices generated by an AI, right?

spk_0 But those things are all right.

spk_0 Like we had the translation from CPUs to GPUs largely

spk_0 for those recommended engines.

spk_0 And now they're going.

spk_0 That's fairly new.

spk_0 I would say.

spk_0 In the last three or four years,

spk_0 yeah, Zuck would tell you,

spk_0 I was at SIGRAF and Zuck would tell you,

spk_0 you know, they were late getting to shows.

spk_0 For sure.

spk_0 GPUs for Meta is what, a couple of years?

spk_0 Okay.

spk_0 A year and a half.

spk_0 It's pretty new.

spk_0 Search would GPUs?

spk_0 For sure.

spk_0 Brand spanking new.

spk_0 For sure.

spk_0 Brand spanking new.

spk_0 Search for GPUs on GPUs.

spk_0 So your argument would be the probability that we're going to have a trillion

spk_0 dollars of AI revenues by 2030 is near certain because we're all

spk_0 already.

spk_0

spk_0 Already.

spk_0

spk_0 Okay.

spk_0 Let's just talk about incremental.

spk_0 Incremental.

spk_0 Now we can talk about incremental.

spk_0

spk_0 Right.

spk_0 As you do your bottoms up or your tops down,

spk_0 I just heard your tops down about percentage of global GDP.

spk_0

spk_0 What is the percentage probability that you think will have a glut,

spk_0 will run into a glut in the next three or four or five years?

spk_0 Right.

spk_0 It's a distribution of,

spk_0 we don't know the future.

spk_0 It's a distribution of properties.

spk_0 Until.

spk_0

spk_0 We.

spk_0 Fully.

spk_0 Convert.

spk_0 All.

spk_0 General purpose computing.

spk_0 To accelerate a computing and AI.

spk_0 Until we do that.

spk_0 Yes.

spk_0 I think the chances are extremely low.

spk_0 Okay.

spk_0 And that will take a few years.

spk_0 They'll take a few years.

spk_0 Yeah.

spk_0 Let me ask one more.

spk_0 And that I'll kill all recommender engines are AI based until all content

spk_0 generation is AI based because content generation.

spk_0 Consumer oriented content generation is very largely

spk_0 recommender systems and so on so forth.

spk_0 And all of that's going to be AI generated until until all of this stuff,

spk_0 what classically was hyper scale now transitions to AI,

spk_0 everything from shopping to e-commerce to all that stuff until everything goes over.

spk_0 But all this new build,

spk_0 right?

spk_0 When we're talking about trillions,

spk_0 we're investing ahead of where we are.

spk_0 You know,

spk_0 is that like at will?

spk_0 Are you obliged to invest the money even if you see a slowdown or a kind of a glut coming?

spk_0 Or is this one of these things that you're just waving the flag

spk_0 to the ecosystem to say get out and build?

spk_0 And at some point in time, if we see some of this slowdown,

spk_0 we can always pull back on the level of investment.

spk_0 Actually, it's the other way because we're at the end of the supply chain, right?

spk_0 And so we respond to demand.

spk_0 Okay.

spk_0 And right now,

spk_0 all the VCs will tell you, you guys know, the demand, the short, there's a shortage of

spk_0 compute in the world.

spk_0 Not because there's a shortage of GPUs in the world.

spk_0 Okay.

spk_0 If they give me an order, I'll build it.

spk_0 Right.

spk_0 We've over the last couple of years, we've really plumbed the supply chain.

spk_0 So all of the supply chain behind me from way for starts to co-host HBM memories,

spk_0 you know, all of that technology, we've really geared up.

spk_0 Yeah.

spk_0 If we need to double or double.

spk_0 Yes.

spk_0 Okay.

spk_0 So the supply chain is ready.

spk_0 Now we're just waiting for demand signals.

spk_0 And when the CSPs and the hyperscalers and our customers do their annual plan and they give us,

spk_0 you know, their forecast,

spk_0 we respond to that and we build to that.

spk_0 Now what's going on, of course, is that every one of their forecasts that they provide us,

spk_0 it turns out to have been wrong.

spk_0 Right.

spk_0 Because they under-forecast it.

spk_0 And so now we're always in a scramble mode.

spk_0 And so we've been in the scramble mode now for, you know, a couple of years.

spk_0 And it's whatever forecast we've been given has been always significant increase from last year.

spk_0 But not a, not a no.

spk_0 Such a last year seemed to be pulling back a little bit, you know, seemed to be,

spk_0 you know, some people called them the adult in the room, tamping down kind of some of these,

spk_0 these expectations a few weeks ago.

spk_0 He said, hey, I've also built two gigs this year and we're going to accelerate in the future.

spk_0 Do you see some of the traditional hyperscalers that may have been moving a little slower than,

spk_0 let's call it a core weave or, or, or a Elon X or maybe a little slower than Stargate?

spk_0 Do you see them all? It sounds like to me, they're all leaning in more now and they're all also.

spk_0 Because of the second exponential.

spk_0 Okay.

spk_0 We've already had one exponential we were experiencing, which was the adoption rate of AI,

spk_0 the engagement of AI was growing exponentially.

spk_0 Yes.

spk_0 The second exponential that kicked in was reasoning.

spk_0 Yeah.

spk_0 That was the conversation we had one year ago.

spk_0 One year ago.

spk_0 Yeah.

spk_0 It said, hey, listen, the moment you take AI from one shot,

spk_0 memorizing an answer and generalizing that's basically portraying.

spk_0 Memorizing an answer, what's eight times eight?

spk_0 Just memorize it.

spk_0 Memorizing an answer and generalizing that was one shot AI.

spk_0 Now a year ago, reasoning came about, research came about,

spk_0 tool use came about, and now you're thinking AI.

spk_0 One billion X.

spk_0 It's going to use a lot more compute.

spk_0 Certain hyperscale customers to your point had internal workloads that they had to migrate

spk_0 anyways from general purpose computing to accelerated computing.

spk_0 So they built through the cycle.

spk_0 I think maybe some hyperscalers had different workloads.

spk_0 So they weren't quite sure how quickly they could digest it.

spk_0 That's right.

spk_0 Everyone has now concluded that they dramatically underbuilt.

spk_0 One of the applications that my favorite is just good old-fashioned data processing.

spk_0 Structured data and unstructured data.

spk_0 Just good old-fashioned data processing.

spk_0 And very soon we're going to announce a very big initiative

spk_0 of accelerated data processing.

spk_0 Data processing represents the vast majority of the world's CPUs today.

spk_0 It still completely runs on CPUs.

spk_0 If you go to data breaks, it's mostly CPUs.

spk_0 If you go to snowflakes, mostly CPUs.

spk_0 A SQL processing at Oracle, mostly CPUs.

spk_0 Everybody's using CPUs to do SQL, structured data.

spk_0 In the future, that's all going to move to AI data.

spk_0 That is one gigantic massive market that we're going to move to.

spk_0 But you need everything that Nvidia does requires.

spk_0 Acceleration layers and requires domain-specific data processing.

spk_0 Recipes.

spk_0 We got to go and build that.

spk_0 But that's coming.

spk_0 So one of the pushbacks I turned on CNBC yesterday.

spk_0 They were like, oh, glut, bubble.

spk_0 When I turned on Bloomberg, it was about round tripping and circular revenues.

spk_0 Okay.

spk_0 And so for the benefit of people at home,

spk_0 know these arrangements are when companies enter into a misleading transaction

spk_0 that artificially inflates revenue without any underlying economic substance.

spk_0 So in other words, gross propped up by financial engineering,

spk_0 not by customer demand.

spk_0 And the canonical case everybody's referencing, of course,

spk_0 is Cisco and Norteil from the last bubble 25 years ago.

spk_0 So when you guys are Microsoft or Amazon are investing in companies

spk_0 that are also your big customers, in this case, you guys investing in OpenAI.

spk_0 While OpenAI is by intense ability to chips,

spk_0 just remind us and remind everybody else like, what is it?

spk_0 What are the analysts on Bloomberg and otherwise getting wrong

spk_0 when they're hyperventilating about circular revenues or about round tripping?

spk_0 10 gigawatts is like $400 billion.

spk_0 Right.

spk_0 Something like that.

spk_0 And that $400 billion will have to be largely funded by their offtake,

spk_0 right?

spk_0 Their revenues, which is growing exponentially.

spk_0 It has to be funded by their capital, the money they've raised through equity,

spk_0 and whatever debt they can raise.

spk_0 Those are the three vehicles.

spk_0 And the equity that they could raise and the debt that they could raise

spk_0 has something to do with the confidence of the revenues that they could sustain.

spk_0 For sure.

spk_0 And so smart investors and smart lenders will consider all of these factors.

spk_0 Fundamentally, that's what they're going to do.

spk_0 That's their company.

spk_0 It's not my business.

spk_0 And of course, we have to stay very close to them to make sure that we build in support of their

spk_0 continued growth.

spk_0 And so there's the revenue side of it and it's nothing to do with the investment side of it.

spk_0 The investment side of it's not tied to anything.

spk_0 It's an opportunity to invest in them.

spk_0 And as we were mentioning earlier,

spk_0 this is likely going to be the next multi-trillion dollar

spk_0 hyper scale company.

spk_0 And who doesn't want to be an investor in that?

spk_0 And my only regret is that they invited us to invest early on.

spk_0 I remember those conversations.

spk_0 And we were so poor.

spk_0 We were so poor we didn't invest enough.

spk_0 And I should have given them all my money.

spk_0 And the reality is, if you guys don't do your jobs and keep up with,

spk_0 if Vera Rubin doesn't turn into a good chip,

spk_0 they can go get other chips and put them in these days.

spk_0 That's right.

spk_0 There's no obligation that they have to use your chips.

spk_0 And like you said, you're looking at this as an opportunistic equity investment.

spk_0 The other thing I would say,

spk_0 and we've made some great investments.

spk_0 I got to put it out there.

spk_0 We invested in XAI.

spk_0 We invested in CoreWeave.

spk_0 Incredible.

spk_0 Yeah.

spk_0 How smart was that?

spk_0 As I go back to this, the other fundamental thing it seems to me is,

spk_0 you're putting it out there.

spk_0 You're saying this is what we're doing.

spk_0 And the underlying economic substance here, right?

spk_0 It's not that you're just somehow sending revenues back and forth between the two companies.

spk_0 We got people sending money every month for Chatchee PT,

spk_0 a billion and a half monthly users using the product.

spk_0 You just said, every enterprise in the world is either going to do this or they will die.

spk_0 Every sovereign views this as existential

spk_0 to their national security and economic security is nuclear power.

spk_0 What person,

spk_0 company or nation says, intelligence is basically optional for us.

spk_0 For us.

spk_0 I mean, it's fundamental to them.

spk_0 Well, I've been to the automation of intelligence.

spk_0 I beat the demand question to death.

spk_0 So let's jump in a little bit to system design.

spk_0 And I'm going to turn to Clark here in a second on that.

spk_0 But in 2024, you switched to your annual release cycle,

spk_0 right with Hopper.

spk_0 You then had a massive upgrade which required significant data center overhaul

spk_0 with Grace Blackwell in 2025.

spk_0 And in the back half of 26, we're going to get Vera Rubin,

spk_0 27 will get Ultron and 28 Feynman.

spk_0 How is the annual release cycle going?

spk_0 Okay.

spk_0 What were the main goals of going to an annual release cycle?

spk_0 And did AI inside Nvidia allow you to execute the annual release cycle?

spk_0 Mm-hmm.

spk_0 Yeah, the answer is yes.

spk_0 On the back, the last question.

spk_0 Without it,

spk_0 Nvidia's velocity or pace or scale would be limited.

spk_0 And so without AI these days,

spk_0 it's just simply not possible to build what we build.

spk_0 Now, why do we do it?

spk_0 There's something that, remember,

spk_0 Eddie said it at his earnings call or his conference.

spk_0 Satya has said it.

spk_0 Sam has said it.

spk_0 The token generation rate is going up exponentially.

spk_0 Yeah.

spk_0 And the customer use is going up exponentially.

spk_0 I think there are 800 million weekly active users or something like that.

spk_0 Yes.

spk_0 I mean, that's about two years from Chanchi PT, right?

spk_0 And each of those users is generating massively more tokens because they're using

spk_0 inference time reasoning.

spk_0 That's right.

spk_0 Exactly.

spk_0 And so the first thing is because the token generation rate is going up so incredibly,

spk_0 two exponentials on top of each other.

spk_0 We have to, unless we increase the performance

spk_0 at incredible rates, the cost of token generation will keep growing because more's loss dead,

spk_0 right?

spk_0 Because transistors basically cost the same every single year now.

spk_0 And power is largely the same between those two fundamental laws,

spk_0 unless we come up with new technologies to drive the cost down.

spk_0 Even if there's a slight difference in growth,

spk_0 you give somebody a discount of a few percent.

spk_0 How's that going to make up for two exponentials?

spk_0 And so we have to increase our performance annually at a pace that keeps up with that exponential.

spk_0 So in the case of going from, I guess, a Kepler to,

spk_0 all the way to Kepler, all the way to Hopper was probably a 100,000X.

spk_0 That was the beginning of the AI journey for Nvidia, 100,000X in 10 years.

spk_0 Okay. Between Hopper and Blackwell, we increased because of MV-Link 72,

spk_0 30X in one year.

spk_0 And then we'll get another X-factor again with Ruben.

spk_0 And then we'll get another X-factor with Feynman.

spk_0 And the way we do that is because the transistors aren't really helping us very much,

spk_0 right?

spk_0 More's loss largely.

spk_0 The density is growing up, but going up, but the performance is not.

spk_0 If that's the case, one of the challenges that we have to do is we have to break the entire

spk_0 problem down at the system level and change every chip at the same time and all the software

spk_0 stack and all the systems all at the same time.

spk_0 The ultimate extreme code design.

spk_0 Nobody has ever code designed at this level before.

spk_0 We change the CPU, revolutionize the CPU, a GPU, the networking chip, the MV-Link scale up,

spk_0 the SpectrumX scale out.

spk_0 Somebody said, I heard somebody said, oh yeah, it's just Ethernet.

spk_0 Yeah, right.

spk_0 Okay. So SpectrumX Ethernet is not just Ethernet and people are starting to discover,

spk_0 oh my God, the X-factress is pretty incredible.

spk_0 Right.

spk_0 You know, MV-S Ethernet business, the just Ethernet business is the fastest growing Ethernet

spk_0 business in the world.

spk_0 And so scale out.

spk_0 And of course, now we have to build even larger systems so we scale across multiple AI factories

spk_0 connected together.

spk_0 And then we do this at an annual pace.

spk_0 And so we now have an exponential of exponentials going ourselves for technology.

spk_0 And that allows our customers to drive the cost of tokens down,

spk_0 keep making those tokens smarter and smarter with pre-training and post-training and thinking.

spk_0 And as a result, when the AI gets smarter, they get more used.

spk_0 When they get more used, they're going to grow exponentially.

spk_0 For people who may not be as familiar.

spk_0 What does extreme code design mean?

spk_0 Extreme code design means that you have to optimize the model, algorithm,

spk_0 system, and chip at the same time.

spk_0 You have to innovate outside the box.

spk_0 Because Moore's Law said, you just have to keep making the CPU faster and faster.

spk_0 Everything got faster.

spk_0 You were innovating within a box.

spk_0 Just make that chip faster.

spk_0 Yeah.

spk_0 Well, if that chip doesn't go any faster, then what are you going to do?

spk_0 Innovate outside the box.

spk_0 And so, Nvidia really changed things because we did two things.

spk_0 We invented CUDA, invented GPUs, and we invented the idea of code design at a very large scale.

spk_0 That's why there's all these industries we're in.

spk_0 We're creating all these libraries and code design.

spk_0 Number one, full stack.

spk_0 Extreme is even beyond software and GPUs.

spk_0 It's now at the data center level, switches and networking and all of that software

spk_0 in the switches and networking in the nicks, the scale up, the scale out,

spk_0 optimizing across all of that.

spk_0 As a result of that, Blackwell to Hopper is 30X.

spk_0 No Moore's Law could possibly achieve that.

spk_0 And so, that's extreme.

spk_0 And that comes from the extreme code design.

spk_0 That's because Nvidia has, that's why we got into networking and switching and scale up and

spk_0 scale out of scale across and building CPUs and building GPUs and building nicks.

spk_0 You know, that's the reason why Nvidia is so rich in software and people.

spk_0 We check in more open source software in the world than just about anybody except one other

spk_0 company. I think it's AI too or something like that.

spk_0 And so, we have such enormous richness of software.

spk_0 And that's just in AI.

spk_0 Don't forget computer graphics and digital biology and times vehicles and

spk_0 you know, the amount of software we produce as a company is incredible.

spk_0 That allows us to do deep and extreme code design.

spk_0 I heard from one of your competitors, you know, yes, he's doing this because it helps

spk_0 drive down the cost of token generation.

spk_0 But at the same time, your annual release cycle makes it almost impossible for your competitors

spk_0 to keep up.

spk_0 The supply chain gets locked up more because you're giving three year visibility to your supply

spk_0 chain. So now the supply chain has confidence as to what they can build to.

spk_0 So do you think about this way before you before you ask the question, think about this.

spk_0 In order for us to do several hundred billion dollars a year of AI infrastructure build out,

spk_0 yes, think about how much capacity we had to go start a year ago.

spk_0 Yes.

spk_0 We're talking about building hundreds of billions of dollars of

spk_0 wafer starts and DRAM buys and are you guys talking?

spk_0 This is now at a scale that hardly any company can keep up.

spk_0 So would you say your competitive moat is greater today than it was three years ago?

spk_0 Yeah, you know, first of all, there's just more competition than ever before,

spk_0 but it's harder than ever before.

spk_0 And the reason why I say that is because wafer costs are getting higher,

spk_0 which means that unless you do co-design at an extreme scale, you're just not going to be able

spk_0 to deliver the X factor growth.

spk_0 Number one, number and so you, you know, unless you're working on six, seven, eight chips a year,

spk_0 that's amazing thing.

spk_0 It's not about building an ASIC.

spk_0 It's about building an AI factory system.

spk_0 And this system has a lot of chips in it and they're all co-designed.

spk_0 And together, they deliver that 10X factor that we get almost regularly.

spk_0 Okay, so number one, the co-design is extreme.

spk_0 The second thing is that the scale is extreme.

spk_0 When your customers deploy a gigawatt, that's a 400,000, 500,000 GPUs.

spk_0 Getting 500,000 GPUs to work together is a miracle.

spk_0 I mean, it's just a miracle.

spk_0 And so your customers are taking an enormous risk on you to go buy all of this.

spk_0 You got to ask yourself, what customer would place a $50 billion P.O.

spk_0 on an architecture?

spk_0 Right.

spk_0 On an unproven architecture.

spk_0 That's right.

spk_0 A new one.

spk_0 A new architecture.

spk_0 You just taped out a whole new chip.

spk_0 You're as excited as you are about it, you know, and everybody's excited for you.

spk_0 And you just showed it first, Silicon.

spk_0 Who's going to give you $50 billion P.O.?

spk_0 And why would you start $50 billion with a wafer for a chip that just taped out?

spk_0 But for Envidia, we could do that because our architecture is so proven.

spk_0 So the scale of our customers, so incredible.

spk_0 Now the scale of our supply chain is incredible.

spk_0 Who's going to start all of that stuff, pre-build all of that stuff for a company,

spk_0 unless they know that Envidia can deliver through, isn't it right?

spk_0 And they believe that we can deliver through to all of the customers around the world.

spk_0 They're willing to start several hundred billion dollars at a time.

spk_0 This is the scale's incredible.

spk_0 To that point, you know, one of the biggest key debates and controversies in the world

spk_0 is this question of GPUs versus ASICs.

spk_0 Google's TPUs, Amazon's Trainiam.

spk_0 And it seems like everyone from ARM to OpenAI to Anthropic are, you know, rumored to be building one.

spk_0 Last year you said, you know, we're building systems, not chips.

spk_0 And you're driving performance through every single part of that stack.

spk_0 You also said that many of these projects may never get to production scale.

spk_0 But given like the seeming-

spk_0 Most of them.

spk_0 Given the seeming success of Google's TPUs,

spk_0 you know, how are you thinking about this evolving landscape today?

spk_0 Yeah.

spk_0 And-

spk_0 First of all, the advantage that

spk_0 that Google had is foresight.

spk_0 Remember, they started TPU 1

spk_0 before everything started.

spk_0 You know, this is no different than a startup.

spk_0 You're supposed to build a startup.

spk_0 You're supposed to create a startup before the market grows.

spk_0 You're not supposed to come up as a startup when the market's a trillion dollars large.

spk_0 You know, this fallacy and all VCs know this, this fallacy,

spk_0 that a large market, if you could just take a few percent market share,

spk_0 you could be a giant company.

spk_0 Right.

spk_0 That's actually fundamentally wrong.

spk_0 You're supposed to take 100 percent of a tiny company, a tiny industry,

spk_0 which is what Nvidia did, right?

spk_0 Which is what TPUs did.

spk_0 There were only the two of us.

spk_0 But you better hope that that industry gets really big.

spk_0 You're creating an industry.

spk_0 That's right.

spk_0 Right.

spk_0 That's right.

spk_0 And I mean, the Nvidia story,

spk_0 you know, which is-

spk_0 And so that's the challenge for the-

spk_0 For people who are building A6 now.

spk_0 It looks like a juicy market.

spk_0 But remember, this juicy market has evolved from a chip called the GPU.

spk_0 So I just described an AI factory.

spk_0 And you guys just saw I just announced a chip called CPX for context processing

spk_0 and diffusion video generation.

spk_0 A very specialized workload.

spk_0 But an important workload inside the data center.

spk_0 I just pre-looted to maybe AI data processing processors.

spk_0 Because guess what?

spk_0 You need long-term memory.

spk_0 You need short-term memory.

spk_0 The KV cache processing is really intense.

spk_0 AI memory is a big deal.

spk_0 You know, you kind of like your AI to have good memory.

spk_0 And just dealing with all the KV caching around the system,

spk_0 really complicated stuff.

spk_0 Maybe it wants to have a specialized processor.

spk_0 Maybe there's other things, right?

spk_0 So you see that Nvidia's our-

spk_0 Our viewpoint is now not GPU.

spk_0 Our viewpoint is looking at the entire AI infrastructure.

spk_0 And what does it take for these incredible companies

spk_0 to get all of their workload through it,

spk_0 which is diverse and changing?

spk_0 Look at the transformer.

spk_0 The transformer architecture is changing incredibly.

spk_0 If not for the fact that KUDA is easy to operate on and iterate on,

spk_0 how do they try all of their vast number of experiments

spk_0 to decide which of the transformer versions

spk_0 what kind of attention algorithm to use?

spk_0 How do you disaggregate KUDA helps you do all that?

spk_0 Because it's so programmable.

spk_0 And so the way to think about our business now is

spk_0 you look at when all of these ASIC companies

spk_0 or ASIC projects start three, four, five years ago,

spk_0 I got to tell you, that industry was super adorable and simple.

spk_0 There was a GPU involved.

spk_0 But now it's giant and complex.

spk_0 And in another two years, it's going to be completely massive.

spk_0 The scale is going to be so large.

spk_0 And so I think that the battle of getting into

spk_0 a very large market as a nascent player, it's just hard, as you guys know.

spk_0 Even for the customers who perhaps are successful with ASICs,

spk_0 isn't there an optimal balance in their compute fleet?

spk_0 I think investors are very much binary creatures.

spk_0 They just want a yes or no black and white answer.

spk_0 But even if you get the ASIC to work,

spk_0 isn't there an optimal balance because you think

spk_0 I'm buying the Nvidia platform.

spk_0 CPX is going to come out for pre-fill, for video generation,

spk_0 maybe a decode platform.

spk_0 A video transmitter.

spk_0 Exactly.

spk_0 Yeah.

spk_0 So there will be like many different chips or parts to add to the Nvidia ecosystem.

spk_0 Accelerated compute fleet.

spk_0 As new workloads are born, and people are trying to tape out new chips today,

spk_0 are not really anticipating what's happening a year from now.

spk_0 They're just trying to get a chip to work.

spk_0 That's right.

spk_0 I said it another way.

spk_0 Google is a big GPU customer.

spk_0 If you look at, and Google is a very special case.

spk_0 We just have to show respect,

spk_0 where respect is very sure.

spk_0 We're really deserved.

spk_0 TPU is on TPU 7.

spk_0 Yes.

spk_0

spk_0 And so,

spk_0 and it's a challenge for them as well.

spk_0 Right?

spk_0 And so the work that they do is incredibly hard.

spk_0 So I think the first thing to let me do it.

spk_0 You know, remember, there are three categories of chips.

spk_0 There's the category chips that are architectural.

spk_0 X86 CPUs, ARM CPUs, and video GPUs, architectural.

spk_0 And it has an ecosystem above.

spk_0 And the architecture allows,

spk_0 has rich IP and rich ecosystem, very complicated technology.

spk_0 It's built by the owners like us.

spk_0 Okay?

spk_0 There's A6.

spk_0 I worked for the original company,

spk_0 LSI Logic, who invented the idea of A6.

spk_0 As you know, LSI Logic is not here anymore.

spk_0 And the reason for that is because A6 is really fantastic

spk_0 when the market size is not very large.

spk_0 It's easy to have somebody be a contractor to help you

spk_0 put the packaging of all that stuff together

spk_0 and do the manufacturing on your behalf.

spk_0 And they charge you 50, 60 points of margin.

spk_0 But when the market gets large for an A6,

spk_0 there's a new way of doing things called COT.

spk_0 Customer on tooling.

spk_0 And who would do something like that?

spk_0 Apple's smartphone chip, the volume is so large,

spk_0 they would never go pay somebody else 50, 60% gross margin to be an A6.

spk_0 They do customer on tooling.

spk_0 And so, where will TPUs go when it becomes a large business?

spk_0 Customer on tooling.

spk_0 There's no question about it.

spk_0 And so, A6, but there's a place for A6.

spk_0 Video trans coders will never be too large.

spk_0 Smart Nix will never be too large.

spk_0 And so, when there's 10, 12, 15 A6 projects going on in an A6 company,

spk_0 I'm not surprised by that.

spk_0 Because they're probably five smart Nix and four trans coders.

spk_0 And are they all AI chips?

spk_0 Of course not.

spk_0 And if somebody were to build an embedded,

spk_0 embedding processor for a specific recommender system,

spk_0 and that was an A6, of course, you could do that.

spk_0 But would you do that as the fundamental compute engine for AI

spk_0 that's changing all the time?

spk_0 You've got low latency workload.

spk_0 You've got high throughput workload.

spk_0 You have token generation for chat.

spk_0 You have thinking workload.

spk_0 You have a video generation workload.

spk_0 Is there a, you know, now you're talking about the very...

spk_0 That's very...

spk_0 The workhorse backbone of your acceleration.

spk_0 That's what a video is all about.

spk_0 Yeah.

spk_0 Again, done this down.

spk_0 So, like playing chess and checkers.

spk_0 Right, the fact of the matter is,

spk_0 the folks who are starting A6 today,

spk_0 whether it's Trainiam or whether it's some of these other

spk_0 intrinsic accelerators, etc.

spk_0 They're building a chip that's a component of a much larger machine.

spk_0 You've built a very sophisticated system, platform, factory,

spk_0 whatever you want to call it.

spk_0 And now you're opening up a little bit, right?

spk_0 So, you mentioned CPX GPU, right?

spk_0 That is, it seems to me that in some ways,

spk_0 you're disaggregating the workloads to the best slice of the hardware

spk_0 for that particular demand.

spk_0 And we did.

spk_0 We announced this thing called Dynamo.

spk_0 Disaggregated AI workload orchestration.

spk_0 And we open-sourced it

spk_0 because the future AI factory is disaggregated.

spk_0 And you launched MV Fusion.

spk_0 That even said to your competitors.

spk_0 Including Intel, which you just invested in.

spk_0 That's right.

spk_0 You know, the way in which you participate in this factory

spk_0 that we're building, because nobody else is crazy enough

spk_0 to try to build the entire factory.

spk_0 But you can plug into that if you have a product

spk_0 that's good enough compelling enough that the end user says,

spk_0 hey, we want to use this instead of an ARM GPU

spk_0 or we want to use this instead of your inference accelerator, etc.

spk_0 Is that correct?

spk_0 We're delighted to connect you in.

spk_0

spk_0 Tell us a little bit more.

spk_0 MV link fusion, such a great idea.

spk_0 And we're so happy to partner with Intel on that.

spk_0 It takes the Intel ecosystem.

spk_0 You know, most of the world's enterprise still runs on Intel.

spk_0 It takes the Intel ecosystem,

spk_0 takes the Nvidia AI ecosystem,

spk_0 makes all the computing and we fuse it together.

spk_0

spk_0 And we did that with ARM.

spk_0 Right?

spk_0 And there are several others we're going to be doing with.

spk_0 That opens up opportunities for both of us.

spk_0 It's a win for both of us.

spk_0 Great, great win.

spk_0 I'll be a large customer of theirs.

spk_0 And they're going to expose us to a much,

spk_0 much larger market opportunity.

spk_0 Yeah.

spk_0 It's deeply related to this idea.

spk_0 Is the argument you've made that kind of

spk_0 shocks some people where you say our competitors building A6.

spk_0 They could literally,

spk_0 all their chips are cheaper already today.

spk_0 But they could literally price them at zero.

spk_0 Our objective is they could price them at zero.

spk_0 And you would still buy an Nvidia system

spk_0 because the total cost of operating that system,

spk_0 power, data center, land, etc.

spk_0 The intelligence out is still a better bet than buying.

spk_0 Because the lamp power in shell is already $15 billion.

spk_0 Right.

spk_0 So we've taken a crack at kind of the math on that.

spk_0 But walk us through your math.

spk_0 Because I think for people who don't spend as much time here,

spk_0 it just doesn't compute.

spk_0 How could it possibly be that you were pricing

spk_0 your competitors chips at zero given the expense of your chips

spk_0 and it still is a better bet?

spk_0 There's two ways to think about it.

spk_0 One way is, let's just think about it from a perspective of revenues.

spk_0 So everybody's power limited.

spk_0 And let's say you were able to secure two more gigawatts of power.

spk_0 Well, that two gigawatts of power,

spk_0 you would like to have translate to revenues.

spk_0 Yes.

spk_0 So your performance or tokens per watt

spk_0 was twice as high as somebody else's token per watt.

spk_0 Because you deep, I did deep and extreme co-design.

spk_0 And my performance was much higher per unit energy.

spk_0 Then my customer can produce twice as much revenues

spk_0 from their data center.

spk_0 Who doesn't want twice as much revenues?

spk_0 And if somebody gave them a 15% discount,

spk_0 the difference between our gross margins was called 75 points.

spk_0 And somebody else's gross margins,

spk_0 called it 50 to 65 points.

spk_0 It's not so much as the make up for the 30 times difference between black,

spk_0 wall and hopper.

spk_0 Let's pretend hopper hoppers an amazing chip and amazing system.

spk_0 Let's pretend somebody else's asick is hopper.

spk_0 Black walls 30 times.

spk_0 So you've got to give up 30x revenues in that one gigawatts.

spk_0 It's too much to give up.

spk_0 So even if they gave it to you for free,

spk_0 you only have two gigawatts to work with.

spk_0 Your opportunity cost is so insanely high,

spk_0 you would always choose the best per per watt.

spk_0 So I heard this from one of the CFAs at one of the hyperscalers

spk_0 that given the performance improvement,

spk_0 that's coming out of your chips, again, precisely to that point, tokens per gig,

spk_0 and power being the limiting factor, that they had to upgrade to the new cycle.

spk_0 So when you look ahead, Ruben Ultra, a Feynman, does that trajectory continue?

spk_0 We're building what, six, seven chips a year now.

spk_0 And you just wanted that system.

spk_0 That's right.

spk_0 And those systems, software is everywhere and takes the integration and the optimization

spk_0 across all of those six, seven chips to deliver on the 30x black wall.

spk_0 Now imagine I'm doing this every single year.

spk_0 Bam, bam, bam, bam, bam.

spk_0 And so if you build one asick in that soup of asicks,

spk_0 in that soup of chips and we're optimizing across that,

spk_0 you know, it's a hard problem to solve.

spk_0 This does bring me back to where we started about the competitive mode.

spk_0 We've been covering this in investors for a while.

spk_0 We're investors throughout the ecosystem in competitors of yours,

spk_0 you know, from Google to Broadcom.

spk_0 But when I really just first principles around this and say,

spk_0 are you increasing or decreasing your competitive mode?

spk_0 You move to an annual cadence.

spk_0 You're co-developing with the supply chain.

spk_0 The scale is massively bigger than anybody anticipated,

spk_0 which requires scale, both a balance sheet and of development.

spk_0 Right?

spk_0 The moves you made both through acquisition and organically with things like envy fusion,

spk_0 or CPX, which we just talked about.

spk_0 All of those things together,

spk_0 cause me to believe that your competitive mode is increasing, Veeza V.

spk_0 At least in so far is building out the factory or the system.

spk_0 It's at least surprising.

spk_0 But I think it's interesting that your multiple is much lower than most of those other people.

spk_0 And I think part of that has to do with this law of large numbers,

spk_0 a four and a half trillion dollar company couldn't possibly get any bigger.

spk_0 But I asked you this a year and a half ago.

spk_0 As you sit here today,

spk_0 if the market's going to AI workloads are going to 10X or 5X,

spk_0 we know what CAPX is doing, etc.

spk_0 Is there any conceivable world in your mind

spk_0 where your top line in five years isn't two or three X bigger than it is in 2025?

spk_0 Like what's the probability that it's actually not much higher than it is today, given those

spk_0 advantages? I'll answer it this way.

spk_0 Our opportunity, as I described it, is much larger than the consensus.

spk_0 I'll say it here. I think Nvidia will likely be the first 10 trillion dollar company.

spk_0 And I've been here long enough. It wasn't that long ago, just a decade ago,

spk_0 as you well remember, the people said there could never be a trillion dollar company.

spk_0 Now we have 10.

spk_0 Right. And today, the world's bigger.

spk_0 Right. And today, this is back to the exponentials around GDP and the growth rate.

spk_0 The world is bigger.

spk_0 People misunderstand what we do.

spk_0 They remember we're a chip company.

spk_0 Right.

spk_0 And we build chips. Boy, do we build chips and build the most amazing chips in the world.

spk_0 But Nvidia is really an AI infrastructure company.

spk_0 We are your AI infrastructure partner.

spk_0 And our partnership with OpenAI is a perfect demonstration of that.

spk_0 That we are their AI infrastructure partner.

spk_0 And we work with people in a lot of different ways.

spk_0 We don't require anybody to buy everything from us.

spk_0 We don't require that they buy the full rack.

spk_0 They could buy a chip. They could buy a component.

spk_0 They could buy our networking.

spk_0 We have customers just buying only our CPU.

spk_0 Just buy our GPUs and buy somebody else's CPUs and somebody else's networking.

spk_0 We're kind of okay

spk_0 selling any way you like to buy.

spk_0 My only request is just buy a little something from us.

spk_0 You said, this isn't just about better models.

spk_0 We also have to build.

spk_0 We have to have world-class builders.

spk_0 And you said, you know, the most world-class builder maybe that we have in the country is Elon Musk.

spk_0 And we talked about Colossus 1.

spk_0 And what he was doing there standing up a couple hundred thousand

spk_0 at the time, H-100's, H-200's in a coherent cluster.

spk_0 Now he's working on Colossus 2, which may be 500,000 GBs,

spk_0 millions of H-100 equivalents in a coherent cluster.

spk_0 I would not be surprised if he gets to a gigawatt before anybody else there.

spk_0 Right.

spk_0 In one way or another.

spk_0 Yes, so say a little bit about that.

spk_0 The advantage of being, you know, the builder who, you know,

spk_0 isn't just building the software and the models,

spk_0 but understands what it takes to build those clusters.

spk_0 Well, you know, these AI supercomputers are complicated things.

spk_0 The technology is complicated.

spk_0 Procuring it is complicated because of financing issues.

spk_0 Securing the lamp power and shell.

spk_0 Powering it is complicated.

spk_0 Building it all.

spk_0 Bring it it all up.

spk_0 I mean, these are, this is unquestionably the most complex

spk_0 systems problem humanity has ever endeavored.

spk_0 And so Elon has a great advantage that in his head,

spk_0 all of these systems are interoperating

spk_0 and the interdependencies are, you know, resides in one head,

spk_0 including the financing.

spk_0 Yes.

spk_0 And so he's a big GPT.

spk_0 He's a big supercomputer himself.

spk_0 He's the ultimate GPU.

spk_0

spk_0 Yeah.

spk_0 And so he has a great advantage there.

spk_0 Yeah.

spk_0 And he has great sense of urgency.

spk_0 Yeah.

spk_0 He has a real desire to build it.

spk_0 And so when, when will comes together with, with skill,

spk_0 yeah, you know, unbelievable things can happen.

spk_0 Yeah.

spk_0 Yeah, quite unique.

spk_0 Something you've been so involved in is,

spk_0 I want to talk about sovereign AI.

spk_0 I want to talk about China and the global AI race

spk_0 that's going on, you know.

spk_0 And when I look back at you 30 years ago,

spk_0 you couldn't have imagined you were going to be hanging out in palaces with

spk_0 amirs and the king this week.

spk_0 And you're at the White House all the time.

spk_0 The president has said that you and Nvidia are critical to US,

spk_0 you know, national security.

spk_0 So when, when you look at that,

spk_0 first just contextualize for me.

spk_0 Like, it's hard to believe that you would be in those places.

spk_0 If sovereigns didn't view this at least as existential,

spk_0 as important as maybe we did nuclear in the 1940s.

spk_0 Right.

spk_0 If we don't have a Manhattan project today,

spk_0 at least funded by the government,

spk_0 but it's funded by Nvidia, it's funded by OpenAI,

spk_0 it's funded by Meta, it's funded by Google.

spk_0 We have companies today, the size of nation states,

spk_0 and thank God for America.

spk_0 Right.

spk_0 Who are funding something that it appears to me,

spk_0 presidents and kings think our,

spk_0 think is existential to their future economic and national security.

spk_0 Would you agree with that?

spk_0 Nobody needs atomic bombs.

spk_0 Everybody needs AI.

spk_0 Well said.

spk_0 Okay.

spk_0 Here, here.

spk_0 Yeah.

spk_0 Here.

spk_0 And so that's a very, very large difference.

spk_0 AI, as you know, it's modern software.

spk_0 I just, that's where I started.

spk_0 From general purpose computing to accelerated computing,

spk_0 from human-ridden code line of time to AI-ridden code,

spk_0 that foundation can't be forgotten.

spk_0 We've reinvented computing.

spk_0 There's not a new species on earth,

spk_0 which is reinvented computing,

spk_0 and everybody needs computing.

spk_0 It needs to be democratized.

spk_0 Which is the reason why

spk_0 everybody, all of these countries realize

spk_0 they have to get into the AI world.

spk_0 Because everybody needs to stay in computing.

spk_0 There's nobody in the world that says,

spk_0 guess what?

spk_0 You know, I used to use computers yesterday.

spk_0 I'm pretty good with clubs and fire tomorrow.

spk_0 And so everybody needs to move into computing.

spk_0 It's just being modernized, that's all.

spk_0

spk_0 Number one,

spk_0 it is the case that,

spk_0 that in order to participate in AI,

spk_0 you have to encode within AI your history,

spk_0 your culture, your values.

spk_0 And of course, AI is getting smarter,

spk_0 so that even the core AI is able to learn these things fairly quickly.

spk_0 You don't have to start from the ground, you know, from ground zero.

spk_0 And so I think that every country

spk_0 needs to have some sovereign capability.

spk_0 I recommend that they all use OpenAI,

spk_0 they all use Gemini,

spk_0 they all use, you know,

spk_0 these open models use GROC.

spk_0 And I think they, I recommend they all do that.

spk_0 I recommend they all use Anthropic.

spk_0 But they should also dedicate resources to learn how to build AI.

spk_0 And the reason for that is because they need to learn how to build it,

spk_0 not just for language models,

spk_0 but they need to build it for industrial models,

spk_0 manufacturing models,

spk_0 natural security models.

spk_0 Natural security models.

spk_0 There's a whole bunch of intelligence,

spk_0 they have to go cultivate themselves.

spk_0 So they ought to have sovereign capability.

spk_0 Every country should develop it.

spk_0 And is that what you see?

spk_0 Is that what you're hearing around the way they all realize it?

spk_0 They all realize it.

spk_0 And they all are going to be customers of OpenAI

spk_0 and Thropic and GROC and Gemini.

spk_0 But they all really need to also build their own infrastructure.

spk_0 And this is the big idea that what Envitya does

spk_0 is we're building infrastructure,

spk_0 just as every country needs energy infrastructure.

spk_0 The communications and internet infrastructure.

spk_0 Now every single country needs AI infrastructure.

spk_0 So you, let's start with the rest of the world.

spk_0 You know, our good friend David Sacks, the AI's are.

spk_0 Yeah, you know, we're doing a heck of a job.

spk_0 We are so lucky to have David and Shreemram in Washington, D.Z.

spk_0 Doing AI in the AI SAR.

spk_0 What a smart move by President Trump

spk_0 to put them in the White House.

spk_0 Because during this pivotal time,

spk_0 the technology is complicated.

spk_0 Shreemram is the only person in Washington, D.C.

spk_0 that I think knows Kuda.

spk_0 Yeah.

spk_0 And which is strange anyways.

spk_0 But I just love the fact that during this pivotal time

spk_0 when technology is complicated,

spk_0 policy is complicated,

spk_0 the impact to the future of our nation is so right,

spk_0 that we have somebody who is clear minded,

spk_0 dedicating the time to understand the technology

spk_0 and thoughtful to help us through that.

spk_0 And it would seem to me,

spk_0 you know, going back to the Manhattan Project analogy.

spk_0 Yeah, right.

spk_0 That you have a president who understands,

spk_0 how existential this is.

spk_0 You have governors like Greg Abbott and Texas

spk_0 who want to remove regulations to accelerate

spk_0 because they understand how important it is.

spk_0 You have secretaries right at energy

spk_0 and Doug Bergram at interior and Latinac at commerce

spk_0 who also understand how important this is.

spk_0 How pro-energy they are.

spk_0 Could you imagine?

spk_0 Could you imagine the alternative?

spk_0 If we had an administration right now

spk_0 who is not pro-energy and want energy to grow in our nation

spk_0 so that we could have AI.

spk_0 I find it.

spk_0 I just can't even think about it.

spk_0 I find it ironic that just a couple years ago

spk_0 we were saying China's building 100 nuclear reactors.

spk_0 They're so far ahead of us.

spk_0 Like that's the primitive to AI.

spk_0 But now you have people when we go to build it,

spk_0 everybody says, oh, it's a glut.

spk_0 Like it seems to me that this is something

spk_0 that the government,

spk_0 it is in their interest.

spk_0 And we have industry and government working together

spk_0 in a way that I haven't seen a long time.

spk_0 You've been around a long time.

spk_0 You're very close with President Trump at this stage.

spk_0 Help us understand like what is the nature

spk_0 of industry government relationships?

spk_0 We saw that dinner last week with all the CEOs.

spk_0 You spend a lot of time.

spk_0 Is it unique?

spk_0 Have you seen anything like this in your career

spk_0 over the last 30 years?

spk_0 It was hard to go to DC in the past, as you know.

spk_0 Getting an appointment is almost impossible.

spk_0 President Trump has an open door to leaders

spk_0 who wants to come in

spk_0 and help them understand the future.

spk_0 This is an administration that believes in growth.

spk_0 Fundamentally, President Trump wants America to grow.

spk_0 If we can grow economically,

spk_0 we will be strong militarily.

spk_0 If we could grow economically, we will be secure.

spk_0 I've never met somebody who is secure, who is poor.

spk_0 Being rich as a nation is an essential part

spk_0 of national security.

spk_0 And he knows that.

spk_0 He also wants America to win the AI race.

spk_0 This is going to be a very long-term race.

spk_0 And he understands that this is a pivotal time.

spk_0 He wants the technology industry to run.

spk_0 He wants everybody in the world to be built

spk_0 on American technology.

spk_0 These are sensible, logical things.

spk_0 The opposite is strange to me.

spk_0 I take everything and I just reversed it.

spk_0 We want our country not to grow.

spk_0 And because we don't want our country to grow,

spk_0 we don't need energy because we know we need energy to grow.

spk_0 And so let's not have any energy.

spk_0 And in fact, we don't want our technology industry to lead.

spk_0 He understands that our technology industry

spk_0 is our national treasure.

spk_0 And that technology,

spk_0 like corn and steel and things in the past,

spk_0 are now such fundamental trade opportunities.

spk_0 It's an essential part of trade.

spk_0 And why would you not want American technology

spk_0 to be coveted by everyone so that it could be used for trade?

spk_0 Right. So let's talk about, you know,

spk_0 the internet, Google spread around the world.

spk_0 We had democratic values spread around the world by way of search.

spk_0 And Google didn't have to go to Washington to get permission to do it.

spk_0 It just happened.

spk_0 We diffused our technology around the world.

spk_0 David Sacks has been crystal clear of the need

spk_0 to accelerate export licenses

spk_0 so that the American AI stack wins around the world.

spk_0 Right? We're talking chips, we're talking models,

spk_0 we're talking data centers, etc.

spk_0 We know a year and a half ago that wasn't happening.

spk_0 It was a concept that was called small yard tall fence

spk_0 or something like that, small yard tall fence.

spk_0

spk_0 And the irony of it was it was described in such a way

spk_0 and it was recommended in policy in such a way.

spk_0 It was a small yard tall fence around America.

spk_0 That was the strange part.

spk_0 I think President Trump's got to write that we want to maximize exports.

spk_0 We want to maximize American influence around the world.

spk_0 We're supposed to maximize those things.

spk_0 And do you see those licenses coming?

spk_0 Are you seeing the acceleration in Washington?

spk_0 I know it's being set at the top,

spk_0 but are you seeing it flow down through government

spk_0 that's accelerating us around the world?

spk_0 Secretary Ludnick was all over it.

spk_0 Great. Yeah.

spk_0 So now let's talk about China.

spk_0 You know, what most people may not realize

spk_0 is I think you understand China

spk_0 as well as any leader in the United States.

spk_0 We've been there for 30 years.

spk_0 Been there for 30 years.

spk_0 What most people don't realize is up until a couple of years ago,

spk_0 you had dominant market share within China in terms of...

spk_0 95% market share.

spk_0 95% market share in the most important thing, arguably.

spk_0 And you have said that our biggest own goal

spk_0 that we as a country could have

spk_0 under the guise of somehow trying to slow them down

spk_0 is we've unilaterally disarmed.

spk_0 We forced Nvidia out of China,

spk_0 which is allow Huawei to accelerate

spk_0 on the back of monopoly profits within China.

spk_0 And I just saw this morning,

spk_0 you're seeing announcements out of Huawei and Baba

spk_0 and others that they're going to build data centers

spk_0 around the world now.

spk_0 Huawei has a three-year plan to pass Nvidia

spk_0 funded by the monopoly profits

spk_0 in the biggest AI market in the world.

spk_0 So it's looking like your admonition

spk_0 that this is a huge mistake to hand China

spk_0 monopoly markets is coming true.

spk_0 The president said,

spk_0 after the ban on H20s,

spk_0 now we have a situation where you can sell chips to China,

spk_0 but there's a 15% export tax.

spk_0 But now it appears that the Chinese perhaps

spk_0 offended by statements out of the United States

spk_0 are saying, no, Nvidia is not allowed to sell here now.

spk_0 Where do we stand today between Nvidia and China?

spk_0 And can you reiterate kind of what you think

spk_0 we as a country should be doing

spk_0 to put ourselves in a best position

spk_0 to win the AI race around the world?

spk_0 We have a competitive relationship with China.

spk_0 We should acknowledge that China,

spk_0 rightfully, should want their companies to do well.

spk_0 I don't for a second be grudge him for that.

spk_0 They should do well.

spk_0 They should give them as much support as they like.

spk_0 It's all their prerogative.

spk_0 And don't forget that China has some of the best entrepreneurs in the world

spk_0 because they came from some of the best STEM schools in the world.

spk_0 They're the most hungry in the world.

spk_0 Yes.

spk_0 996, as you know, this is a very producing the most AI engineers in the world.

spk_0 The audience knows nine and more into nine at night six days a week.

spk_0 That is their culture.

spk_0 Okay, we're up against a formidable, innovative, hungry, fast moving,

spk_0 under-regulated.

spk_0 Okay, people don't realize this.

spk_0 They are very lightly regulated.

spk_0 Let's regulate it.

spk_0 Ironically, we are in the capitalist system.

spk_0 That's right.

spk_0 People think that they're centrally governed.

spk_0 But remember, the genius of China was distributed economic systems.

spk_0 And so all of these 33 provinces and all the mayor economy has driven an enormous amount

spk_0 of internal competition, internal economic vibrancy, which of course has some of its side effects.

spk_0 But this is a vibrant entrepreneurial, high-tech, modern industry.

spk_0 And to one, some of the things that I heard, they could never build AI chips.

spk_0 That just sounded insane.

spk_0 Two, that China can't manufacture.

spk_0 China can't manufacture.

spk_0 If there's one thing they could do is manufacture.

spk_0 Three, their years behind us is a two years, three years.

spk_0 Come on, they're nanoseconds behind us.

spk_0 Nanoseconds behind us.

spk_0 Yeah, they're nanoseconds behind us.

spk_0 And so we've got to go compete.

spk_0 We've got to go compete.

spk_0 So the question then becomes, what's in the best interest of China, of course,

spk_0 is that they have a vibrant industry.

spk_0 They also publicly say, and rightfully, I believe they believe this,

spk_0 is that they want China to be an open market.

spk_0 They want to attract foreign investment.

spk_0 They want companies to come to China and compete in the marketplace.

spk_0 And I believe that they, I hope, I believe in I hope, that would return to that

spk_0 in our context, answering your question, what do I see in the future?

spk_0 I do hope.

spk_0 Because they say it, their leaders say it, and I take it face value.

spk_0 And I believe it, because I think it makes sense for China,

spk_0 that what's in the best interest of China is for foreign companies to invest in China,

spk_0 compete in China, and for them to also have vibrant competition themselves.

spk_0 And they would also like to come out of China and participate around the world.

spk_0 That is, I think, is a fairly sensible outcome.

spk_0 And what we need to do as a country is to enable our technology industry, which today is the

spk_0 I'm privileged to be working in an industry.

spk_0 That is our national treasure.

spk_0 We have to acknowledge it is our national treasure.

spk_0 It is our best industry.

spk_0 It is our single best industry.

spk_0 Why would we not allow this industry to go compete for its survival,

spk_0 for this industry to go and proliferate the technology around the world so that we could

spk_0 have the world be built on top of American technology, so that we can maximize our economic success.

spk_0 Magnum maximized our geopolitical influence, maximized this technology industry,

spk_0 doing such a vibrant time, such a pivotal time to allow it to thrive.

spk_0 The skeptics says,

spk_0 Jensen just wants to sell more chips, and if he can sell them to China, great, he'll sell them to China.

spk_0 He doesn't care about what that means for America. That's a skeptic.

spk_0 Now, I just can I just address the skeptics just because I want America

spk_0 ecosystem and economy to grow. It doesn't make me wrong.

spk_0 Right. Okay.

spk_0 So first of all, everything that has been said so far, that's been made up so far,

spk_0 about China has proven to be wrong. The facts are just the ground truth is wrong.

spk_0 And so just because we want America to win, just because we want this industry to grow

spk_0 doesn't make me wrong.

spk_0 Correct.

spk_0 And I think anybody who knows you, and now the president, certainly myself,

spk_0 you deeply care about the country.

spk_0 You deeply want the United States of America to win the global AI race.

spk_0 You just happen to believe, and I think you have as much experience or more experience in anyone,

spk_0 that it enours to our advantage.

spk_0 The probability of us winning the global AI race actually goes up if you are competing in China.

spk_0 That's right.

spk_0 Because it allows us to tap in to half of the world's AI engineers,

spk_0 keeping them, you know, in this ecosystem.

spk_0 And let's be clear with the companies we're talking about here, by dance,

spk_0 allie, Baba, etc.

spk_0 These are companies that are largely owned by American investors.

spk_0 Yeah.

spk_0

spk_0 Right.

spk_0 Like these are global companies that are building recommender engines.

spk_0 And by the way, extraordinary technologies.

spk_0 Incredible companies.

spk_0 And so I think, and I'm hopeful,

spk_0 that the argument that you're making vis-à-vis China, which is a harder argument than

spk_0 diffusion to the rest of the world, I understand that.

spk_0 And that's why I thought when the president said, you know, I don't know, it's a flip of a coin.

spk_0 Maybe Jensen's right.

spk_0 Maybe the other guys are right.

spk_0 But if Jensen's willing to put a little bit of 15% into the US treasury

spk_0 as a hedge on that, then I'll go for it.

spk_0 But I was really disappointed on the heels of that.

spk_0 I think if the Chinese feel like they're being taken advantage of,

spk_0 that we're going to send them chips that are, you know, 10 years old or something,

spk_0 then I get why they had that response.

spk_0 H20 is really quite spectacular still, then.

spk_0 Yeah.

spk_0 Of course, it's not as good as, but well.

spk_0 And I get that.

spk_0 I look, you know, we're impatient.

spk_0 And I believe that they're wise.

spk_0 Or they're thinking through their situation.

spk_0 They have larger agendas to deal with.

spk_0 Relatively, you know, vis-à-vis the United States.

spk_0 There are a lot of discussions going on.

spk_0 But I'll come back to the ground truth, fundamental truth.

spk_0 I believe that is in the best interest of China,

spk_0 that Nvidia is able to serve that market and compete in that market.

spk_0 I fundamentally believe it's the best interest of China.

spk_0 It is, of course, fantastic interest in the United States.

spk_0 It is fundamental.

spk_0 But those two truths can coexist.

spk_0 It is possible for both to be true.

spk_0 And I believe it is both true.

spk_0 And so I, I'm rather, you know, even though I tell all of our investors

spk_0 that our guidance includes no China.

spk_0 And I appreciate all of our investors to include no China in any of our guidance.

spk_0 We've got plenty of growth opportunities outside.

spk_0 And we've got all of that.

spk_0 It's true.

spk_0 It doesn't make China not important to us.

spk_0 It's very important to us.

spk_0 Anybody who thinks that the Chinese market is not important

spk_0 has their head deep in the sand.

spk_0 And so this is one of the most important markets in the world.

spk_0 Smart markets, as you know.

spk_0 Smart people doing smart things and we want to be there.

spk_0 Yeah.

spk_0 And I think it's in the best interest of both countries that we are there.

spk_0 And so I think when I take a step back,

spk_0 I am confident that ultimately the wisdom will prevail.

spk_0 Yes.

spk_0 I've always been confident that wisdom prevails.

spk_0 I've always been confident that truth prevails.

spk_0 And it's taken me this far.

spk_0 And I believe that to be fundamentally true now.

spk_0 And so these things will get sorted out.

spk_0 And we will have the opportunity to go compete in that China market.

spk_0 I'm not very political, but very topical is the administration's decision to charge

spk_0 100,000 per H1B visa.

spk_0 You've spent a lot of time with the president.

spk_0 You've called him our secret weapon in AI.

spk_0 I also know you want to recruit the best and brightest to our country.

spk_0 Yeah.

spk_0 So how do you think about the decision to charge 100,000 per H1B visa?

spk_0 Does this make it easier or harder to recruit talent?

spk_0 And you know, it's perhaps a little different for large companies or small companies.

spk_0 Like how do you think about it?

spk_0 I'm going to start with it's a great start.

spk_0 What on you said it's a great start.

spk_0 It's a great start.

spk_0 I'm just going to start there.

spk_0 And the reason for that is this.

spk_0 That implies I hope it's not the end.

spk_0 But I think it's a great start.

spk_0 I just hope it's not the end.

spk_0 Here's what I fundamentally believe.

spk_0 America has one a singular brand reputation that no country in the world has.

spk_0 And no country in the world is in the position or in the horizon to be able to say

spk_0 come to America and realize the American dream.

spk_0 What country has the word dream behind it?

spk_0 Yes.

spk_0 It's part of its brand.

spk_0 We are utterly singular.

spk_0 And you're talking to somebody who represents the American dream.

spk_0 My parents didn't have any money sent us over here.

spk_0 We started from nothing.

spk_0 You guys know I, you know, bus tables,

spk_0 wash dishes, clean toilets.

spk_0 And here I am.

spk_0 This is the American dream.

spk_0 Prison Trump knows that.

spk_0 We want legal immigrants.

spk_0 There's a difference between legal immigrants and illegal immigrants.

spk_0 But the idea that it's a country that's free for all doesn't make sense.

spk_0 And so now the question is how do we go from

spk_0 the idea that we want to protect fundamentally the American dream

spk_0 to dealing with illegal immigrants at such a large scale,

spk_0 how do we find a logical pragmatic solution?

spk_0 Right.

spk_0 So the idea that we put a $1,000 price tag on H1B

spk_0 probably sets the bar a little too high.

spk_0 But that's the first bar.

spk_0 It at least eliminates illegal immigration.

spk_0 And that's a good start.

spk_0 How does it eliminate illegal immigration?

spk_0 Well, at least it eliminates abuse of the H1B.

spk_0 Of the H1B.

spk_0 At least.

spk_0 And that's a good start.

spk_0 And at least we can have a conversation.

spk_0 So one of the things that we know about prison in Trump, he's a good listener.

spk_0 He actually listens to me.

spk_0 I mean, he listens to me.

spk_0 He listens to me and he doesn't have to.

spk_0 And he listens to a lot of people.

spk_0 And he's integrating a lot of information.

spk_0 And this is obviously a very complicated issue.

spk_0 And so I think that this is a fine start.

spk_0 It's a fine start.

spk_0 But I'm not confused that anyone in the administration,

spk_0 anyone in the White House is confused.

spk_0 That legal immigration,

spk_0 immigration is the foundation of the American dream.

spk_0 And it's the ultimate brand that we want to protect.

spk_0 And that's the future we want to protect.

spk_0 And I would also say,

spk_0 it seems to me that certainly SACs and other people in the administration know

spk_0 that we have to recruit the world's best and brightest.

spk_0 We should not sacrifice the greatness of the brand.

spk_0 Charging $100,000 or let's say, you know, it got lowered to 50 or whatever the case is.

spk_0 It does seem like it tilts the plain field in favor of big companies who can effectively sponsor

spk_0 all these people.

spk_0 And it's more challenging for the startup ecosystem where people are already super expensive.

spk_0 And now I got to pay this fee on top of it.

spk_0 It also has an unintended consequence.

spk_0 It might accelerate investment outside the United States.

spk_0 And so there are unintended consequences.

spk_0 But like I said, start somewhere, move towards the right answer.

spk_0 You know, oftentimes people want to go directly from a wrong answer, wrong condition.

spk_0 We don't want this condition where we're at.

spk_0 Right.

spk_0 And directly jump to the perfect answer is hard to find.

spk_0 Just start somewhere.

spk_0 It's the entrepreneurial way.

spk_0 It's important to me, and the president talked about before when he was running for office,

spk_0 he wanted a staple of green card to the diplomas of these STEM students.

spk_0 You know, smart.

spk_0 So people coming to the United States from China AI researchers study in at Stanford.

spk_0 We want to keep them here.

spk_0 We want to get, you know, and by the way, if their families can't get here,

spk_0 they're going to leave after a few years.

spk_0 So you might even want to make it easier for their families to come here.

spk_0 And others, are you confident that we have a strategic plan in this administration?

spk_0 You know, this is a start.

spk_0 But your conversation, say, give you confidence that we have a broader strategic plan

spk_0 to make sure we're recruiting the best in the brightest.

spk_0 I don't know that I have an answer for that.

spk_0 Okay.

spk_0 But I understand that where we're at is not where we want to be.

spk_0

spk_0 And I don't think anybody's lost their focus on, you know, the American dream,

spk_0 the importance of immigration, the importance of attracting all of the world's best talent to United

spk_0 States, create the conditions for them to stay here.

spk_0 There are things that are done from time to time that works against what I just described.

spk_0 Right.

spk_0 Making foreign students uncomfortable

spk_0 and being here in the US.

spk_0 The brand threatens the brand.

spk_0 Let's not forget that it's okay to be competitive with China,

spk_0 but be careful not to be tough on Chinese.

spk_0 And so we need to make sure that that slippery slope isn't crossed.

spk_0 You know, and so there are all of these things that goes along with

spk_0 finesse and nuance.

spk_0 Yes.

spk_0 But the fact of the matter is we know where we want to be.

spk_0 We know we're in a difficult situation.

spk_0 We don't want to be here.

spk_0 And President Trump doesn't have much time to move us in that direction.

spk_0 Right.

spk_0 And so to the extent that we move in that direction,

spk_0 I believe it's a good start.

spk_0 Agreed.

spk_0 Yeah.

spk_0 I heard from a Chinese researcher leading one of our leading labs in the US

spk_0 that three years ago,

spk_0 90% of the top AI researchers graduating from universities in China

spk_0 wanted to come to the United States and did come to the United States

spk_0 to work in our leading labs.

spk_0 And he guessed that today that's closer to 10 or 15%.

spk_0 Right.

spk_0 So seeing a precipitous drop, that's precisely a concern that we have.

spk_0 Right.

spk_0 So have you seen this?

spk_0 Have you, you know, you're paying attention to both markets?

spk_0 Do you see this?

spk_0 And what are the things we need to do in order to reverse that?

spk_0 Definitely see a greater concern of Chinese students

spk_0 who come here and remain here.

spk_0 And or many of them who come here for school

spk_0 and are thinking about going elsewhere.

spk_0 Right.

spk_0 Many of them thinking about Europe.

spk_0 Right.

spk_0 And so I think we need to be super, super concerned about this.

spk_0 This is this is this is a source of existential crisis.

spk_0 This is definitely the early indicators of a future problem.

spk_0 Right.

spk_0 You know, smart people's desire to come to to America

spk_0 and smart people smart students desire to stay.

spk_0 Those are what I would call KPIs.

spk_0 Yes.

spk_0 Early indicators of future success.

spk_0 Yes.

spk_0 I think of it a bit like the warriors,

spk_0 you know, if they have an advantage of recruiting all the best players in the NBA,

spk_0 right, then they can continue to win championships.

spk_0 Yeah.

spk_0 But the second that recruiting pipeline,

spk_0 right, because of the brand of the warriors gets to

spk_0 manage or something else happens,

spk_0 then they're not going to be able to recruit the best future players

spk_0 and you're not going to win championships.

spk_0 And when I you talk about the American dream so eloquently,

spk_0 that being brand USA, right, the right to come here and to do what you've done.

spk_0 And, you know, so I hope that the feedback to this administration is not just the administration.

spk_0 It's also just how we as a country talk about immigration.

spk_0 That's right.

spk_0 Right.

spk_0 This needs to be the place that welcomes the best and the brightest,

spk_0 that attracts as a strategic plan for recruiting the best and the brightest and making sure

spk_0 that this is the place that they want to work.

spk_0 As you know, there's a phrase, and I didn't even hear about this phrase until just a few years ago,

spk_0 China hocks.

spk_0 Yes.

spk_0 And apparently, if you're a China hawk, you get to wear that label with pride.

spk_0 It's almost like a badge of honor.

spk_0 It's a badge of shame.

spk_0 There's no questions a badge of shame.

spk_0 There's no question that although they want what's in the best interest of our country,

spk_0 and we all want what's in the best interest of our country,

spk_0 destroying that pipeline of the American dream is not patriotic.

spk_0 Right.

spk_0 They think they're doing the right thing for our country, but it's not patriotic.

spk_0 Not even a little bit.

spk_0 And so we need to continue to be the great country we are to have the confidence

spk_0 of a great country.

spk_0 Yes.

spk_0 And to have the confidence of a great country and have somebody who wants to compete with us

spk_0 and to have the attitude, bring it on.

spk_0 Right.

spk_0 Bring it on.

spk_0 Because I believe in our people.

spk_0 I believe in our people.

spk_0 I believe in the people that are here.

spk_0 I believe in our culture.

spk_0 I believe in our country.

spk_0 I believe in our system.

spk_0 Bring it on.

spk_0 And is it your take that that's where the president is?

spk_0 He's a pragmatist.

spk_0 He's a believer in the growth and the ability of the United States to compete.

spk_0 It seems to me that's where he is.

spk_0 There's no question President Trump is the bring it on president.

spk_0 Right.

spk_0 And he doesn't seem to me like the reason I'm confident and I've said on this pod

spk_0 that I think he'll get a big deal done with China.

spk_0 I really, really do hope so.

spk_0 Yeah.

spk_0 And I think he speaks positively

spk_0 with great respect and great eloquence about his relationship and the importance of China.

spk_0 Not one time have I ever heard him say the word decouple, which we heard a lot in the last

spk_0 administration.

spk_0 Right.

spk_0 You can't decouple against the single most, the two most important relationships for the next

spk_0 century.

spk_0 Right.

spk_0 That doesn't make any sense at all.

spk_0 Yeah.

spk_0 De-coupling is exactly the wrong concept.

spk_0 Right.

spk_0 I mean, it seems to me he and Scott Besson are saying, listen, we need to make America great.

spk_0 We need to re-industrialize America.

spk_0 We need to balance and make sure that we have fair trade.

spk_0 That we protect industries that we need to help build.

spk_0 That China helps us do that, recognizing that we have helped them do it over the course of the

spk_0 last 25 years.

spk_0 But that ultimately said, the best way to understand me is I'm a great deal maker.

spk_0 I make deals.

spk_0 Right.

spk_0 Whereas I think in other camps, there are people who are iconoclastic or dogmatic.

spk_0 You know, it's the mere shimer view of China that there's a great power struggle.

spk_0 One must must win and one must lose.

spk_0 Versus this idea that every country has to look exactly like ours.

spk_0 Right.

spk_0 We want to versus.

spk_0 Right.

spk_0 You want America to win, but that doesn't have to come at the expense of poking an eye

spk_0 and telling somebody else they have to lose.

spk_0 Because we're that confident.

spk_0 Yeah, we're that confident.

spk_0 Because we're that mighty.

spk_0 Because we're that incredible.

spk_0 I've got no trouble.

spk_0 As you know, I've got no trouble working with all my colleagues in the ecosystem.

spk_0 Right.

spk_0 And notice, we just did the ultimate deal.

spk_0 Right.

spk_0 Partnering with Intel, a company that spent most of its life trying to put us out of business.

spk_0 Right.

spk_0 And I had no trouble partnering with them.

spk_0 Right.

spk_0 You know, and so and the reason for that is because

spk_0 number one, bring it on.

spk_0 Yes.

spk_0 And number two, the future is so much greater.

spk_0 It doesn't have to be all us or them.

spk_0 It could be us and them.

spk_0 Yeah.

spk_0 Well, nonetheless, bring it on.

spk_0 Yeah.

spk_0 Read.

spk_0 You know, you you mentioned something that's profoundly important to both of us.

spk_0 You and I've talked a lot about this, the American dream.

spk_0 You know, and it was I think Abraham Lincoln who said,

spk_0 fundamental to the American dream is the right to rise.

spk_0 Yeah.

spk_0 That's the belief that your kids can do better than you did.

spk_0 That's right.

spk_0 Right.

spk_0 And you you've experienced the right to rise.

spk_0 We've all experienced the right to rise in America.

spk_0 So now you go to Wikipedia, you go look up American dreams,

spk_0 come to my picture.

spk_0 Right.

spk_0 Yeah.

spk_0 And yet, we live at this time where because of the nature

spk_0 of these technological systems, we have companies that are going to be worth 10 trillion.

spk_0 We'll probably have individuals that are worth a trillion.

spk_0 Those are the incentives that give people the encouragement to rise.

spk_0 But at the same time, when we head into this age of abundance,

spk_0 something that I was deeply worried about was that too many people get left behind.

spk_0

spk_0 Right.

spk_0 And they feel left out and left behind.

spk_0 So it makes sense for them to attack this system of capitalism.

spk_0 Something that you and I worked on together and I'm deeply grateful for was the idea of

spk_0 West America that we have to start every kid at birth on the capitalist right to rise journey.

spk_0

spk_0 Give them a thousand bucks in great companies like

spk_0 the Indian Social Security and OpenAI, etc.

spk_0 And they benefit.

spk_0 Right.

spk_0 As the country wins, they win.

spk_0 And they own it individually.

spk_0 They can see it on their phone.

spk_0 Every kid is a shareholder in the future of America.

spk_0

spk_0 So on the 200 because of your support.

spk_0 And I wanted to take the chance on this podcast.

spk_0 And the support of...

spk_0 Well, I want to thank you for starting it, for driving it.

spk_0 Yeah.

spk_0 Thank you.

spk_0 What a great idea.

spk_0 And, you know, so this...

spk_0 You're genius.

spk_0 The...

spk_0 Please.

spk_0 This past

spk_0 in the big beautiful bill.

spk_0 Yeah.

spk_0 Most people don't even realize that yet.

spk_0 Starting in 2026, every child born forevermore in the history of this country.

spk_0 We'll start off with an investment account at birth,

spk_0 see it in a thousand bucks in the best American companies.

spk_0 And your company has agreed to add to the accounts

spk_0 of not only the kids who work for your employees,

spk_0 but maybe even kids of others.

spk_0 I'm going to adopt schools, you know,

spk_0 and lots of philanthropist and companies.

spk_0 We think every company across America...

spk_0 Wonderful way for companies to give back.

spk_0 Right.

spk_0 Yeah.

spk_0 As part of the 401k,

spk_0 this seems to me to be part of the change in the social contract that needs to occur.

spk_0 Because if we're seeing this exponential progress,

spk_0 we know that the evolution of government and the social contract needs to keep up with it.

spk_0 Obviously, President Trump and bipartisan group in the House and Senate passes into law.

spk_0 So maybe just talk to us a little bit when you think about

spk_0 the pace and magnitude of changes that are coming.

spk_0 Right.

spk_0 I know you believe it would be a net good,

spk_0 but there are also going to be a bunch of people displaced along the way.

spk_0 We probably need things like this and other things.

spk_0 Right. In order to, you know,

spk_0 bring everybody along for the journey.

spk_0 There's several things that President Trump has done.

spk_0 And let me just start there.

spk_0 It has done that is incredibly good for bringing everybody along.

spk_0 The first thing is re-industrializing America.

spk_0 President Trump, Secretary Lutnik,

spk_0 you know they're all in behind that,

spk_0 all the work that they're doing.

spk_0 Encouraging companies to come build here in the United States, investing in factories,

spk_0 and re-skilling and upskilling that skilled labor workforce,

spk_0 incredibly valuable to our country.

spk_0 The idea that we no longer make it only that you get a PhD or you go to,

spk_0 you know, one of the great schools.

spk_0 And only in that way can you build a great life and deserve to have a great living.

spk_0 We've got to change all that.

spk_0 That doesn't make any sense.

spk_0 We love craft.

spk_0 Right.

spk_0 I love people who make things with their hands.

spk_0 And we're now, we're not going to go back and build things.

spk_0 Yes.

spk_0 Build magnificent, incredible things.

spk_0 I love that.

spk_0 That's going to transform America.

spk_0 There's no question about that.

spk_0 There's a whole band of an economy, a whole band of society

spk_0 that has been largely left behind because we outsourced everything.

spk_0 Right.

spk_0 Now, I'm not suggesting we insource everything.

spk_0 Right.

spk_0 You know, all the people arguing about, you know,

spk_0 manufacturing tennis shoes and toothpicks.

spk_0 I mean, you know, that's that's denigrating a perfectly good discussion into some insane level.

spk_0 You know, we've got to recognize that reindustrial and unites,

spk_0 reindustrial and unites America is just fundamentally going to be

spk_0 transformed, transformative.

spk_0 Number one, number two.

spk_0 And aspirational.

spk_0 Oh, it's fantastic.

spk_0 Elon taking us to Mars, watching spaceships caught with, you know,

spk_0 chopsticks out of the sky.

spk_0 This is not only great for the industrializing base of America.

spk_0 It's aspiration for America.

spk_0

spk_0 That's right.

spk_0 And then, of course, AI.

spk_0 It is the greatest equalizer.

spk_0 Just think, everybody can have an AI now.

spk_0 The ultimate equalizer.

spk_0 We've closed the technology divide.

spk_0 Remember, the last time that somebody had to learn,

spk_0 wants to use a computer for their economic or career benefit,

spk_0 they have to learn C++ or C or at least Python.

spk_0 Now they just have to learn human.

spk_0 I know.

spk_0 And so, and if you don't know how to program an AI, you tell the AI,

spk_0 hi, I don't know how to program an AI.

spk_0 How do I program an AI?

spk_0 And the AI explains it to you or does it for you?

spk_0 It does it for you.

spk_0 And so it's incredible.

spk_0 It's and we've now closed the technology divide with technology.

spk_0 This is something that everybody's got to engage.

spk_0 Open AI has 800 million active users.

spk_0 Gosh, it really needs to be 6 billion.

spk_0 It really needs to be 8 billion soon.

spk_0 And so I think that's number one, the number two.

spk_0 And the number three, I think the AI will change tasks.

spk_0 The thing that people confuse is there are many tasks that will be eliminated.

spk_0 There are many tasks that will actually be created.

spk_0 But it is very likely that for many people, their jobs are gainfully protected.

spk_0 And so for example, I'm using AI all the time.

spk_0 You're using AI all the time.

spk_0 My analysts are using AI all the time.

spk_0 My engineers, every one of them use AI continuously.

spk_0 And we're hiring more engineers.

spk_0 We're hiring more people.

spk_0 We're hiring across the board.

spk_0 The reason for that is because we have more ideas.

spk_0 We can now go pursue more ideas.

spk_0 The reason for that is because our company became more productive.

spk_0 And because we became more productive, we became more rich.

spk_0 We became more rich.

spk_0 We can hire more people to go after those ideas.

spk_0

spk_0 The concept that AI comes along and therefore there's going to be a mass destruction of jobs

spk_0 starts with the premise that we have no more ideas.

spk_0 Right.

spk_0 It starts with the premise we have nothing left to do.

spk_0 Everything we're doing in our lives today, this is the end.

spk_0 And if somebody else were to do that one task for me,

spk_0 I have one task less.

spk_0 Now I have to sit there and wait for something.

spk_0 You know, wait for retirement.

spk_0 Sit on my rocking chair.

spk_0 That idea doesn't make sense to me.

spk_0 And so I think that intelligence is not a zero sum game.

spk_0 The more intelligent people I'm surrounded by,

spk_0 the more geniuses I'm surrounded by,

spk_0 surprisingly the more ideas I have,

spk_0 the more problems I imagine that we can go solve,

spk_0 the more work we create, the more jobs we create.

spk_0 And so I think for I don't know what the world looks like in a million years,

spk_0 that's going to be left for my children.

spk_0 But for the next several decades,

spk_0 my sense is that economy is going to grow.

spk_0 Lots of new jobs are going to be created.

spk_0 Every job will be changed.

spk_0 Some jobs will be lost.

spk_0 And we're not going to be writing horses on streets.

spk_0 And those things will be fine.

spk_0 You know, humans are famously skeptical

spk_0 and terrible at understanding compounding systems.

spk_0 And they're even worse at understanding exponential systems

spk_0 that accelerate with size.

spk_0 We've talked about exponentials a lot today.

spk_0 And the great future is Ray Kurzweil said,

spk_0 in the 21st century, we're not going to have 100 years of progress.

spk_0 We're likely to have 20,000 years of progress.

spk_0

spk_0 You said earlier, we're so fortunate to be living at this moment

spk_0 and contributing to this moment.

spk_0 I'm not going to ask you to look out 10 or 20 or 30 years

spk_0 because I think it's so challenging.

spk_0 But when we think about 2030,

spk_0 things like robots.

spk_0 30 years is easier than 2030.

spk_0 Oh, really?

spk_0 Yeah, yeah.

spk_0 OK.

spk_0 So I'll get a great two license to go out 30.

spk_0 As you think out over the course of,

spk_0 I like these shorter time frames because they have to marry bits and atoms.

spk_0 More important.

spk_0 Bits and atoms.

spk_0 The hard part of building this stuff, right?

spk_0 Because everybody's saying it's going to happen.

spk_0 It's not interesting, but not helpful.

spk_0 Exactly.

spk_0 But if we have 20,000 years of progress,

spk_0 reflect on that statement by Ray,

spk_0 reflect on exponentials.

spk_0 And how all of our listeners,

spk_0 whether you work in government,

spk_0 whether you're in a startup,

spk_0 whether you're running a big company,

spk_0 need to be thinking about the accelerating rate of change,

spk_0 the accelerating rate of growth,

spk_0 and how you will be co-intelligent in this new world.

spk_0 Well, there are a lot of things that many people have already said.

spk_0 And they're all very sensible.

spk_0 I think in the next five years,

spk_0 one of the things that is really cool that's going to get solved

spk_0 is the fusion of artificial intelligence and electronics, robotics.

spk_0 And so we're going to have,

spk_0 we're going to have, you know,

spk_0 AIs that are going to be wandering around us.

spk_0 And that everybody knows.

spk_0 We all know that we're going to all grow up with our own R2D2.

spk_0 Yeah.

spk_0 And that R2D2, remember everything about us and coach us along the way

spk_0 and be our companion.

spk_0 We already know that.

spk_0 Okay.

spk_0 And so the idea,

spk_0 and the idea that every human will have their own GPUs associated with them in the cloud,

spk_0 and that they're 8 billion people, 8 billion GPUs,

spk_0 that's a, you know, viable outcome.

spk_0 Yeah.

spk_0 And so,

spk_0 And he's having their own model.

spk_0 Is fine tuned for them?

spk_0 Fine tuned for them.

spk_0 And that AIs in the cloud is also embodied in a whole bunch of,

spk_0 it's embodied in your cars,

spk_0 embodied in your own robot.

spk_0 It's everywhere with you.

spk_0 And so that I think that that future is a very sensible thing.

spk_0 The idea that we're going to understand the infinite complexity of biology

spk_0 and understanding the system of biology,

spk_0 and how it, how to predict it and have digital twins of everybody,

spk_0 our own digital twin for healthcare,

spk_0 like we have a digital twin for shopping at Amazon.

spk_0 Why wouldn't we have our digital twin at healthcare?

spk_0 Of course we would.

spk_0 And so, you know, a digital system that predicts how we're going to age,

spk_0 what disease we're likely going to have,

spk_0 and anything that's about to happen maybe even next week,

spk_0 or, you know, tomorrow afternoon, and predict it early,

spk_0 of course, we wouldn't have all that.

spk_0 And so I think all of that is a given.

spk_0 I think the part that I'm asked a lot by CEOs that I work with,

spk_0 about now given all of that, what happens?

spk_0 What do you do?

spk_0 And this is the common sense of things that move fast.

spk_0 If you have a train that's about to get faster and faster and go exponential,

spk_0 the only thing that you really need to do is get on it.

spk_0 Yeah.

spk_0 And once you get on it, you'll figure everything else out along the way.

spk_0 And so to predict where that train's going to be,

spk_0 and try to shoot a bullet at it,

spk_0 or predict where that train's going to be,

spk_0 and it's going exponentially faster every second,

spk_0 and go figure out what intersection to wait for it,

spk_0 that's impossible.

spk_0 Just get on it while it's going kind of slowly,

spk_0 and go exponential along the way.

spk_0 A lot of people think this just happened overnight.

spk_0 You know, you've been at this for 35 years.

spk_0 I remember hearing Larry Page say probably around 2005 or 2006,

spk_0 that the end state of Google will be when the machine can predict the question before you

spk_0 even answer it, before you even ask it, and give you the answer without having to look.

spk_0 Right. I heard Bill Gates say in 2016.

spk_0 Because contextually, you must be asking him up.

spk_0 Well, you must be wondering about that.

spk_0 Right.

spk_0 I heard Bill Gates say in 2016 when somebody said,

spk_0 it hasn't all the things been done.

spk_0 We've had the internet, we've had cloud, we've had mobile,

spk_0 social, et cetera.

spk_0 He said, we haven't even started.

spk_0 He said, what do you think?

spk_0 Why would you say that?

spk_0 We won't even begin until machines go from being dumb calculators

spk_0 to beginning to think for themselves, to think with us.

spk_0 Right.

spk_0 Kind of that is the moment that we're in.

spk_0 I think to have leaders like you, leaders like Sam,

spk_0 and Elon, Satchet, et cetera, it's such an extraordinary advantage for this country.

spk_0 Right.

spk_0 And to have the cooperation that we see between a system of risk capital that I take,

spk_0 you know, that I'm part of, which can provide the risk capital for people to do.

spk_0 We're not relying on government having a Manhattan project.

spk_0 We can actually do this ourselves and together for the benefit of the country.

spk_0 It's an extraordinary time.

spk_0 And at a scale that's unimaginable.

spk_0 Right.

spk_0 It's an extraordinary time.

spk_0 But I also think, you know, one of the things that I'm just grateful

spk_0 is that we have leaders who also understand their responsibility

spk_0 to the fact that we are creating change at an accelerating rate.

spk_0 And we know, while it will most likely be great for the vast majority,

spk_0 there'll be challenges along the way.

spk_0 And we'll deal with those as they come and raise the floor for everybody and make sure

spk_0 that this is a win, not just for some elite plutocrats at the top hanging out in Silicon

spk_0 Valley.

spk_0 And don't scare them.

spk_0 But it's a win.

spk_0 It's a win for them.

spk_0 Don't scare them.

spk_0 That's right.

spk_0

spk_0 We will.

spk_0 Yeah.

spk_0 So thank you for that.

spk_0 Exactly.

spk_0 As a reminder to everybody, just our opinions, not investment advice.

NVIDIA: OpenAI, Future of Compute, and the American Dream | BG2 w/ Bill Gurley and Brad Gerstner

Interactive Transcript

Topics Covered

Related Episodes

Out on a Limb | Nov 2nd | Hillhurst United Church

How Do You Respond to Jesus? John 5:16–47 :Daily Devotional, Daily Bible Study

Episode 240: Otto Aerospace CEO Paul Touw talks teardrop, laminar flow, and Phantom 3500

Burnout, Breakthroughs, and the Power of Getting Back Outdoors | Gun Talk Hunt

Share Episode